Native fishing in the Great Lakes â. A multidisciplinary approach to zooarchaeological remains from precontact Iroquoian villages near Lake Simcoe, Ontario.
RIJKSUNIVERSITEIT GRONINGEN
NATIVE FISHING IN THE GREAT LAKES – A MULTIDISCIPLINARY APPROACH TO ZOOARCHAEOLOGICAL REMAINS FROM PRECONTACT IROQUOIAN VILLAGES NEAR LAKE SIMCOE, ONTARIO
Proefschrift
ter verkrijging van het doctoraat in de Letteren aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. D.F.J. Bosscher, in het openbaar te verdedigen op dinsdag 7 september 1999 om 14.15 uur
door Suzanne Julia Needs-Howarth
Promotores:
Prof. dr. A.T. Clason Prof. dr. G.J. Boekschoten
Referent:
Dr. D.C. Brinkhuizen
ISBN 90 367 1133 9
Native fishing in the Great Lakes – A multidisciplinary approach to zooarchaeological remains from precontact Iroquoian villages near Lake Simcoe, Ontario
Suzanne Needs-Howarth
© Suzanne Needs-Howarth, Toronto, 1999. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior permission of the author.
TABLE OF CONTENTS
LIST OF TABLES
vii
LIST OF FIGURES
viii
ACKNOWLEDGMENTS
ix
CHAPTER 1: CONTEXT Introduction Research context and objectives Environmental context Cultural context
1 1 2 3 3
CHAPTER 2: SETTLEMENT PATTERNS Introduction Feature types Barrie Dunsmore Carson
6 6 6 6 7 8
CHAPTER 3: ANALYSIS PROCEDURES Identification Quantification
9 9 10
CHAPTER 4: TAPHONOMY Introduction Butchering and processing Food preparation Consumption Discard Burial Recovery
14 14 14 16 17 17 17 18
CHAPTER 5: INTERPRETING FISH REMAINS DATA Historical data Fisheries science data Biogeography Fish habitat Fishing techniques Zooarchaeological indicators of method and location of capture Age and season of death The Three Fisheries Model
23 23 23 23 25 25 27 29 34
CHAPTER 6: SAMPLE DESCRIPTION AND ANALYSIS Barrie Dunsmore Carson
39 39 42 46
CHAPTER 7: SYNTHESIS Compiling the evidence Timing of capture Location of capture Technique of capture Fish processing Diachronic change in fish catches Evidence for fishing at other local sites Other subsistence indicators
50 50 50 50 51 52 52 54 55 v
Subsistence change in the Kempenfelt Bay area Concluding remarks and recommendations for future research
57 59
SUMMARY
61
SAMENVATTING
66
TABLES
71
FIGURES
104
BIBLIOGRAPHY
128
vi
LIST OF TABLES
1. Archaeological cultural chronology of research area 71 2. Dates of occupation of the Barrie, Dunsmore and Carson sites 71 3. Overview of contexts analysed and recovery methods at the Barrie site 72 4. Overview of contexts analysed and recovery methods at the Dunsmore site 72 5. Overview of contexts analysed and recovery methods at the Carson site 72 6. Binomials and generic names of taxa identified 73 7. Taxonomy, biology and ecology of the major fish families 76 8. Representation of 14 common fish cranial elements 79 9. Relationship of 8 diagnostic elements to family NISP by site 79 10. Relationship of 14 common and 8 diagnostic elements to NISP by species by site 80 11. Summary of NISP, NUSP and BW by recovery method at the Barrie site 82 12. Summary of NISP, NUSP and BW by recovery method at the Dunsmore site 84 13. Summary of NISP, NUSP and BW by recovery method at the Carson site 86 14. Proportion of cleithrum breakage in brown/yellow bullhead and yellow perch 88 15. Summary of fish NISP and NUSP by recovery method 88 16. Ubiquity of flotation and average NISP per feature 88 17. Fragmentation rates of fish and mammal bones 89 18. Fish osteometrics employed 89 19. Proportions of fish bones with proportional element size data and osteometrics 90 20. Relationship of stretched net mesh size to fish length 90 21. Size and age at maturity of northern pike, brown bullhead and yellow perch 90 22. Calcified structure edge conditions relating to season of capture 90 23. Fish bone distribution by body area 91 24. Overview of NISP and BW, all classes 92 25. Overview of brown bullhead pectoral and dorsal spine age and growth analysis 96 26. Overview of Percidae scale age and growth analysis 97 27. Fish assemblage by major feature at the Barrie site 98 28. Fish assemblage by major feature at the Dunsmore site 99 29. Fish assemblage by feature at the Carson site 100 30. Examples of fish caloric content 102 31. Number of cranial bones identified in relation to one vertebra at the Barrie, Dunsmore, Carson, Keffer, Over, Molson and Hunter's Point sites 102 32. Summary of average proportional element size of northern pike, brown bullhead and yellow perch 102 33. Contribution of non-fish taxa organized by probable procurement habitat 103 34. Summary contribution of dog and deer 103
vii
LIST OF FIGURES
1. Regional geographic context 2. Location of the Barrie, Dunsmore and Carson sites 3. House 1 at the Barrie site as an example of feature types 4. Settlement patterns at the Barrie site 5. Settlement patterns at the Dunsmore site 6. Settlement patterns at the Carson site 7. Relative abundance of eight diagnostic elements 8. Abundance of eight diagnostic elements in relation to the most common element by family 9. Proportional element size of all fish by recovery 10. Proportional element size of yellow perch by recovery 11. Walleye scale osteometrics 12. Selected fish bone osteometrics 13. Fish taxa ranked by habitat preference 14. Modern Percidae growth 15. Proportional element size of selected fish species at the Barrie site 16. Proportional element size of selected fish species at the Dunsmore site 17. Proportional element size of selected fish species at the Carson site 18. Relative contribution to the fisheries model 19. Relative contribution to the fisheries model, including vertebrae 20. Brown bullhead age distribution based on dorsal and pectoral spines 21. Walleye age distribution based on scales 22. Fish cranial bone identified by family at the Barrie, Dunsmore, Carson, Wiacek, Hubbert and Molson sites 23. Fish cranial bone and vertebrae identified by family at the Barrie, Dunsmore, Carson and Molson sites 24. Taxonomic class distribution by NISP 25. Taxonomic class distribution by BW 26. Taxonomic class distribution by NISP at the Barrie, Dunsmore, Carson, Wiacek, Hubbert and Molson sites 27. Species richness of fish, bird, mammal 28. Cumulative frequency of fish species identifications
viii
104 105 106 107 108 109 110 111 113 113 114 114 115 116 117 118 120 122 122 123 123 124 125 126 126 126 126 127
ACKNOWLEDGMENTS
I owe many thanks to many people. For their patience, support, suggestions and feedback I thank my supervisor, Prof. A. T. Clason, Groningen Institute for Archaeology, University of Groningen, the Netherlands; my co-supervisor, Prof. G. J. Boekschoten, Faculty of Earth Sciences, Free University of Amsterdam; and my “referent”, Dr. Dick Brinkhuizen, Stichting Monument en Materiaal, Groningen. Prof. Clason kindly agreed to take me on as her student six years ago, even though this would certainly mean a substantial commitment into her official retirement. Dr. Brinkhuizen's involvement in my research, first informally, and now in an advisory capacity, has substantially broadened my horizons. For agreeing to become involved in my work and read the manuscript I thank my reading committee, which, in addition to Prof. Clason, consisted of Prof. Charles Cleland, Department of Anthropology, Michigan State University, East Lansing; Prof. Arturo Morales Muñiz, Department of Biology, Universidad Autónoma de Madrid, Spain; and Prof. David Noakes, Department of Zoology, University of Guelph, Ontario. For their helpful comments and suggestions on earlier drafts of this thesis I thank Dr. James Barrett, Department of Anthropology, University of Toronto; Dr. Steven Crawford, Axelrod Institute of Ichthyology, University of Guelph, Ontario; Prof. Conrad Heidenreich, Department of Geography, York University, Ontario; Mr. Michael McMurtry, Ontario Ministry of Natural Resources, Lake Simcoe Fisheries Assessment Unit, Sutton, Ontario; and Dr. Ronald Williamson, Archaeological Services Inc., Toronto, and Department of Anthropology, University of Toronto. I received information and support from zooarchaeologists around the world: Mr. Chris Andersen, Archaeology and Heritage Planning Unit of the Ontario Ministry of Citizenship, Culture and Recreation, Toronto; Dr. James Barrett; Dr. Jim Burns, Quaternary Paleontology, Provincial Museum of Alberta, Edmonton; Dr. Evelyne Cossette, Department of Anthropology, Université Laval, Quebec city; my fellow-student at Groningen, Mr. Lambertus van Es; Dr. Leslie Hartzell, Department of Anthropology, Bishop Museum, Hawaii; Ms Heather Henderson of Historic Horizon, Toronto; Dr. Foss Leach, Archaeozoology Laboratory, Museum of New Zealand Te Papa Tongarewa; Dr. Patrick Lubinski, then of the Department of Anthropology, University of Wisconsin, Madison; Dr. Beverley Smith, Department of Anthropology/Sociology/CJ, University of Michigan, Flint; Mr. Stephen Cox Thomas of Bioarchaeological Research, Toronto; and Dr. Jørn Zeiler of Archaeobone, Groningen. I owe a debt of gratitude to the late Dr. Howard Savage of the Howard Savage Faunal Archaeo-Osteology Collection at the University of Toronto Department of Anthropology, who taught me to the basics of zooarchaeology and first got me excited about fish bones. Mr. Stephen Thomas was a strict mentor in the early years, and continues to provide stimulating discussion. Certain aspects of the fishing strategies model presented in this thesis were developed in collaboration with Mr. Thomas for a co-authored conference presentation and resulting publication. Mr. Thomas kindly allowed me to use the data resulting from his analysis of part of the Dunsmore site zooarchaeological assemblage, as well as the taxonomic and element coding systems and conventions he developed for use with the GFAUNA2 zooarchaeological input and analysis system. I especially acknowledge my friend and colleague Dr. James Barrett, who has been an inspiration throughout my graduate career. His support, advice and editing have been invaluable. Thanks are also owing to several of my anthropology colleagues for information and helpful suggestions: Ms Christine Dodd, Ontario Ministry of Citizenship, Culture and Recreation, London; Dr. Alexander von Gernet, Department of Anthropology, University of Toronto at Mississauga; Mr. Kenneth Cassavoy, at the time at Department of Anthropology, Trent University; Mr. Ed Koenig, Department of Anthropology, McMaster University; Dr. Jim Molnar, Lion's Head, Ontario; Dr. Stephen Monckton, Bioarchaeological Research, Toronto; Mr. John Steckley, Humber College, Toronto; and Dr. Rick Sutton, at the time at the Department of Anthropology, McMaster University. I am grateful also to all my “fishy” colleagues for answering my endless questions about the specifics of fish behaviour and habitat, and for providing me with a biological and ecological perspective on fishing: Mr. Dave Buritt and Mr. Fred Dobbs, Nottawasaga Valley Conservation Authority, Angus, Ontario; Mr. David Loftus and Mr. Lloyd Mohr, Ontario Ministry of Natural Resources Lake Huron Management Unit, Owen Sound, Ontario; Mr. Robin Craig, Ontario Ministry of Natural Resources, Midhurst, Ontario; Dr. John Casselman and Mr. Ken Scott, Ontario Ministry of Natural Resources Lake Ontario Fisheries Assessment Unit, Glenora, Ontario; Mr. Frank Amtstaetter, Mr. Michael McMurtry, Mr. Mark Stirling, Mr. Cam Willox and Mr. Jakob La Rose, Ontario Ministry of Natural Resources Lake Simcoe Fisheries Assessment Unit, Sutton, Ontario; Mr. Ken Nicholls, Aquatic Science Section, Standards Development Branch, Ontario Ministry of Environment and Energy, Toronto; Mr. Alwyn Rose, Ontario Federation of Anglers and Hunters, Peterborough, Ontario; Prof. Eugene Balon, Prof. F. W. H. Beamish, Dr. Stephen Crawford, Mr. Greg LeBreton, Institute of Ichthyology and Department of Zoology, University of Guelph, Ontario; Mr. David Brown, Watershed Ecosystems Graduate Program, Trent University, Peterborough, ix
Ontario; Dr. Thomas Whillans, Department of Biology, Trent University; Ms Lyn Bergquist, GIS & Web Applications for Water Resources, Minnesota Department of Natural Resources; Dr. Edwin Crossman (emeritus), Dr. Tony Harold, Dr. Erling Holm, and Ms Marty Rouse, Centre for Biodiversity and Conservation Biology, Royal Ontario Museum, Toronto. Many of these people provided me with valuable insights on the local fish populations based on their own field experience and the data contained in OMNR databases. The fact that these data are not available in published form is reflected in the numerous “personal communication” citations in the text. I should like to single out a few of these individuals for the practical help they gave me: I thank Dr. John Casselman for sharing with me his insights on age and growth studies, and for familiarizing me with basics of computer-assisted interpretation, and Mr. David Brown for taking on the analysis of the fish scales; these collaborations resulted in two coauthored conference presentations that form the basis for the discussions on age and growth in this thesis. I also thank Mr. Greg LeBreton for sectioning the sturgeon spines and carrying out preliminary interpretations (disappointing though they were!); and Dr. Erling Holm for diving into barrels of formalin-soaked fish to obtain fish scales for comparative purposes. Special thanks are owing to Prof. Richard Hoffmann, Department of History, York University, Ontario, for feedback on contemporary sport fishing techniques and for taking me out one cold winter's day in 1996 to show me the Nottawasaga River from an angler’s perspective. For allowing me to use their laboratory facilities I thank the Lake Ontario Fisheries Station, Ontario Ministry of Natural Resources, Glenora. I especially thank the University of Toronto Department of Anthropology for accommodating my needs as an external researcher for the duration of my lab research. For allowing me access to the reference collections in their care I thank the late Dr. Howard Savage and Dr. Max Friesen of the University of Toronto Department of Anthropology; Dr. Kevin Seymour, Dr. Erling Holm and Dr. Jim Dick of the former departments of Vertebrate Palaeontology; Ichthyology and Herpetology; and Ornithology, respectively (now Centre for Biodiversity and Conservation Biology), of the Royal Ontario Museum; and Dr. Darlene McCuaig-Balkwill and Ms Anne Rick of the Zooarchaeological Analysis Programme of the Canadian Museum of Nature in Ottawa. For access to the archaeological collections, logistical support and site information I thank Dr. Rick Sutton, then of McMaster University; Ms Beverley Garner, Mr. David Robertson, and Dr. Ronald Williamson of Archaeological Services Inc, Toronto; and the staff at Huronia Museum, Midland, Ontario, including Mr. Jamie Hunter. I thank Ms Della Saunders of the University of Toronto Department of Anthropology for processing the Carson flotation samples. I thank my University of Toronto Department of Anthropology student volunteers, Ms Sara Cherubin and Ms Ildiko Horvath, and my Ontario Archaeological Society “Passport to the Past” volunteers, especially Mrs. Gisela Curwen and Ms Caroline Puzinas, also both at University of Toronto Department of Anthropology, for their help with weighing, counting, sorting and bagging. I thank Mr. G. Delger of the Groningen Institute for Archaeology for drawing Figure 1 to 6, and Mr. H. J. Waterbolk, of the same institution, for scanning and emailing them. Mrs. A. C. Bardet of Gronigen effected a transformation of my rusty written Dutch in her diligent editing of the Dutch summary. I thank the Ontario Heritage Foundation for its financial support of this research under the now-defunct Archaeology Programme. The Barrie and Carson site zooarchaeological analyses (1993 and 1995) were funded in part through Archaeology Research grants to the respective excavators, whereas my portion of the Dunsmore site analysis (1994) was funded by an Archaeology Student Research Grant. The processing of the bulk soil samples from the Carson site was also funded under the Ontario Heritage Foundation Grant for the Carson assemblage.
x
CHAPTER 1: CONTEXT
“It is probably true that the imaginative efforts of most ... fishermen who spent time thinking about options, such as how, when, and where they should employ their efforts to gain food, far surpassed those of modern archaeologists, who view this process as a static exercise in energetics, or who are constrained by the formal properties of idealized models.” (Cleland 1982:777)
Introduction Fishing was an important component of the precontact economy in much of the Great Lakes area. In the lower Great Lakes this has become especially evident recently, with improved archaeological recovery of fish remains (e.g., Cooper 1996; Junker-Andersen 1988; Lennox et al. 1986; Needs-Howarth 1997b; Needs-Howarth and Thomas 1998; Stewart 1991b; Thomas 1996a; Thomas 1996b; Thomas 1996d). There is need to develop analytical methods specific to these fish remains, in order to understand when, where, and how the fish originated: the time, location and method of catch.
and in the numerous student reports that, until about ten years ago, were the main source of local zooarchaeological data and interpretation. The approach described above is employed to investigate within-site and between-site differences in the temporal and spatial use of the local environment by the inhabitants of three precontact longhouse villages located near Lake Simcoe, Ontario. These sites range in date from the end of the thirteenth century to the turn of the sixteenth century A. D. (Table 1, 2). The inhabitants, who are broadly known as Ontario Iroquoian, based their subsistence on slash-and-burn maize horticulture, gathering, hunting and fishing (Dodd et al. 1990).
This thesis presents several lines of investigation, which, in combination, offer a way to understand in detail fish subsistence strategies in the Great Lakes area. These involve using palaeoenvironmental data to understand the ancient landscape, using biogeographical data to understand local fish distribution and habitat, and using fisheries science data to understand seasonal variation in fish behaviour. They also include examining the ethnohistoric literature for descriptions of the techniques and social customs surrounding fishing in the early contact period. Moreover, they involve re-thinking the zooarchaeological data themselves – getting more information from relatively small collections of fish bones by looking at species distribution, co-occurrence, element size and state of maturity within features as well as at the site level. The approach adopted here combines information on location, season and method of resource procurement, allowing us to understand how precontact native people scheduled their time, energy, material and labour resources.
A prerequisite to zooarchaeological interpretation is the ability to differentiate between taphonomic effects (i.e. the processes operating on an animal from its death, through butchering, deposition, burial, recovery, sampling and interpretation of its bones) and real differences in time and space. Prior to interpreting fishing strategies, therefore, this thesis presents a comparative analysis of site-specific taphonomic issues, with an emphasis on fragment weights, element sizes, fish vertebrae and fish scales. The methodological potential and limitations of the fish component of the collections are discussed in detail because some of these issues have not been explored previously in an Ontario context. The application focuses mainly on the fish remains; other aspects of the zooarchaeological assemblage are discussed primarily as a way of understanding fish bone taphonomy and fishing strategies. In order to get a broader picture of fishing strategies in the area, data from other sites in southern Ontario are also discussed. Particular attention is paid to the origins of a specialized fishery for autumnspawning lake trout (Salvelinus namaycush), lake herring (Coregonus artedi) and lake whitefish (Coregonus clupeaformis), documented for the cultural descendants of these people, the Huron of the early contact period.
Parts of this approach have been employed by researchers working in other contexts. The multidisciplinary, comparative approach advocated here offers some additional and different ways to understand fish subsistence strategies. While it is tempting to think that it is easier to understand fish procurement because it is often more restricted in habitat and season than that of some mammals and birds, the situation is probably more complex. Key questions remain about when and where fish were caught: during the spawning-run, at a predictable location, or during other times of the year, perhaps more opportunistically or incidentally? The fisheries science literature and the zooarchaeological assemblages themselves indicate that some fish were usually caught during the spawning-run, and others usually not. In other words, procurement of many fish species was not necessarily as seasonally restricted as has sometimes been assumed, both in the published literature
As is often the case, this research raises as many questions as it answers. It is not easy to quantify the factors that cause faunal assemblage variability, especially taphonomy and human choice. However, detailed inter- and intra-site comparisons can extend our knowledge of Iroquoian subsistence and at the same time further our understanding of the possibilities and limitations of our methods.
1
Research context and objectives Despite over 50 years of intensive research on precontact sites in the area, there are still comparatively' few animal bone collections from Iroquoian sites in the Lake Simcoe area that have been excavated using dry screening and/or flotation and that have been analysed by experienced zooarchaeologists. Studies of fish bones in Ontario have been characterized by a general lack of methodological and theoretical considerations. Interpretations of subsistence are mostly limited to a discussion of faunal abundance based on number of identified specimens (hereafter NISP), minimum number of individuals (hereafter MNI) or, rarely, bone weight (hereafter BW). There have been no in-depth investigations of the taphonomic factors that influence interpretation. Intra-site comparisons are often limited to broad contextual units, such as houses and middens. There have been few inter-site comparisons, apart from two theses (Hamalainen 1981; Stewart 1997), and one very general article (Prevec and Noble 1983); these focus more on mammals and birds than on fish. No comprehensive comparative studies are available for the Lake Simcoe area.
sample size. While some consultants are voluntarily complying with more specific guidelines for zooarchaeological analysis put forth recently (Cooper et al. 1995), including the requirement that “analyses must be conducted and/or closely supervised by an experienced zooarchaeologist”, these guidelines have not yet been included in any government regulations. Collections in the study area have anywhere from 42 (Hamalainen 1974) to 1526 (Christine Dodd, personal communication, based on Lennox et al. 1986) fish bones identified below the taxonomic level of class, excluding vertebrae and scales. These fish bones are usually divided among many different deposits, with different taphonomic histories. Because there may be up to 25 different fish taxa at each site, it is rare to have a statistically valid sample per context per species – even less so per element. There are six major causes of observed variability between the zooarchaeological samples: availability of resources; human subsistence choices; butchering and processing; preservation; recovery and analytical methods. Any of these factors alone, or in combination, can result in significantly different datasets, and it is difficult to control for this variation.
Recent work on material from the Wiacek (Lennox et al. 1986; Thomas 1993), Barrie (Needs-Howarth 1995a; Needs-Howarth and Sutton 1993; Needs-Howarth and Thomas 1998), Hubbert (Thomas 1996b), Dunsmore (Needs-Howarth 1994; Needs-Howarth and Thomas 1994a; Needs-Howarth and Thomas 1998; Thomas 1996d), Carson (Needs-Howarth 1997b) and Molson sites (Cooper 1996) indicates that precontact animal food subsistence in the Lake Simcoe area differed substantially from the generalized descriptions in recent syntheses (Dodd et al. 1990; Ramsden 1990). Subsistence strategies appear more varied in time and space, and more concentrated on fish. This calls for a refocusing of (zoo)archaeological research aims and methods to include ways of dealing with fish remains in detail, using additional and different approaches to those used on other classes, such as birds and mammals.
Discussions of precontact human settlement location, duration and population movement have focussed on population dynamics (Warrick 1988), conditions for maize cultivation (Campbell and Campbell 1992), secondary growth habitats (Monckton 1994), or wood for heating and house construction (Fecteau et al. 1991). The availability of animal resources traditionally has not been considered as a motivating or limiting factor in settlement and subsistence choices. There are, however, clear differences in faunal profiles, both within and between sites in the area. These differences probably result at least in part from people making choices regarding their resources, even going to some lengths to procure preferred species. It is, therefore, important to understand fishing as part of the overall subsistence strategy by determining when, where and how the occupants of these sites fished, and the relationship of fishing to other food-getting activities. Only then will it be possible to understand to what extent resources such as fish played a role in Iroquoian settlement and subsistence choices.
Fish bones have only been systematically recovered since the implementation of sieving procedures in the 1970s. Excavators on both academic and commercial projects usually aim for complete recovery through coarse mesh dry sieving. The government-mandated mesh size is 6.4 mm (Task Force on Self Regulation 1992); only a few archaeologists use 3.2 mm. The intensity of wet screening and flotation sampling is uneven and neither are mandatory. Less than optimal recovery is probably a contributing factor to relatively small zooarchaeological sample sizes. Another limitation to sample size is the salvage context of most excavations; only those portions of the site to be impacted by development are excavated.
The samples used in this application derive from three sites, the Barrie (BcGw-18), Dunsmore (BcGw-10) and Carson (BcGw-9) sites, which differ in their temporal and spatial distribution. While these differences offer the opportunity to track changes in time and space, they also pose certain problems in interpretation. Small sample sizes and great diversity of species pose problems for statistical manipulation. Small numbers of finds in individual features are a problem, particularly at Carson, the latest site. Barrie, the earliest site, is located on a creek, immediately south of an upland area, with easy access to a large bay. The two later sites are located immediately north of the same upland area,
Analysis of the recovered zooarchaeological remains is uneven in scope, although draft guidelines concerning mitigative excavation (Task Force on Self Regulation 1992) do recommend that some analysis be carried out for each project. Until recently the bulk of analysis was carried out as undergraduate student projects with imposed limits on 2
close to a small lake and creek, and were probably occupied within the same 100 year period. Slight environmental differences may complicate the differentiation between resource availability and human choice at the Barrie site on the one hand, and the Dunsmore and Carson sites ones on the other hand.
Lowlands. These consist largely of poorly drained silts and clays with eastern hemlock, maple and basswood swamps. The section of the Simcoe Lowlands located in the basin of the Nottawasaga River, however, contains sandy and loamy soils of low natural fertility, with cedar (Thuja sp.) swamps (Heidenreich 1971:70), including the extensive Minesing Swamp.
The sample sizes are similar, thus minimizing differential sample size as a cause of variation. Taphonomic variation also has to be taken into consideration. While it is very difficult to control for differential preservation, it appears possible to understand some of the effects of differential recovery. The latest assemblage, from the Carson site, was recovered with a smaller dry screen mesh size, and intensity of flotation also varies. It was obvious from the start that sorting out recovery differences from real differences in resource exploitation was essential. As is the case with so much zooarchaeological research, taphonomy is a recurring theme throughout this work.
The largest river in the area is the Nottawasaga River, which flows northward into Nottawasaga Bay. The numerous creeks and rivers in the area all drain into either Nottawasaga Bay, Severn Sound or Lake Simcoe and adjacent Lake Couchiching, which in turn drains via the Severn River into the Severn sound. The area around the three sites is drained by creeks flowing into Kempenfelt Bay of Lake Simcoe, and Willow Creek, which drains into the Nottawasaga River. The area south-west of the sites is drained by the Boyne, Pine and Mad rivers, which flow eastward into the Nottawasaga River. The uplands to the north of the sites are drained by the Wye, Sturgeon, Coldwater and North rivers, which flow northward directly into Severn Sound.
Environmental context The archaeological sites discussed in this thesis are situated between Lake Simcoe and Georgian Bay of Lake Huron, in and around the contemporary city of Barrie, Ontario (Figure 1, 2). The area that would have been exploited on a regular basis by the site inhabitants roughly coincides with the contemporary boundaries of Simcoe County. This area is bounded to the north by the Severn River, the Severn Sound (part of Georgian Bay) and the southern extremity of the Canadian Shield; to the east by Lake Simcoe and connecting Lake Couchiching; to the south by the Holland Marsh and connecting Holland River and the Oak Ridges Moraine (left by receding glaciers in the last glaciation); and to the west by Nottawasaga Bay (part of Georgian Bay) and the Niagara Escarpment (a geological formation originating over 400 million years ago). In the following discussion, Nottawasaga Bay refers specifically to the area of Georgian Bay closest to the sites.
Cultural context Southern Ontario has been occupied by humans since the early Holocene (Table 1). While there are few zooarchaeological finds from earlier periods, data from the last millennium are much more abundant. At this time the area between Lake Simcoe and Nottawasaga Bay was densely occupied by (semi-) sedentary communities. Settlement in the historic period consisted of villages, satellite villages, hamlets and special purpose camps for activities such as corn horticulture and fishing. It is likely that at least parts of this pattern go back to the precontact period.
The sites lie in the maple-hemlock section of the Great Lakes/St. Lawrence Forest Region, in a transitional area between the deciduous and the coniferous/ deciduous Canadian Biotic Zone. According to early Euro-Canadian surveyor records, vegetation was dominated by maple (Acer sp.), beech (Fagus grandifolia) and basswood (Tilia sp.), with eastern hemlock (Tsuga canadensis), white pine (Pinus strobus) and oak (Quercus sp.) as secondary dominants (Heidenreich 1971:63).
Over the last 15 years many previously known and newly discovered village sites have been excavated in the context of developer-funded mitigative excavation projects. This study focuses on faunal assemblages from three recently excavated late precontact Iroquoian horticultural villages near Kempenfelt Bay of Lake Simcoe and Little Lake, excavated in a salvage context: the Barrie, Dunsmore and Carson sites. These sites range in date from the end of the thirteenth to the beginning of the sixteenth century A. D. The site occupants are the cultural antecedents of the Huron, who were living in the area at the time of European contact, until they were dispersed by Iroquois groups from New York State starting in 1649 (Table 1).
The major physiographic regions in the immediate vicinity of the sites are the “Simcoe Uplands” and the “Simcoe Lowlands.” The uplands consist of a series of broad rolling till plains about 85 metres above the adjacent lowlands (Wilson and Ryan 1988:209). The sandy soils of these uplands are well drained, with low to moderate natural fertility. The main sources of water are the numerous springs that feed the permanent lowland streams. The Simcoe Uplands are separated by a series of steep- sided, flat-floored valleys and basins known as the Simcoe
The period from A. D. 500 to A. D. 1300 saw great cultural and economic change in Southern Ontario and New York State. The timing and correlation of maize horticulture, semi-permanent settlement and matrilocal residence is still being debated (e.g., Chapdelaine 1993; Crawford and Smith 1996; Snow 1994; Snow 1995; Snow 1996). Recent data from transitional/early Late Woodland Princess Point complex sites in south-western Ontario (Crawford and Smith 1996; Crawford et al. 1997) indicate that maize horticulture was being practised there by A. D. 540. 3
Unlike in other areas of southern Ontario and New York State, there appears to have been no in situ development of Iroquoian villages in the area between Lake Simcoe and Nottawasaga Bay in the first millennium A. D. (Table 2). There were Middle Woodland temporary spring fishing camps (Sutton 1996a:46) and there were some Early Iroquoian warm season fishing camps, but no villages. Archaeological surveys and excavations have been very intensive. Having ruled out all other explanations, Sutton (1996a) argues that the most likely reason for the lack of Early Iroquoian villages in this area is that a new population migrated into the area during the Uren substage of the Middle Iroquoian period.
mentioned late thirteenth-early fourteenth century Holly site, two early-mid fourteenth century sites, and one midlate fourteenth century. There is also a cluster of sites slightly further south, along the tributaries of Lover's Creek, and there are two early-mid fourteenth century sites on the north side of Little Lake. Many of these sites are known only through surface survey or small-scale test excavations (Robertson and Ramsden 1996a). In general, Iroquoian settlements consisted of large, permanently occupied villages with many multi-family longhouses and large external refuse deposits. The three sites that are the focus of this work were probably each occupied for at least 20-30 years, after which time the communities moved to a new location (Warrick 1988). In the contact period, it is thought that villages were relocated for a combination of reasons related to slash-and-burn horticulture, including soil exhaustion, construction and firewood depletion, refuse accumulation, insect infestation and disease. The close spacing of some protohistoric and contact period village clusters suggests that the new settlement catchment may have purposefully included the anthropogenic community of berry plants from the previous village (Monckton 1994). Smaller special purpose sites were located away from the villages in corn fields, or close to seasonally available resources such as fish and whitetailed deer (Odocoileus virginianus) (Ramsden 1990). Because all the excavated sites from the period A. D. 1300-1500 are villages, it is not clear yet to what extent the Huron settlement typology can be extrapolated back in time.
The Uren substage of the Middle Iroquoian period (Table 1) saw the initial establishment of semi-permanent villages in the Lake Simcoe area. The Barrie site represents a pioneering community that originated south of Oak Ridges Moraine and migrated to the area between Lake Simcoe and Nottawasaga Bay at the end of the thirteenth century or the start of the fourteenth century, after a local gap in permanent occupation of some 500 years (Sutton 1996). The migration followed a leap-frog pattern, bypassing the area between the Oak Ridges Moraine and the southern end of Lake Simcoe. Most Middle Iroquoian sites in the area are, in fact, located within a 10 km radius of the head of Kempenfelt Bay. The Barrie site is located at the end of the most direct canoe route from the north shore of Lake Ontario, via the Humber River, the bush trail across the Oak Ridges Moraine, the Holland River and Cooks Bay. It is suggested (Sutton 1996a) that the improved climate of A. D. 1000 to 1200 (Baerreis et al. 1976) was one of the inducements for some Middle Iroquoian groups to move north into the area between Lake Simcoe and Nottawasaga Bay. The only other site that may date to the Uren sub-stage of the Middle Iroquoian period is the recently excavated Holly site (BcGw-58), located a few kilometres south of the Barrie site (Ron Williamson, personal communication 1999).
Subsistence in both the contact and precontact periods was based on slash-and-burn horticulture involving non-local domesticates: maize (Zea mays), and to a lesser extent beans (Phaseolus vulgaris), squash (Cucurbita pepo) and sunflower (Helianthus annuus). Tobacco (Nicotiana rustica) was used for medicinal and ritual purposes, although most tobacco may have originated via trading with other Iroquoian groups outside the area (Heidenreich 1972:69). While the area between Lake Simcoe and Nottawasaga Bay lies close to the northern limit for corn horticulture, the mean annual frost free period and mean annual temperature are, in fact, sufficient for the Eastern Complex maize variety to mature (Monckton 1992).
The Middle Iroquoian period was a time of rapid changes in population size, the emerging of new settlement-subsistence systems, an increasing formalisation of socio-political organization, and an increasing homogenization of certain aspects of material culture throughout southern Ontario (Dodd et al. 1990). The Dunsmore site, dating from the midto late fifteenth century or very early sixteenth century, and the Carson site, dating somewhat later, from the late fifteenth century to the early sixteenth century, were part of a proliferation of permanent villages in the area. This increase in village sites is attributed to a population explosion in the fourteenth century (Wright 1966), hypothesized to result from the decreased infant and juvenile mortality and increased fertility that accompanied a markedly increased reliance on corn horticulture (Noble 1975:37; Trigger 1985:99; Warrick 1988:343-346).
Based on ethnohistorical records, Heidenreich (1972:58) estimates that corn represented about 65 percent of the Huron diet, beans and squash 15 percent, gathered plant foods five percent, meat five percent and fish 10 to 15 percent. Palaeobotanical data from five Huron villages (Monckton 1992), however, indicate that up to 25 percent of dietary calories may have come from non-cultivated plant foods, especially dried fruit. Indeed, deposition of the seeds of wild berries such as bramble (Rubus sp.) and strawberry (Fragaria sp.) in refuse and human faeces around the village may have fostered anthropogenic plant communities (Monckton 1994). Ratios of 13C/12C in human bone indicate that by A. D. 1400 maize may have contributed 50 percent of the carbon in the diet. Some of the 13C enrichment may
Middle and Late Iroquoian sites located close to the Barrie, Dunsmore and Carson sites include the previously 4
have resulted from human consumption of the meat of dogs (Canis familiaris), which may have been fed maize, and deer, which may have browsed on stands of corn. The 15N content in human bones indicates that beans were not a major food source. It appears, therefore, that fish and meat remained the main sources of protein in the human diet even after the adoption of horticulture (Schwarz et al. 1985). The only domesticated animal resource was the dog. In the contact period this animal was eaten by the Huron, apparently mostly in the context of religious ceremonies (Sagard 1939:220, 226). Ongoing isotopic analyses are aimed at further elucidating the dietary role of maize in southern Ontario (Ron Williamson, personal communication 1999).
Ontario.
It has been suggested (Chapdelaine 1993:178) that Iroquoian men were very mobile between April and November, while women seldom left the village for a lengthy period. Based on ethnographic sources it is likely that women and children focussed on local gathered resources, while men were more involved with hunting and fishing. Both men and women may have been involved in the trapping of fish and smaller mammals, such as squirrels and muskrat. The contact period Huron engaged in mass deer hunting drives in areas further south, towards Lake Ontario (Heidenreich 1971:207). There is no evidence from precontact sites in the area for such intensive deer exploitation (Robertson et al. 1995). Heidenreich (1972:71) suggests that in general hunting by the contact period Huron brought small returns. In contrast, he argues that fish were a plentiful resource that reproduces rapidly. Fish were in many ways more predictable and reliable in terms of location and seasonal habits, they were easy to catch, and could be dried and stored for long periods of time. According to Heidenreich it is not surprising that for the Huron fishing was more important than hunting (Heidenreich 1971:212). Some of the numerous trails used by the Huron (Hunter 1906) may have been in use in earlier times, thus allowing for easy movement between different locations in the area between Lake Simcoe and Nottawasaga Bay. It is likely that the communities in the area between Lake Simcoe and Nottawasaga Bay were in contact with other Iroquoian communities south of the Oak Ridges Moraine. Ethnohistoric accounts indicate that the contact period Huron traded ceramic pots, chert, maize, fishnets and tobacco to the hunter-gatherer Algonkian-speaking people on the Canadian Shield to the north, in exchange for native copper, dried berries, dried fish, furs, reed mats and meat (Heidenreich 1971). The Algonkian-speaking Odawa, located on the Bruce Peninsula, appear to have played a pivotal role as middlemen in the trade between the Iroquoian groups and Algonkian groups in the French period (Smith 1996). The precontact origins of regional trade contacts are reflected in finds at the Barrie site of a native copper needle, small amounts of Hudson Bay Lowland and Detour chert that must have come from northern Ontario, probably via the Odawa (Sutton 1996a:128), as well as Onondaga chert from the Iroquoian- speaking people in south-western 5
CHAPTER 2: SETTLEMENT PATTERNS
Introduction For readers not familiar with Iroquoian villages, some key features of the settlement pattern are discussed below. An example of a house plan with interior feature types is provided in Figure 3. At Barrie and Dunsmore feature types were designated by the field archaeologists, using criteria developed by the consulting firm Archaeological Services Inc. Some aspects of these feature types were discussed in a previous publication (Needs-Howarth and Thomas 1998). Feature type interpretations for the Carson site have not yet been made available; instead, tentative interpretations were made by the author, based on the archaeologists’ original field notes.
Ash pits are usually smaller and shallower than regular pits, and contain mostly ash in the fill. They are almost always located near a hearth. They were probably temporary containers for ash removed from hearths. The ash may have had industrial uses, such as hide tanning, corn preparation (Waugh 1973) or sanitation (i.e. odour suppression). Positioned next to hearths, ash pits did not interfere with traffic; they could, therefore, be left uncovered, allowing some floor debris to accumulate in them. Because ash pits are so shallow, the contents may reflect a short depositional period (Ron Williamson, personal communication 1994). Large flat-bottomed pits surrounded by a ring of peripheral posts of a small diameter, with interior hearths and an entrance sloping upward from the bottom of the pit, are argued to be the remains of sweatlodges, used for ceremonial and ritual activities (MacDonald 1986; MacDonald 1992). They are located inside or adjacent to longhouses, with an entrance from the longhouse. These semi-subterranean features often contain a top layer of dark organic soil mixed with artifacts, resulting from re-use as a refuse pit. The reasons for their abandonment are not clear. In some cases, lack of layering in the fill indicates they were filled in rapidly after their original intended use had ended. Because of their large size, semi-subterranean features which do contain layered fill could represent a more prolonged period of deposition.
Village boundaries are often marked by a natural feature, such as an abrupt break in slope. Villages typically consist of a number of longhouses with central hearths. Each hearth is thought to have accommodated two families. House walls are usually evidenced by a dense pattern of postmoulds. House walls were often repaired, rebuilt, extended or contracted. Feature types The most recognizable and bone-rich features at Iroquoian sites are the outdoor refuse dumps, or middens. All features, and especially middens, may contain superimposed evidence of many different procurement and refuse deposition events. In the absence of clear horizontal layering and/or piece plotting (as recommended in a footnote by Stewart 1991a:69), faunal remains from midden contexts are not all that informative of fine-scale human activity. Middens may relate more to community-based processing (Smith 1996). As Ramsden (1996:106) puts it “the complex catchment patterns of middens make them all but useless as sources of artifactual samples, for any but the most general of analytical purposes.” Complaints about middens are not restricted to Ontario; writing about Iroquoian sites in New York, Kuhn (1986) suggests that greater attention to both the spatial patterning of artifacts and the systemic processes of midden formation is warranted.
Each of these feature types implies a different refuse deposition history. It is likely that pits used for storage and other specific activities were filled in fairly quickly once their original intended use had ended. An empty, disused pit may, therefore, contain refuse from subsistence activities occurring over a short time period. In contrast, a midden accrues refuse over the course of many years. Middens are more likely to contain fill from diverse and unrelated activities that occurred over a long period of time. Of all types of cultural deposit, middens are least well suited for fine-grained study of individual subsistence events or clusters of temporally related events. The fill of some larger features, especially those with recognisable stratigraphy, while providing suitably large sample sizes, may contain superimposed evidence of more than one procurement and refuse deposition event. Furthermore, a certain amount of faunal material dropped on the earthen floor of a longhouse will be trampled into the surface, and traces of this material will find its way into any open feature. Finally, any feature could contain food debris representing the remains of fish caught and preserved months before they were actually consumed (Needs-Howarth and Thomas 1998).
Pits were used to store food and personal effects, or to bury refuse (Chapdelaine 1993). It appears that pits used for storage and other specific activities were filled in with refuse once their original intended use had ended. Pits are therefore somewhat similar to middens in faunal composition, although they are usually smaller, and their duration of use may have been shorter. Storage and refuse areas may be distinguished on the basis of relative artifact heterogeneity (Kent 1999). However, a filled in storage pit may contain refuse produced just after it was last emptied. While this may represent a small window of time, there would also be a lot of “noise” from incomplete cleaning out of previous pit contents (Chapdelaine 1993:187; Thomas 1997b).
Barrie The Barrie site is located on the outskirts of the contemporary city of Barrie on a sandy loam terrace overlooking Minesing Swamp, a lowland cedar swamp (Figure 2). The terrace occupied by the site is 25 metres 6
above the floor of a wide, flat-bottomed valley that is part of the Simcoe Lowlands. The Simcoe Uplands rise 55 m higher behind the site (Sutton 1996b). The site is bisected by Dyment’s Creek, which issues partway up the upland slope just north of the site. This creek flows southward down onto the valley floor and then turns eastward to drain into Kempenfelt Bay, currently three km east of the site (Sutton 1996a).
The recovered faunal material was packed in paper bags and kept separate from other artifacts to minimize postexcavation damage and deterioration. All faunal material was washed and air dried. After analysis, each identified specimen was re-packed in a small, labelled zipper-style bag with an integrated label for long-term curation. The recent Barrie site collections were donated for curation to Huronia Museum in Midland by the excavator.
Test excavations by James Hunter in 1976 sampled one of the middens (Hunter 1978). The eastern half of the site was excavated in 1991 and 1992 by Richard Sutton as part of his doctoral research at McMaster University in Hamilton, Ontario (Sutton 1996a; Sutton 1996b).
Ceramic seriation and associated cluster diagrams place the site firmly in the Uren sub-stage, A. D. 1280-1330 (Sutton 1996a) (Table 2). The wood charcoal sample is dominated by maple and beech, followed by elm (Ulmus americana), pine, ironwood (Ostrya virginiana), ash (Fraxinus sp.), tamarack (Larix laricina), oak and birch (Betula sp.), suggesting the presence of a mature maple-beech forest in the area (Monckton 1993). However, substantial quantities of fleshy fruits indicate that disturbed and forest edge habitats were also present near the site. Maize represents 20 percent of the plant remains. The absence of beans, squash and sunflower may be due to small sample size and/or poor preservation of the former two taxa (Monckton 1993).
The ploughzone was highly disturbed. Following local excavation convention, it was removed with the use of a “Gradall” earth moving machine. Ploughzone stripping with this accurate type of equipment allowed subsurface settlement features to be exposed and did not damage material in the undisturbed portions of features and middens. All in situ deposits were exposed and excavated using shovels and trowels. Fills of ash pits, hearths and other sensitive features were trowelled, allowing for the documentation of in situ articulations. The remaining features and middens were excavated by shovel.
Dunsmore The Dunsmore site is located about 4.5 km northwest of Kempenfelt Bay on a broad flat promontory overlooking a small tributary of Willow Creek, which connects to the Nottawasaga River via Minesing Swamp (Figure 2). The site is north-west of the city of Barrie, about 2.5 km west of Little Lake. To the north, the site boundary is defined by the presence of a gully, while to the south and west the land slopes more steeply to the tributary and associated wetlands (Figure 5) (Robertson and Ramsden 1996b).
All fill was screened on a 6.4 mm shaker screen suspended from a tripod, except for a total of 751 litres of bulk soil samples subjected to flotation. The flotation samples were processed on-site, using the two-bucket method. The heavy fraction was screened through a 2 mm mesh geological sieve (Monckton 1993). The site was probably .8 or .9 hectares in size, with an estimated population of 308 people (Sutton 1996a:174). About 17 percent of the total site area was uncovered (Figure 4). House 2, 17.6 m long and 6.1 m wide, was completely excavated. Interior feature density suggests that House 2 was “occupied for a considerable length of time” (Sutton 1996a:196). House 1, 32.2+ m long, 6.7 m wide and House 3, which is overlapped by the south end of House 1, were partly excavated.
Test excavations by James Hunter in the early 1970s sampled one of the middens (Hunter 1978). Dunsmore was salvage excavated in 1989 by Richard Sutton (field supervisor) and Ronald Williamson (project director) of Archaeological Services Inc. (ASI), an archaeological consulting firm in Toronto, retained by the Rose Corporation, a commercial developer. Excavation and bulk sampling procedures were identical to those employed at the Barrie site (Robertson and Ramsden 1996b), although flotation sampling was less intense (Table 4).
The faunal sample from the 1991-92 salvage excavations derives from the undisturbed areas of three middens (18 m2), 29 features within the three houses, and four exterior refuse pits (Figure 4, Table 3) (preliminary data in NeedsHowarth 1995a). While five midden areas were sampled, zooarchaeological analysis was restricted to samples from three of them. Midden A covers about 85 m2, of which 18 m2 was excavated. Zooarchaeological remains from 12 m2 were analysed. The northwest section was undisturbed. Midden B extends beyond the 20 m2 that were excavated. Zooarchaeological remains from 5 m2 were analysed. Most of this midden was undisturbed. Midden D appears to have started out as a deep pit, 4.3x3.6 m, which was gradually expanded and filled with refuse. The zooarchaeological sample from one square was analysed.
The size of the Dunsmore site is 1.9 ha, of which 1.5 ha was investigated during the 1989 salvage excavations. These uncovered 16 structures and four middens, relating to three house clusters. While some of the structures in the northeast cluster may have been seasonally occupied, the houses from which the zooarchaeological remains derive were most likely permanent residences (Robertson and Williamson 1996). The houses in the south-central and west clusters that were the subject of zooarchaeological analysis were all permanently occupied. The largest structure is House 7, at 54 m long (Robertson and Ramsden 1996b).
7
Again, summary identifications of bone material from the earlier test excavations (Thomas 1978) are excluded from this discussion because the method of quantification is not compatible. Analysis of the recently excavated faunal sample concentrated on a non-random sample selected by the excavator, including the undisturbed areas of three middens (24 m2), and 81 features within 10 houses (Table 4). It was analysed by the author (preliminary data in NeedsHowarth 1994) and Stephen Cox Thomas (1996d) (employed by ASI). The laboratory analysis was conceived and carried out as a collaborative project.
different time or for a different purpose” (Archaeological Research Associates Ltd. 1990:10). This assemblage was previously analysed in part by two students as part of an introductory course in zooarchaeology at University of Toronto. The current work is effectively a partial re-analysis of a portion of the sample analysed by Crane (1990) and Dompierre (1990). It also includes additional contexts that were not previously analysed or that were not previously available (i.e. the heavy fractions) (preliminary data in Needs-Howarth 1997b). The faunal sample under discussion here was intended to comprise all the finds from three houses (Table 5). Because of inconsistent field laboratory bag labelling procedures it was, however, not possible to ascertain whether all bags from these houses were, indeed, examined. House 1 and House 5 are both 52 m long, House 3 is 59 m long (Varley and Cannon 1994). The only feature with significant numbers of bones is House 5 Feature 1, a refuse pit measuring about 55 cm across and 23 cm deep. For the purposes of comparison with other sites, one midden context (Midden 4) was also examined. The generally excellent state of preservation is attested by the ubiquity of fish scales. The collections were donated to Huronia Museum for curation in 1989 by the landowner.
After washing, bones were dried in a cabinet heated by a light bulb. This is not considered to have had any detrimental effect on the faunal remains. After analysis, bones were repacked together by context in paper bags. The Dunsmore collections were donated for curation to Huronia Museum in Midland. An AMS radiocarbon date on corn of A. D. 1430-1510 (Cal.) concurs with ASI ceramic seriation (David Robertson, personal communication 1998) (Table 2). Cultigens made up a small percentage of the seeds in the analysed floatation samples; fleshy fruits appear to have been more important (Monckton 1996). The wood charcoal assemblage is dominated by maple, beech and white pine (Pinus strobus), with ironwood, ash, and elm of secondary importance, suggesting a secondary successional mixed forest (Monckton 1996).
A radiocarbon date from the site of A. D. 1507±27 is argued (Varley 1993) to be in agreement with ceramic seriation (Table 2). Wood charcoal data are not yet available. The ceramic assemblage contained substantial numbers of Lalonde High Collar rimsherds, which occur in a small area of the area between Lake Simcoe and Nottawasaga Bay in the fifteenth century.
Carson The Carson site is located close to the Dunsmore site on a former beach ridge near Little Lake (Figure 2). The site was salvage excavated by Archaeological Research Associates Ltd. (1990) in 1988 and 1989, prior to construction of a proposed housing sub-division. Parts of the site were excavated by supervised high school students enrolled in an archaeological field school. The disturbed ploughzone was stripped with the use of a bulldozer. All subsurface features were excavated with shovels and trowels. Small flotation samples were taken from half the fill of all features. Ninety-seven flotation samples, representing 241 litres, were processed in 1997 by Della Saunders with the aid of the SMAP style flotation system at Department of Anthropology, University of Toronto at Mississauga. Heavy fractions were recovered with a 2.4 mm geological sieve. The remainder of the sectioned fill was dry screened through 3.2 mm mesh. The village was approximately 2.8 hectares in size, consisting of eight houses and two large middens (Figure 6). Excavation concentrated on six of these houses, ranging in length from 52 m to 75 m, and both middens. Construction details of House 6 suggest “it may have been built at a 8
CHAPTER 3: ANALYSIS PROCEDURES
Identification The objective of the laboratory component of this analysis was to identify as many bones as possible to species or genus, and to fully describe them with respect to element, portion (both verbal and coded), skeletal development and alterations. The number of specimens of all classes that could be identified below class (number of identified specimens, or NISP) at the Barrie site is 737; at Dunsmore 905; and at Carson 852.
In addition to all non-artifactual (unworked) bone identified below class, the database and spreadsheet include bone artifacts and tools (worked bone) that could be identified below class for those contexts also containing unworked identifications. These selected worked bone artifacts were included to enable detection of the use of larger mammals and birds whose flesh may have been eaten before the bones were modified into scrapers, awls or beads. In other words, only those worked bones that are comparable in level of taxonomic detail to the unworked bone were included. Some non-fish bone could be re-fitted. These joined specimens were allocated a single catalogue number in order to reduce the degree of interdependence between specimens.
All bones were examined and identified to element and taxonomic level by comparison to a reference collection. Most identifications were made at the Howard Savage Faunal Archaeo-Osteology Laboratory at the University of Toronto. Material that could not be satisfactorily identified at this facility was taken to the Royal Ontario Museum Centre for Biodiversity and Conservation Biology in Toronto. Some specimens were identified at the Zooarchaeological Analysis Programme, Canadian Museum of Nature, in Ottawa.
The identifications were entered in a faunal analysis entry program originally developed by Thomas (1991), parts of which have subsequently been modified by the author for use in Corel Paradox 8©. NISP and BW for the identified and unidentified portions of each sample, organized by taxon and context, were entered into a Corel Quattro Pro 8© spreadsheet, summaries of which appear in this thesis. While the database distinguishes between the subassemblage recovered through dry screening and the subassemblage recovered through floatation, these have been combined in some instances to increase sample size. Unless otherwise noted, data pertain to this combined sample. The full electronic datasets are curated by the author. To simplify the spreadsheet tables, less certain identifications are included with positive genus or species identifications. Taxonomy follows Mandrack and Crossman (1992) for fish; Logier and Toner (1961) for amphibians and reptiles; Godfrey (1986) for birds; and Honacki et al. (1982) for mammals. Because of the GFAUNA2 coding system, the position in the phylogenetic sequence of the family Esocidae and the family Salmonidae continues to follow an earlier source (Scott and Crossman 1973).
Stephen Thomas (1996d) was responsible for approximately one third of the below-family identifications of the Dunsmore site; the remainder of the Dunsmore collection (including the “unidentified” component), as well as all of the Barrie and Carson collections, were analysed by the author. Salmonidae vertebrae from Barrie and Dunsmore were identified by the author and Stephen Thomas (NeedsHowarth and Thomas 1998). Non-Salmonidae vertebrae from these two sites, as well as all vertebrae from Carson, were analysed by the author. To ensure consistency in analysis and descriptions between the collections, the three collections were examined a second time by the author, in the same sequence in which they were examined initially (first Barrie, then Dunsmore, then Carson). This allowed insights gained during the analysis of the later collections to be applied to the earlier ones.
Binomial and common names for all taxa are presented in Table 6. Details on taxonomy, biology, ecology and osteology of the major fish families are presented in Table 7. While it might be considered preferable to use common names throughout, in the case of fish family names the scientific name will be used to avoid ambiguity. For example, the “salmons” family (Salmonidae) in this context does not include any species with “salmon” in their common name. Scientific family names are used to indicate familylevel identifications (e.g., “suckers” indicates white sucker and longnose sucker, whereas Catostomidae could also include members of the genus Moxostoma).
Anatomical terminology follows Miller (1965) for mammals, Gilbert, Martin and Savage (Gilbert et al. 1981) for birds, and Lepiksaar (1983) for fish. Following convention in nomenclature in Ontario, some of the spelling of fish osteology has been retained from Mujib (1967), except that angular is here called articular, following Lepiksaar (1983) and Courtemanche and Legendre (1985). All fish cranial and pectoral girdle elements have been identified whenever possible. Fish vertebrae and scales were assigned to family or genus where possible. Size qualifiers are included whenever a vertebra was significantly smaller (< or >) than the range in the standard reference specimen. Because fish vertebrae and scales are not routinely identified by Ontario zooarchaeologists, these identifications are here excluded from the fish NISP and tabulated separately.
The database employs a number of different levels of identification to guarantee distinction from other taxa at the same level of taxonomic specificity (Driver 1992). It should be noted that, while there are many osteological similarities between domestic dog and wolf (Canis lupus), evidence from dog burials and other sources suggests Iroquoian 9
domestic dogs in this area were much smaller than wolves, and there does not appear to have been any size overlap in adult individuals (Stephen Thomas, personal communication 1995). In an Iroquoian context, therefore, it is possible to distinguish between these two species on the basis of absolute element size and state of maturity. This is reflected in the relative lack of Canis sp. identifications. Thomas measured a suite of dog teeth and metapodia from the Carson site and confirmed that all the Canis cf. familiaris material is most probably domestic dog, rather than wolf.
balance probably would have revealed considerable variation in the weights of individual fish bones, this added precision could be misleading for material that was subject to differing rates of bone decomposition and sub-fossilisation, and that sometimes includes some adhering soil matrix. It is explicitly acknowledged here that both below-class BW and NISP, “will depend a great deal on the ability to identify all anatomical components of the different taxa in the collection with equal confidence” (Leach 1997:6). The identification of a subset of bones that are robust, speciesdiagnostic and/or relevant to the interpretation of fish butchery patterns is becoming common practice among zooarchaeologists working in Europe and the Pacific (e.g., Barrett 1997; Colley 1990; Enghoff 1991; Leach 1986), but applications involving North American fish bone assemblages are less common (e.g., Butler 1996; Hartzell 1992).
Animal species which are separable by zoologists on the basis of external markings or genetics are not necessarily separable on the basis of osteology (Driver 1992:39). This is especially true of fish (see Parmalee et al. 1972 for an explicit recognition of this problem). In addition, while the osteology of fish is well described (Cannon 1987; Courtemanche and Legendre 1985; Lepiksaar 1983; Rojo ), inter-individual variation among species is not (see Brinkhuizen 1989:42-53). Species in some genera are known to hybridise, especially lake herring and lake whitefish; Esocidae; and pumpkinseed and related taxa (see Table 7).
A growing awareness that fish bone datasets from archaeological sites in southern Ontario are uneven in terms of element representation, element completeness, and/or degree of taxonomic specificity (Needs-Howarth 1995b), prompted an investigation into the relationship between these factors, and their effect on NISP and MNI (NeedsHowarth 1999). The development of a set of diagnostic elements for Great Lakes fish was initiated in 1997 using data from the Barrie, Dunsmore and Carson sites. Qualitative and quantitative assessment of this variability at the family level has resulted in a preliminary list of eight cranial and pectoral girdle bones that are robust and speciesdiagnostic in these collections (Needs-Howarth 1999). The following section consists of excerpts from the 1999 paper.
Identifying fish to family level may be sufficient for some research questions, however, it will become clear that species identifications are necessary to assess fish subsistence in detail. More detailed taxonomic identifications also allow researchers to spot rare taxa (e.g. Colburn 1997). The biology and behaviour of certain species, especially Centrarchidae, is poorly known for these bodies of water and the precontact species distribution is highly uncertain. Depending on sample condition, quality of the reference collection, and minor differences in osteology through time and space, it is not always certain what species the archaeological material represents. This is a problem in the genus Esox, and also among several of the Centrarchidae. Especially with fish, it is often possible to distinguish species within the same genus by one element, but not another (this seems to be particularly problematic with the genus Catostomus). Some fish, especially yellow perch (Perca flavescens), are easily identified to species because many of the bones are distinctive. Because fish are the focus of this study, this issue receives some further consideration in the following section.
With the exception of lake sturgeon (Acipenser fulvescens), the fish families represented on archaeological sites in the Lake Simcoe/Nottawasaga Bay area are osteologically similar. Sturgeon have few ossified cranial bones, which may vary in morphology and number (Brinkhuizen 1986). Only 17 cranial and pectoral girdle bones were (tentatively) identified to element in a collection of 85 lake sturgeon remains from the Barrie site (Needs-Howarth 1995a). Sturgeon do, however, possess highly distinctive ossified dermal scutes, which are analogous to scales. Scales, however, are more fragile, less likely to be recovered using standard 6.4 mm aperture dry screen mesh, and less readily identifiable to species than sturgeon scutes. Despite this bias, lake sturgeon scutes were explicitly included in the below family NISP to compensate for the inherent lack of ossified sturgeon cranial bones. In addition, the pectoral spines of lake sturgeon are completely ossified, which is not the case in most other fish species represented in these samples. In light of these major osteological differences, it is problematic to define diagnostic elements for lake sturgeon.
Quantification As was recently reiterated by Leach et al. (1997), the first step towards understanding past fishing behaviour from archaeological remains is to establish the relative abundance of different types of fish. The bones from the three assemblages under consideration here were quantified in terms of NISP and BW. BW is used to elucidate taphonomic issues, rather than to establish taxonomic abundance per se. BW was recorded for each individual identified element, using an electronic balance with a precision of ± .1 g. Not surprisingly, most identified fish bones weighed .1 g. Even larger groups of unidentified fish bones from a single context frequently weighed .1 g. While a more precise
Ictaluridae also have highly distinctive ossified pectoral and dorsal spines (over 35 from the three sites combined). The disproportionate number of pectoral spines (9-12 percent of 10
Ictaluridae NISP) found at all three sites indicates brown bullhead (Ameiurus nebulosus) skeletons were probably favoured as a source for tools. In addition, both Ictaluridae and Catostomidae have distinctive vertebrae relating to the swim bladder (the Weberian apparatus), fragments of which, including the modified ribs, can often be identified to family or genus. Catostomidae and Centrarchidae possess ossified pharyngeal arches/plates that also can be identified to family, genus and occasionally to the level of species.
elements. However, Esocidae, Catostomidae and Ictaluridae, with a comparatively large NISP for the 14 commonly identified elements, each lack several of these elements. In that light, the fact that Salmonidae only lack four of the 14 commonly identified elements is surprising. This may relate to the distinctive surface appearance of Salmonidae, especially lake trout, bones. As will be discussed in more detail below, the combined effects of off-site processing and poor preservation properties of Salmonidae cranial bone have almost certainly caused Salmonidae vertebrae to be both more numerous and more ubiquitous than cranial bones (Needs-Howarth and Thomas 1998). Because the taphonomic history of the Salmonidae likely differs considerably from that of other families, it might be advisable not to use diagnostic elements for this taxon, although the eight diagnostic elements do appear to be well represented. Other taxa that are less frequently identified in this part of Ontario, such as longnose gar (Lepisosteus osseus), bowfin (Amia calva) and burbot (Lota lota), may require a different approach. One could decide to identify all bones from rarer taxa (Nicholson 1998:18), or, if a taxon is represented only by non-diagnostic elements, it could be marked simply as present.
As has been observed by other researchers (Gobalet and Jones 1995), some elements of some taxa are over-represented in zooarchaeological assemblages because they are inherently robust or easily distinguished from the same element in other taxa. Despite general similarities in bone shape, size, and structure among most of the Great Lakes fish families, bones from the cranial and pectoral girdle are neither consistently recovered nor consistently identified to genus or species. This is likely the result of a number of factors, including taphonomy; osteological similarities and differences relating to phylogeny; and the previously mentioned possible hybridisation within the genera Coregonus, Esox and Lepomis. However, even if special elements such as the Weberian apparatus, the pharyngeal arches, pectoral and dorsal spines are eliminated from consideration, some major differences in taxonomic and element representation remain. It appears that the comprehensive identification of most cranial and pectoral girdle bones, as it has been traditionally carried out by researchers in the area, while producing a very complete dataset, may not necessarily result in the most bias-free data-set for the purpose of inter-taxa comparisons of relative taxonomic abundance.
Elements that are well represented in the remaining five major fish families may be good candidates for diagnostic elements (Table 8, Figure 7). However, given that the current research questions necessitate the use of species-level identifications whenever possible, determination of diagnostic elements must also take into consideration the level of specificity of within-family taxonomic designations, as well as the coded and verbal descriptions of element completeness in the original databases. In other words, abundant representation and tight clustering in Figure 7 is not the only criterion for selection of diagnostic elements. Qualitative assessments of bone breakage patterns for each element by family (NeedsHowarth 1999) suggest that the most important determinant of the suitability of diagnostic elements is morphology, rather than preservation or element completeness. Of particular importance are intra-taxon similarities and differences in morphology, and element robusticity, which is here defined as relating to the shape and size of the bone. This is in accord with Nicholson’s (1992) finding that fish bone survival is mediated not so much by bone density, but by bone shape and size.
In order to increase the clarity of patterning, the definition of diagnostic elements was based on the combined fish NISP by family of the three samples. Osteological similarities between closely related species may justify determining diagnostic elements at the family level. Similarly, osteological similarities within families facilitate the use of the same set of diagnostic elements throughout the Great Lakes area, although biogeographic differences between watersheds would necessitate a modification of the taxa included, for example the addition of American eel (Anguilla rostrata) and Atlantic salmon (Salmo salar) for the Lake Ontario/St. Lawrence River watershed. Based on a qualitative assessment of eel remains from the Steward site (BfFt-2), Junker-Andersen (1984:114) suggested the dentary, vomer, parasphenoid and operculum are the most readily identifiable and abundant.
Based on the current data, it appears that eight elements are, for the most part, reliably and consistently identified to species in the six major fish families: the articular, ceratohyal, cleithrum, dentary, hyomandibular, operculum, preoperculum and quadrate (see designation “*” in Table 8). There is some variability in ease of identification. For example, Ictaluridae hyomandibulars are not always identifiable to species, something that was noted also for the Middle Woodland Hector Trudel assemblage on the St. Lawrence River in Quebec (Cossette 1995:395) (while this site is culturally distinct from the three sites under discussion here, the in-depth analysis of the fish remains
As a preliminary step, elements that occurred 15 times or more in the combined sample were tabulated. Diagnostic elements were selected from these 14 commonly identified elements (Table 8). As might be expected, there is a general trend towards more consistent representation of different elements with increasing family NISP. Centrarchidae and Percidae, which are best represented in terms of NISP, are the only two families to include all 14 commonly identified 11
provides useful comparative data on aspects of osteology and taphonomy). These 8 diagnostic elements should be compared in a systematic manner to other Great Lakes assemblages to establish to what extent the findings at the Barrie, Dunsmore and Carson sites hold true at other sites. It is likely that this list of 8 elements is to some extent assemblage-specific, relating to the sizes of fish, butchering and processing, preservation conditions and recovery.
interesting to note that the relationship for Percidae is the most consistent between sites. Very likely this relates to the large numbers of this taxon recovered in all three assemblages. As has been noted above, similarities at the family level mask some differences at the species level, highlighting again the need for larger sample sizes. Esocidae elements at the Barrie site are less evenly represented than at Dunsmore and Carson. Articulars are substantially under-represented with respect to the adjacent element, the dentary, while at Dunsmore the relationship is 1:1. Esocidae dentaries are also over-represented at the Hector Trudel site (Cossette 1995:374). While not all the Dunsmore articulars and dentaries match into real pairs, the taphonomic histories of Esocidae do appear to vary between sites. Alternatively, the discrepancy may relate to sampling error.
Combined, these eight diagnostic elements comprise 62 percent of the cranial and pectoral girdle bones identified below family for the six major fish families, and 81 percent of the 14 commonly identified elements. The fact that these eight elements make up much less than 62 percent of the total suite of elements in the fish head that are considered to be “identifiable” by many zooarchaeologists may confirm their utility for. The remaining six commonly identified elements are problematic because of uneven representation across the six major families. It appears that this can be attributed mostly to differences in osteology, mechanical strength and associated susceptibility to fragmentation and other taphonomic factors, rather than small sample size.
Catostomidae represent some contrasts. The cleithrum is most abundant at Dunsmore, yet absent at the two other sites. All Catostomidae cleithrum identifications at Dunsmore were made by Steve Thomas (1996d), which opens the possibility that this element was not recognized in fragmentary form by the author. Once again, articulars are under-represented in relation to the dentary.
The comparatively low proportion that the 14 commonly identified elements represent of the total NISP of Catostomidae and Ictaluridae results from inclusion in the unmodified NISP of Weberian vertebrae and modified ribs for the former and dorsal and pectoral spines for the latter. For Catostomidae and Ictaluridae there is a considerable difference between unmodified NISP and the 14 common elements, while there is a much more consistent relationship between the 14 common elements and the eight diagnostic elements.
Ictaluridae element representation is very even and very similar between the three sites. Note that Figure 8 includes the pectoral spine, so that a comparison may be made between representation of this and the articulating element, the cleithrum. Three elements at Carson have high representation: pectoral spines, opercula and cleithra. Once again, articulars are under-represented. Pectoral spines at Barrie and Dunsmore are under-represented with respect to the articulating element, the cleithrum, indicating that they were perhaps curated for tool use, preventing them from being discarded at the site. Spine fragments lacking the medial articulation may have fallen through the screens. Given the inclusion of the palatine in other lists of diagnostic elements, it should be noted that no brown bullhead palatines were identified. This concurs with recovery tests conducted at the Hector Trudel site assemblage (Cossette 1995:249), where brown bullhead palatines were only found in mesh sizes smaller than 6.4 mm.
Salmonidae articulars and quadrates tend to be over-represented (Table 8, Figure 7, 8), as are Esocidae articulars and dentaries; and perch opercula (Table 8, Figure 8). Over-representation of Esocidae articulars and dentaries was also noted at the Hector Trudel site (Cossette 1995:374). In addition, the robusticity of Ictaluridae and the osteological distinctiveness of yellow perch will likely cause the overall relative abundance of these taxa to be exaggerated with respect to what was originally deposited, even after diagnostic elements are applied. In general, these eight diagnostic elements are probably more applicable to larger samples, because individual elements of the rarer taxa will be less well represented in smaller samples. However, while it may be tempting to expand the suite of diagnostic elements for smaller samples, this would almost certainly result in a more distorted relative taxonomic representation.
Centrarchidae distribution is also quite similar between sites, although Barrie has notably fewer cleithra. Numbers of dentaries and articulars are similar at Dunsmore and Carson, while at Barrie articulars are under-represented.
Table 9 shows the relationship of the eight diagnostic elements to the conventional NISP at each site. Esocidae are on average well represented by the eight elements, as are Ictaluridae, especially considering that the pectoral spine is not among the diagnostic elements. The low agreement between conventional NISP and diagnostic elements for Catostomidae relates, at least in part, to the exclusion from the diagnostic elements of the Weberian complex. It is
Percidae distribution is very even and similar. The relationship between dentaries and articulars is similar to Centrarchidae. It is likely that this relates, at least in part, to similarities in osteology in the order Perciformes, and associated breakage and recovery patterns. Table 10 presents a summary of the NISP per species/genus for the 14 commonly identified elements and the eight 12
diagnostic elements. While diagnostic elements represent between 74 and 60 percent of the conventional NISP, use of these eight elements instead of all potentially identifiable elements does not alter the rank order of the taxonomic identifications. Indeed, the proportional contribution is minimally changed; it is reduced by at most four percent and increased by at most six percent. On the one hand, these differences are not as marked as might have been feared based on the above discussion of differential representation of elements. On the other hand, Table 10 is a good advertisement for the ability of the eight diagnostic elements to predict rank order, and, to a large extent, proportional contribution of individual taxa. It is interesting to note that the same species can gain or lose in proportional contribution. This is no doubt assemblagespecific, and depends on which other species were over- or under-represented in the traditional NISP. The utility of these proposed eight diagnostic elements is in conjunction with a more traditional analysis, as a way of understanding how samples may be biassed. For samples of different sizes analysed by zooarchaeologists with different training, using different reference collections and operating under varying financial and time constraints, the use of diagnostic elements can increase the validity of both interand intra-site comparisons. Data from other Great Lakes assemblages are needed to confirm that the patterning observed at Barrie, Dunsmore and Carson is indeed related to osteology and morphology, rather than butchering or recovery. During the current work, diagnostic elements are used to qualify rather than quantify the data.
13
CHAPTER 4: TAPHONOMY
Introduction Before discussing the possibilities of fish bone interpretation, it is essential to reflect on the limitations imposed by taphonomy – extending from bias introduced by butchering practices, to bias introduced by soil preservation conditions, to bias introduced by field recovery techniques. Some of the issues discussed below are relevant to fish bone analysis in general, some to zooarchaeological analysis in general. This chapter, however, concentrates on taphonomic factors that are specific to the fish taxa recovered, to the culture area, and to the excavation context.
vertebrae belonged are accounted for by cranial bones in the identified component of the sample. There are some contexts with only a few vertebrae. Most features contained under 10 identified vertebrae per family. There are no features with more vertebrae identified to family than could be accommodated in a single fish. Can the cranial to vertebral ratios enlighten us about off-site processing, perhaps at a fishing camp? Establishing “expected” cranial to vertebral ratios is fraught with difficulty. The number of cranial bones varies between taxa, and the number that are routinely identified will also vary between taxa and among researchers. For cranial bones, the “expected” number of bones relates to what is considered analytically useful, which taxa are present, and how much time and expertise is available. The “expected” number of cranial bones will usually include the eight diagnostic elements discussed above, and a variable number of other elements per species.
Butchering and processing Many Iroquoian assemblages exhibit little evidence for butchering or burning on fish remains. It is likely, however, that fish underwent varying degrees of processing at the catch site, at the village, or inside the houses. Sagard (1939:186, 316-318) describes how the Huron gutted lake sturgeon and Salmonidae to dry them in the sun, or if the weather was unfavourable, to smoke them on hurdles or poles. The oily flesh of lake sturgeon and Salmonidae could be preserved by smoking. Natural freezing has also been suggested as an option (Cleland 1982: 779; Rostlund 1952:137), although Molnar (1997:30) argues that temperatures were too inconsistent to practice this method effectively in the Georgian Bay area. Sometimes the Huron boiled lake whitefish in kettles, skimming off the oil with a spoon and putting it into hollowed-out gourd containers (Sagard 1939:186). It appears preservation through salting was unknown.
Taphonomy plays a major part, as many bones will be destroyed or fall through the dry screen mesh. Vertebrae will be lost if their diameter and length are less than the screen mesh aperture. Caudal vertebrae and vertebral centra can be lost because they lack the projections that may prevent them from falling through the screen (e.g. Prevec 1985). For vertebrae, therefore, the “expected” number of elements is in large part dependent on recovery. If all vertebrae are recovered and are in good condition, it should be possible to identify many of them at least to the family level. The most difficult distinction in these assemblages is between Centrarchidae and Percidae, something noted previously by Cossette (1995:403).
Off-site processing Catch site processing is an important source of taphonomic bias. One way of establishing whether fish were decapitated or otherwise butchered at the catch site is to look at the ratio of cranial bone to vertebrae at the village. Fish vertebrae have been used to evaluate the importance of lake trout, lake herring and lake whitefish, and to examine butchering patterns by several local researchers (Cooper 1996; Molnar 1997; Thomas 1996a).
A total of perhaps 40-60 “expected” number of cranial bones appears to be a reasonable estimate for Esocidae, Catostomidae, Ictaluridae, Centrarchidae and Percidae. If fish were brought to the site whole, the ratio of cranial bones to vertebrae for these taxa can be expected to range from 2.1:1 to .9:1. However, this ratio is likely much lower for Salmonidae (Salmonidae taphonomy is discussed in more detail below). Given that lake trout possess up to 69 quite robust vertebrae and have especially fragile cranial bones, the “expected” ratio may be only .4:1. Lake herring and lake whitefish have somewhat fewer (50-64) and less robust vertebrae than lake trout, so the ratio there would be intermediate.
If the village inhabitants were decapitating and filleting fish at the catch site, and bringing only the cleaned fish back to the village, the posterior pre-caudal and all caudal vertebrae may be expected to be over-represented at the village. Unfortunately, relatively few vertebrae were recovered from the three sites. An attempt was made to identify the majority of vertebrae to family, and those of Salmonidae to genus. The proportion of vertebrae that could be identified below class is much higher at Barrie (80 percent) than at either Dunsmore (46 percent) or Carson (48 percent).
Overall, cranial bones outnumber vertebrae: Barrie 3:1, Dunsmore 14:1, and Carson 16:1 (Table 11, 12, 13). There are, therefore, substantially fewer vertebrae than would be expected based on the range of realistic natural ratios. All else being equal, Carson should have relatively more vertebrae because of improved recovery of small vertebral centra and caudal vertebrae in the 3.2 mm mesh. A comparison with Carson is problematic because vertebra
The number of vertebrae represented in the taxa identified range from 28 to 69 (Table 6) (Oates et al. 1993; Scott and Crossman 1973). Most of the individual fish to whom these 14
size distribution as it may relate to recovery was not investigated. The cranial to vertebral ratios may indicate that more off-site processing was practised at Barrie than at Dunsmore, although none of the sites have strong evidence for off-site processing.
All the fish scales are Perciformes. While this correlates to some extent with the proportions of identified cranial bones, certain taxa, like Esocidae and Catostomidae, are conspicuously absent. In modern samples of comparablesized fish to those represented by the cranial material, these cycloid scales appear somewhat less robust than the ctenoid scales of the Perciformes. Indeed, the very large Catostomidae scales recovered from a midden at the Steward site (Junker-Andersen 1984) were all fragmentary, although the condition of large Catostomidae scales at the Hector Trudel site (Cossette 1995) was much better. Salmonidae scales are fragile and also very small relative to body size, so that they are more likely to break and/or slip through the screen.
At Barrie, vertebrae represent 23 percent of all fish bones in the houses, 37 percent in the middens. At Dunsmore the proportions are 12 percent vs. 10 percent; at Carson six percent vs. 20 percent (although the sample size from the midden is very small). If fish scales had been recovered at Barrie, the relative contribution of fish vertebrae would be proportionately less. Nevertheless, it is obvious that Barrie contains substantially more vertebrae than the two later sites. This does not appear to be related to the increased heavy fraction sample at this site, as 19 percent overall came from flotation, and 38 percent from the 6.4 mm dry screen. At Barrie and Carson, vertebrae are more numerous in middens than in houses, perhaps indicating that at least some vertebrae were removed before cooking and preferentially deposited in external refuse deposits.
The main taphonomic difference between Carson and the other two sites is screen mesh aperture. While fish scale recovery may not be enhanced by fine mesh dry screening, the recovery of many small perch bones from the Carson 3.2 mm mesh and flotation samples and, more recently, the recovery of very small perch bones and small cycloid scales from the flotation heavy fractions from the Grafton site by Thomas and Needs-Howarth (Thomas 1997a), may well be a compelling argument for the use of small aperture wetscreening or more extensive flotation on sites with sandy soil. It is possible that the smaller dry screen mesh size employed at Carson has allowed for the recovery of greater numbers of fish scales.
The high proportion of vertebrae that could be identified to family at Barrie requires explanation. It may relate to the overall larger size of the fish at this site, since vertebrae of very young or small fish may be harder to identify. It may also relate to the degree of fragmentation, since incomplete vertebrae are also harder to identify. This in turn may relate to fish size: the larger the fish, the more likely it is to be filleted, rather than pounded up in its entirety (see section on food preparation, below). Alternatively, processing and cooking methods at the Barrie site may have favoured preservation of vertebrae.
There are some differences between houses and middens. At Dunsmore, fish scales represent four percent of the total bone recovery from the houses, and seven percent of the middens. At Carson fish scales represent nine percent of the total from the houses, and 16 percent of the middens. This may suggest that refuse from primary processing in the form of descaling was deposited more frequently in external refuse deposits. Taphonomic effects may have been more severe in the houses because of burning, trampling and delayed burial.
Skinning Fish scales are less numerous in the zooarchaeological collections than vertebrae, despite the fact that each fish possesses many more of the former than the latter. This is probably owing to the relative fragility of scales, and perhaps also to off-site fish processing and skinning. Fish scale recovery varied considerably between the three sites. Only one fish scale was recovered from the Barrie site (Midden D heavy fraction). This may be in part due to the slightly inferior preservation at this site, indicated by extensive iron staining observed on some bones. The Dunsmore sample contained 105 scales (70 from screens, representing seven percent of fish by recovery, 35 from flotation, representing five percent). The Carson sample contained 376 scales (342 from screens, representing 10 percent, 34 from heavy fraction, representing four percent). Additional scales may be present in the light fractions; those of the Carson site have not yet been sorted.
The fact that the rest of the Carson fish assemblage is so similar to the Dunsmore assemblage indicates that perhaps there are real differences in the frequency of fish processing at these sites. Fish scales of several species tend not to survive being buried after they have been cooked (Nicholson 1996a). Cooking method and feature deposition history combined probably play a major role. It is quite possible, for example, that the Carson material simply included more refuse from fish that were skinned prior to cooking and less from fish that were cooked or roasted with the scales still attached. Additional support for this hypothesis is found in the fact that the proportion of brown bullhead, a species that does not possess scales, is higher at Carson than at Dunsmore, so that there would be relatively fewer scales associated with the Carson fish catch, based on taxonomic distribution. It is hypothesized here that the Dunsmore people were more likely to descale their fish at the catch location or boil or
The Barrie assemblage was subjected to the most intensive flotation, whereas the Carson site underwent only limited flotation. This discussion shows that more extensive flotation does not necessarily result in the recovery of more fish scales; burial conditions and the degree of fossilisation are also important. 15
roast their fish in the skin, which would result in fewer fish scales being deposited or preserved at the site. On-site dog scavenging cannot be ruled out, but this would affect fish heads, vertebrae and other animal remains as well as fish skins with scales.
have been conducive to bone preservation. Fish might be preserved by drying or smoking, depending on the season (Thwaites 1896-1901:10:101, 34:215). The Huron sometimes packed the smoked fish in “casks” to have at feasts or use in soup, especially in winter (Sagard 1939:186). Writing about the New York Iroquois in the mid-nineteenth century, Morgan (1962:64-65) says that “whatever was gained by any member of the household on hunting or fishing expeditions, or was raised by cultivation, was made common stock. Within the house they lived from common stores.” Most of the fish at these three sites were small (compared, for example, to some marine fish), and may well have been eaten during a single meal by one or more families in the same longhouse.
Filleting Another way to look at fish butchering is to examine the relative representation of the cleithrum, a bone that may well have been broken during decapitation. Brown/yellow bullhead and yellow perch cleithra are frequently broken. It is almost always the dorsal part of the element that survives to be identified (Table 14). It is interesting to note that five of the complete yellow perch cleithra from Dunsmore are from H8 F206 (MNI 4), and that all of these are much smaller than the site average. Perhaps these smaller individuals were cooked whole. For brown/yellow bullhead broken cleithra almost always include all or part of the articulation for the pectoral spine, while for yellow perch they include mainly the dorsal “wing.” This is likely related to element morphology and robusticity.
Huron foods consisted mainly of corn-based stews, cooked in large ceramic pots, or corn bread (Heidenreich 1972:57). Ordinary meals frequently consisted of corn soup, flavoured by berries, fish or meat (Sagard 1939:106-107, 230). A glimpse of the cooking process was seen at the Peace Bridge site at the north end of Lake Erie near Niagara Falls where Thomas (1998) analysed the contents of a ceramic cooking vessel dating to about A. D. 675. It contained remains of walleye (Stizostedion vitreum) and other fish, as well as the scapula of a whitetailed deer. While the exact relationship between the Princess Point people who were responsible for this pot and the second millennium occupants of the area between Lake Simcoe and Nottawasaga Bay is unknown, some cultural continuity in cooking practices may be assumed.
While the anterior/ventral cleithrum in brown/yellow bullhead is much more robust than the corresponding part in yellow perch, anterior/ventral portions are conspicuously absent in both taxa. The range of dimensions of the anterior/ventral portion in each species is fairly similar, and their sizes would in many cases have enabled recovery in 6.4 and 3.2 mm mesh. Recovery bias is probably not the only explanation. A further examination of relative element representation of these two taxa in relation to element robusticity, however, reveals that these two taxa experienced quite different taphonomic histories. Percidae have relatively fragile vertebrae that look somewhat similar to Centrarchidae, causing some, perhaps, to have been lumped inadvertently with Centrarchidae. In contrast, Ictaluridae vertebrae are more easily recognisable and quite robust. The high number of Ictaluridae cleithra (and other cranial bones) in relation to vertebrae is, therefore, hard to explain. A similar dearth of Ictaluridae vertebrae was noted at the Keffer site (Stewart 1991b). Breakage of brown/yellow bullhead cleithra may be a side-effect of extraction of the pectoral spine, either to avoid injury or to enable curation of the spines as tools. Given the extreme robusticity of brown/yellow bullhead cranial bones in relation to other fish, it is possible that the heads were avoided by the dogs, thus leading to the observed relative over-representation in relation to post-cranial elements. Pike heads, containing very little meat and sharp teeth, may have been similarly unattractive for consumption by humans (Noe-Nygaard 1983:130; Noe-Nygaard 1987:28) and dogs. In conclusion, the ratio of cranial bone to vertebrae can only be used as direct evidence for butchering if both identification and taphonomic biases are considered, including dog scavenging.
Some food preparation techniques may have resulted in considerable loss of fish bones. Smaller fish were likely cooked whole and may have been eaten in their entirety (e.g. Brinkhuizen 1989: 267). This may explain the dearth of small Cyprinidae, and other taxa that are common in the area today (Latta 1998:10). In discussing Mesolithic midden refuse from Denmark, Enghoff (1991:116) offers the scenario that “bones and scales have simply been skimmed off potfuls of fish soup.” Cooking or stewing may have adversely affected the preservation of fish remains, because heating causes morphological change in bone (Richter 1986); boiled fish bone loses much of its mechanical strength (Jones 1990). Considerable loss and breakage of fish bones occurs during roasting, although “block-shaped” bones (such as vertebrae) and dense, hard bones survive better than others (Spennemann and Colley 1989). In any event, if animal bones survived the stewing process to be buried and recovered, there would not necessarily be many butchery marks. Iroquoian researchers have made much of Sagard’s (1939:316-318) description of the Huron taboo on burning fish bones, which also holds for Algonkian groups such as the Ojibway (Stephen Crawford, personal communication 1999): “They take special care not to throw any fishbone into the fire [because the bones represent]...spirits of the fish themselves..., which would
Food preparation Iroquoian food preparation techniques in general may not 16
warn the other fish not to allow themselves to be caught....” As was noted above, however, burnt and calcined fish bones do show up in zooarchaeological assemblages. These are often bones of the pectoral girdle, which may have been exposed during cooking. A number of the lake sturgeon dermal plates from the Barrie site were burnt, though not calcined, perhaps because they were exposed during roasting.
one undisturbed Iroquoian site indicated that refuse inside houses accumulates near the outer walls (Finlayson 1985). This patterning is, of course, obscured at a plough-disturbed site. Since processing activities relating to large catches of animals may be expected to be conducted preferentially in outside work areas, middens may be expected to contain more refuse resulting from mass-capture events (1996b; Thomas 1996d). House feature deposits, on the other hand, are more likely to contain refuse from activities carried out inside the house.
Consumption Fish at Iroquoian sites were most likely consumed by both humans and dogs. Dogs were kept by Iroquoian people as pets and as a food source (Sagard 1939:111, 117, 123, 226, 227). They may have been fed kitchen refuse, and since they were probably allowed to roam free at least some of the time, they had easy access to refuse in middens and pits. Dogs, therefore, are likely to have been an important taphonomic agency. Carnivore gnawing was observed on three of the identified bones from Barrie, 10 from Dunsmore, and four from Carson, i.e. roughly in proportion to the total NISP. No carnivore damage was observed on fish bones. This does not, however, mean dogs did not eat food refuse. They probably were fed or scavenged from middens scraps of kitchen refuse, of which nothing identifiable would remain.
Feature fill may include redeposited material from unrelated activities, and food debris representing, for example, preserved fish. It is also conceivable that food refuse in secondary contexts has become mixed with food excreted in faecal matter of humans and dogs. A feature may, therefore, contain superimposed evidence of more than one procurement and refuse deposition event. One of the problems on Iroquoian sites is distinguishing between primary and secondary refuse contexts. Much of the bone in middens and reused sweatlodges is probably secondary refuse. There is always a certain amount of bone in ash pits, which may contain hearth sweepings. A comparison of thermally altered bones between feature types at Barrie and Dunsmore demonstrates that ash pits do not contain more burned and calcined bones than other feature types (Needs-Howarth and Thomas 1994a). It was argued that this may indicate that ash pits do not consist of unmodified hearth sweepings, which could be expected to include bones thermally altered from overheating or being dropped in the fire. Pits, on the other hand, contain over twice as much thermally altered bone as the other feature types. It seems likely that pit fill consisted in part of normal hearth sweepings (Needs-Howarth and Thomas 1994a).
There is a considerable amount of literature on the effects of human consumption and dog scavenging on the destruction or (in the case of dogs) displacement of animal bones. The degree to which bones become modified by the eating and digestion process is related to a number of factors, including the species, age, condition of the tissue, structural density, shape, completeness of skeleton and bones, pretreatment of bones, and the disposal method (Butler and Schroeder 1998; Jones 1986; Kent 1981; Lyon 1970; Payne and Munson 1985; Stallibrass 1990; Walters 1984). Smaller fish especially may have been eaten whole or reduced to unidentifiable fragments through cooking or chewing (Rojo 1987). In one feeding experiment (Jones 1986), less than 10 percent of ingested bones of fish 24 to 35 cm total length (hereafter TL) survived passage through the digestive system of a human being and a dog, while 25 percent survived in another study using fish of 6.5 - 8.5 cm standard length (SL) (Butler and Schroeder 1998).
Burial Bone preservation at all three sites is facilitated by the sandy, non-acidic soils and the fact that natural and cultural taphonomic processes have been allowed little time to destroy the zooarchaeological evidence. The collections have undergone only about 500 to 700 years of postdepositional disturbance and metal plough agriculture has been a significant factor in site disturbance for less than 200 years. All categories of bone at these sites are capable of being preserved without carbonisation.
Discard Details on in-house refuse deposition are obscured by plough damage to the house floors; the excavated refuse deposits represent only the truncated remains of the original features. In general, sedentary populations discard items outside their use locations (Gifford 1980; Murray 1980). Middens mostly contain a random mix of byproducts of activities conducted over unspecified periods of time, from one or several adjacent structures. It is also possible that refuse was deposited down-slope, outside the house and midden areas. However, the abundant food refuse in interior pits, hearths and reused semi-subterranean sweatlodges suggests that the inhabitants of these three sites did not discard all their refuse outside their houses. Excavations at
Preservation of the bones was generally good; diagnostic features were preserved on many of the bones and they exhibited only minimal excavation damage. There are some unquantified differences in preservation between the Barrie sample and the two later ones. The Barrie bone appears to have been subjected to iron staining. Some of the surface texture of mammal longbones appears somewhat more porous than comparable bones at the other sites. There was also more soil and/or root etching. The differences in 17
preservation became quite apparent when viewing sectioned pectoral spines of brown bullhead from each of the sites. The sections from Barrie are noticeably more stained and display a loss of some of the bone interior. These phenomena were also observed on the lake sturgeon spine sections, although here comparable material from the two later sites was not available. The paucity and poor condition of fish scales from Barrie may also relate to differential preservation. Preservation at the Dunsmore site was generally very good, particularly in Midden A, Midden B and most of the major features. Preservation at Carson was also generally very good. Intact, well-preserved fish scales were recovered from both Dunsmore and Carson. Storage of the Carson flotation samples may have affected bone preservation of the heavy fraction samples. Samples from Barrie and Dunsmore were processed on-site, using the two-bucket method. In contrast, the Carson flotation samples were stored for six years in an unheated garden shed before being processed. The plant remains in the flotation samples suffered considerable deterioration during storage (Della Saunders, personal communication 1996). Some fish bones had assumed a rubbery consistency, suggesting some deterioration of the zooarchaeological remains as well. Deterioration was apparently not uniform among samples, since some flotation heavy fractions contained substantial numbers of intact fish scales. Nicholson's (1996b) actualistic studies indicate that “bone loss between taxa frequently does not follow easily defined rules, for example based on bone size or bone density, and [...] soil environment such as pH and drainage.” While the latter do play a role in bone preservation, a major factor, at least in the initial stages, may be the range and types of soil microorganisms available (Nicholson 1996a). Autolysis by the fat contained in the bone may have caused additional taphonomic loss (Mézes and Bartosiewicz 1994).
(Oncorhynchus tshawytscha) cranial bones do not preserve as well as their vertebrae (Butler and Chatters 1994). Lake whitefish cranial bones are more fragile and less likely to survive being cooked or roasted than their vertebrae (Lubinski 1996). Recovery Introduction Some aspects of zooarchaeological recovery were previously discussed in Chapter 2. This information needs to be expanded upon in order to understand certain aspects of taphonomic bias. The zooarchaeological samples from Barrie and Dunsmore were recovered with the aid of 6.4 mm dry screen, from Carson with 3.2 mm dry screen (this refers to the mesh grid; actual mesh apertures are somewhat smaller). All three assemblages derive a proportion of their bone fragments from flotation. The aperture of the smallest heavy fraction nested sieve is 2 mm for Barrie and Dunsmore, 2.4 mm for Carson. Flotation samples are a subsample of the feature fill in any given refuse context, the remainder of which is screened (Table 3, 4, 5). The intensity of flotation varies and soil volumes are available only for the floated components. Since screened volumes are not available (except in a general way, by multiplying feature dimensions), it is not possible to calculate the number of fragments per soil volume. The lack of comprehensive volume density information leaves a big gap in our understanding of taphonomy. Screen mesh size is undoubtedly a significant bias in the recovery of faunal remains, especially smaller taxa such as fish. Because of the differences in mesh sizes and intensity of floatation and the lack of screened volume data it is difficult to judge to what extent each assemblage is subject to recovery bias. There are further complicating factors, which are discussed below. In general, however, the combinations of different recovery methods generate certain expectations. One would expect both the kind and number of remains recovered and their fragment sizes to correlate to some extent with mesh aperture. In the absence of soil volume data, our only guide to recovery biases are fragment sizes, and the proportions of NISP, number of unidentified specimens (i.e. usually those bones not identified below the taxonomic level of class, hereafter NUSP) and BW as they relate to the different recovery methods. The cultural and temporal similarities between Dunsmore and Carson may be expected to have resulted in somewhat similar original bone assemblages. If major differences were found, these might be attributed to recovery. Determining whether variability in the zooarchaeological samples can be attributed to differential recovery is essential to inter-site comparisons.
Bone loss also relates to processing and cooking method, and may differ between species (Nicholson 1996a). Buried fish bones decompose more rapidly when the flesh has been removed and when they have been boiled (Nicholson 1996a; Nicholson 1996b). As Nicholson (1992) established, fish bone survival is mediated not by bone density, rather by bone shape and size. In general, one would expect the thin, flat cranial bones to be destroyed more quickly than the cylindrical vertebrae, simply because they have less mechanical strength and relatively more surface area (Morales Muñiz 1984; Rojo 1987; Spennemann and Colley 1989). Taphonomy of Salmonidae bones requires some further consideration. Salmonidae may be subject to two biases not affecting other taxa. The first is a depositional bias relating to off-site processing. This is explored below. The second is a taphonomic bias against representation of Salmonidae cranial bones, indicated by recent actualistic experiments (Butler and Chatters 1994; Lubinski 1996). Because they are relatively less dense than vertebrae, chinook salmon
Cautionary tales from the literature In general, the relationship between sieve sizes and frequency of material recovered is curvilinear; flotation is considered by some to be the most cost-intensive means of intensive data recovery (e.g., Ball and Bobrowsky 1987). A comparison between 6.4 mm and 3.2 mm dry screening showed that the larger aperture biases against species 18
diversity (Gordon 1993). A much-cited experiment by David Thomas (1969) showed that large quantities of small mammal bones were lost in 6.4 and even 3.2 mm dry screen mesh. Only 1.6 mm dry screen mesh resulted in full recovery of mammal bones, while fish were not recovered at all from the larger mesh sizes. An experiment by Casteel (1972) indicated that all freshwater fish vertebrae (species and size not specified) were lost in the approximately 5.6 mm mesh. A comparison of recovery methods for Hawaiian material (Gordon 1993; 1994) concluded that sediment screened at 3.2 mm produced about 250 times as much fish bone per volume as sediment screened at 6.4 mm. Another comparison, using Hohokam material, indicated the 3.4 mm mesh produced between 1.6 and 5 times more bone fragments, and that 90 percent of fish remains may be lost through the 6.4 mm mesh (James 1997). Recovery tests at the Middle Woodland Hector Trudel site (Cossette 1995) showed that fish bones were particularly under-represented in the 6.4 mm field recovered sample and that smallest fish were only collected in the 2 mm and 1 mm fine screens and flotation samples. A recent publication (Cannon 1999) indicates that, at least for mammals, mathematically correcting for differential recovery is extremely problematic.
contexts were sampled at Carson than at Dunsmore, but the effects of the increased flotation are not that noticeable. This may relate to the absolute volume of the samples, which was 2 litres or less at Carson, and usually more than 2 litres at Dunsmore. There appears to be a relationship between intensity of flotation and average number of bone finds per context; at Barrie, the number of finds per context is greater than at Dunsmore and Carson. How does ubiquity of flotation and mesh aperture of the geological sieve used to recover the heavy fraction relate to fish recovery? Although the ratio of screened to floated volumes is not available, it is known that the flotation samples are only a fraction of the screened volumes. The flotation samples, however, contribute a disproportionately large number of bones. At Barrie flotation accounts for 41 percent of the overall number of bone fragments recovered; at Dunsmore 31 percent; at Carson 36 percent. This conforms to expectations. One would also expect to see a correlation between the proportion of fragments derived from the flotation heavy fraction and the proportion of fish. Initially, this appears not to be the case, with Barrie having the lowest proportion of identified fish (53 percent vs. Dunsmore 75 percent, and Carson 67 percent). The proportion of unidentified fish is surprisingly similar between the sites (Barrie 74 percent, Dunsmore 78 percent, Carson 82 percent of NUSP). However, the proportion of fish finds within the flotation sample shows that flotation does result in increased fish remains, at least at Barrie, where fish represent 85 percent of the heavy fraction and fish represent only 57 percent of the screened material. For Dunsmore the figures are 81 percent and 75 percent, for Carson and 75 percent and 80 percent.
Effects of dry screening All else being equal, a smaller mesh aperture should result in more fish, and even more unidentified fish than a larger aperture. The 3.2 mm dry screening mesh used at Carson is expected to result in increased recovery of (small) fish remains compared to the 6.4 mm mesh used at Barrie and Dunsmore. In addition, the finer dry screen mesh size is expected to result in a much greater number of unidentified fish at Carson than the other two sites. Carson lags behind Dunsmore in proportion of screened fish NISP (Table 15). The effects of the smaller mesh size are, however, noticeable in the huge number of unidentified fish fragments from the screen. At Barrie and Dunsmore, 63 percent and 57 percent of the fish from the screen, respectively, could not be identified. At Carson 85 percent could not be identified (with the exception of fish vertebrae, which, while frequently identified to family, are included in the NUSP). Considering only the fragment counts, it appears, therefore, that the smaller mesh size has not really increased the recovery of identifiable fish bones; it has, however, substantially increased the number of unidentifiable fish bones.
These differences between recovery methods are not as marked as at other sites. At the Keffer site (AkGv-14), located just north of Lake Ontario, the number of bone fragments derived from 6.4 mm screening and flotation were roughly similar. In the former sample fish represented only 30 percent, while in the latter they represented 70 percent (Stewart 1991b). Animals bones from the 1983 excavations at the Wiacek site (BcGw-26), analysed by Christine Dodd (Lennox et al. 1986), derived mostly from flotation. Fish represent 69 percent of the screened sample, and 84 percent of the floated sample. For this site soil volumes are available; screened material represents 92 percent by volume, but only three percent by number of fragments. At the St. Lawrence Iroquoian Steward site (BfFt-2) (Junker-Andersen 1984), flotation also markedly increased bone recovery; 36 percent of all analysed elements were recovered from four percent of the soil volume.
Note that the number of lake herring or lake whitefish vertebrae is low compared to lake trout vertebrae at Dunsmore and Carson. This may be because the former are less robust and generally much smaller; as noted above, lake herring caudal vertebrae in particular would fall right through the 6.4 mm mesh.
Comparison of screened and floated samples Consistent with expectations, Barrie shows a substantial improvement in fish recovery through flotation. Almost half of the fish bones in the assemblage derive from the heavy
Effects of flotation The Barrie sample was subjected to substantially more flotation than the other two samples (Table 16). More 19
fraction. Also as expected, the proportion of unidentified fish in the whole floated sample is greater than in the screened sample. Fish are 87 percent of floated NUSP, only 13 percent of floated NISP. The Dunsmore sample has more equal proportions of fish from the screens (Table 15). In terms of ratios of identified to unidentified fish, Dunsmore, however, is similar to Barrie. Fish are 91 percent of floated NUSP, only nine percent of floated NISP. At Carson, these numbers are almost identical: 92 percent and eight percent. Again, it is necessary to look at ratios of fish in the whole sample to identify an anomaly. Identified fish at Carson contribute 94 percent to the heavy fraction of all five classes combined, compared to a 65 percent contribution to the screened component. This is very odd considering that the difference in mesh aperture for screened and floated material at Carson is less than 1 mm: 3.2 mm vs. 2.4 mm. If screen size is the only determinant of recovery, there should be little difference between the screened and floated components. The question arises: “Why are there not more fish in the Carson screened sample?”
Support for this hypothesis may be found in the presence of fragile fish scales in both the screened and floated components. Flotation as a recovery strategy, in general, may be more effective because the heavy fraction bone can be recovered under controlled circumstances in a laboratory, by either a trained zooarchaeologist or palaeobotanist. Fragment weights Fragment weights can be used as a proxy for fragment size, and thus aid in understanding taphonomy (Landon 1992). As noted above, many identified fish bones weighed less than the precision limit of the balance. It is not possible, therefore, to elucidate taphonomic issues relating to BW from the identified fish bones. The combined unidentified fish bones, including vertebrae, were weighed per context; here average fragment weights can provide some insight into taphonomy. It should be pointed out that the Carson assemblage does include more small features with small sample sizes, so that the precision limits of the electronic balance may result in the recording of .1g weights for (groups of) small unidentified fragments that in fact weigh less than .1g.
The answer may lie in how effectively bones were retrieved from among other finds and dirt in the screens. The Barrie and Dunsmore sites were excavated by experienced field crews, but Carson was excavated in part by inexperienced field school students. Following recovery tests at the Hector Trudel site (Cossette 1995), it was suggested that field crews may miss smaller, incomplete or less recognisable bones that would be recognized under laboratory conditions by trained zooarchaeologists. Although this study also found significant biases relating to mesh aperture, it was concluded that screen operator bias or screening conditions were in many respects more important than mesh size. It is hypothesized here that field recovery from the dry screens at the Carson site is biassed against the recovery of fish bones in general, and of small and fragmented fish bones in particular.
As expected, fragment weights correlate to some extent with intensity of recovery. Screened material tends to weigh more than heavy fraction material. This difference is particularly obvious at the Carson site, where the midden material was recovered entirely through screening. Floated mammal remains at Barrie form an exception because of the inclusion of a large section of bear (Ursus americanus) humerus, weighing 32.1 g. The absolute fragment weight of all classes is substantially larger at Barrie than at either Dunsmore or Carson. Surprisingly, this is the site with the most intensive flotation. While differences in mammal BW may relate to taxa included (i.e. more large mammals at Barrie), this is far less of a problem for fish. The large number of sturgeon bones from the middens at Barrie probably do account for much of the difference in average identified fish BW between Barrie and the two later sites. However, since it is unlikely that any sturgeon bones are included in the unidentified material, the difference there must be attributed to other factors. It appears from proportional element size observations that the occupants of the Barrie site caught slightly larger fish (see below), but the difference is not huge. Therefore, the cause must be sought in increased predepositional fragmentation and/or post-depositional trampling, especially at Dunsmore.
Support for this hypothesis is found in fragmentation rates (Table 17). The number of broken fish bones within the flotation samples is fairly consistent between the three sites, indicating that the collections may have started out being similarly fragmented. Indeed, the proportions of complete and broken bones are similar between the screened and the floated components, with one exception. It is interesting to note that there are fewer identified fish bones with major damage in the screened component at Carson than at the other two sites, whereas the number of complete fish bones is more similar. This may indicate that field crews did not retrieve broken fish bones that could have been identified. If this is so, the real proportion of unidentified fish bones could be expected to be much higher than what was retained for analysis. The relatively high proportion of complete mammal bones compared to Barrie and Dunsmore supports the hypothesis that field recovery may have been biassed towards complete bones. The lower fragmentation rates at Carson do, however, suggest that the unexpectedly low fish NISP is not the result of overly aggressive screening or forcing sediment through the mesh.
Fish element size To further identify taphonomic bias, fish cranial bone size can be used to assess fragment sizes, as a proxy of fish size. Some zooarchaeologists record fish size as small, medium, or large in relation to a reference specimen of known size (e.g. Cooper 1996). Other techniques involve comparing archaeological material to several reference skeletons of known length and grouping them into size classes (e.g. 20
Martin and Colburn 1989). Unfortunately, the reference collection used for the current research did not contain sufficient specimens to be able to use a size class method, although efforts have been made to expand the fish collection in the last two years . The best results are achieved by fitting measurements of multiple modern specimens to a regression equation. While regression equations relating fish bone size to fish length have been developed outside the Great Lakes area (e.g. Brinkhuizen 1989; Butler 1996; Desse and Desse-Berset 1994; Leach et al. 1996), watershed-specific equations are not yet available for species in the Great Lakes area.
The proportional element sizes were recorded at 10 percent intervals, except for some intuitive fractions, such as 1/3 or 3/4. The back-calculated proportional element sizes based on osteometrics were recorded at one percent intervals. The osteometrics facilitated a verification of the proportional element size. The difference between the proportional element size and the back-calculated proportional element size based on osteometrics on an individual bone was small (usually less than 10 percent). Where there were discrepancies, the back-calculated proportional element size was adopted in favour of the proportional element size. The proportional element size refers only to relative bone size, not to fish size. Intra-individual asymmetry, as well as inter and intra-individual differences in the relationship between element measurements and TL (Brinkhuizen 1989:66-67), make it difficult to decide the range of percentage points that could signify a “real” individual.
In order to obtain a general idea of fish size, a proportional method was adopted during the Barrie analysis whereby macroscopic size observations are expressed as a percentage of a single reference specimen of known length (Needs-Howarth 1995a). This type of proportional method is not considered ideal, especially when only one specimen is used for the estimations. In the absence of numerous reference specimens and regression equations relating bone size to TL, it may, however, be considered an acceptable way of getting a general idea of fish size (Heinrich 1981).
Figure 9 shows the proportional element size distribution of all sized and/or caliper-measured fish bones by site. Because this graph relies on size in relation to a reference specimen of known size, the size distribution is to some extent dependent on species composition. For example, the archaeological northern pike (Esox lucius) bones tended to be smaller than the reference specimen, whereas the yellow perch and brown bullhead were fairly similar in size to the reference specimen. This means that the graph for the site with the most pike is going to be shifted towards the left. This graph also masks some of the variation resulting from differences in taxonomic distribution between recovery methods. This combined size information at the site level can, however, be used in a very general way to explore differences in fish proportional element size between recovery methods. As one would expect, the sub-sample derived from flotation includes bones from smaller individuals. Comparison of fish relative element size within each site indicates that there is a discrepancy in average fish bone size between the screened and floated material at all three sites. At Barrie the difference is negligible, at Dunsmore it is 18 percent, and at Carson 20 percent.
This method was further developed during the Dunsmore analysis with the addition of osteometrics (Needs-Howarth and Thomas 1994b; Needs-Howarth and Thomas 1998). It will be show below that the osteometrics on individual elements can provide some clues to taphonomy. However, since not all fish cranial and pectoral girdle bones were measured using dial calipers (Table 19), and since the measured sample size per element per species was small, it was desirable to somehow integrate these data with the proportional element size observations. The sample size was increased by proportionately converting the osteometrics to a size percentage of the reference specimen of known size, and by combining data from all elements per species (Needs-Howarth and Thomas 1998). The resulting loss of detail and lack of control over MNI was argued to be compensated by the increased sample size per species (Table 19). The osteometrics are based on what is practical, given the often fragmentary state of the collections, not necessarily on what may give most accurate size conversions (Table 18). Metrics follow those suggested by Morales Muñiz and Rosenlund (1979) where possible, but recorded to .1 mm rather than .5 mm, as recommended by Brinkhuizen (1989:73). Owing to bone breakage, many osteometrics had to be restricted to the compact, but smaller, areas of articulation and symphysis. It is acknowledged here that measuring errors have a greater effect on these smaller surfaces (Bartosiewicz and Takács 1997:15; Brinkhuizen 1989:71). It should be noted that a measurement used by zoologists to calculate yellow perch TL, taken from the opercular articulation to the margin, so that the angle of the intersecting line is 900, (Bardach 1955) appears vulnerable to warping in archaeological specimens.
To verify this pattern Figure 10 shows the proportional element sizes of the most abundant taxon at each site, yellow perch. At Barrie there is a discrepancy of five percent between screened and floated material, while at Dunsmore it is 17 percent and at Carson it is 19 percent. The small difference at Barrie may relate to the fact that the overall size of yellow perch is larger at this site, so that the small end of the spectrum would be less adversely affected by loss through screening. The differences in proportional element size between recovery methods at Dunsmore probably relate to the 4.4 mm difference in mesh aperture between the two recovery methods. The differences between the two recovery methods at Carson are hard to explain. If screen mesh size was the major determinant of bone (fragment) size the two sub-assemblages should show little difference in average 21
size . This parallels the unexpected differences in percentages of identified fish in the heavy fraction compared to the screened component noted above.
Absolute fragment size As will be discussed below, age and growth analysis of fish scales and pectoral spines recovered by dry screening demonstrated that the average age of two fish taxa is less at Carson than at Dunsmore. The average size of various cranial elements is also less. This is exactly what one would expect if screen size is biassing towards smaller fragments and/or fish size at Carson. Although the data will be discussed later, in Chapter 6 and 7, it is necessary at this point to briefly discuss whether these differences might be attributable to differential recovery bias. The dimensions of the brown bullhead pectoral spines, even when only the medial/proximal 25% survives, are large enough to have be retrieved in both the 6.4 mm dry screen mesh used at Dunsmore and the 3.2 mm mesh used at Carson. An important clue to recovery bias between Dunsmore and Carson comes from the size of fish scales. To test whether the decreased age of fish scales at Carson was a function of recovery, the anterior-posterior length of the scales from Dunsmore and Carson was measured. This dimension is easier to measure than the greatest height. While the relationship between height and length varies with placement on the body, the height is usually the greater of the two. In other words, the length is a conservative measure of fish scale size. The size distribution in Figure 11 indicates that there are some smaller scales in the Carson sample. However, only the very smallest Carson scales would have fallen through the 6.4 mm mesh size used at Dunsmore. This would suggest that differences in age-atdeath between the two sites are not entirely attributable to recovery bias. Osteometrics of cranial bones of major fish taxa also show consistent differences. In order to examine differential recovery as an explanation for differing fish element sizes, Figure 12 shows dial-caliper measurements of two abundant elements, the dentary and operculum, of brown bullhead and yellow perch (Figure 12). The average size at Carson is slightly smaller, and it is possible that some smaller brown bullhead dentary fragments may have been preferentially recovered in the smaller mesh size used at Carson. However, bearing in mind the generally triangular shape of the operculum and the magnitude of the caliper measurements, it appears that the opercula at Carson would also have been caught in the larger mesh size in use at Barrie and Dunsmore. Sorting out recovery biases The combination of NISP, BW, and measurements of fish cranial bones and scales has provided some insights into taphonomy. Barrie and Dunsmore differ quite substantially. Given that recovery is so similar (except for flotation volumes), the explanation is likely to relate at least in part to culture or environment. Findings at Carson did not conform to expectations. Differences between the Barrie or Dunsmore samples, and the Carson sample may be more difficult to interpret.
22
CHAPTER 5: INTERPRETING FISH REMAINS DATA
Historical data The practice of late precontact (zoo)archaeology in southern Ontario is enriched because of abundant ethnohistoric descriptions of the Huron. As Ramsden (1996:104) has recently pointed out, the extensive written observations on the Huron provide archaeologists with “precisely what archaeologists in other areas try to create models for: a set of plausible hypotheses about the social, political, economic and ideological processes that produced at least part of the observable archaeological pattern.” In some ways this is a drawback, because it distracts attention from archaeological data (Ramsden 1996), especially topics such as subsistence.
(Heidenreich 1972:9-10). Explorer, trader and coloniser Samuel de Champlain (Biggar 1922-36) spent the winter of 1615 in the area between Lake Simcoe and Nottawasaga Bay. The writings of Récollet Friar Gabriel Sagard (1939), who spent the winter of 1623-24 among the Huron, are more similar to those of an anthropological participantobserver, making his writings in some ways more comprehensive (Tooker 1964:6). A follower of St. Francis, Sagard was more interested in the animal world than either the Jesuits or Champlain (Heidenreich 1972:8-9). These sources were written about an alien society by European males whose primary interest was in converting the native inhabitants to Christianity or finding trade opportunities. Extrapolating back in time 50 or 100 years the precontact Huron (if they can even be called Huron at that point) is problematic enough using some form of Direct Historical Approach. Extrapolating back 325 years to the date of occupation of the Barrie site must be done with great caution, even though these people may have direct ancestral links to the Huron. While underlying customs and beliefs may have remained fairly stable over time, it appears likely that the subsistence choices would have varied with chronological and geographical changes in the environment
Ethnohistorical descriptions of the Huron, other Iroquoianspeaking peoples, and their neighbours from the early contact period are useful, but they are probably quite biassed and limited in scope. The Jesuit Relations (Thwaites 18961901) were written by Fathers Jean de Brébeuf, F. J. Le Mercier, Jerome Lalemant, Paul Ragueneau and Francesco Bressani. Their accounts pertain to their work as Jesuit missionaries in southern Ontario from A. D. 1634 to 1650 and emphasize mythology and religion. Correspondingly, their descriptions of animals and the natural world are weak 23
and subsistence base. The seasonal cycle described for the Huron is closely linked to the men’s roles as fur traders, which took them away from the villages for long periods of time in spring and summer. While precontact Iroquoian people in the area probably traded furs to some extent with neighbouring groups, this was not as important as it was in the French period. Huron women spent most of the spring and summer tending the corn fields (perhaps at specialpurpose horticultural cabins), until the harvest in late August or early September.
Palaeobotanical and zooarchaeological research confirms that prior to Euro-Canadian settlement of the area there was substantial forest cover. Iroquoian settlement likely resulted in substantial but patchy forest clearance by different villages for agriculture and construction wood. This may be expected to have caused erosion and associated changes in stream flow. Euro-Canadian settlement likely caused additional changes, including modification of the local wetlands habitats. Contemporary species distributions are influenced by construction of dams and other obstacles in various streams and rivers surrounding Lake Simcoe. Discussions with fisheries biologists and conservation officers suggest that the flow regimes of watercourses and marshes are likely to be so changed, even by the early 1800s, that extrapolating hydrobiology back in time before Euro-Canadian settlement is problematic. Even minor differences in flow, depth, width, or substratum of the water sources in the area could have major implications for fish ecology, especially species composition and spawning behaviour. This argument has been used (Yerkes 1981b) to critique the use of species composition from modern fish catch for the purpose of archaeological analogy (Limp and Reidhead 1979). In other words, historic environmental data are of limited use to this kind of micro-scale investigation. While there may exist some potentially relevant information on the biogeographical distribution of fish in the form of nineteenth century fishing records, this information can only be used as proof of the existence of a taxon in a particular watershed. It cannot be used as negative evidence for lack of taxa, since absent taxa may not have been sought after by Euro-Canadian fishers. Blind reliance on current biogeographical data can result in circular arguments. The zooarchaeological record provides our only empirical evidence for fish species distribution and diversity in the period immediately before contact. However, using zooarchaeological data to extrapolate precontact biogeography is problematic, because the extent of human selectivity is unknown.
The use of ethnographic analogy brings its own problems. Iroquoian archaeologists recognise that the archaeological record “departs significantly from ethnohistoric descriptions” (Jamieson 1989:309). The kinds of questions now being asked about Iroquoian behaviour involve topics not included in the abundant ethnohistorical data (Trigger 1982). Ethnohistorical sources provide some detail on food processing, yet are uninformative about discard behaviour (Needs-Howarth 1992); they do not provide the kind of information necessary to identify formation processes or the cultural context of foods. Fortunately, however, the ethnohistorical sources do contain some useful details on techniques of capture for fish. These will be discussed below. Fisheries science data The fish identified at the three sites include representatives of all the major fish families one would expect to find in the area. Small Cyprinidae, such as shiners and minnows, are lacking, but these are rarely recovered at archaeological sites in the area. Fisheries biologists have described in detail the biology, growth and behaviour of some of the taxa commonly found on archaeological sites, such as lake sturgeon, Salmonidae, longnose sucker (Catostomus catostomus), white sucker (Catostomus commersoni), and Percidae. However, there is a lack of fisheries data for species that are not important to the twentieth century Ontario commercial or sport fishery, such as yellow and brown bullhead. This problem was also noted by Cossette (1995). Local data are essential, since fish biology and behaviour, especially seasonal movements, vary with latitude and environment. Wherever possible, general observations have been replaced by, or complemented by, local data provided by fisheries researchers (Table 7).
In theory, it should be possible to determine the water body of origin through DNA signatures that are unique to a particular fish population. DNA analysis is currently being applied to modern fish material by fish biologists and environmental scientists. Dr. Thomas Whillans of Trent University and colleagues at the Ontario Ministry of Natural Resources are planning to obtain DNA signatures from lake herring and lake whitefish vertebrae from the Warminster site to determine whether these fish derived from the Georgian Bay/Nottawasaga River drainage or the Lake Simcoe drainage (Thomas Whillans, personal communication 1998).
Biogeography Introduction Having local data about the fish identified in the collections is helpful to some extent. However, in order to be able to understand to what extent humans were selective in their subsistence choices it is essential to know the biogeographic distributions of all taxa available at that time. Modern environmental data are not refined enough for this purpose, and cannot be extrapolated back in time. It has been recommended archaeologists do their own palaeoecological reconstructions (Lovis and MacDonald 1997). Unfortunately, while this might be feasible for mammals or birds, for fish this is quite problematic.
Lake Simcoe Lake Simcoe is part of the Trent-Severn Waterway system that connects Lake Ontario to Georgian Bay of Lake Huron (Figure 1). Lake Simcoe covers an area of 722 km2, with an average depth of 6 m (Johnson 1997). It is a cold water system, while adjacent Lake Couchiching is shallower, and hence warmer. Kempenfelt Bay on the west side of Lake 24
Simcoe (Figure 2) is steep-sloped, with a maximum depth of 41 m (Evans et al. 1996). In the past, Kempenfelt Bay may have extended further westward into a marsh-like lagoon (Robin Craig, personal communication 1998). Cook's Bay, at the southern end of the lake is up to 15 m deep (Evans et al. 1996); fish habitat in and around Cook’s Bay may have been altered by the draining of the Holland Marsh in the 1930s.
year. With sufficient volume, the main channel is navigable by fish (and humans pursuing them in canoes) from the mouth at Nottawasaga Bay to at least 40 km south of Minesing Swamp, near the branching of the Boyne River (Robin Craig, personal communication 1998).
Recent research has shown that phosphorous loading from soil erosion, agriculture and effluent has resulted in increased phytoplankton production and subsequent depletion of hypolimnetic dissolved oxygen (Johnson and Nichols 1988; Nichols 1997). While the increase in primary production may have helped some of the warm water near-shore taxa (Michael McMurtry, personal communication 1998), depleted levels of hypolimnetic dissolved oxygen in late summer constitute a major stress for the cold water fish species in the lake, including lake trout, lake herring and lake whitefish. The effect on cool water species, such as yellow perch, is less certain because they are more flexible in their habitat requirements (Michael McMurtry, personal communication 1998). Euthrophication is not the only factor in the current decline of lake whitefish and lake trout. Heavy harvesting and invasion of the lake by rainbow smelt (Osmerus mordax, Osmeridae) have also affected these species. Accidental introduction of another non-native species, the common carp (Cyprinus carpio, Cyprinidae), in 1896 is thought to have caused destruction of the wild rice beds and other rooted aquatic vegetation in the Holland River and Cook’s Bay, thereby degrading important habitat for muskellunge (MacCrimmon and Skobe 1970). In short, “eutrophication, increased fishing pressure, habitat destruction and invasion of non-native plants and animals have transformed Lake Simcoe and its assemblage of fishes over the past 150 years” (McMurtry et al. 1998). This transformation includes the loss of lake sturgeon, decline of muskellunge, failure of recruitment of lake trout and lake whitefish, and major fluctuations in abundance of lake trout, lake whitefish, lake herring, and yellow perch. The combination of habitat change and fish introductions poses considerable difficulties for reconstructing precontact fish communities from modern fisheries data. Rivers The Nottawasaga River is a warm water ecosystem. Shifting sand bars at the mouth of the Nottawasaga River (Hunter 1906:30), and silt on old levies in the river, indicate that the lower reaches of the Nottawasaga River were probably almost as muddy in the past as they are today, possibly making the main channel of the river an unsuitable spawning habitat for some species (Robin Craig, personal communication 1998). An influx of groundwater in Minesing Swamp, however, causes the water to be very clear beside the main channel. In spring, the groundwater has sufficient volume to restrict the muddy river water to the main channel (Robin Craig, personal communication 1998). Water levels in the river vary considerably from year to 25
Wetlands The extensive wetlands in the area between Lake Simcoe and Nottawasaga Bay may have been ecologically diverse compared to the upland forests. Indeed, Minesing Swamp may have been more productive than Lake Simcoe itself (Robin Craig, personal communication 1998). Minesing Swamp was more extensive in the precontact period (Sutton 1996a). The name Minesing is an Ojibway word that translates as “at the island” (John Steckley, personal communication 1998), probably referring to an area of higher ground surrounded by marl and swamp just north of Willow Creek and just east of the Nottawasaga River (Hunter 1906). As noted above, the west end of Kempenfelt Bay may have accommodated wetlands species.
journeys on open water. Fishing techniques can be divided into active and passive (e.g. Brinkhuizen 1983). The active techniques, such as spearing or angling, result in one fish being caught at a time. Passive techniques are usually less time consuming and involve netting or fish-trapping, allowing many fish to be caught simultaneously. As might be expected, variation in time, location and method of fishing results in different catches. Angling Angling could have been used to obtain most of the taxa represented. Hand-held lines can be used in the upper Great Lakes on open water and through the ice to catch large predaceous species (Cleland 1982:764). Angling at the Barrie, Dunsmore and Carson sites may not have been as important as netting because it is not as efficient in obtaining large numbers of fish. Fishing lines were most likely made from indian hemp (Apocynum cannabinum) (Sagard 1939:189), which has poor tensile properties when wet (Salls 1989). Indeed, Sagard (1939:189) mentions that hooks, made of wood with a bone bar, tied with hemp cord, were often found in the stomachs of fish, suggesting that the lines were not sufficiently strong. No fish hooks were recovered at the three sites, although two worked bone splinters from Dunsmore may have functioned as gorges.
Fish habitat Fish in lake habitats, such as Lake Simcoe and Nottawasaga Bay, were inaccessible for much of the year, and only abundantly available during their spawning season (Cleland 1982:766). However, given the site locations, most fish species in these three assemblages were obtainable in more than one kind of habitat, including the large lakes, but also rivers, creeks, smaller lakes and marshes. To bring this potentially ambiguous picture into better focus, following Thomas’ work on the nearby and culturally similar Hubbert site (Thomas 1996b), Needs-Howarth and Thomas (1998) turned to a fish habitat preference study conducted by wildlife biologists Jude and Pappas (1992). This study used correspondence analysis of fish abundance statistics from 16 Great Lakes fish census studies from the 1970s, 1980s and early 1990s to rank the habitat preferences of 113 species on an ordinal scale. Three major species complexes were defined on the basis of habitat preference scores: the open water (1-31); transitional, including open water, nearshore, and wetlands (32-66); and coastal wetlands (67113) taxocenes. In these three zooarchaeological assemblages the extremes of the spectrum are represented by lake herring (rank 7, open water) and grass pickerel (rank 109, coastal wetlands) (Figure 13).
Spearing/harpooning Spearing and harpooning was likely practised on many of the fish species recovered, especially the larger individuals. Cleland suggests the harpoon was in use by Middle Woodland times and may eventually have largely replaced spearing, since its detachable head attached to a line vastly improved chances of capture (Cleland 1982:774) . Wooden shafts or leister-type implements discarded at the sites would long since have decomposed. Chert spear heads are frequently recovered at Iroquoian sites and it is possible that some of them were used on fish.
Fishing techniques Introduction It is not known how the site occupants perceived the costs and benefits of the various fishing techniques available to them. In a general way, it can be assumed they used the simplest technology appropriate to the intensity of exploitation, and a location involving a short travel distance (Needs-Howarth and Thomas 1998; Thomas 1996b). While the Great Lakes have a large sustainable production of fish, it is spread out over a huge volume of water, most of it inaccessible to people travelling in small canoes. Both Sagard (1939) and Joutel (Kinietz 1965) indicate that native people did not travel more than a league from shore. The location of Odawa camps on islands in Georgian Bay indicates that the Odawa at least were capable of covering 8 km of open water. However, because creek, river, wetland and lake habitats are found within a five km radius of each site, most fish species represented at the sites would be convenient to exploit, without necessitating long canoe
Netting Evidence for the use of netting technology in the Great Lakes dates to at least the Late Archaic/Early Woodland (Petersen et al. 1984). Large quantities of fall-spawning lake herring, lake whitefish and lake trout bones indicate a redesign of the existing net technology, leading to the development of the gill-net (Cleland 1982; Cleland 1989). While it has been argued (Martin 1989) that nonhorticultural groups in Michigan were using gill-nets during the Middle Woodland, the zooarchaeological data from Ontario appear to concur with Cleland's Late Woodland timing. The use of nets was a cooperative enterprise, requiring a considerable labour investment for making, setting and tending nets and also for processing and/or preserving the catch (Cleland 1982). Iroquoian fishing nets were commonly made from Indian hemp (Sagard 1939:240) or nettle (Urtica holosericea) (Hennepin 1903:522). It is unlikely that many nets and their 26
attached netsinkers would be found at the village. They were probably maintained at the fishing locations. No netsinkers were recovered from the sites, and the nets themselves would not be preserved.
plant fibre nets may have displayed some variation in mesh aperture within a single net and, therefore, probably resulted in a somewhat less restricted size distribution. The effectiveness of the net is related to girth (McCombie and Berst 1969; McCombie and Fry 1960). In general, deepbodied fish are represented by shorter individuals than the more elongated fish from the same netting event, although swimming speed and gill morphology also play a role. Fish of a given size are taken most efficiently when their girth is approximately 25 percent greater than the perimeter of the mesh (McCombie and Berst 1969).
When spring-spawning fish travelled from Kempenfelt Bay up the tributary streams in large numbers, they may have been caught in shallow water in seine-nets or fish-traps (Needs-Howarth and Thomas 1998). A broader size range would be expected in each depositional context if the site inhabitants used techniques of mass capture such as seinenetting or fish-trapping. Seine- (or impounding) nets are long, deep, fine-meshed nets that have floats at the top and weights at the bottom so they rest on the substrate. A seinenet is used as a fence to actively corral fish towards the shore. The finer the mesh, the more small-sized fish will be taken. Compared to gill-nets, seine-nets trap a wider range of sizes. Seine-nets continued to be used after the adoption of gill-nets because they were better suited to the capture of shallow water fish (Cleland 1982:778). Fish caught in seinenets would usually still be alive when the fishers came to claim them. This means undesirable fish could released alive.
Unlike impounding nets, gill-nets are lethal to most things that become stuck in them; fish in gill-nets are often dead by the time the net is pulled up. The Algonkian-speaking Ojibway who currently live on the Bruce Peninsula have a taboo against returning dead fish, blood and guts to the water (Molnar 1997, citing a personal communication from Stephen Crawford). Paradoxically, the use of gill-nets may, therefore, result in a more varied species and size distribution at the processing location than the less selective method of seine-netting (Molnar 1997, citing a personal communication from Stephen Crawford). This taboo may have also pertained to the Huron and their ancestors.
Ice fishing was conducted by means of angling and seinenets (Sagard 1939:98).
The presence of diving birds may suggest the use of gillnets. In discussing merganser (Mergus merganser) bones found at a Middle Neolithic lake-side site in Switzerland, Studer (1992:81) describes how the merganser “exploits the same areas as the fishermen and enjoys abstracting some of the fish entangled in the nets and, if caught in the net [it drowns].” Large quantities of bones derived from lakedwelling deep-water fish may be another indication of the use of gill-nets.
“They make several round holes in the ice and that through which they are drawn the seine is some five feet long and three feet wide. Then they begin to set their net in this opening; they fasten it to a wooden pole six or seven feet long, and place it under the ice, and pass this pole from hole to hole, where one or two men put their hands through and take hold of the pole to which one end of the net is tied, until they come back to the opening five or six feet wide. Then they let the net drop to the bottom by means of certain small stones fastened to the end of it. After it has been to the bottom they draw it up again by main force by its two ends” (Biggar 1922-36(3):166-168) .
Gill-net fishing could be carried out by two or three people from a canoe. Net fishing can be quite a dangerous activity, as was indicated by an interment of two intermingled skeletons of adult men together with numerous net sinkers at a seventeenth century Iroquoian component near the Niagara River at Fort Erie (Granger 1976-77). In this case the current may have proved lethal, whereas on open lake water it would most likely be the high winds and waves.
In this quote, Champlain records how the Huron used nets for ice fishing. Cleland (1982:762) sees this as an example of gill-netting, while others interpret this activity as seinenetting (Heidenreich 1972:72; Molnar 1997:24).
Sagard (1939:185-186) recounts joining in a fishing expedition with the Huron, apparently involving gill-nets. They went by canoe to an island in the “Fresh-Water Sea” (Georgian Bay of Lake Huron) and “every evening they carried the nets about half a league or a league out into the lake, and...at daybreak they went to draw them in, and always brought back many fine big fish such as Assihendo [lake whitefish], trout, lake sturgeon, and others.”
Gill- (or entangling) nets are set to form an underwater curtain for entangling, gilling and wedging fish, with the latter two being the dominant modes of capture. Floaters and sinkers keep the net vertically stretched at a set depth. This technology is suited to offshore fishing in the deeper waters of the Great Lakes (Cleland 1982). The size distribution of fish from a gill-net is often more restricted than a seine-net (Hamley 1975). Modern gill-nets are size selective, with most fish being within 20 percent of the size median (Hamley 1975). Gill-nets made from cotton (and other plant fibres by inference) may be less efficient than modern nylon nets (Berst 1961; McCombie and Fry 1960). Handmade
Having observed gill-netting at the straights of Mackinac in Lake Michigan in 1687, Joutel (Kinietz 1965:29) provides the following description: “There are fish of various kinds which they catch with nets...and, although they only make them with 27
ordinary sewing thread, they will nevertheless stop fish weighing over ten pounds. They go as far as a league out into the lake to spread their nets... They have nets as long as 200 fathoms, and about two feet deep. At the lower part of these nets they fasten stones, to make them go to the bottom; and on the upper part they put pieces of cedar wood. Such nets are spread in the water...the fish being caught as they pass... The nets are sometimes spread in a depth of more than thirty fathoms, and when bad weather comes, they are in danger of being lost.”
Zooarchaeological indicators of method and location of capture Introduction It is possible to infer the technique of capture, location and time of capture, from the fish bones themselves, by going beyond simple reliance on habitat or spawning behaviour. As Wheeler and Jones suggest, “fishing methods are usually targeted to catch a particular species or range of species of a limited size range. Thus, by considering the species composition and size of fishes represented in archaeological assemblages, it may be possible to suggest which fishing technique was used in the past” (Wheeler and Jones 1989:168).
Fish-traps and weirs Fish-traps made of twigs or wicker can be custom-built to catch fish within a certain size range. They are often efficient for catching smaller, more plentiful fish (Brinkhuizen 1983:25). Other methods for catching fish involve the use of impounding gear or a weir. Residents of the river and wetlands, like northern pike and brown bullhead, may have been caught with fish-traps and weirs (Needs-Howarth and Thomas 1998). There are fish weirs between Lake Couchiching and Lake Simcoe, which were described by Champlain in 1615 (Biggar 1922-36(3) :5657).
This research presents a model for identifying the refuse of discrete or related fishing events through evaluation of fish biology, co-occurrence of taxa in refuse deposits, relative fish element sizes and seasonality attributes of fish and other animal remains. The basic premise for this method was presented by Needs-Howarth and Thomas (1998). It is explicitly aimed at elucidating both the nature of the deposits and the season of deposition. Fishing for food is a sensible option only when you can be reasonably sure to encounter fish. This involves targeting specific locations or specific times. “If a fisherman knows the habits of non-migratory fishes he wishes to catch and the places where they live during the year... and the techniques needed to catch them, he can obtain these fishes throughout the year” (Brinkhuizen 1989:111). This statement, while formulated in a European context, is probably equally valid for North American freshwater taxa. During their spawning times many fish congregate in predictable locations. Sagard says that the Huron knew within one or two days when each fish would start its spawning-run (Sagard 1939:231). Since fish lipid content, and correlated to that, caloric content, is at its peak at spawning, this may have been a time of intensified fish exploitation. The fact that the exploitation of certain taxa is considerably more efficient around their spawning time does not, however, justify the explicit or implicit assumption underlying many zooarchaeological analyses in southern Ontario that fish were almost exclusively caught during spawning time. In other words, optimisation of the resource does not necessarily imply spawning time exploitation. It is also possible that not all fishing activity was aimed at maximum efficiency.
“There is another lake immediately adjoining [Lake Simcoe],...draining into the small one [Lake Couchiching] by a strait [Atherley Narrows], where the great catch of fish takes place by means of a number of weirs which almost close the strait, leaving only small openings where they set their nets in which the fish are caught; and these two lakes empty into the Freshwater Sea [Georgian Bay, by way of the Severn River].” The weir has been investigated by archaeologists (Cassavoy 1993; Johnston and Cassavoy 1978) and was found to have been in intermittent use from the Late Archaic, around 2500 B. C. (Johnston and Cassavoy 1978). The orientation of the stake pattern indicates that the weir was used to obstruct fish swimming in both directions, towards Lake Couchiching and towards Lake Simcoe. Excavator Kenneth Cassavoy (personal communication 1995) suggests that the construction indicates that users of the weir were primarily fishing for large quantities of smaller fish. Both weirs and fish-traps require frequent inspections to avoid losing the contents to fish-eating aquatic mammals (G. J. Boekschoten, personal communication 1999). A permanent fish weir would have demanded labour cooperation and coordinated visits (Martin 1989:596). It is quite likely that the weir at Atherley Narrows was operating by people living in the immediate surrounding area, and that the occupants of these Barrie, Dunsmore and Carson sites did not have access. No other archaeological weirs are known locally.
The use of spawning time and location as an explanatory device is most applicable to species that aggregate during spawning (Needs-Howarth and Thomas 1998). With some exceptions (Thomas 1996b) much previous local research (D'Andrea et al. 1984; Lennox et al. 1986; Naylor and Savage 1984; Stewart 1991a:Figure 4; Stewart 1992) has not fully utilized the distinction between those species whose spacing behaviour makes them more available and/or abundant during spawning, and those whose spacing behaviour does not. For example, during most of the year lake sturgeon and 28
longnose sucker occupy deep water habitat (Scott and Crossman 1973, 86, 532-4) that is generally inaccessible to exploitation with precontact Iroquoian fishing technology. During their spring spawning-runs these species aggregate in harvestable quantities and move into watercourses, making them accessible to people equipped with simple technology (Cleland 1982). The lake sturgeon represented at the Barrie site were likely only purposefully procured during their spring spawning-run up the Nottawasaga River (NeedsHowarth 1996). The open water taxocene thus includes more taxa that were seasonally restricted in availability than either the large bay and estuary or coastal marsh taxocene (Figure 13). In contrast, the spawning behaviour of other taxa – such as brown bullhead, Rock bass (Ambloplites rupestris), and Pumpkinseed (Lepomis gibbosus) – does not involve aggregation or movement that significantly increases their availability (Scott and Crossman 1973, 589603, 703-6, 716-7). Instead, these species are available throughout warmer weather for stream-based and onshore exploitation (Needs-Howarth and Thomas 1998).
condition and reproductive state, whereas the length will stay the same or increase. The size estimations and metrics derived from the archaeological samples, therefore, should relate better to total length than to total weight (Brinkhuizen 1989; Owen and Merrick 1994; Shawcross 1968). Since it is not known a-priori whether a given bone results from a gill-net catch, a seine-net catch, or some other method, and since accurate TL estimates are lacking, we are limited to analysing netting events in terms of similarly shaped taxa or single species, rather than net type or mesh size. In addition, to obtain somewhat reliable information, this can only be done for taxa with relatively large NISP values.
Between these two extremes are species like white sucker, yellow perch, and Smallmouth bass (Micropterus dolomieu) (Needs-Howarth and Thomas 1998). These fish congregate in nearshore shoals and in local streams during their spring spawning-runs, but they are also available in various watercourse or nearshore habitats during the rest of the fishing season (MacCrimmon and Skobe 1970, 67, 101, 118-9).
Fish bone size and state of maturity Most of the species represented at these sites could be caught at different times of the year. For example, yellow perch are more easily obtained during their spawning-run from mid-April to early May, but can also be caught throughout the warmer weather or during ice-fishing on the lake. In addition to informing about recovery bias and technique of capture, fish bone size has yet a third application. Fish cranial bone size can be used as a general guide for maturity and hence time, location and mode of capture. Mature fish may have been part of a spawning-run catch, whereas most immature fish were not. Immature fish could still have been caught at the time of the spawning-run, but, for most taxa this would probably represent a different intensity of fishing effort, in a different location.
Table 20 shows what size range of each species could be expected in different stretched mesh sizes, illustrating clearly the effects of fish shape and body depth. Esocidae, with their torpedo-like shape are less likely to get stuck in a given mesh size than perch, with their deeper body, which in turn are less likely to get stuck than brown bullhead with their wide body and protruding pectoral and dorsal spines.
Fish bone size and technique of capture For fish that may have been caught with techniques of mass capture (e.g. seines, weirs), NISP or MNI alone are inadequate in examining fishing techniques. The size of lake whitefish at Algonkian sites along Lake Huron and Michigan has been used to infer method of capture (Smith 1996). Fish size has been used in conjunction with known size and weight of fish of various sizes intervals in the reference collection to estimate size distribution (Martin and Colburn 1989).
Because yellow perch constitute such a large proportion of the fish NISP it was important to find out whether the fish represented at the sites were part of a spawning-run catch, or part of a more generalized fishery throughout the warmer weather. Indeed, the interpretation of the nature of fish catches represented in refuse deposits hinges to a significant extent on these yellow perch. Using relative-size observations as a very general guide to fish size it is possible to estimate at what proportional element size percentage bones were likely large enough to belong to sexually mature individuals.
Unless some fish were released alive, a broader size range in each depositional context should result if the site inhabitants used techniques of mass capture (NeedsHowarth and Thomas 1998). Because a seine-net acts like a fence, the finer the mesh, the more small-sized fish will be taken. The disparity in the distribution of very small perch bones between Barrie and Dunsmore (Figure 10), may indicate that, on average, finer net mesh sizes were used at Dunsmore (Needs-Howarth and Thomas 1998). It is obvious that fish proportional element size observations, in addition to being useful in clarifying taphonomic issues (as discussed in Chapter 4), are essential in interpreting fishing techniques.
Modern Lake Simcoe male perch reach sexual maturity by about 150 mm fork length (hereafter FL); females by about 170 mm FL (McMurtry 1991:30-31). Mature female yellow perch caught by anglers in 1983 at Atherley narrows had an average FL of 198 mm (Arndt 1989:12-13). It is, therefore, cautiously suggested that archaeological bones measuring 80 percent or more of the size of the reference specimen element belong to sexually mature individuals (Needs-Howarth and Thomas 1998) (Table 21).
Proportional element size were used to identify groupings of bones of similar size that may constitute the catch from a net. Fish size can be quantified as TL or total weight (hereafter TW). The weight will vary because of seasonal 29
Size at maturity of yellow perch fluctuates in relation to latitude and watershed productivity, which in turn is related to climate, composition of the fish population, absolute numbers of fish, and predation (David Brown, Michael McMurtry, personal communication 1999). The presence of many Percidae scales in the Dunsmore and Carson assemblages appeared to afford the opportunity to get independent data on fish age and associated sexual maturity.
reflect changes in rate of growth. Studies aimed at testing the accuracy of age determinations from calcified structures show that there is normally good agreement between assessed age and actual age for young and fast-growing fish (Casselman 1983; Casselman 1987). Age interpretation from scales becomes less reliable when the checks or zones associated with annuli are indistinct, variable in appearance, or coalesce. Crowding of annuli occurs in older fish, which grow more slowly, for a shorter period of the year (Casselman 1983; Casselman 1987). As the body of the fish approaches asymptotic, or theoretical ultimate, length, the scales almost or completely stop growing, making linear growth of scales an unreliable mechanism for recording growth in older fish (Casselman 1983; Casselman 1987). In contrast, the cleithrum of northern pike, muskellunge and lake trout continued to grow at an increasingly faster rate than the body as the fish approached it asymptotic length, possibly as a functional response to give structural support to a disproportionately greater body mass (Casselman 1990:686).
Age and season of death Introduction Parts of this section were previously presented (NeedsHowarth and Brown 1998; Needs-Howarth and Casselman 1996). The study of age and growth from calcified structures has a long history in fisheries research, where it is used to obtain important physiological and environmental information. Most research concentrates on describing seasonal growth in order to establish population structure and growth patterns. The technique that is most widely employed involves deciphering age from cyclic growth evidenced in optically different seasonal growth zones in calcified structures, and using this seasonal growth history to infer the passage of time.
Age and growth interpretation is easiest when the fish is young and when it is growing fast, early in the growing season (Appelget and Smith 1951; Bennett 1937; Creaser 1926; Hille 1941; Hubbs and Cooper 1935). A striking example of the late formation of annuli in older fish is found in bluegill (Lepomis macrochirus) from Missouri (Lane 1954), where the annulus appeared three months earlier on scales of yearlings than on scales of fish aged four years.
The main premises for calcified structure analysis is that fish grow throughout their life and that fish body temperature is inconstant. Body temperature varies in a positive manner, within certain limits, with the temperature of the surrounding water. Fish therefore display a marked seasonal regularity that relates to their cold-bloodedness. Temperature is a major influence on growth, development and spawning (Hokanson 1977). “Fish grow in response to their biotic and abiotic environment and in accordance with some function of size or age already attained” (Weisberg 1993:1229). An individual growth increment is thus the sum of growth related to the characteristics of the calendar year in which it formed, and growth related to the age (size) of the fish when it formed (Casselman 1983). To interpret calcified tissue correctly, these underlying mechanisms of fish growth must be understood and applied.
The maximum reliable age determination thus varies by element and by species (Casselman 1990). The kind of structure used in fisheries research depends on the aims of the study, and the ease of collection. Scales are the most frequently used element (Beamish 1973; Harkness 1922; Hatch 1961; Hogman 1968; Lane 1954; Ovchynnyk 1962; Ovchynnyk 1965; Regier 1961), in part, at least, because it is not necessary to kill the fish to obtain them and they are easy to prepare. Other elements discussed in the North American fisheries literature include otoliths (Yosef and Casselman 1995), pectoral spines (Beamish 1973; Cuerrier 1951; Marzolf 1955; Rien and Beamesderfer 1994), vertebrae (Marzolf 1955; Ovchynnyk 1962; Ovchynnyk 1965), opercula (Bardach 1955; Ovchynnyk 1962; Ovchynnyk 1965), and cleithra (Casselman 1974).
Interpretations of age and growth rely on the presence, continuity and spacing of circuli and interruptions in scales, and the optical density, opacity and translucity of zones in other structures. The annulus is associated with the distal edge of a concentric ring in the form of a check in scales or translucent zone in spines and other structures, which can be detected in all regions of the structure. It is formed when the fish is growing slowly, or not at all; in temperate regions this is usually during the autumn, winter and early spring (Creaser 1926; Hubbs and Cooper 1935). Validation of age and growth interpretation and assessment of variability are important parts of any study, whether this involves establishing growth curves for modern or archaeological fish. A study of elemental composition and appositional growth of northern pike cleithra injected with tetracycline (Casselman 1974) confirmed that opaque zones form at a much faster rate than translucent zones, and hence that optically different zones in calcified tissue do indeed
Recent research on fish biochronology holds much promise, for both biological and archaeological applications. By detrending growth increments for the effects of age-related allometry, it is possible to establish a history of environmental effects (Weisberg 1993). In fish species located close to the northern limit of their range, these environmental effects may be used as indicators of climate change. Long-lived fish that shared a few years of their lives as contemporaries can then be cross-matched in a manner analogous to dendrochronology to obtain a population biochronology. This method has been successfully applied 30
to populations of freshwater drum (Aplodinotus grunniens, Sciaenidae) (Pereira et al. 1995a; Pereira et al. 1995b).
interpretation, such as erosion or resorption, to be described. It also permits recording of negative observations, which can be just as important as positive observations. The digitized measurements are stored on disk as absolute measurements, but they are also visually recorded on a relative scale, so that the growth cycles of individuals of different ages and sizes may be compared on the screen or in a print- out. The software is used to standardize interpretation procedures and, in younger individuals of well-studied species, to convert the growth conditions on the edge of the structure to a calendar age.
CSAGES fish growth interpretation system Fish age and growth have been recorded with the use of a microcomputer and digitizer since the early nineteen eighties (Cailliet et al. 1996; Frie 1982; McGowan et al. 1987; Small and Hirschhorn 1987). The Calcified Structure Age and Growth Extraction System (CSAGES) and related software (Casselman and Scott 1994) currently being developed at the Ontario Ministry of Natural Resources Fisheries Research Station, Glenora, makes age and growth interpretation more objective and more accurate (Casselman and Scott 1994). Interpretations are made on the anterior field, measuring from the focus (or centre) of the scale to the outer margin, at 50x or 100x magnification. The position of growth interruptions is recorded using a hand-operated digitizing tablet. The extent, nature and temporal significance of each check or zone is described and classified using confidence ratings that range from 1 to 9. The appearance of the edge of the structure in relation to the seasonal growth cycle is coded in a standard manner (Table 22). The edge conditions described in Table 22 require some additional clarification. Assigning a date of capture based on the most recent growth history is somewhat more difficult. The absolute and relative chronology of growth cessation, spawning, and growth resumption varies, even within genera. Interpretation of the margin in relation to the fish's life cycle, therefore, requires determination of the expected time of annulus formation and spawning at different ages for each taxon represented in the sample. Depending on the species and the age, a scale in the “o” condition (annulus incompletely formed, translucent check or zone present on the edge) (Table 22), could have been obtained any time from late autumn to late spring. Depending on the age of the fish, and whether growth resumed before or after spawning, it may be difficult to establish whether a fish was caught around spawning time. The “*” condition is very restricted in time, and therefore rarely encountered. It is the most difficult to interpret, owing to the fact that the annulus becomes visible only when new bone tissue is laid down. The variability in rate of growth within a season, and between seasons, entails that the “+” first new growth, annulus just completing formation) and “++” (growth less than half of previous annual zone ) conditions do not necessarily equate to the absolute amount of time elapsed.
Potential of age and growth analysis on archaeological fish remains The utility of age and growth analysis to archaeologists was first described in detail by Casteel (1974). The principle of uniformity allows us to assume that the parameters that affect modern fish also operated in the past. Provided that adequate local comparative samples are available, calcified structures from archaeological sites offer us the potential of establishing age structure of the catch as well as season of capture. The age of the fish can be used to infer whether the individual was sexually mature or not. This is a major benefit, since age is a more reliable indicator of maturity than size (Brown, personal communication 1998). An important additional benefit of age and growth analysis is only just now being explored (Bergquist 1996). Using biochronology, growth records of long-lived fish may be used to obtain a minimum duration of site occupation. If appropriate climatic data, or a proxy thereof, is available, fish growth sequences may even provide an absolute chronology of site occupation, using the temperature signal as the common variable to anchor the chronology in time. Age and growth analysis is a useful tool to archaeologists; CSAGES is sophisticated enough to provide considerably more reliable results than have been obtained in previous archaeological studies of calcified structures. In order to get realistic results, however, the limitations of age and growth analysis in general must be considered. Archaeological age and growth studies face additional challenges, since they are based on a different premise than those in fisheries science. Archaeologists lack the basic, apriori information that is usually available to the fisheries biologist: species, date of capture, and location of capture. In addition, it is not always possible to choose the most useful element; the practical application is limited by what is preserved and recovered (Cossette 1995:543). Because otoliths grow at a faster rate than the body during slow somatic growth they are excellent structures for recording growth in slow-growing and old fish (Casselman 1990:686). Thus lake trout and northern pike otoliths continue to record seasonal growth after the scales have started to fail, resorb or erode (Casselman 1990). The suitability of otoliths in recording growth of old fish has also been demonstrated in freshwater drum (Pereira et al. 1995b). However, it appears that otoliths were not preserved and/or recovered on sites in the Lake Simcoe area. Indeed,
CSAGES (Casselman and Scott 1994) increases replicability of results because it relies on standardized descriptions and qualifiers. It goes beyond simply marking the location of checks and zones. It incorporates standardized procedures for describing specific criteria, and more precise and detailed interpretive systems for data extraction and interpretation. The CSAGES computer program permits conditions on the edge that may affect 31
Freshwater drum is rare on the earlier sites, although it was identified in reasonable quantities at the Auger site (BdGw3), a Huron village located near the Coldwater River, north of Bass Lake (Latta 1998).
In addition to any problems specific to archaeological studies, there are many problems surrounding age interpretation that are encountered by both modern and archaeological interpretations. As noted above, because of the cyclicity of growth, it is easiest to establish seasonality from individuals that were taken during periods of rapid growth. The difficulty in pinpointing time of catch in cases where a fish was caught during the period of growth arrest is explicitly acknowledged by many zooarchaeological researchers (Carlson 1988; Morey 1983; Rojo 1987; Van Neer 1993). It has led one researcher (Carlson 1988:69) to conclude, perhaps overly pessimistically, that it is “impossible to distinguish the time of death within that eight month period” of slow growth. While all methods, including CSAGES, are hampered when the calcified structure is experiencing growth arrest, CSAGES provides the analyst with a system to look in more detail at growth at the scale edge. This forces the analyst to examine the calcified structure more critically, and to be explicit about criteria used to evaluate checks and zones. The condition of the margin can be described as different levels within the “++” and “o” condition (Table 22). In addition, the accurate description of the edge condition of other scales in the sample can be used for corroboration.
The opercula and cleithra of all species at Barrie, Dunsmore and Carson are too uniformly opaque and/or mineral stained for interpretation, although these elements have been used successfully in biological studies (Casselman 1974; Ovchynnyk 1965) and on archaeological material from the Warminster site, a Huron village located north-west of Dunsmore and Carson (Whillans, personal communication 1998). Fish scales are among the more fragile zooarchaeological remains, although there is considerable variation among taxa and body sizes. As noted above, cycloid scales are expected to be under-represented in relation to the more durable ctenoid scales. In general, Salmonidae scales are very small and fragile in relation to body size, although some intact lake whitefish scales were successfully used to indicate autumn capture at the Gros Cap site (Martin 1982:83). Given the relatively aggressive method of recovery (dry screening vs. water screening), it is entirely likely that the sample of fish scales from the sites under discussion here is biassed towards larger scales from taxa with the most durable scales. Imperfect preservation and edge damage on scales from archaeological sites can result in many rejected samples, causing additional bias. When working with fish scales, rather than paired cranial elements, archaeologists are further hampered because they do not know the exact location of the scale on the fish body. It is argued (Tesch 1978:96) that “for comparability the scales in any one study series should be taken from the same location on all the fish....” Archaeological scales could be from any part of the body, which limits the potential for comparing scales in a sample. For this reason also, it is not possible to calculate MNI, and thus rule out duplication of individuals. If two scales show different ages or different growth patterns on the margins, they are likely to represent two individuals (for an example, see Penmann and Yerkes 1990), however, the opposite is not necessarily true.
The nature of fish growth means that scales from older animals will be harder to interpret, although it is still possible to obtain a reliable minimum age. Erosion or resorption of the edge in older fish can also complicate interpretation of season of death. Another problem arises from regenerated scales, which are formed when the original scale is lost, often in the first years of life, or during spawning. Regenerated scales have an amorphous centre, but possess edges characteristic of the species. Because they lack a focus, they can only be used to establish a minimum age, not an absolute age. While regenerated scales are readily recognisable (Casselman et al. 1986), they do not allow for precise age determination. It has been suggested by some (Artz 1980:51; Casteel 1976:71) that all archaeological scales ages four or five and over should be eliminated from analysis. This is, however, unnecessarily cautious for most species, and would bias the samples with respect to longevity (Carlander 1987).
Identification of archaeological scales below family level using drawings and photos (Daniels 1996; Lagler 1947; Oates et al. 1993) or a reference collection may be difficult because size and shape of scales vary with age, TL and placement on the body of the fish. Because cranial and pectoral girdle bones can often be identified to species, the identified taxa from the faunal sample will narrow the choices. It is, however, conceivable that a taxon is represented only by its scales. Such scales must be confidently identified to species in order to maintain control over MNI (see Brinkhuizen 1997 for a discussion). As will be shown in Chapter 6, in some special cases a detailed age and growth analysis has the potential to confirm a species identification.
Problems with age and growth interpretation encountered by fisheries biologists are compounded for those working with archaeological material (Cossette 1995:494; Martin 1982:84; Wheeler and Jones 1989:157). Archaeologists are advised (Wheeler 1978:73) to “have their results checked by competent fishery biologists, at the outset of their study” and to “not expect simply to examine one or two scales or otoliths and establish seasonality of capture on this basis” (Wheeler and Jones 1989:156). Indeed, some archaeological researchers discuss how their own inexperience with age and growth interpretation may have led to bias (Hanna 1981; Junker-Andersen 1984), and have their material re-examined by a fisheries biologist (JunkerAndersen 1984).
32
The method described here has the potential of greatly enhancing the amount and quality of age and growth information extracted from archaeological fish bones and scales. Experience with CSAGES teaches the interpreter to be careful and increasingly observant. Experience in interpreting material from different species of different ages allows the interpreter to achieve much greater precision than is usual for archaeological age and growth studies. This is because the method goes beyond simply recording whether the annulus has formed yet, or has formed recently; it also quantifies and qualifies the status of the margin. Confidence codes allow for an evaluation of the assigned season of capture. CSAGES therefore satisfies the important criteria of efficiency, accuracy and precision identified by Monks and Johnston (1993), although not exactly in the manner they recommend.
CSAGES co-author John Casselman. A random sample of 37 larger scales was analysed by David Brown using CSAGES. Sample preparation was carried out by the author and sample interpretation was carried out by David Brown. Initially, the archaeological scales were gently cleaned in warm water and liquid soap and dry-mounted between microscope slides. Surprisingly, warped scales were easily flattened by taping the two slides together. While dry-mounted archaeological specimens can be interpreted in some cases (Scott 1989), often they are much more opaque than modern scales and therefore more difficult to interpret (Artz 1980; Hanna 1981). The scales were discoloured and appeared to be covered in a thin film of mineral concretions. Placing them in 60 percent glycerol (often used for modern samples) did little to enhance the image. A sonic cleaner was also used, although there are concerns (Yerkes 1980) that this may cause breakage at the annulus. This too did little to enhance the image. Finally it was decided to impress them. This involves placing the scale sculptured side down on a thin slide of warmed, clear cellulose acetate. The slide and scale are then passed through a roller press. Although this is the technique preferred by fisheries biologists, there was a certain reticence towards impressing the more degraded archaeological material. Tests showed, however, that despite the mineral concretions that covered the scales, the impressed images were very clear. Either the concretions affected just the valleys between the circuli, or they covered the entire scale uniformly. If just enough pressure was used, rolling the slides left a perfect impression, which could then be viewed under a microscope or in a microfiche reader. This indicates that scales do not necessarily have to be rejected because they appear degraded during macroscopic examination. This in itself is an important finding of this study. Discussion of the results is provided in Chapter 6.
Informative skeletal structures at Barrie, Dunsmore and Carson As has been discussed above, inter-site distributions of fish remains indicate that fish scales do not always survive precontact butchering, cooking, long-term burial and subsequent excavation. All bone material from the Barrie site was noticeably less well preserved. This may explain why only one scale was recovered from this site, despite the fact that it was subject to the most intensive flotation sampling. Improved preservation at Dunsmore and Carson has resulted in much better scale recovery (Table 23). As noted above, no otoliths have been recovered from Barrie Dunsmore or Carson. If otoliths were preserved, some should have been recovered from the flotation heavy fraction. The residues from Barrie and Dunsmore were sorted by the palaeoethnobotanist (Monckton 1993; Monckton 1996); the residue from Carson was sorted by the author. Since none of the flotation residues contained otoliths, operator bias during sorting is probably not a factor. And since other cranial elements were recovered in abundance for most taxa, off-site disposal of fish heads is not a sufficient explanation for the absence of this particular element. Instead, the lack of otoliths likely results from a combination of large dry screen mesh aperture; the relative and absolute small size of the otoliths of taxa represented (especially Salmonidae); and the unique chemical composition of this element.
Brown bullhead spines As mentioned in relation to the discussions of bone tools and diagnostic elements, brown bullhead dorsal and pectoral spines have a better than average change of being recovered and identified. Pectoral spines of Ictaluridae have been used successfully to establish age and approximate time of capture (Cossette 1995; Morey 1983). While both the fisheries data discussed above and the zooarchaeological data discussed in Chapter 6 suggest that brown bullhead at Barrie, Dunsmore and Carson were caught throughout the warmer months of the year, the recovery of substantial numbers of pectoral and dorsal spines from the three sites offered the opportunity to test this hypothesis. The entire sample of pectoral and dorsal spines recovered from Barrie, Dunsmore and Carson was analysed using CSAGES. Sample preparation was carried out by the author under supervision of David Brown. Interpretation was carried out on a contract basis by David Brown at the Ontario Ministry of Natural Resources, Glenora.
Percidae scales Age and seasonality interpretations based on archaeological fish scales usually give the time of catch as a range of several months (Artz 1980; Benecke and Thue 1995; Penmann and Yerkes 1990; Scott 1989). In addition, many problematic scales are rejected (Hanna 1981; Scott 1989). In recognition of the problems surrounding fish scale interpretation, and in order to get precise and reliable data, fish scale interpretation for this project has been carried out by David Brown, a specialist who has worked for eight years as research biologist, examining fish growth and fish community dynamics at the Ontario Ministry of Natural Resources, Glenora Fisheries Station, under direction of 33
To avoid the section breakage experienced by other researchers (Cossette 1995), it was decided to embed all spines. Brown bullhead pectoral and dorsal spines with intact proximal ends were hardened in high resolution epoxy resin matrix for two days. Two sections of about 325 micron thickness were cut from each spine with an Isomet low-speed blade wafering saw operating in a tap-water blade bath. The location of the first section was immediately below the v-shaped basal groove, ensuring that the total circumference of the spine was present (cut #2 in Harvey and Fortin 1982). Sectioning at the correct angle and thickness is necessary to avoid obscuring annuli. The best of each of the sets of two thin-sections was then polished on fine lapping film and fixed with epoxy resin to microscope slides. After another day of curing the reverse was polished on 600 and 800-grit Carborundum paper until the zonation was clearly visible.
suggest that lake sturgeon pectoral spines continue to record growth to an advanced age. Archaeological material may be considerably harder to interpret; inexperienced interpreters working with older individuals from the Hector Trudel site exhibited considerable intra-reader variation in age estimates (Cossette 1995:556). Lake sturgeon age at sexual maturity varies between bodies of water, and by latitude (Harkness and Dymond 1961; Roussow 1957). First spawning may take place some years after maturity is reached (Roussow 1957:560). It is probably safe to assume that if the sturgeon represented at the sites were older than 25 years, they were of spawning age (Harkness and Dymond 1961:32). This does not imply the fish were actually on their spawning-run when caught, but it would allow for that possibility. Five fragments of the first (ossified) fin ray, or pectoral spine, of lake sturgeon were recovered from the Barrie site. An introduction by Dr. David Noakes of University of Guelph led to a collaboration in 1996 between the author and Mr. Greg LeBreton, who at the time was completing his M.Sc. on growth histories of lake sturgeon in Canadian waters. While CSAGES analysis has been successfully applied to this taxon previously, budgetary constraints made it desirable to take up LeBreton’s offer of complimentary analysis, using a method he developed.
The translucent zones on spines appeared less variable in appearance than the circuli on scales, and were therefore easier to interpret. The sections were viewed under transmitted light at 50x and l00x magnification, examining the posterio-lateral section of the medial surface of the cut. The radius of measurement was oriented from the centre of the lumen along the maximum radius for growth. For pectoral spines, this runs to the outer edge of the dorsal/posterior area of the spine (i.e. longest length) (Cossette 1995: 545; Marzolf 1955: 244; Sneed 1951:178).
The size of the elements indicated that they were quite old, and that it would not be possible to obtain season of death information. Instead, it was hoped to establish a minimum age in order to establish whether the bones derived from sexually mature individuals. The “clumping” of annuli in older fish in the years preceding spawning could be another indicator of reproductive status (Roussow 1957). Caution is required, however, because annulus pairing also happens in immature fish of some populations (Rossiter et al. 1995), and it is entirely absent in other populations (Dumont et al. 1987). In addition to getting clues to reproductive status, it was hoped the spine data could be used to build a biochronology (Bergquist 1996), using LeBreton’s modern lake sturgeon samples as a model.
Focussing required some precision so as not to become confused by zones lower down, which are closer together, because the element tapers. Deterioration of tissue around the lumen in spines from older individuals of a closely related taxon, the channel catfish (Ictalurus punctatus), can partly obscure the first annuli (Marzolf 1955:245). This was found to be true for our Bullhead samples. It appears that spine growth is accomplished by seasonal deposit of tissue on the outside. The basal groove continues within the spine as a central lumen; as the spine grows, the lumen enlarges at the expense of the earliest bone deposits (Marzolf 1955:246). The basal groove also lengthens and enlarges, so that the point of sectioning has to be moved distally along the spine for older fish. Discussion of the results is provided in Chapter 6.
A successful application of fish biochronology must satisfy three biological criteria (Bergquist 1996). The first criterion is that the calcified structures of the species in question must strongly reflect the temperature signature (or other fluctuating environmental signal) of the surroundings. This is the case with lake sturgeon spines. The second criterion is that the species must be long-lived, in order to increase the likelihood of cross-matching with other individuals and a temperature series. Lake sturgeon is more suitable than most species because it is so extremely longlived, although age estimates do become less accurate with age. The third criterion is that the species must be living at the limit of its range, so that it is more sensitive to temperature fluctuations. In this respect, lake sturgeon is noticeably less suitable than the species on which fish biochronology was first developed, the freshwater drum.
Lake sturgeon spines Thin sections of lake sturgeon spines can be used to establish age and approximate time of capture (Cuerrier 1951; Cuerrier and Roussow 1951; Rossiter et al. 1995; Roussow 1957). Only one previous archaeological application has involved this particular species of sturgeon (Cossette 1995). In a recent validation study, age estimations based on annulus counts were highly consistent with known ages for fish up to 15 years old. Agreement declined slightly at age 18 and older, but never fell below 80 percent (Rossiter et al. 1995). The oldest fish was 36 at recapture. These data 34
The range of lake sturgeon extends to Hudson’s Bay; individuals living in the Great Lakes region cannot be considered to be close to the northern limit of their range.
1998). Species that regularly occur together in archaeological deposits were thus used to establish the most likely time and location of capture for the species recovered in each archaeological context. Co-occurrence of taxa to interpret fishing events from deposits has been used in Odawa and Ojibway contexts (Molnar 1997; Smith 1996).
A biochronological study must also satisfy sample size criteria (Bergquist 1996). In a pilot study involving freshwater drum otoliths, several individuals in both the modern and the archaeological samples exhibited low correlation between otolith patterns and temperature series. Larger samples are needed to assure consistent and strong cross-matching and to obtain a statistically valid correlation with temperature, both as a way to validate the crossmatched series, and to anchor it in absolute time. Small sample size alone was likely to preclude crossmatching of our lake sturgeon spines and correlation to a temperature series. Against the odds, it was hoped we would be able to cross-match the five pectoral spine fragments to obtain a minimum duration of site occupation, using LeBreton's modern samples of known capture date as a model for cross-matching and de-trending (i.e. factoring out age effects) of the archaeological samples. LeBreton was also exploring the possibility of using old growth cedar trees on the Niagara Escarpment (some of which are 800 years old) as a surrogate of temperature, so that he might anchor the “floating” relative growth chronology in time, and thus obtain an absolute date for site occupation. Again, the aim was to use the matching of modern population data series to air temperature as a model for matching the archaeological spines to the cedar tree ring data.
For the current application, the model has been expanded to incorporate identifications of all vertebrae at the genus or family level, and data on age and season of death. The model presented here combines fish habitat and spawning data, fish behaviour, relative element frequencies per taxon, age and growth data, associations between taxa within features, and seasonality inferences related to those features. The three fisheries complexes defined for the current work are: 1) Spring Spawning-run Fishery: a spring time watercourse-oriented inland fishery that focuses on intensive exploitation of spring-spawning taxa such as lake sturgeon, suckers, yellow perch and walleye; 2) Generalized Warm Weather Fishery: a generalized bay or inland fishery for opportunistic warm weather exploitation of resident taxa that do not aggregate in harvestable quantities during their spawning-runs, such as Esocidae, brown bullhead and Centrarchidae, and of immature and non-spawning yellow perch; and 3) Lake Fishery: a lake-oriented fishery on Kempenfelt Bay and Nottawasaga Bay that included inshore exploitation of autumn-spawning Salmonidae. Fish may be found together in a deposit because they inhabit the same waters and/or they spawn together and/or they are amenable to the same techniques of capture. An understanding of fishing strategies was facilitated by a consideration at the feature level and site level of detailed species identifications, fish biology, scales, vertebrae, element size, age and season of capture. Some components of this model have been employed by other researchers working with freshwater fish remains in a North American context. Cleland's discussion on the inland shore fishery in the upper Great Lakes (Cleland 1982) highlighted the uniqueness of the autumn Lake Fishery in terms of habitat, spacing behaviour, method of capture and preservation. Species composition and size has been used in conjunction with seasonality data from scales to identify fishing practices at an Oneota site in Wisconsin (Yerkes 1981a). Cossette’s (1995) interpretation of fishing at the Hector Trudel site incorporated both ethnohistoric and local ecological and fisheries biology data. She explicitly recognized that fish exploitation does not necessarily correlate with spawning season. She used diversity and richness indexes to assess variability among samples, and used fin spines to estimate the ages of four fish species.
In the end, it was the poor state of preservation of the archaeological material that proved fatal to the project. Initial examination of the sections indicated they suffered from delamination, decalcification and probable ironstaining. The longest consecutive segment was 10 years. This precluded all but a rudimentary estimation of minimum age. The research plans are presented here so that they might be tried by other researchers on better preserved samples. The Three Fisheries Model Introduction Three kinds of fishing strategies were proposed by Thomas (1996b) for the Hubbert site (BbGw-9), located south of Kempenfelt Bay near Lover's Creek: 1) a spring upstream fishery that concentrated on species which migrated a substantial distance upstream; 2) a long-term watercourse fishery to exploit resident populations of fish living upstream, in the low gradient lower reaches, and in the estuary; and 3) a lacustrine fishery to exploit resident populations of fish throughout Kempenfelt Bay, perhaps with emphasis on fish spawning on shoals. These fisheries complexes were further refined by Needs-Howarth and Thomas (Needs-Howarth and Thomas 1994b; NeedsHowarth and Thomas 1998) by the inclusion of detailed information on taxa co-occurrence within single refuse contexts, by categorising yellow perch bones based on proportional element size, by identifying and quantifying Salmonidae vertebrae; and by emphasizing the temporal components of each fishery (Needs-Howarth and Thomas
The issue of fishing strategies has been investigated recently in two studies by Ontario archaeologists. Using student analyses of fish remains from a midden at the Huron period Auger site, near Bass Lake, Latta (1998) relates fish remains to “opportunistic”, “flexible” and “structured” procurement patterns that are based on method of 35
exploitation; fish spacing behaviour; spawning; feeding behaviour; habitat; travel distance and organisation of labour. Seasonal variability in fish behaviour and human exploitation play a more limited part in this model. Molnar’s (1997) interpretation of an unstratified, multi-component fish bone assemblage at the early seventeenth century Odawa Hunter's Point site, a temporary camp located next to a deep bay in Georgian Bay, uses statistics to define refuse deposits. Fishing strategies are described on the basis of the kinds and numbers of elements; the number of species; fishing equipment and location of catch. A restricted range of fishing activities results from the seasonal occupation and the limited range of environments. Fishing strategies at Barrie, Dunsmore and Carson, which were likely more variable in terms of location and time of capture, cannot be explained by this model. An overview of fishing activities as they relate to the fish taxa identified at the sites is presented below. Lake sturgeon The biology, behaviour and capture of lake sturgeon were discussed in an earlier paper (Needs-Howarth 1996). Lake sturgeon are bottom-feeders that usually live in large lakes and large rivers (Table 7). It is likely that lake sturgeon were abundant in Lake Simcoe and connecting waters during the time of occupation of Barrie, Dunsmore and Carson. Because of overfishing there was no significant lake sturgeon population left in Lake Simcoe by the beginning of this century (MacCrimmon and Skobe 1970:30). Lake Couchiching and the Holland and Severn rivers also had a commercial lake sturgeon fishery (Harkness and Dymond 1961:99). Fisheries biologists (David Loftus, William Beamish, personal communication 1995) confirm there may have been a resident population in the Nottawasaga River in the past. There certainly is now, and probably was in the past, a resident population in Georgian Bay of Lake Huron, which spawns in the Nottawasaga River (David Loftus, personal communication 1995). Lake-resident populations of lake sturgeon were not really available to the site occupants for most of the year, because they were dispersed and inhabited deeper water, away from the lakeshore (Scott and Crossman 1973:86). During the spawning season, however, lake sturgeon move into shallow waters or rivers. River-resident populations also become more concentrated and more accessible at spawning. Lake sturgeon spawn in rivers at depths of about .5 m to 5 m, in areas of swift water or rapids, or at the foot of low falls that prevent further migration (Harkness and Dymond 1961:17, 38). In the lower Great Lakes, there were also populations of lake-spawning sturgeon, which spawned in shallow water over rocky ledges close to shore or around rocky islands (Harkness and Dymond 1961:40). The spawning date appears to be associated with water temperature, which has to have reached between 13.9 and 15.50 C (Harkness and Dymond 1961:36, 38). For the lower great lakes, this means sometime in May. Because of differential heating of the water mass, spawning temperatures will be reached earlier in
rivers than in lakes (Harkness and Dymond 1961:40). Although lake sturgeon can travel great distances, they have a strong homing instinct (Lloyd Mohr, William Beamish, personal communication 1995 and Harkness 1923:19; Scott and Crossman 1973:87). Return to the same spawning locations is confirmed by recent tagging experiments (Lloyd Mohr, personal communication 1999). The mean age at first spawning for females in a St. Lawrence River population was 19 years (Guenette et al. 1992).Year classes of males spawn every one, two or three years, females every four to six years (Dumont et al. 1987; Magnin 1962; Roussow 1957). A group of spawning sturgeon may, therefore, be composed of individuals of different states of maturity, and different ages and sizes (Roussow 1957). In any given year, some year classes would not spawn (Lloyd Mohr, personal communication 1995). It is thought that the periodicity of spawning may vary within a year class; CSAGES studies are currently being implemented to confirm this hypothesis (Lloyd Mohr, personal communication 1999). It should be noted that both immature sturgeon (Cuerrier 1951; Dumont et al. 1987; Harkness and Dymond 1961) and non-spawning mature sturgeon (McKinley et al. 1993) go up rivers in spring at the time mature fish of certain age classes are on their spawning migration. It is likely that members of the same population spawned in more or less the same locations, minimizing the year to year variation in which spawning sites were actually being used (William Beamish, personal communication 1995). In his overview of precontact fishing in the upper Great Lakes, Cleland argues that lake sturgeon were probably only purposefully procured during their spring spawning-run up larger streams and rivers (Cleland 1982:766). This is substantiated by earlier investigations by the author (NeedsHowarth 1996). Several fisheries biologists have suggested that lake sturgeon can be caught individually from the shore, especially if they are in shallow water. Spearing may also have been practised from canoes on either Lake Simcoe or Nottawasaga Bay. The French explorer Charlevoix (1923:236) described the following technique for catching sturgeon, relating to one of the Great Lakes (name not specified), during his visit to New France in 1720: “Two men place themselves in the two extremities of a canoe; the [one] next [to] the stern steers, the other standing up holding a dart to which is tied a long cord, the other extremity whereof is fastened to one of the cross timbers of the canoe. The moment he sees the sturgeon within reach of him, he lances his dart at him and endeavours, as much as possible, to hit in the place that is without scales. If the fish happens to be wounded, he flies and draws the canoe after him with extreme velocity; but after he has swam the distance of an hundred and fifty paces or thereabouts, he dies, and then, they draw up the line and take him.”
36
Lake sturgeon can also be captured by means of set lines (Harkness and Dymond 1961:61; Rostlund 1952:11), as implied in the description of sturgeon fishing by Sagard (1939:189), who mentions that hooks were often found in the stomachs of fish. In the same section he mentions that in Georgian Bay there were fish “of such monstrous size that nowhere are they to be found bigger.” This appears to indicate that the lines the Huron used to catch sturgeon were not always strong enough to hold the larger individuals. For this reason, Cleland (1982:778) argues that spears and harpoons probably continued to be the dominant mode of capture.
result of a spawning-run catch. The closest spawning locations of either a river-resident or lake-resident population are in the Nottawasaga River. There are no significant barriers that form obvious spawning locations on the main part of the Nottawasaga River today (Buritt, personal communication 1995). There are, however, ripples close to the mouth of the Nottawasaga River at Montgomery Rapids that may have provided a suitable spawning ground, especially at lower water levels (Robin Craig, personal communication 1995, Fred Dobbs, personal communication 1999). There is also a set of riffles or small rapids north of Baxter, some distance south of the branching of the Pine River (Fred Dobbs, personal communication 1999), and between Baxter and Alliston, around the branching of the Boyne River (Fred Dobbs, Robin Craig, personal communication 1995).
Sturgeon are apparently easily entangled in gill-nets because their pectoral spines get stuck, even in wide-meshed nets (Von Brandt 1984:170). Father Louis Hennepin (1903:522) mentions that the Iroquois in New York fished for lake whitefish and lake sturgeon with large gill-nets that required two men at either end to draw them into shore. Taking into consideration Sagard's comments about fish breaking the fishing line, it is likely that larger lake sturgeon would sometimes break these nets made from hemp or nettle fibre. Nevertheless, netting probably was a effective manner of catching lake-dwelling populations of lake sturgeon.
If there was a river-resident population, its spawning location may have been known to the occupants of the Barrie, Dunsmore and Carson sites. The mouth of the Nottawasaga would have been the most convenient and predictable place for humans to intercept spawning-run sturgeon of the Georgian Bay population, even if they didn't actually spawn at Montgomery Rapids. Indeed, early archaeological investigations (Hunter 1906:30) noted an “Indian camp” at this location.
The weir at Atherley Narrows was suggested as a possible catch location by several fisheries biologists (William Beamish, Lloyd Mohr, personal communication 1995), but it is not known whether any sturgeon remains were found at the associated multi-component Dougall site (Burns, personal communication 1995). It is likely that the resident Lake Simcoe and Lake Couchiching populations moved between their respective lakes, especially if they were part of the same genetic population (William Beamish, personal communication 1995). It appears likely, therefore, that sturgeon could be caught in the weir, but not necessarily in great numbers at a predictable time of year, unless they happened to pass through the weir on their spawning migration.
Salmonidae Lake trout, lake herring and lake whitefish spend most of their time in the pelagic zones of lakes, where they often concentrate in large numbers or in known zones (McCrimmon 1958). In spring and early summer they move below the thermocline, where they remain until the upper waters cool. In autumn they home to near-shore locations (McCrimmon 1958). In the middle of this century, 50 percent of the shoreline of Lake Simcoe was considered good spawning habitat for lake trout (McCrimmon 1958). Georgian Bay also supports large numbers of Salmonidae. As was noted above, very few lake trout, lake whitefish or lake herring cranial bones were identified, but lake trout vertebrae were found in many features, indicating at least some emphasis on these lake-dwelling taxa. Contemporary sport fishers catch lake trout and yellow perch from the ice in January, February and March by angling or spearing (MacCrimmon and Skobe 1970:32, 50). Lake whitefish and lake herring can also be found on shoals in spring after breaking up of the ice. Lake trout occur in surface waters immediately after ice melt (Scott and Crossman 1973:225). During summer, these fish are generally too dispersed and occupy waters too deep to be caught efficiently and in significant numbers without nets. In early October, lake trout approach the shore, and in mid-October they spawn over shoals close to shore (MacCrimmon and Skobe 1970:54) (Table 7). Somewhat later, in mid-November and late November, respectively, lake whitefish and lake herring also spawn over shoals close to shore (MacCrimmon and Skobe 1970:38; Scott and Crossman 1973:239, 269). Spawning concentrates large numbers of both species in
In addition to the arguments noted above against exploitation of lake sturgeon in (deep) lake waters, there are strong arguments in favour of riverine exploitation. Even outside the spawning season a river provides more opportunities for netting and somehow confining the fish. During the spawning-run the sturgeon are already concentrated in a group, and it would be even more efficient and convenient to intercept them with fish-traps, nets or spears. spawning-run sturgeon were caught in fish-traps by eighteenth century native people on a river in the boreal forest region of northern Ontario (Michalenko et al. 1991). More recently they used spears or 92 m long and 20.5 or 23.0 cm aperture set nets, which they checked once or twice a day (Michalenko et al. 1991:450-451). If there was indeed a river-resident population of lake sturgeon in the Nottawasaga, it is possible that lake sturgeon were caught outside the spawning season. Large numbers of sturgeon remains, however, are more likely to have been the 37
shallower waters that would be accessible using canoes and gill-nets or spears (MacCrimmon and Skobe 1970:50, 54; Scott and Crossman 1973:239).
Neither the channel catfish (Ictalurus punctatus) nor the bullheads are noted for spawning migrations (Scott and Crossman 1973). It is likely that brown bullhead was caught throughout the warm seasons in river, stream and wetland habitats. Given bullhead nocturnal feeding habits and generally solitary nature (except when schooling with juveniles), it is likely that this species was primarily caught using passive technology like fish-traps, trap-nets and gillnets (McMurtry 1991). Cossette (1995:549-550) points out that brown bullhead could also have been caught on a baited line without hooks. The co-occurrence of brown bullhead with yellow perch in many major features supports the hypothesis advanced by Needs-Howarth and Thomas (1998) that perch exploitation was not necessarily limited to the spawning season.
Needs-Howarth and Thomas have argued (1998) that it is unlikely that many resources would be allocated to a lacustrine fishery during the highly productive spring spawn season of other taxa; extensive exploitation of these Salmonidae species was probably mostly limited to their autumn spawning season. This is in accord with the ethnohistoric sources (Sagard 1939:186). An autumn Lake Fishery may have offered some overall advantages over the Spring Spawning-run Fishery in terms of timing. As Cleland argues, “while the spring fishery may have come at a good time to relieve late winter food deficits, it could not forstall them”(Cleland 1982:779). In an Iroquoian context, the autumn Lake Fishery occurs after the corn harvest, thus avoiding conflicts with labour requirements for horticulture.
Centrarchidae do not tend to engage in extensive spawningruns (Scott and Crossman 1973, 589-603, 703-6, 716-7). It is likely that these fish were exploited throughout the warm weather. Exploitation may have coincides with their summer spawning time, but may not have been aimed specifically at spawning-run exploitation.
Other taxa The longnose sucker prefers deeper, cooler waters and would be most easily accessible during the spawning-run (Table 7). White suckers do occur in Lake Simcoe, but also extend into shallower lake and downstream habitats (MacCrimmon and Skobe 1970:66). Older white suckers occupy deeper waters further from shore (MacCrimmon and Skobe 1970:542; Scott and Crossman 1973:542). Proportional element size data indicate that the majority of longnose and white suckers from the archaeological collections are mature. White suckers co-occur in features with fish that were probably part of the Spring Spawning-run Fishery, but also with fish that were more likely part of the Generalized Warm Weather Fishery. The literature confirms that “some late spawning suckers may have overlapped early with spawning lake sturgeon” (Harkness and Dymond 1961:39). White sucker may, therefore, have been obtained together from the Nottawasaga River with lake sturgeon and longnose sucker during a period of overlap in their spawning-runs in early May, or together with spawning yellow perch from Lake Simcoe, or together with brown bullhead and/or smaller sunfish at any time during the warm weather (Needs-Howarth and Thomas 1998). Given their size and their frequent co-occurrence with lake sturgeon, it is suggested here that most suckers are part of the Spring Spawning-run Fishery. Suckers on their spawning-run are easily speared (Cleland 1982:774).
Yellow perch are a cool water species that swim in sizegraded schools. Schools of younger fish swim closer to shore than schools of larger, more mature fish (Scott and Crossman 1973:759). This schooling habit makes smaller yellow perch susceptible to capture in nets and inshore fishtraps. Yellow perch may not have occurred in the Nottawasaga River in the past (Fred Dobbs, personal communication 1995). Yellow perch engage in extensive spawning-runs into the rivers and small streams around the lake in late April and May (MacCrimmon and Skobe 1970:101) (Table 7), where they can be caught using nets or weirs. Unless some fish were released alive, refuse from spawning-run seine-net catches is expected to lack bones from small, sexually immature individuals. Yellow perch could have been caught together with other species during the spring spawning-run, but since their availability is not as tied to the spawning season as that of some other taxa, like the lake sturgeon, Salmonidae or suckers, they were probably also an important component of the Generalized Warm Weather Fishery (Needs-Howarth and Thomas 1998). Walleye generally live in cool, large, turbid lakes, large streams or rivers at moderate depth (Table 7). The walleye is a sight predator and lives in schools, sometimes with white suckers (Scott and Crossman 1973:772). It is also often associated with yellow perch, northern pike, muskellunge and smallmouth bass (Scott and Crossman 1973:772).
Northern pike co-occur with brown bullhead and yellow perch in many of the major features examined. Northern pike spawn after the ice melts (Scott and Crossman 1973:357) and finish spawning in the Lake Simcoe drainage by early May (MacCrimmon and Skobe 1970:92). The later part of the spawning-run may just overlap with the start of the spawning-run of yellow perch and suckers (Table 7). However, northern pike size distribution and consistent cooccurrence with brown bullhead (see Chapter 6) suggests that northern pike were exploited throughout the warm seasons as part of the Generalized Warm Weather Fishery, most probably in fish-traps.
There are several likely locations of capture of walleye. If population densities were greater in the past, walleye may have been found in what are now considered marginal habitats. Today, walleye are rare in Lake Simcoe. The resident population used to spawn in the Holland River, at the south end of the lake (MacCrimmon and Skobe 1970), but sometime after the mid-1930s they started to spawn in the Talbot River, on the east side of the lake, possibly as a 38
result of deteriorating water quality in the Holland River (Michael McMurtry, personal communication 1998). These locations are quite far from the sites, even by canoe. The Nottawasaga system appears a more probable location for exploitation than either the Holland or Talbot rivers. The Nottawasaga River is a nursery for young fish of the resident walleye population in Nottawasaga Bay (Robin Craig, personal communication 1998). The adults live in the bay, and only come to the river to spawn in Minesing Swamp. The main part of the river is unsuitable spawning habitat (Robin Craig, personal communication 1998). As noted earlier, an influx of groundwater in Minesing Swamp, however, causes the water there to be very clear. In spring, the groundwater keeps the muddy river water in the main channel (Robin Craig, personal communication 1998). The walleye spawn in this shallow, clear, moving ground water, where they are very visible. Walleye often move into tributary rivers as soon as the rivers are ice-free and while the lake is still ice-covered, several weeks before spawning (Scott and Crossman 1973:771). Because the distance from Nottawasaga Bay to the spawning location is quite far, some of the mature fish may move into river in October and November so that they are ready to spawn in the swamp as soon as the ice breaks up (Robin Craig, personal communication 1998). When conditions are right, the spawning group rushes into shallow water and stops while they release their spawn and eggs (Scott and Crossman 1973:771). Contemporary walleye spawning-runs are often very large, with fish crowding a shallow part of a river, where they can be easily speared or netted. After spawning, some fish linger in the rivers, but most return to open water and disperse. In autumn walleye once again return to the shoals of the lake shore (MacCrimmon and Skobe 1970:107-111). It has been suggested that some walleye at the Hubbert site were captured together with lake trout in autumn (Thomas 1996b:116). Depending on flow conditions and other obstructions, such as beaver dams, it is possible for spawning walleye of the Nottawasaga Bay resident population to reach Willow Creek, where they can also be easily taken (Robin Craig, personal communication 1998). There is also a resident population in Little Lake, which spawns in riffles upstream from the lake (Robin Craig, personal communication 1998). The fish here are easy to catch, more concentrated than in the swamp, though less abundant. Little Lake is, as the name indicates, quite small, and is unlikely to have been the source for the oldest of our fish. These are most likely from the Nottawasaga Bay population, possibly by way of Willow Creek (Robin Craig, personal communication 1998).
39
CHAPTER 6: SAMPLE DESCRIPTION AND ANALYSIS
Barrie Faunal assemblage As noted earlier, the faunal sample derives from the undisturbed areas of 29 features within two houses, four external refuse pits and three middens (Figure 4, Table 3). Some details on the unidentified component of the sample have already been provided in the section dealing with taphonomy (Table 11). No Euro-Canadian domesticates were identified, except for a pig mandible in one feature that has been excluded from analysis; the sample therefore appears to be free of intrusive skeletons. While it is possible that some bones of the major fossorial taxon identified, the woodchuck (Marmota monax), represent in situ deaths in a burrow, heat alteration on some of the bones, and their uniform state of preservation indicates this material probably represents food refuse. The current section provides more detail on the portion of the fish assemblage identified below the taxonomic level of class.
further, immature, individual is present. As noted above, lake sturgeon spawning migrations can include immature individuals. It is quite possible that these 85 elements derive from a single catch of a minimum of four individuals. However, the extensive spatial distribution of the finds suggests that fishing for lake sturgeon may have been more than an incidental activity. A successful biochronological application of the pectoral spines might have provided answers in the form of calendar year of capture.
At the Barrie site, the 380 fish elements identified below class constitute 53 percent of the combined NISP for all classes. In order to allow comparison with other southern Ontario assemblages, NISP is calculated as a percentage of the combined fish, amphibian, reptile, bird and mammal sample. The sixth commonly recovered taxonomic class, Pelecypoda, or freshwater clams, has been excluded here because of unevenness in recovery and/or preservation between the three sites, and because a disproportionately large number appear to have been curated as tools or ornaments. Clams contribute three percent to the NISP of all six classes at Barrie, and even less at the other sites. The fish assemblage is quite rich, but not very even; it is dominated by lake sturgeon and yellow perch, which together make up over half of the fish NISP.
As noted above, during microscopic examination of the sections it became obvious that the spines suffered from extensive demineralisation, delamination and probable ironstaining. Similar effects were observed also on the Hector Trudel material (Cossette 1995:544). As a result, our interpretations were limited to minimum age estimates on three of the sections. The sections exhibited undamaged segments of at least ten years of interpretable growth patterns. The age of these three individuals was in excess of 20 years. They can, therefore, indeed be assumed to have been sexually mature.
The two spine fragments that appeared, macroscopically, to be best preserved were selected for sectioning. As the fragments were immersed in cold cure epoxy and hardener (Industrial Formulators, Canada), air bubbles escaped as air in the bone was being replaced by epoxy. The amount of air escaping was greater than in modern spines; this may relate to the fossilisation process. Thin-sections were made with a diamond blade saw operating in a tap-water blade bath.
Salmonidae Lake trout, lake herring and lake whitefish represent two percent of the fish NISP. At Barrie, mostly lake trout cranial bones were identified, although lake herring or lake whitefish is represented by one cranial bone and limited numbers of vertebrae. Lake whitefish can sometimes be identified specifically based on absolute size, since they grow larger than the other common member of the genus, lake herring. However, since these species have been known to hybridize and since dwarf populations of the former sympatrically co-exist with regular-sized fish (Scott and Crossman 1973:270), species identification based on archaeological material is problematic.
Acipenseridae Lake sturgeon elements represent almost a quarter of the fish NISP at the Barrie site. Such an abundance of sturgeon elements is unusual for village sites in the region. As noted above, ossified dermal scutes were included in the NISP to compensate for the paucity of ossified cranial bones in this species. While elements of the reference specimen at University of Toronto were not labelled, some cranial bones were tentatively identified to specific element using illustrations (Bartosiewicz and Takács 1997; Desse-Berset 1994). It is hard to know how much food these 85 elements represent. The five pectoral spines represent a minimum of three individuals. Their size indicated that they belonged to mature individuals.
Two contexts contained only cranial bones. Five contexts, including Midden A and Midden B, contained only vertebrae. Two additional contexts, H1 F133 and Midden D, contained both Salmonidae cranial bones and vertebrae. Lake herring or lake whitefish (Coregonus sp.) contributed two percent to the identified vertebrae, while lake trout contributed 15 percent. This supports the statements made in Chapter 4 regarding greater ubiquity of vertebrae vs. cranial bone, although the larger, more robust lake trout vertebrae dominate.
A dermal scute with a pronounced, sharp ridge, suggests at least one of the individuals was considerably younger than the individuals represented by the pectoral spines. Between 20 and 25 years of age, the sharp points on the dermal scutes disappear, associated with attainment of sexual maturity. This pointy scute, therefore, suggests at least one 40
Esocidae Pike family fishes represented 12 percent of identified fish (NISP). The osteological similarities between the three local members of the genus and the lack of reference specimens covering the entire spectrum of age and sex are reflected in some less certain identifications. Vertebrae represent 21 percent of identified fish. The vertebrae appear to be mostly from larger individuals, whereas the cranial bones are mostly from smaller individuals.
identified fish. Only the more common taxa are represented. This is the only taxon for which proportions of vertebrae identifications more or less match cranial bone identifications. Percidae The current representatives of the family Percidae in the Lake Simcoe drainage are yellow perch and walleye, which comprises two sub-species, walleye (Stizostedion v. vitreum) and blue walleye (Stizostedion v. glaucum). The status of sauger (Stizostedion canadense) in precontact Lake Simcoe is uncertain.
Catostomidae White and longnose sucker combined represent 11 percent of the identified fish, with white sucker four times as plentiful as longnose sucker. Because of the behavioural and habitat differences between the two species (Table 7), a special effort was made to identify each bone to species. Problems in differentiating the two species on many elements are reflected in the high proportion of genus level identifications. Vertebrae represent 39 percent of identified vertebrae. The proportional element sizes indicate mature individuals.
Yellow perch is well represented at 32 percent of the NISP. Only one element was identified as Stizostedion sp. Percidae vertebrae represent only four percent of identified vertebrae. Since Perciformes vertebrae are quite similar, it is possible that some yellow perch were incorrectly identified as Centrarchidae, but this is not likely to account for all of the discrepancy. The single Percidae fish scale recovered from the site was subjected to CSAGES analysis. A first requirement was to identify the species. As expected based on the cranial bone taxonomic distribution and robusticity of the different scale types, initial macroscopic examination of the scales from Barrie, Dunsmore and Carson indicated they belonged to the order Perciformes, which includes Centrarchidae and Percidae. Further examination showed that those selected for analysis were all Percidae. Based on the cranial bone NISP at all three sites (Table 24), it was assumed that the Percidae scales recovered would be from yellow perch. Less likely options would be the genus Stizostedion, either walleye or blue walleye, or, even less likely, based on current distribution, the Sauger, which was not specifically identified at any of the sites.
Ictaluridae Catfish family fishes represent eight percent of the fish NISP. Only two of the 25 identifications pertain to pectoral spines. A single element was identified as channel catfish; two further elements were indeterminate bullhead or catfish. Vertebrae represent only two percent. This is surprising given the distinctiveness, robusticity and solid projections of Ictaluridae vertebrae. Two brown bullhead pectoral spines from Midden A at the Barrie site were subjected to age and growth analysis (MNI 2) (Table 25) (Needs-Howarth and Brown 1998). Poor preservation, in conjunction with damage on the edge, made it impossible to obtain an age for one spine from the Barrie site, although the edge condition could be somewhat confidently interpreted as autumn. The other spine was somewhat confidently interpreted as being from a fish aged four, caught in spring.
With the exception of the single scale from the Barrie site, the growth pattern around the focus was very uniform, possibly relating to juvenile feeding movements. The overall growth rate is similar. This suggests that the scales from Dunsmore and Carson belong to just one species. The scales of the three members of the family Percidae are similar in appearance, however, they can be differentiated. The focus of yellow perch and walleye scales is filled with ridges, whereas that of Sauger is largely empty, with only a few scattered ridges (Daniels 1996:80). The scales under examination were obviously yellow perch or walleye.
Data from Northern Wisconsin (Scott and Crossman 1973:601-602) suggest brown bullhead reach maturity at three years. In the brown bullhead population in the Rivière aux Pins near Montreal back-calculated lengths at time of annulus formation for fish age three years averaged 201 mm for males and 194 mm for females. For fish aged four years, average sizes were 254 mm for males and 240 mm for females (Harvey and Fortin 1982). Maximum age in this population was 10+ years, with TL 350 mm. In a study from a eutrophic lake in New York State back-calculated lengths at time of annulus formation were 220 mm for males age 3 years, 215 mm for females age three (Sinnott and Ringler 1987: Table 3). In this population the maximum age was 7 years, with TL 317 mm.
Perch has a more defined, and hence uniform, growing season because it is close to the northern limit of its range. walleye has a more northerly distribution and hence is more tolerant of cold water; as a result, walleye has a less uniform growing season, which is reflected in more sub-annular checks. Walleye also grow much faster initially, and they live much longer (Figure 14). Based on these differences in growth pattern, David Brown is confident that the partial scale from Midden D at the Barrie site is yellow perch, and that all analysed scales from Dunsmore and Carson are
Centrarchidae The Centrarchidae, consisting mostly of small panfish, is moderately well represented, totalling 14 percent of 41
walleye.
refuse was deposited in autumn, when deer are most likely to have been hunted. It is, therefore, likely that this feature contains at least two discrete episodes of deposition.
The yellow perch scale from the Barrie sites belonged to an individual caught in spring. It was probably five years old at time of death (Table 26) (Needs-Howarth and Brown 1998).
Barrie Feature 136, an ash pit in House 1 While the fish NISP, consisting of only two taxa, would suggest the conclusion that the fish assemblage of this feature is not very diverse, the vertebrae represent an additional two taxa. The 11 yellow perch bones show a restricted size distribution (Figure 15), including mostly individuals of spawning size. The only other zooarchaeological finds from this feature are a single clam shell fragment and some unidentified bird and mammal fragments. This admixture of finds conforms with the feature type interpretation.
Fish by feature Introduction As was discussed in Chapter 5, there is some overlap in species composition within the Three Fisheries Model, particularly between the Spring Spawning-run Fishery and the Generalized Warm Weather Fishery. By utilising information on fish co-occurrence in features, together with fish proportional element size distribution, an attempt can be made to establish when the fish found in the archaeological deposits were caught, and hence what kind of fishing activity they represent. The following is a description and interpretation of the fish finds in those features rich in fish bone (here arbitrarily defined as those containing more than 10 fish bones identified below family (Table 27)). These features are described here to illustrate the intra-site variability in fish deposits. Proportional element sizes for the Barrie site are provided in Figure 15.
Barrie Feature 156, a pit in House 1 This feature contains a mixture of fish, with no dominant taxon. It may include secondary refuse from several fishing events. Seven painted turtle (Chrysemys picta) elements indicate warm weather activity. This feature has an unusually high ratio of identified remains. As was the case for Feature 136, Feature 156 contained no identifiable bird or mammal remains.
Barrie Feature 133, a refuse-filled depression in House 1 This context includes heavy fraction. A clustering of larger yellow perch bones (Figure 15) and the presence of lake sturgeon, white sucker and cf. Catostomus sp. (including many Catostomidae vertebrae) may indicate net fishing during the Spring Spawning-run Fishery. Four lake trout cranial bones (representing at least two individuals) and 17 vertebrae, as well as four lake herring or lake whitefish vertebrae, may indicate a autumn Lake Fishery, which may have included the suckers as well (Needs-Howarth and Thomas 1998). Following the Three Fisheries Model outlined in Chapter 5, the northern pike and brown bullhead cranial bones and the Centrarchidae vertebrae may have been part of the Generalized Warm Weather Fishery around the site or in Kempenfelt Bay (Needs-Howarth and Thomas 1998), although the status of northern pike in pre-twentieth century Lake Simcoe is uncertain (MacCrimmon and Skobe 1970:85, 90, 91, 92).
Barrie Feature 163, a pit in House 1 The presence of lake sturgeon indicates a Spring Spawningrun Fishery in the Nottawasaga River. Some of the yellow perch, totalling 90 percent of the fish NISP in the feature, may derive from a spawning-run catch in Lake Simcoe tributaries (Figure 15) (Needs-Howarth and Thomas 1998). The presence of three common merganser bones may relate to fish exploitation; it is possible that these diving, fisheating were unintentionally caught when they became entangled in nets set over shoals (e.g. Studer 1992). Attributes of immaturity on the common merganser bones indicate autumn exploitation of the lake shoals (NeedsHowarth and Thomas 1998). Nine Esocidae and three Centrarchidae vertebrae may have been obtained as part of the Generalized Warm Weather Fishery. Barrie Feature 216, a pit in House 2 Almost three quarters of the fish bones in this feature are from lake sturgeon and suckers, indicating it mainly represents a Spring Spawning-run Fishery (Needs-Howarth and Thomas 1998). Suckers are also well represented in the vertebrae. A single large lake trout vertebra from this feature may represent a stored resource or an incidental early spring capture from the lake shoals (Needs-Howarth and Thomas 1998). This feature is notable for the large numbers of black bear (Ursus americanus) remains, consisting of thoracic and lumbar vertebrae, as well as phalanges, probably all from the same immature individual. While finds of (calcined) bear distal extremity bones at other sites have been interpreted as the possible remains of bearskin cloaks (Needs-Howarth 1992; Thomas et al. 1998), the inclusion in Feature 216 of vertebrae as well may indicate food or sacrificial refuse. There is mention in the ethnohistoric sources of young bears being kept in captivity
Given the overall fish contents of the feature, however, it is possible these taxa also derived from the lake or river, as a by-product of a Spring Spawning-run Fishery for yellow perch and suckers in or a Lake Fishery for Salmonidae in spring or autumn. The Salmonidae and lake sturgeon may have been caught in the lake in spring as well as part of the spring component of the Lake Fishery. Alternatively, the autumn component of the Lake Fishery may have included the suckers. Either hypothesis can be supported by a humerus from a diving duck (probably bufflehead (Bucephala albeola)), a taxon that migrates through the area in April and early May, and again in late October and November. Lake sturgeon, however, were most likely obtained during the Spring Spawning-run Fishery, whereas the five whitetailed deer bones (MNI 2) found in this context would indicate that at least some of the faunal 42
by the Huron and subsequently being sacrificed during feasts (Trigger 1976:41).
floated) is 92 percent, with the size distribution suggesting that the majority were large enough to have been sexually mature when caught (Figure 10). Although the large numbers of suckers and yellow perch in the Barrie fish bone assemblage do not necessarily represent exclusive exploitation during the spawning-run, the contents of certain features do appear to be the result of mass-capture events. This is especially noticeable in F133, with its fairly tight clustering of perch bone sizes.
Fishing events Individual features sometimes contain taxa indicative of different seasons of exploitation. This admixture can be explained by often prolonged, and intermittent use, causing several procurement events to be represented in a single archaeological context. Fish may also have been dried and stored for later use, adding to the confusion. An attempt will be made, nevertheless, to generally characterize the Barrie fish assemblage in terms of fishing events.
As was noted in Chapter 4, the ratio of cranial bone to vertebrae may indicate that more off-site processing was practised at Barrie than at Dunsmore and Carson. Off-site processing may suggest exploitation of waters away from the site, such as the Nottawasaga River, Lake Simcoe, or Nottawasaga Bay, which may have made catch-site decapitation more desirable. Fishing at Barrie appears to make use of rivers and the lake shore, rather than the local stream, Dyment’s Creek. This is also reflected in the absence in the zooarchaeological collections of positive identifications of brook trout (Salvelinus fontinalis), which was apparently common in the cold streams in the city of Barrie earlier this century (Robin Craig, personal communication 1995). Sutton (1996a) suggests that the site occupants had prior knowledge of the area. The fact that the occupants knew of, and were able to successfully exploit, spawning longnose sucker and lake sturgeon lends support to this idea. Perhaps the Barrie people were familiar with lake sturgeon in their previous village, probably located south of the Oak Ridges Moraine, since there were lake sturgeon in Lake Ontario in the precontact period (Ontario Ministry of the Environment 1998).
Lake sturgeon appear in all but one of the major features discussed above. As noted under “diagnostic elements”, assessing abundance based on NISP is somewhat problematic. The 85 sturgeon bones identified from the Barrie site may only represent four individuals. As noted above, the fact that local sturgeon spawn in May does not mean they were exclusively caught during that time. These finds may derive from gill-nets used on Lake Simcoe, or from set nets used on the Nottawasaga River outside spawning season. The zooarchaeological, archaeological, zoological and ethnohistoric data, together with intra-site bone distributions, however, indicate that it is likely that the site occupants were mostly exploiting spawning populations (Needs-Howarth 1996). While lake sturgeon does not consistently co-occur with suckers, all features with suckers do have lake sturgeon bones as well. Knowledge of lake sturgeon behaviour, seasonal distribution and local geography, together with the substantial number of finds, suggests that this taxon was targeted by the occupants of the Barrie during its spawning-run (Needs-Howarth 1996). The pectoral spines belonged to sexually mature individuals. This may support the hypothesis of spawningrun exploitation, although immature individuals do travel along with mature individuals on their spawning-run. The most productive and predictable place for a sturgeon fishing expedition would probably be at or close to the mouth of the Nottawasaga during the spring spawning-run. The site occupants may not have ventured out just for the lake sturgeon, but if they knew, from prior experience, approximately when the spawning-run would start, they could have combined lake sturgeon fishing with other activities.
Dunsmore Faunal assemblage As noted earlier, the faunal sample from the Dunsmore site derives from the undisturbed areas of three middens (24 m2), and 81 features within 10 houses (Figure 5, Table 4). Information on the “unidentified” component is provided in Table 12. The current section provides more detail on the fish assemblage. The 665 fish bones identified constitute 75 percent of the assemblage identified below family, expressed as NISP (Table 28). In terms of BW, however, fish represent only 40 percent. The assemblage is more diverse than at the Barrie site. It should be noted that discussions of the assemblage as a whole may obscure some of the variability between the three segments of the site.
While there are too few sized northern pike bones in individual features to make any clear statements on fish size distribution, the average size of northern pike bones suggests the kind of limited size distribution from a small fish-trap. The average proportional element size suggests that many of the northern pike identified at Barrie were of sexually immature individuals. These were most likely caught with passive technology, such as fish-traps, or nets. In contrast to northern pike, the distribution of brown bullhead sizes is much wider, with a distinct peak.
In terms of NISP, yellow perch is the most important taxon, followed by brown bullhead, pumpkinseed, northern pike and white sucker. Each of these fish exceed the NISP of the next most abundant taxon, the muskrat (Ondatra zibethicus). Acipenseridae Lake sturgeon is represented by only four bones, or one percent of fish NISP; this is comparable to recovery at other sites in the region, with the exception of the Barrie site.
The average size of all yellow perch elements (screened and 43
Lepisosteidae Longnose gar is only represented by two of its highly distinctive vertebrae from Midden B, which are not included in the NISP.
fish over hot coals, or, in the case of burnt spines, the selective burning of potentially hazardous debris. Some spines, however, were curated as expedient awls (Thomas 1996d:146).
Salmonidae At Dunsmore, in addition to lake trout, both lake herring and lake whitefish were positively identified. Salmonidae represent three percent of identified fish. Lake herring or lake whitefish vertebrae are represented in similar quantities to their cranial bones, whereas lake trout vertebrae represent 28 percent. This may be partly explained by the large size and robusticity of lake trout vertebrae compared to lake herring or lake whitefish vertebrae and all Salmonidae cranial bones. Three contexts contained only cranial bones; seven contexts, including Midden B, contained only vertebrae. Only two contexts, H8 F230 and H11 F427, contained both cranial bone and vertebrae, again suggesting Salmonidae may have been subject to unique taphonomic processes.
Because Ictaluridae pectoral and dorsal spines are readily identified to family or genus, and because there are proportionately many of them at Dunsmore, the NISP of this family is over-represented with respect to other fish in the NISP (see Table 10, Figure 8).
Esocidae Pike family fishes represent nine percent of identified fish. The positive identifications pertain to grass pickerel and northern pike only. Unlike at Barrie, vertebrae, representing 12 percent of those vertebrae identified to family, are not grossly over-represented.
Gadidae Four elements were identified as burbot. It is generally a deep water species. Burbot spawn in mid-winter under the ice in lakes (Scott and Crossman 1973:643). Likely times of capture would be during the post-spawning movement into tributary rivers, or when it moves into shallower water on summer nights. Burbot may also have been caught during the autumn spawning-run when preying on Salmonidae roe (Molnar 1997: 171; Scott and Crossman 1973:226, 274; Smith 1985:101).
Spines from Dunsmore did not appear decalcified and the edges were intact. A total of six brown bullhead pectoral spines were subjected to age and growth analysis (Table 25), five of which are from a single square in Midden B. Each spine represents a different individual. Ages range from three to six years, with an average of 3.8 years. The majority of the spines, therefore, derive from mature individuals. Season of capture ranged throughout the warm weather, but with concentration on late autumn and spring.
Cyprinidae A single pharyngeal arch was identified as a probable creek chub or fallfish (Semotilus sp.). Lack of reference specimens precluded a more detailed identification. This identification does indicate that the Dunsmore people occasionally obtained smaller carp family fishes. Alternatively, this element may represent stomach contents of a piscivorous fish.
Centrarchidae As at Barrie, Centrarchidae are well represented (20 percent of NISP), with vertebrae proportionate to cranial bones. Following the same logic as with Ameiurus, the Lepomis sp. identifications probably mostly represent pumpkinseed, rather than bluegill. Note that there are few Micropterus sp. identifications. In contrast to many other multi-species genera, the two representatives in this genus, the smallmouth and largemouth bass (Micropterus salmoides), are perhaps more readily distinguishable than the taxonomic lists in other works imply (Cooper 1996:22; Molnar 1997:145). Three elements were identified to the less commonly identified genus Pomoxis.
Catostomidae Suckers represent 12 percent of cranial bone. As at Barrie, there are many more white suckers than longnose suckers. Catostomidae vertebrae are very numerous, representing 39 percent of identified vertebrae. The proportional element sizes indicate mature individuals. Ictaluridae Catfish family fish contribute over one quarter of the NISP; most of the identifications at Dunsmore probably represent brown bullhead. In spite of concerted efforts to identify other species in the genus, only one element from a black bullhead (Ameiurus melas) was noted. Large Ameiurus sp. identifications could include yellow bullhead (Ameiurus natalis), but given the dominance of brown bullhead in the species identifications, this is unlikely. Five elements were identified as the larger, river-dwelling, member of this family, channel catfish (Ictalurus punctatus). As at Barrie, vertebrae are under-represented at only one percent.
Percidae With 188 elements, representing almost a third of the fish NISP, the Dunsmore yellow perch are the most numerous taxon at any of the three sites. Surprisingly, not a single Percidae vertebrae was identified. The recovery of many scales from several features afforded the opportunity to examine seasonality of catch in detail for discrete refuse deposits (Needs-Howarth and Brown 1998). A total of 15 scales were impressed from the Dunsmore site (Table 26). This sample comprised the scales from the major features at the site and several others. Since the objective was to assess feature seasonality, it was decided
Ictaluridae display a high proportion of thermal alteration, that has been linked by Thomas (1996d:145) to roasting of 44
to exclude the scales from the middens. Only 12 impressions were analysable; using CSAGES, they were identified as walleye on the basis of both growth patterns and absolute age.
for females (Scott and Crossman 1973:772). This means that all walleye scales analysed belonged to sexually mature individuals. MNI is equal to the number of year classes represented in the sample (Rojo 1987), but can be increased by including growth conditions at the margin. Based on the distribution of ages and growth conditions on the margin, it is likely that each scale at the Dunsmore site represents a different individual.
The age distribution of the archaeological specimens thus provides additional confirmation for the species identifications. Figure 14 provides yellow perch and walleye growth rates for several bodies of water. The graphs average out the often considerable sexual dimorphism in size, which increases with age in both species (e.g. McMurtry 1991). While growth patterns and rates may have been different in the oligotrophic Lake Simcoe of the 1300s and 1400s, this graph, nevertheless, illustrates a marked difference in relative growth patterns, which is likely to have pertained in prehistory. Some of the fish from Dunsmore and Carson were in excess of 14 years old, which is three years older than the maximum age for yellow perch in the modern Lake Simcoe populations.
Approximately one in four scales was regenerated. This proportion is similar to that observed in modern samples (David Brown, personal communication 1998). Some radii had broken off; this does not, however, affect interpretation, which is done in the area immediately lateral to the radii. Several of the oldest scales had stopped growing; the edge was failing and could not be interpreted with confidence (Table 26). Some scales exhibit the “*” condition (Table 22), indicating first new growth, with the annulus just completing formation. In calendar months, this means approximately early May to early June, depending on the age of the fish, growth rate, and temperature. Some exhibited the “+” condition, with some growth beyond the annulus, indicating capture from early June to late July, depending on age, growth rate, and temperature. Of the four scales exhibiting the “++” condition, three had stopped growing, making season of death interpretation impossible. One scale with the “++” condition could be attributed to a very late autumn or very early spring capture. Most of the Dunsmore scales, however, exhibited the “o” condition: an incompletely formed annulus with a translucent check or zone present on the edge. This condition happens during the period of slow growth or growth arrest, which falls between November to May, depending on age, growth rate, and water temperature.
Absolute scale size provides complementary support of the species identification, independent of the CSAGES analysis. While yellow perch scales are larger in relation to body size than walleye (about 2 percent of FL, vs. 1.6 percent, based on data of key scales on file, OMNR Sutton), their absolute size-at-age is much smaller. For the sake of (conservative) argument it is assumed that all the scales in our samples represent the largest scales on the fish body, which would derive either from the area posterior to the operculum, or between the dorsal fin and the lateral line. The anteriorposterior length of the scales from Dunsmore and Carson was compared to a non-random sample of the largest scales of the Lake Simcoe Perch caught in 1980, taken from behind the operculum (data on file at Ontario Ministry of Natural Resources, Sutton). Almost all the scales are larger than those of 10 and 11-year-old Perch. Bearing in mind the fact that modern populations are probably faster growing because of eutrophication of the lake, and that the archaeological scales are not necessarily the largest ones on the fish body, this effectively rules out yellow perch. The size-at-age and growth pattern of some of the younger individuals is so similar to that of the older individuals that they can safely be identified as walleye.
Fish by feature Introduction The following is a description and interpretation of the fish finds in the features containing more than 10 fish bones identified below the level of family (Table 28). Relative bone sizes are also provided (Figure 16). Dunsmore Feature 110, an irregular shaped, shallow pit in House 1, northeast cluster While the large sucker component (51 percent of fish NISP) (Table 28) might be taken to imply exploitation of spring-spawning fish, the fish assemblage from this feature may be more consistent with deep-water net fishing activity during the autumn component of the Lake Fishery (NeedsHowarth and Thomas 1998). This feature contained 10 bufflehead duck (Bucephala albeola) elements, attributable to at least three individuals. This non-resident duck migrates through the area in April and early May, and again in late October and November (Saunders 1947, 361). Attributes of immaturity noted on eight of these elements indicate at least some of these ducks were obtained during the autumn migration (Needs-Howarth and Thomas 1998). As hypothesized for merganser earlier, these diving ducks may
While the two sub-species of walleye cannot be distinguished based on cranial osteology, a marked difference in growth rate would be evidenced in the scales. The largest blue walleye on record for Ontario was only 11 years old and 375 mm long (Scott and Crossman 1973:772). No blue walleye are known from Lake Simcoe (Michael McMurtry, personal communication 1998, MacCrimmon and Skobe 1970; Rawson 1930) or Nottawasaga Bay and connecting waters (Robin Craig, personal communication 1998), so that all the scales from Dunsmore and Carson can safely be identified as the subspecies Stizostedion vitreum vitreum. Average age of the Dunsmore scales is 11.2 years. None of the fish were younger than age eight years (Table 26). Age at maturity is two to four years for males and three to six years 45
have been accidentally taken when they became entangled in fishing nets. The Lake Fishery is consistent with the large proportion of bottom feeders, such as suckers and bullheads (totalling 73 percent of the feature assemblage), and the presence of 12 sucker vertebrae. These taxa could have been caught in nets that touched the lake bottom. Lake fishing is also consistent with the average proportional element size of the yellow perch bones from this feature (Figure 16), which is well above the site average (Figure 10) (NeedsHowarth and Thomas 1998). The opercula of some of these yellow perch appear particularly robust, with an irregularly ridged surface.
sexually mature, but this should not be taken to imply that the individual was actually caught while on its spawning-run. Two lake trout vertebrae and a possible lake herring or lake whitefish vertebra may represent a stored resource, or may have been procured from the lake with the six sucker bones and the larger yellow perch during either the autumn of the spring component of the Lake Fishery (Needs-Howarth and Thomas 1998). To get an idea of how many individuals are represented among the 60 yellow perch bones in this feature, MNI values were calculated using element duplication, L/R matches and proportional element size. It must be noted that paired elements are not necessarily perfectly symmetrical (Brinkhuizen 1989:67, 73), so that matching of L/R pairs must be done within a certain range of proportional element size. Finding hypothetical “real individuals” in the database is of necessity somewhat arbitrary.
Evidence from the fish scales in this feature suggests a certain amount of fishing activity in spring. Two of four walleye scales, aged 10 and 11 years, indicate a spring kill. The season of capture of the other two scales, both aged 14 years, could not be interpreted because the edge was failing (Table 26) (Needs-Howarth and Brown 1998).
The conventional MNI, based on element duplication of the left operculum, is 13. With the inclusion of proportional element size (which automatically involves looking at all elements, not just the most numerous one), this figure is dramatically increased. With the arbitrary cut-off set at less than 5%, the MNI becomes about 24. If it is set more generously (or conservatively), at less than 10%, the resulting MNI is still increased by 25%, to about 16. With an NISP of 60, this suggests that, on average, only 2.5 to 3.75 cranial bones of each individual fish may survive to be identified. While some bones from these fish may have ended up in adjacent and associated Feature 230, these figures are, nevertheless, a dire reminder of taphonomic loss and/or the effects of element fragmentation on the identification rate.
Wheeler and Jones (1989:174) suggest that a spawning-run catch of northern pike would include the larger females and several more associated smaller males. The northern pike size distribution (Figure 16) might indicate exploitation during the spawning-run, including two or three recently mature males and a single associated female of the same year class; however, with a sample size of at most four individuals it would be difficult to argue against incidental capture. The brown bullheads are also large enough to have been caught during spawning, but again, it would be difficult to argue against incidental capture. Apart from a single ruffed grouse (Bonasa umbellus) element, the fish and duck bones were the only identified bone.
Dunsmore Feature 230, a filled-in semi-subterranean structure in House 8, west cluster This feature contained several autumn indicators: bones of Salmonidae, a migratory diving bird and a passenger pigeon. It contained three lake trout cranial bones, 10 lake trout vertebrae, and one basipterygium attributable either to lake whitefish or lake herring. Considered together, these items are good evidence for the autumn component of the Lake Fishery (Needs-Howarth and Thomas 1998). These autumnspawning species are associated with one bufflehead duck bone. Instead of a deliberate capture, it may represent a bird that became accidentally entangled in a fish net set over lake shoals (Needs-Howarth and Thomas 1998). Brown bullhead and yellow perch account for almost half of the fish assemblage, suggesting also a substantial emphasis on wetlands, stream and nearshore species. While the measured sample size is very small, it is suggested that the larger proportional element sizes of both brown bullhead and yellow perch (Figure 16) may be consistent with procurement of fish from a lacustrine, rather than an upstream, habitat (Needs-Howarth and Thomas 1998).
Dunsmore Feature 128, an ash pit in House 1, northeast cluster The small quantity of fish bones from this feature was a varied mixture. The single lake trout vertebra may have been a stored resource, or more likely, the feature represents an admixture from several different fishing and consumption events that got swept into this secondary deposit. Apart from two dog bones, fish were the only zooarchaeological find. Dunsmore Feature 206, a probable filled in sweatlodge in House 8, west cluster The midden-like nature of this feature results in a more ambiguous picture. Yellow perch bones make up 39 percent of this feature. They are on the small side (Figure 16), indicating they probably were not procured exclusively during spawning (Needs-Howarth and Thomas 1998). A substantial warm weather fishery is indicated by the quantity of Centrarchidae, Esocidae and Ictaluridae remains, representing one third of the fish NISP (Needs-Howarth and Thomas 1998). One brown bullhead right pectoral spine was sectioned. It represents an individual age three, killed in spring (Table 25). At this age, this individual may have been
This feature contained one barred owl (Strix varia) bone, and many bear bones. This may reflect the original 46
ceremonial use of the feature. As in Barrie Feature 216, the bear bones were all from the extremities, and were perhaps part of a tanned hide, used in ceremonies, or simply discarded into the feature as refuse.
on average smaller than those at Barrie (Figure 10), and represent a wider TL range. The difference is most marked in the heavy fraction, suggesting that more intensive flotation would have weighted the graph further in favour of smaller individuals.
Dunsmore Feature 347, a large filled-in semisubterranean structure in House 7, south-central cluster This assemblage more likely reflects the Generalized Warm Weather Fishery (Needs-Howarth and Thomas 1998). The most salient trait of this feature is the large component of brown bullhead and yellow perch (comprising 60 percent of the fish). The perch bones represent a wide range of proportional element sizes (Figure 16). Some of these could safely be considered sexually mature. Based on cranial bone and vertebrae, it has been argued by Needs-Howarth and Thomas (1998) that, with the exception of two white sucker and two Catostomidae bones, this feature lacked convincing cranial bone evidence for a Spring Spawning-run Fishery or autumn Lake Fishery. As in Feature 110, evidence from the fish scales (Needs-Howarth and Brown 1998), however, suggests a certain amount of fishing activity in spring. Three of four walleye scales, aged 8, 11 and 12 years, indicate spring capture. The season of capture of the remaining scale, probably in excess of age 12 years, could not be interpreted because it had regenerated and was no longer growing. This feature also has the largest amount of forest edge taxa. Its original function as a sweatlodge may be reflected in two hawk bones (Accipiter gentilis and Buteo jamaicensis), bone beads and dog bones (Needs-Howarth and Thomas 1994a). A single sandhill crane (Grus canadensis) element was modified into a bead. While the raw material for this artifact was probably obtained during the spring or autumn migration (Cadman et al. 1987:160-161), the finished item may have been curated and cannot be tied to season of deposition (Thomas 1996d).
As Thomas (1996d) has noted, the Dunsmore fish assemblage seems characteristic of the immediate local environs. Based on modern fish data (MacCrimmon and Skobe 1970), spawning-run exploitation of Cook’s Bay and adjacent Holland Marsh would be expected to have produced large northern pike, and even larger muskellunge, together with walleye and largemouth bass. While more recent research (Needs-Howarth and Brown 1998) has shown the walleye component of the assemblage to be more substantial than had previously been assumed, the small size of the pike and the dearth of muskellunge do not favour extensive use of the southern part of the lake. The walleye scales likely represent fish caught during the spring. While the “o” condition walleye may have been killed in autumn, this is less likely because at this time they mostly inhabit inaccessible, deeper waters and may have been too inactive to be vulnerable to gill-netting. The limited amount of growth on the scales in the “+” condition indicates that these fish were probably caught during a narrow time frame in spring (Needs-Howarth and Brown 1998). Their advanced age made the scales harder to interpret, but it does provide useful information on fishing. It indicates that the site occupants were exploiting sexually mature individuals. The narrow time-frame of capture indicates they probably were all caught during their spawning-run. Spring spawning-run exploitation conforms to our knowledge of walleye behaviour and habitat. The hypothesis that walleye were caught locally fits with other aspects of the fish assemblage. Walleye scales are distributed over four different features at the Dunsmore site. While three of these are in close proximity in House 1, further scales found in House 7 suggest that walleye fishing might not have been an isolated incident. The lack of walleye vertebrae suggests, however, that filleting may have been carried out away from the village. The alternative, though less plausible, explanation is that walleye were descaled and then thoroughly pounded up for inclusion in fish stew.
Fishing events Fishing events at Dunsmore are less concentrated on the Spring Spawning-run Fishery than at Barrie. The cranial bones can all be interpreted in terms of the Generalized Warm Weather Fishery or the Lake Fishery. Unequivocal evidence for Spring Spawning-run Fishery exploitation comes from CSAGES data on brown bullhead (H8, F206) and walleye (H1 F104, 106, 110 and H7 F347) (Table 25, 26), although the brown bullhead do not necessarily represent spawning-run exploitation.
Carson Faunal assemblage The faunal sample derives from 141 features within seven houses, and 6 m2 in a midden (Table 5). The 558 fish bones identified constitute 67 percent of the assemblage identified below family, expressed as NISP. In terms of BW, however, fish represent only 19 percent. The NISP is dominated by brown bullhead, pumpkinseed and yellow perch.
Northern pike, brown bullhead and yellow perch are ubiquitous in the major features. Both the northern pike and brown bullhead size distributions show a marked curve (Figure 16). The brown bullhead are of a size at which both males and females are mature. The northern pike are quite small; contemporary female northern pike in Georgian Bay are not often sexually mature at this size (Wainio 1966).
Acipenseridae Lake sturgeon is represented by only five bones, or less than
The Dunsmore yellow perch proportional element sizes are 47
one percent of identified fish.
than at Dunsmore.
Lepisosteidae Longnose gar is only represented by eight of its highly distinctive vertebrae from H1/F307, hence it is not included in the NISP (Table 13, 24).
Centrarchidae The Centrarchidae, consisting mostly of small panfish, are well represented, totalling 21 percent of identified fish. Vertebrae are over-represented at 34 percent.
Salmonidae lake trout and lake herring or lake whitefish cranial bones were identified from five contexts (2 percent of NISP), whereas only vertebrae were recovered from 12 contexts. Only H1/F168, a large, stratified house pit measuring over 100 cm wide and 125 cm deep, which had the highest NISP of any context at the site, contained both cranial bones and vertebrae. While several squares in Midden 4 contained Salmonidae vertebrae, no cranial bones were identified. As at the other sites, lake herring or lake whitefish vertebrae were a very minor presence, whereas lake trout vertebrae were well represented at 19 percent of identified vertebrae.
Percidae Perch family fish represent 25 percent of the NISP. As at Dunsmore, it is likely that the five Stizostedion sp. cranial elements are walleye. Not quite as numerous as the brown bullhead, the yellow perch at Carson constitute 23 percent of identified fish. Percidae vertebrae (13 percent) are under-represented, but not to the extreme extent seen at the other sites. Since these vertebrae are quite small, this may be a function of the smaller dry screen mesh aperture. The Carson site provided large numbers of fish scales. A non-random sample was selected, comprising 24 of the most complete scales from Carson, including a range of sizes, from two major features. All these scales are of walleye. The average age is 9.3 years; none are younger than 6 years. The fact that all analysed scales are walleye does not, of course, exclude the possibility that the smaller scales that were not subjected to CSAGES analysis are yellow perch or even Centrarchidae.
Esocidae Pike family fishes represent eight percent of identified fish. Most certain identifications are of northern pike. Vertebrae identifications are proportionate. Cyprinidae As at Dunsmore, a single pharyngeal arch was identified as a probable creek chub or fallfish.
Fish by feature Introduction The following is a description and interpretation of the fish finds in the features containing more than 10 fish bones identified below family (Table 29). Relative bone sizes are also provided (Figure 17).
Catostomidae Longnose and white sucker combined represented seven percent of the identified fish, which is slightly lower than Barrie and Dunsmore. Here white sucker bones are six times more common than those of longnose sucker. Unlike at Barrie and Dunsmore vertebrae identifications were proportionate. The proportional element sizes indicate mature individuals.
Carson House 1, Feature 70D Level 2 This is a stratified house pit 70x60x55 cm deep containing fish bone, scale, pottery and corn. Most of the fish derive from Level 2. The fish assemblage from this level is varied and otherwise also typical for the site, except for the large numbers of smaller pumpkinseed bones. The yellow perch proportional element size distribution is flat, going from very small to very large (Figure 17); this may suggest that some of these perch were caught by angling. There are few vertebrae. With one lake trout identification these do, however, add to the species variety. A river otter (Lutra canadensis) canine may indicate that procurement may have focussed on a river or stream, rather than the open lake or marsh. The alternative option, that this tooth may have been curated, is less likely, given that a river otter fifth cervical vertebra was found in Level 3. Since there were only five river otter finds in the entire Carson sample, it appears probable that some mixing has occurred between the two levels of this pit.
Ictaluridae Fragmentary dorsal and pectoral spines have resulted in a substantial number of non-specific identifications. The proportions in the database summary imply that most of the bullhead and catfish family identifications at Carson probably represent brown bullhead, rather than yellow bullhead or channel catfish. Of the specific identifications, brown bullhead constitutes 29 percent of identified fish, and yellow bullhead less than one percent. Because Ictaluridae pectoral and dorsal spines are readily identified to family, and because there were proportionately many of them at Carson, the NISP of this family is over-represented with respect to other fish. As at Dunsmore, Ictaluridae vertebrae were under-represented at 11 percent of identified vertebrae. Eight brown bullhead pectoral and dorsal spines were analysed from the Carson site (Table 25). Each spine represents a different individual. Age varied from one to six years, whereas season of capture included spring, summer and autumn. The average age of 3.3 years is slightly lower
Evidence from 16 walleye scales (Needs-Howarth and Brown 1998), representing at least nine MNI suggests a certain amount of fishing activity in spring. Because of problems with regeneration and failing edges it was not 48
possible to assign a season of capture to all of these fish (Table 26).
large contribution of brown bullhead, a species that was probably routinely obtained in the same kinds of fishing events as Esocidae. None of the other finds help elucidate the nature of the fish deposit, although a hawk (cf. Accipiter sp.) culmen section, together with some human bone, may indicate a ritual function for the deposit.
Carson House 1, Feature 168 Feature 168 is a large, stratified house pit measuring greater than 100 cm wide and 125 cm deep. The fish assemblage is varied. Suckers represent nine percent of fish NISP, with white sucker dominating. Lake herring or lake whitefish make up eight percent of the assemblage, well above the site average. The proportional element size distribution may suggest that each bone may represent a different individual. The inclusion of eight lake trout vertebrae lends weight to the suggestion that this feature contains refuse from a autumn Lake Fishery event, which may also have included three Catostomidae vertebrae. The yellow perch size range is also very wide and flat (Figure 17), including several larger individuals that may have been caught in the lake.
Carson House 3, Feature 132 This is a probable semi-subterranean sweatlodge, measuring 240x190x90 cm deep. Its fish bone composition is atypical, with a large proportion of Catostomidae, including two vertebrae, and very few perch. The seven northern pike bones represent MNI three (proportional element size included). Although the sample size is very small, the size distribution may suggest this feature includes one female and two smaller males caught together at spawning time. Vertebrae include two lake trout and three Centrarchidae. Once again, the sample size is so small that it is impossible to draw conclusions. This feature also contained seven dog bones, representing 54 percent of the identified mammal remains in the feature.
Evidence from eight walleye scales (Needs-Howarth and Brown 1998), representing at least six MNI suggests a certain amount of fishing activity in spring. Because of problems with regeneration and failing edges it was not possible to assign a season of capture to all of these fish (Table 26). The fish are all eight years or older, indicating that they were sexually mature. Lake Simcoe walleye age eight averaged 613 mm, whereas Lake Huron fish averaged 642 mm (Kushneriuk et al. 1996).
Carson House 3, Feature 164 This ash pit contains smaller brown bullhead (Figure 17). One of these, a 4-year-old individual represented by a left pectoral spine, was caught in spring (Table 25). Two Centrarchidae vertebrae could have been obtained at the same time.
Summer exploitation of lake, marsh or river may be indicated by a horned or red-necked grebe (cf. Podiceps sp.). The scoter (cf. Melanita sp.) derives either from the spring and autumn migration. Alternatively, the grebe and scoter elements may each represent one of the rare overwinterings noted by Godfrey (1986:26-28, 112-114).
Carson House 3, Feature 256 Feature 256 is a large interior pit measuring 140x110 cm on the surface. The faunal sample is dominated by brown bullhead. The strongly curved size distribution with sharp incline starting at 70 percent may indicate a net catch. Only two non-fish bones were recovered from the entire feature.
Carson House 1, Feature 182 This is an interior ash pit measuring 25 cm across and about 45 cm deep, apparently related to a nearby hearth. The complete lack of bird and mammal bones raises the distinct possibility that these were bagged separately and missed due to lack of consistency in bag labelling procedures.
Fishing events As at Barrie and Dunsmore, white suckers co-occur in features with species typical of the Spring Spawning-run Fishery (e.g., House 1 Feature 182) and those typical of the Generalized Warm Weather Fishery (e.g., House 3 Feature 120). Only one feature, House 1 Feature 182, contains unequivocal cranial bone evidence for the Spring Spawningrun Fishery. House 1 Feature F70d and House 1 Feature 168 have CSAGES evidence for a substantial Spring Spawningrun Fishery for walleye. It is likely that at least a proportion of the unanalysed scales from other features are also springcaught walleye. The Carson scale samples were selected to allow for a broad interpretation of fishing in two large features. It is not known, of course, how ubiquitous and numerous walleye scales are in the rest of the site. As at Dunsmore, the finds could have been from a single season’s catch. However, that argument can also be made about any of the cranial bone and vertebrae. Given the paucity of scales likely to survive to be analysed, it is probable that these scales do represent a substantial emphasis on walleye.
This feature contained fewer perch bones than most features. Suckers represent 12 percent, again with white sucker dominating. walleye also represent 12 percent. The yellow perch were mostly large enough to have been spawning when caught. The pike may have represented a spawning-run catch, if those in the 60 percent category are assumed to represent males, those in 80 percent and 90 percent females. In addition, this feature contained one large Percidae vertebra and two possible Catostomidae vertebrae. At least some of the contents of this pit would appear to have derived from spawning-run catches of suckers, yellow perch and walleye in April. Carson House 3, Feature 120 While the sample size for this feature is small, it is interesting to note the absence of pike family fish and the
All major features with northern pike also contain brown bullhead and yellow perch. The later part of the northern 49
pike spawning-run overlaps with the start of the spawningrun of yellow perch and suckers. However, the consistent co-occurrence with brown bullhead suggests that pike were exploited throughout the warm seasons as part of the Generalized Warm Weather Fishery, probably mostly in fish-traps, rather than during their spawning-run. The peak in the size distribution of northern pike, at 60-69 percent, probably represents immature individuals. Bullhead and yellow perch combined constitute at least one third of all major features, but their proportions vary, especially in House 3. The two features with the highest combined total, House 3 Feature 164 and House 3 Feature 256, are dominated by bullhead. These two features may largely represent single fishing episodes during the warm weather. It appears that fishing for brown bullhead at Carson was somewhat more generalized than at Dunsmore, including earlier autumn catches and a wider age distribution. This range of interpretations is not entirely surprising, given that the samples derive from eight separate features. It is likely that brown bullhead was caught throughout the warm seasons in river, stream and wetland habitats. The co-occurrence of brown bullhead with yellow perch in all but House 3 Feature 256, and the presence of some very small yellow perch supports the hypothesis that yellow perch exploitation was not necessarily limited to the spawning season. Yellow perch could have been caught together with other species during the Spring Spawning-run Fishery, but they were also probably an important component of the Generalized Warm Weather Fishery. With the high numbers of brown bullhead and pumpkinseed, the main focus at Carson appears to be on the Generalized Warm Weather Fishery. The substantial numbers of yellow perch, white sucker and northern pike could have resulted either from the same fishery, or from a more specialized Spring Spawning-run Fishery. A few of the Carson fish remains may have derived from a Lake Fishery on Kempenfelt Bay and Nottawasaga Bay for taxa like white sucker, smallmouth bass, yellow perch and walleye, including inshore exploitation of Salmonidae. However, the major emphasis appears to be on the Generalized Warm Weather Fishery, focussing on resident taxa such as northern pike, brown bullhead and Centrarchidae.
50
CHAPTER 7: SYNTHESIS
Compiling the evidence Considerable overlap exists in species composition between the three fisheries complexes: the Spring Spawning-run Fishery, the Generalized Warm Weather Fishery and the Lake Fishery. For example, depending on size, yellow perch may have been part of the Spring Spawning-run Fishery or the Generalized Warm Weather Fishery. Most taxa, in fact, can be interpreted in more than one way. The most likely interpretation is based on a combination of taxonomic abundance, taxonomic composition, size and/or age and season of death. Depending on capture technique or net depth, catches may include taxa not normally associated with each other. It is, therefore, best to speak only in terms of greater or lesser emphasis on a particular fishery (NeedsHowarth and Thomas 1998).
more likely a reflection of greater emphasis on mature fish and spawning-run exploitation. Figures 18 and 19 represent the most straight-forward way of interpreting the fish assemblages from each site, with each taxon being assigned to the fishery from which it would most likely be derived. In order to show the broader trends, these figures do not incorporate some of the nuances and variations identified for individual larger features. They also do not take account of the possible functional variation between the three segments of the Dunsmore site. Bearing in mind the discussion on diagnostic elements, it should be noted that the contribution of brown bullhead, and to a lesser extent suckers, is biassed by the inclusion of unique elements, such as the pectoral spine and Weberian vertebrae.
Evidence for more than one fishery in some archaeological features can be explained by prolonged use of pits and disused semi-subterranean structures, causing several fishing events to be represented in a single archaeological context. Evidence for such reuse was found at the Early to Middle Iroquoian Myers Road site, north of Lake Erie (Ramsden et al. 1998). Oily fish, such as lake trout, may also have been smoked or frozen and stored for later use, adding to the confusion. Feature seasonality data should be used with caution, because of prolonged and intermittent use of features and stratigraphic mixing resulting from contemporary in-filling of pits and semi-subterranean structures with refuse and soil from other areas of the site.
Location of capture Yellow perch were probably caught in Lake Simcoe or the tributary streams at the head of Kempenfelt Bay. It is likely that at least some of the other fish at the Barrie site derived from the very productive Minesing Swamp or the hypothesized lagoon on Kempenfelt Bay. Exploitation of lake sturgeon most likely occurred in the Nottawasaga River (Needs-Howarth 1996). Use of the Nottawasaga River, which can mostly only be navigated by canoe in the main channel, located some kilometres away from the site, indicates that a short travel distance was not the only factor in deciding where to fish. The large numbers of lake sturgeon remains at the Barrie site may hint at a much greater emphasis on the environs near Nottawasaga Bay, which could also have provided the lake trout, lake whitefish, lake herring, yellow perch, suckers and northern pike. While exploitation of fish in Nottawasaga Bay and at the mouth of the Nottawasaga River would entail a substantial canoe journey and probably an overnight stay, it cannot be discounted. Hunter (1906:47) draws a portage trail that runs from the head of Kempenfelt Bay westward to Willow Creek, straight across the portion of the Simcoe Uplands directly north of the Barrie site. Perhaps a precursor of this trail was used by the Barrie people to get from the site to the river.
Timing of capture Despite the great potential for ambiguity, the Three Fisheries Model indicates some definite inter- and intra-site differences in seasonal activity. Both the taxonomic composition and fish proportional element sizes in the major features at Barrie suggest substantial emphasis on the Spring Spawning-run Fishery. In contrast to the Barrie site, several features at Dunsmore and Carson lack compelling evidence for a Spring Spawning-run Fishery. At the two later sites the most solid evidence for a Spring Spawning-run Fishery comes not from cranial bones, but from scales. The fishing events reflected in the selected major features discussed in Chapter 6 may not be representative of the complete range of fishing strategies at the respective sites. However, there are also differences between the site assemblages, both in terms of NISP and fish proportional element size. Strong evidence of a spring emphasis at the Barrie site comes from the large number of lake sturgeon remains from the middens. While Barrie and Dunsmore have very similar proportions of yellow perch (Table 24, Figure 13), at Barrie many more of them are in the size range of sexually mature individuals, indicating probable spawningrun exploitation. The larger average proportional element size of non-perch fish at Barrie, while possibly the result of fishing techniques aimed specifically at larger individuals, is
Locating and catching sturgeon may have been a labour- and time-intensive activity. Why then, is lake sturgeon so common at this site? Was it considered more desirable by the occupants of the Barrie site, which has many sturgeon bones widely distributed across the site, than those of the Dunsmore and Carson sites, which produced only a few sturgeon bones? Or are the reasons more pragmatic? Other archaeological evidence suggests it likely that the Barrie people were familiar with the area before they decided to settle there (Sutton 1996a). Their ability to exploit lake sturgeon may suggest they were also already familiar with the local watercourses.
51
In contrast to the focussed, riverine-oriented fishing strategies at the Barrie site, fishing at Dunsmore and Carson appears to have consisted mostly of the Generalized Warm Weather Fishery, with an emphasis on Little Lake and Willow Creek (see also Thomas 1996d). Fish distribution by family is very similar between these two sites. Ictaluridae increase in importance at Carson at the expense of yellow perch, suggesting a further focus away from Kempenfelt Bay and adjacent waters. It is interesting to note that, while Dunsmore has five channel catfish identifications, Carson has none of these typically riverine fish. The yellow perch at Carson are also somewhat smaller, perhaps suggesting a decrease in emphasis on mass capture during the spawningrun, although the finer dry screen mesh size probably accounts for some of the very small elements sizes.
Since there are no analysed zooarchaeological data from possible fishing camps, it is not known to what extent the assemblages recovered from villages are representative of all fishing activity carried out by the inhabitants. There is also the possibility that some fish were non-local, resulting from trade. This might have involved fish that can be readily preserved, such as lake herring, lake whitefish and lake trout. The relatively low Salmonidae NISP counts at the three sites can, however, be interpreted as indicating a proportionately limited level of local exploitation, probably from the onshore shoals of either Lake Simcoe or Nottawasaga Bay.
Fishing in and around wetlands, either at Little Lake or in Minesing Swamp or at Kempenfelt Bay may have been an attractive option, because fishing with nets or fish-traps could be combined with trapping of small mammals, berry picking, digging for cattail roots, etc. Northern pike, brown bullhead and yellow perch in the wetlands around Little Lake and Kempenfelt Bay may have provided such an abundance of biomass that the later inhabitants of the area did not need to, or want to, fish as intensively as the Barrie occupants during the Spring Spawning-run Fishery. Indeed, the bulk of the northern pike and brown bullhead could have been caught very locally in Little Lake and surrounding area. Age and growth analyses of walleye scales provide some detail that is lacking for the other taxa. When viewed in context with the other fish data, the extensive spring spawning-run exploitation of walleye does not seem incongruent. If the inferred location of catch is correct, exploitation of this taxon becomes simply an extension of a very local pattern of exploitation centred around Little Lake. This emphasizes the importance of inferring not only the season of catch, but also the location.
Technique of capture Techniques of mass capture are most productive when there is a reasonable hope of encountering fish. It has been argued above that for some fish, the most productive time for human exploitation in the past would have been during spawning time. Larger, sexually mature fish would thus be associated with techniques of mass capture. There are some exceptions: in some species, juveniles will travel on the spawning migration (for example lake sturgeon, as mentioned above), and in some species adults will school outside the spawning season, for example lake whitefish, who aggregate in nearshore locations in spring (Scott and Crossman 1973:272). A comparison of average fish cranial bone size would indicate that the Barrie site collections contained a greater number of larger fish. Size distributions in the major features are quite tight and steep at Barrie, and generally more varied at Dunsmore and Carson, with the exception of Dunsmore Feature 206. This would appear to point towards a more extensive use of nets at Barrie. The restricted proportional element size distribution may be the effect of live-release from seine-nets and fish-traps (Stephen Crawford, personal communication 1999). Northern pike, however, is one taxon that appears not to have been caught with nets. Spearing of pike may bias in favour of larger, mature specimens. The use of fish-traps, on the other hand, may result in a more restricted size distribution, because small specimens can escape from fishtraps, whereas large specimens cannot enter (Enghoff 1986; Noe-Nygaard 1983). Modern Lake Simcoe northern pike can become 1140 mm long (MacCrimmon and Skobe 1970:92), although pike in a slow-growing population on Manitoulin Island in Georgian Bay only grow to about half that size (Casselman 1974). Females are sexually mature by age 3-4 (Scott and Crossman 1973) or a range of 457-890 mm TL for Georgian Bay (Wainio 1966), or about 490-530 mm TL on Manitoulin Island (Casselman 1974). Males mature earlier, by age 2-3 or 305-787 mm TL (Wainio 1966) or 380-460 mm TL (Casselman 1974). Minimum size at maturity for both sexes therefore ranges from 305 to 890 mm (Table 21).
Thomas (1996c:121) noted the absence of bowfin and the paucity of longnose gar at Dunsmore, which would be found in same swampy and marshy areas frequented by northern pike. The total absence of longnose gar at Barrie, and its presence at Dunsmore and Carson may give a clue to the exploitation of northern pike. Where northern pike cooccurs at a site with longnose gar, it might reasonably be expected to have been caught from the same environment. Conversely, this may mean that the Barrie northern pike were obtained from a less swampy environment. Some fishing in deeper lake habitats in autumn is indicated by the Salmonidae remains at all three sites. The caloric value of lake trout, and to a lesser extent lake whitefish (Table 30) may have off-set the considerable effort and travel involved in obtaining this species, something which has been suggested for the Dunsmore site by Thomas (1996d). It is likely that the importance of lake trout, lake herring and lake whitefish are under-represented at the sites as a result of off-site processing and preservation biases.
Only one northen pike element from the archaeological collections is larger than the comparable element in the 52
reference specimen of 618 mm. The largest bones in the assemblage may thus belong to individuals of spawning age, but the smaller ones probably do not. If these smaller individuals were not spawning, they probably were not caught in spawning-run mass-procurement events either, and they are probably smaller than the optimal size for spearing (Needs-Howarth and Thomas 1998). The small proportional element sizes and relatively tight size distribution of northern pike appear to indicate extensive use of fish-traps. Some of the larger individuals may have been speared or caught by angling (Needs-Howarth and Thomas 1998).
representation of Salmonidae cranial bone at many sites. And even if the inhabitants were catching large numbers of Salmonidae, as appears to have been the case at the Molson site, other factors may have influenced how many Salmonidae cranial bones ended up being deposited at the village. Catches made in very cold weather could be expected to include more cranial elements because whole (gutted) fishes would have been preserved through freezing. Catch site processing would have been less important on cold days than it would be on warmer days where natural freezing would not occur (Cooper 1996).
Size and age composition of the brown bullhead at each site indicates mostly capture of adults using angling or fish-traps (Figure 15, 16, 17).
Sunfish vertebrae are numerous and ubiquitous at Barrie, Dunsmore and Carson. Molnar (1997) has suggested that the preponderance of cranial bones of Micropterus sp. at the Hunter’s Point site indicates that small- and largemouth bass, with their firm, dry flesh, were filleted and dried at the site for later consumption off-site. It could be argued that the consumption end of this kind of processing pattern accounts for the biassed distribution at Barrie, Dunsmore and Carson.
The advanced size and age of the walleye at Dunsmore and Carson may indicate that the site occupants were using wide-meshed nets or spears. In general the species and associated size distribution at Dunsmore and Carson suggests a greater emphasis on passive technology at these sites. It is interesting to note that all three zooarchaeological assemblages lack very small fish individuals of most species, and also taxa that are generally small, such as minnows, even in the heavy fractions that were sorted to 2 mm. Considering that fishing nets were handmade, there may have been considerable variation in aperture within one net. Some openings may have been small enough to catch really small fish. The general lack of very small fish bones may suggest that tiny fish were either purposefully avoided, discarded at the catch site, or eaten whole. Some of the small fish that were recovered from the zooarchaeological assemblages may have arrived at the site as gut-content of piscivorous fish.
Before attributing these biassed distributions to processing, taphonomic explanations must be ruled out. It is possible, for example, that Centrarchidae heads at Barrie, Dunsmore and Carson were preferentially scavenged by dogs (who may have avoided the sharp teeth of lake trout, northern pike and muskellunge, and the more robust crania of suckers and bullheads), or that they were more subject to decay than other taxa because of their relative osteological fragility. The importance of walleye at Dunsmore and Carson would have been severely underestimated without the fish scale evidence; seven and four walleye cranial bone identifications, respectively, certainly did not suggest a substantial emphasis on this taxon. The survival of so many walleye scales, and almost none of any other taxon, is hard to explain, although relative strength and size must play a role. What is even harder to explain is the paucity of walleye cranial bones. Indeed, none of the six features at Dunsmore and Carson with identified walleye scales contained any walleye cranial bone. With the exception of Carson House 1 Feature 168, which contained one large Percidae vertebra that may be walleye, none of these features contained any Percidae vertebrae either. Since there are no overwhelming numbers of larger Percidae vertebrae anywhere in the sites, it seems likely that walleye were being processed at the catch site. The heads may have been removed, together with some or all of the vertebrae. The skin from the fillets, with adhering scales, may have been deposited in refuse dumps after return to the village.
Fish processing There are some interesting patterns in fish bone distribution that may inform about fish butchering and processing. While off-site processing may be one explanation for the lack of cranial bone, in the case of Salmonidae it is very likely that these numbers also reflect a considerable preservation bias, as was suggested by Molnar (1997:145) for Hunter’s Point and by Needs-Howarth and Thomas (1998) for Barrie and Dunsmore. It has already been noted that the autumn Lake Fishery may be under-represented at Barrie, Dunsmore and Carson in relation to other fishing activities because of taphonomic bias against Salmonidae cranial bone. The same distorted proportions are evident at the protohistoric Huron Molson site (Cooper 1996:40), the late Iroquoian Keffer (Stewart 1991a:86-87) and Over (Thomas 1996a; Thomas 1996c) sites, and the protohistoric Odawa Hunter’s Point site, analysed by Rosemary Prevec (Molnar 1997) (Table 31). In the case of the Over site, there was no cranial bone whatsoever, while Salmonidae represented 40 percent of the vertebrae. As noted earlier, the expected ratios of cranial bones to vertebrae differ between species. This has not been emphasized by other local researchers (e.g., Cooper 1996; Lennox et al. 1986). Differences in osteology and inherent identifiability may account for some of the observed under-
Diachronic change in fish catches A change in fish size can be the result of changes in technique of capture, or changes in location of capture, resulting in a different fish population being exploited (e.g. Sternberg 1994). The latter could be confirmed by age at length information (Sternberg 1994) or potentially by DNA analysis. Alternatively, the size change could result from 53
human predation pressure on the same fish population. Identifying fishing pressure requires distinguishing between fishing and non-fishing sources of mortality, and then determining of the biological significance of the fishing mortality (Stephen Crawford, personal communication 1999). Fish population change may also relate to various non-anthropogenic sources of fish population alterations or to natural fluctuations (Stephen Crawford, personal communication 1999). Since population statistics are not available for fish populations in the past, it will here be assumed that when people start fishing in a pristine environment, they first catch a lot of larger fish (Boddeke 1974). It has been suggested that local fish communities could have changed over a very short period of time, especially under heavy exploitation (Michael McMurtry, personal communication 1998). Fish community structure may have changed under modest levels of exploitation in a limiting environment such as Little Lake (Michael McMurtry, personal communication 1998).
is probably not just a function of recovery bias. Secure osteometrics of brown bullhead pectoral spine length correlate with age in years; this may suggest a decrease in the average age at which fish are being caught at Carson, rather than a decrease in absolute fish growth rate, or stunting. A large number of yellow perch cranial bones were measured at each site (Figure 15, 16, 17); like the bullhead spines, operculum and dentary, these exhibit a decline in size through time. Screen mesh sizes at Barrie and Dunsmore were identical. It may therefore be assumed that differences in yellow perch proportional element sizes among the screened and heavy fraction samples between these two sites are real (Figure 10). Since the sample size is relatively large, sampling error may be less of an issue for yellow perch than it is for other species. The size differences in this case can be explained in terms of the Three Fisheries Model. The yellow perch at Barrie were likely mostly obtained during the Spring Spawning-run Fishery, whereas those at Dunsmore and Carson were likely mostly obtained during the Generalized Warm Weather Fishery, which is hypothesized to have included larger numbers of small fish.
The larger proportional element sizes of fish at Barrie could be argued to be the result of a greater emphasis on Lake Simcoe and the Nottawasaga River, rather than Little Lake. In this respect it is informative to look closely at Dunsmore and Carson, which are so close in time and space. Differential recovery may be a factor, but the discussion on scale diameter and cranial bone metrics suggests it certainly does not explain all the differences in proportional element size.
A decline in walleye size can be inferred from the decline in average age evidenced by the scales. As Figure 21 indicates, the walleye are all six years or older, indicating that they were sexually mature. The age distribution indicates that the Carson sample contains more younger walleye than are found in the Dunsmore sample (Figure 21). Measurements of the anterior-posterior dimension of the scales indicates that this is not attributable to differential recovery (Figure 11). In older fish, the size of fish, and hence scales, does not increase significantly each growing season (Figure 15), so that an average age difference of almost two years would not translate to a significantly bigger or smaller scale or fish. Since neither site has produced any very small walleye scales, it is most likely that the difference in age distribution is not a function of recovery.
Table 32 summarises average size data relating to northern pike, brown bullhead, yellow perch, the taxa which have the most abundant metrics and proportional element size observations. The average size of each taxon is largest at Barrie. This table further demonstrates the similarity in screened fish cranial bone size between Dunsmore and Carson that was apparent from Figure 9. Since these four species make up the majority of fish assemblage, by implication, this means that the sizes of the other fish species exploited are also similar. In conjunction with the size distribution, mean, minimum and maximum, these data can provide clues to changes in fish catches through time.
Inter-site differences in element size and fish age probably relate to some extent to procurement, taphonomy or sample bias, as well as natural population fluctuations and nonanthropogenic factors. For example, the size differences among yellow perch bones between the three assemblages can probably mostly be explained by differing fishing strategies. It is more difficult to account for differences in age and size among fish that were caught as part of the same fishery. The brown bullhead at Dunsmore and Carson were likely all caught during the Generalized Warm Weather Fishery, and the walleye were likely all caught during the Spring Spawning-run Fishery.
Northern pike are largest at Barrie and smallest at Dunsmore. The probable range of fish sizes, however, is not huge. Since this species has an elongated body, which would expand less quickly in circumference than in length, it is quite likely that northern pike in the entire size range could have been caught using the same general fish-trap or net size. Brown bullhead are quite similar in size at the three sites. However, operculum and especially dentary osteometrics (Figure 12) show a trend towards smaller fish at Carson, both in terms of size distribution and average size. Average ages from pectoral spines follow the same trend (Figure 20). As noted in the section on recovery, given the relatively large size of the operculum, dentary and pectoral spine, even when fragmented, the difference in age and size distribution
The biassed and limited nature of the walleye scale samples makes it difficult to draw conclusions about fish size. While there are many large fish represented at either site, the age distribution may, nevertheless, indicate increased pressure on the walleye stocks during the relatively short time 54
interval between Dunsmore and Carson. Percidae, including walleye, can respond rapidly to exploitation (Spangler et al. 1977). The size distribution in the population can shift if larger fish are more vulnerable to the fishing gear. Though the possible mechanisms are complex, a decrease in mean size of individuals is a common response to human exploitation (Spangler et al. 1977).
Amtstaetter, personal communication 1999). The archaeological yellow perch were likely older at a given length than those in the contemporary Lake Simcoe population because of twentieth century eutrophication of Lake Simcoe. Although the archaeological walleye TL is unknown, the scales suggest old and hence large individuals. The average age may relate to recovery method (i.e. use of shaker screens) in favour of larger, more robust scales, and sample selection. The maximum age, however, appears to fit the pattern of large maximum sizes for brown bullhead and yellow perch.
The consistent size and age differences in these two species between the Dunsmore and Carson sites may point to human predation pressure. If the brown bullhead at Dunsmore and Carson were caught in Little Lake, the age and size data might suggests that Iroquoian fishing in this limiting environment was having sufficient impact to cause a small decline in average age/size. Fishing at Dunsmore and Carson may have had some impact on maximum sizes in the local fish stocks, however, the generally old age of the walleye scales may suggest that this impact was by no means severe.
Evidence for fishing at other local sites For comparative purposes it is instructive to examine the fish assemblages at three precontact sites located south of Kempenfelt Bay, to the west of Lover’s Creek. Wiacek has a calibrated date of A. D. 1270-1370 at 1 sigma (Dodd et al. 1990: Table 10.1), in agreement with ceramic seriation (Robertson et al. 1995); a large part of the fish assemblage from the first excavation at the site derives from flotation (Lennox et al. 1986). Additional zooarchaeological remains were recovered during a second excavation (Thomas 1993). Hubbert (Thomas 1996b) was occupied shortly afterwards, with a date calibrating to A. D. 1388-1437 at 1 sigma (MacDonald and Williamson 1996). Ceramic seriation, however, favours an early-mid fifteenth century occupation (Williamson and Powis 1996). The zooarchaeological collection here was recovered in the same manner as at Barrie and Dunsmore. Molson, occupied almost 200 years later, at the turn of the seventeenth century, was mostly recovered by 3.2 mm water screening (Cooper 1996). Like Dunsmore and Carson, Wiacek and Hubbert can be used to assess changes over a very short period of time in a small area, although here comparisons are also hampered by the fact that the assemblages were recovered and analysed in different manners.
The Three Fisheries Model appears to support intra- and inter-site differences in location, timing and technique of fishing activities. The large numbers of lake sturgeon bones at the Barrie site suggest exploitation of the Nottawasaga River. Yellow perch were likely obtained from Kempenfelt bay and tributary streams. The large numbers of yellow perch at all three sites suggest substantial exploitation of Lake Simcoe and tributary streams. Both the species and size/age distributions of the remaining taxa argue for an emphasis on Little Lake at the Dunsmore and Carson sites. A decline in mean fish proportional element size and age of brown bullhead and walleye may suggest that there was some continuity through time in which locations were chosen to exploit particular species. If the directional change in fish size and age is independent of recovery and sample size (which cannot, of course, be proven) it might confirm the relative site chronology of Dunsmore and Carson. From a fisheries science perspective it is interesting to note that the largest fish represented at the archaeological sites are not as large as the largest fish caught in Ontario today. The largest brown bullhead element dimension measured with the dial calipers is 34 percent larger than the same dimension in the reference specimen of known TL. While direct back-calculation may over-estimate TL to some extent, this individual must have been as big (but not necessarily as old) as the fish age 10+ years with TL 350 mm recorded from Quebec by Harvey and Fortin (1982:Table 3). The record for Ontario is 413 mm TL for a fish caught in a pond near Raleigh, Ontario, in 1989 (Alwyn Rose, personal communication 1999).
Since these zooarchaeological assemblages were analysed by other zooarchaeologists, detailed observations are not advisable. The Lover’s Creek sites have approximately the same percentage of yellow perch as Barrie, Dunsmore and Carson, and more suckers (Figure 22). This may suggest a difference in fishing strategies, with a stronger emphasis on the Spring Spawning-run Fishery and/or the Lake Fishery. While fish size was recorded as small, medium and large at Molson (Cooper 1996), and yellow perch opercula were measured at Hubbert (Thomas 1996b), detailed, comprehensive information on fish proportional element size of all species, which might be used to infer maturity, was not recorded.
The largest measured yellow perch bone suggests a fish in the range of 345 mm. The official Ontario record is 375 mm TL for a fish caught in 1995 in Lake Erie (Alwyn Rose, personal communication 1999). An even larger individual is recorded in the Ontario Ministry of Resources Lake Simcoe database. This individual, recently caught by angling, had a length of 400 mm TL, and was at least 11 years old, based growth interpretation of one of its scales (Frank
While Wiacek is located only 4 km south of Kempenfelt Bay, catch site processing may have been carried out for large catches of spawning suckers. At Hubbert Catostomidae NISP are less prominent than at Wiacek (Figure 22). Instead, there is a greater emphasis on Ictaluridae. Molson has many more Catostomidae than 55
Hubbert, but this may be seen as a continuation of an emphasis on this taxon that appears characteristic for the Lover’s Creek sites. What is perhaps most striking is the emphasis on Salmonidae at Molson, signalling, perhaps, the historical start of a large-scale autumn Lake Fishery. Cooper (1996) suggested this on the basis of cranial to vertebrae ratios at the Molson site. The comparison of both cranial bones and vertebrae at Barrie, Dunsmore, Carson and Molson substantially strengthens the argument (Figure 23).
virtually undisturbed habitat. A concurrent focus on birds and mammals preferring riverine environments and deciduous/coniferous forest may indicate that there may have been few clearings or secondary growth areas in the vicinity of the site. If the Barrie site was the only village in the area at the time, a lack of disturbed habitat species and a scheduling emphasis on fish and water-edge taxa would be expected (Needs-Howarth and Sutton 1993). Continued permanent occupation of the area around Kempenfelt Bay could be expected to result in more evidence of forest clearance and forest edge habitats at Dunsmore and Carson.
The Barrie collection does not contain many sucker cranial bones. The Barrie sample does contain cranial bone evidence for a substantial emphasis on spring exploitation of other taxa, namely yellow perch and lake sturgeon, and it contains substantial numbers of sucker vertebrae. Dunsmore and Carson may prove to be atypical for the region in having such a small percentage of sucker bones.
As a result of the varied taphonomic factors operating on each assemblage, it would be unwise to attach much significance to proportions of different taxonomic classes within the entire assemblage. The fact that one site has more fish than another may relate more to taphonomy than to actual differences in procurement emphasis. In order to make sense of the huge variety in the number of non-fish species and their NISP, it is instructive to sub-divide the non-fish sample based on procurement habitat (Table 33). Note that animals that frequented garden habitats probably also frequented non-anthropogenic habitats (Neusius 1996:277). This results in some overlap between categories. For example, Sagard notes that both raccoons (Procyon lotor) and sandhill cranes (Grus canadensis) were major garden pests, but they doubtless were captured away from the village as well.
Paradoxically, the zooarchaeological samples with more intensive recovery – Wiacek and Molson – have more Catostomidae cranial bone identifications. Far from having the kinds of small, fragile bones one would expect to recover in greater numbers using small dry screen mesh or flotation, these taxa possess relatively large and robust cranial bones. This would imply that the emphasis on springspawning Catostomidae at the Lover’s Creek sites is real. However, a consideration of vertebrae identifications is important as well, especially in connection with the Spring Spawning-run Fishery and the Lake Fishery.
The proportion of non-fish derived from marshy/wetland environments is similar between the three sites. This is not surprising, since there were extensive marshes or wetlands nearby each site, and possibly also at the head of Kempenfelt Bay.
Unfortunately, vertebrae identifications at Hubbert were limited to Salmonidae, whereas no vertebrae identifications at all were carried out on the Wiacek sample. All vertebrae were identified to the taxonomic level of family at the Molson site (Cooper 1996). This is the only site of the six under discussion where Salmonidae cranial bone exceeds one or two percent of the fish assemblage (Figure 22). However, when the vertebrae identifications are included (Figure 23), the differences become much more marked. As was obvious already in Figure 19, Salmonidae representation at the Barrie site also increases substantially when vertebrae are included in the NISP. It is interesting to note that Catostomidae vertebrae are also very numerous at Barrie and Molson, perhaps suggesting more off-site processing (i.e. filleting) of spawning-run catches than at the other two sites. Alternatively, the Catostomidae at Molson may have been targeted in autumn, thus effectively being a component of the autumn Lake Fishery.
Following the work of Studer (1992) discussed earlier, the proportion of surface-feeding to diving water birds can help establish whether entangling nets were used. Unfortunately the numbers of duck bones are very small. Dunsmore has notably more diving ducks than either Barrie or Carson. However, these are mostly immature bufflehead elements, which may have resulted from a single incidental catch. The proportion of forest taxa is similar among the three sites, especially when the possible intrusive chipmunk (Tamias striatus) from Carson is excluded from the calculations. Forest edge and clearing-loving taxa are much more prevalent at Barrie than at Dunsmore and Carson, because of the comparatively larger numbers of woodchuck and whitetailed deer.
Other subsistence indicators Through investigations of the cranial bone and vertebrae, as well as other seasonality indicators, it has become apparent that there was a greater emphasis on the Spring Spawningrun Fishery at the Barrie site, while fishing at the Dunsmore and Carson sites appeared to be more opportunistic and generalised, involving Esocidae, bullheads and immature yellow perch in shallower waters. The more seasonally focussed fishing of the Barrie people may relate to the fact that they are the first people to permanently occupy a
Land clearance and soil disturbance associated with Iroquoian settlement construction and agriculture favours greater densities of certain animals than would exist in an undisturbed forest habitat. The numbers of small mammal species that thrive in open or partly-cleared habitat in close proximity to humans should increase in association with land clearance. This issue was explored by Thomas for the Dunsmore site assemblage (1996d); he argues that the 56
emphasis on local, predictable, small-unit resources suggests they were specifically targeted. This kind of “garden hunting” has been documented among tropical (Linares 1976) and temperate (Neusius 1996) horticulturalists. Hunting these animals added high-quality protein to the diet, and at the same time reduced predation on the domestic crops. Barrie and Carson also contain many species that could have been part of a “garden hunting” complex.
contrast, were apparently more inclined to travel further from their village. Given the close proximity of Dunsmore and Carson in both time and space, the similarities in their zooarchaeological assemblages are not surprising. The differences, while subtle, are very informative. Availability of resources probably didn’t change much through time, so that differences in the faunal assemblages may have been a result of changing preferences. This, in itself, is an indication that zooarchaeological assemblages at these sites were not simply a passive reflection of local availability.
While it can be difficult at Iroquoian sites to identify postoccupation burrow deaths during shovel excavation (Thomas 1996d:151), fossorial microfauna is here included in the semi-commensal village and field category, since the elements are distributed among various contexts. While the smaller mesh size at Carson should have been more efficient at retrieving microfauna, no entire or partial mouse (Peromyscus sp.) skeletons were identified. On that basis, the mice and voles (Microtus sp.) from both Dunsmore and Carson are included here as pest species that were hunted in the corn fields and in the village. The Huron reportedly ate either mice or voles (Sagard 1939:227). Some fragile cranial bones of chipmunks from Carson may include a hibernation death. This is reflected in the calculations in Table 33. Contrary to initial expectations, Barrie has more of these disturbed habitat species than the later sites, Dunsmore and Carson.
The deer paradox It is likely that the Barrie people were familiar with the area before they decided to settle there (Sutton 1996a). This means they may have know there were few deer and that the conditions for growing corn were adequate but marginal when they decided to move into the area. It appears plausible that, in addition to the availability of abundant wood for construction and hearths, and the adequate conditions for crop growing, they chose their site location because of the abundant fish and aquatic resources. The Barrie site’s designation as a “pioneer village” appears to be supported by this zooarchaeological analysis. On the one hand, the pristine environment may have offered more subsistence options, on the other hand, the lack of forest edge and clearance habitats may have directed subsistence efforts towards a more limited range of habitats. Only the proportions of deer and small mammals that thrive in disturbed and forest ecotone environments deviate from these expectations (Table 33). The number of deer elements identified at local sites may, however, relate more to biogeography than to human preference. This particular area of Ontario appears to have supported very low deer densities as a result of the local physiography (Robertson et al. 1995). Lack of suitable habitat is argued to be the major reason for the general dearth of deer remains at local sites (Robertson et al. 1995). This hypothesis is supported by the fact that the increase in field and edge habitats in the fifteenth century does not result in an increase in deer (Robertson et al. 1995).
Dog as a food resource A major difference between these three sites is seen in the reliance on domestic dogs (summarized in Table 34). If dog is included in the locally-available species at Dunsmore, over half of the mammal assemblage derives from the cleared fields and the settlement itself (Thomas 1996d). Because of the abundance of dog at Carson, this site has an even stronger village focus. While some of these dogs may have been part of a ceremonial feast, the presence of a dog burial in one of the houses (Dompierre 1990) (not included in this analysis) suggests that dog remains from refuse pits and middens may well have been part of the food cycle. As Thomas (1996d) has pointed out, one of the ways people can make up for a loss of animal protein from large mammal resources is by more intensively exploiting the smaller mammals and the single domesticated resource, the dog. It has been suggested, albeit in a different cultural context (Snyder 1991), that the fattiness and palatability of dog meat may have made it an important alternate resource when wild game resources were at a nutritional low point. It is also possible that the presence of a large population of dogs in the village allowed for more flexibility in other subsistence options. However, if, at the same time, people at these villages were starting to rely more on corn and other domesticated plant resources, the dogs may also have allowed and/or encouraged people to spend more time in and around the village. Overall evidence from non-fish samples actually confirms the trend seen in the fish assemblages of increasingly local subsistence strategies at Dunsmore and especially Carson. The Barrie people, in
While deer remains are generally few in the precontact period, some variation exists, with Barrie having more deer than either Dunsmore or Carson (Table 34). This may be explained by the specifics of deer feeding behaviour. While deer thrive in forest ecotones, they will also readily browse on stands of corn. When viewed in this light, the presence of deer at a pioneer village makes perfect sense – they would have been attracted to the novelty of the corn fields. An alternative or complementary explanation is that deer were occasionally encountered during travels away from the village (Ron Williamson, personal communication 1999). The Barrie people were more than likely familiar with the deer populations in their area of origin, south of Oak Ridges Moraine, and may have obtained deer from there or from other adjacent areas. More substantial numbers of deer 57
bones from precontact sites in Victoria County point to larger deer populations there (Robertson et al. 1995).
BW is not available for the Lover’s Creek sites. The number of dog bones recovered is smaller at Wiacek, Hubbert and Molson than at Barrie, Dunsmore and Carson (Figure 26). Given that this medium-sized mammal is unlikely to be the subject of major taphonomic bias, the lesser emphasis at the Lover’s Creek sites may be real. The Molson zooarchaeological assemblage shows the marked increase in Salmonidae identifications that may mark the start of a large-scale autumn Lake Fishery. The contribution of Salmonidae is the more remarkable considering the fact that this chart only refers to cranial bones, which are arguably the most taphonomically-disadvantaged category of zooarchaeological remains at these sites. The fact that over half the vertebrae at Molson are Salmonidae (Figure 23) suggests an increasing emphasis on off-site processing. This may signal a focussed, large-scale exploitation of Salmonidae on Lake Simcoe or Georgian Bay.
What about the paucity of deer bones at Dunsmore and Carson? Even a modest level of exploitation by the Barrie occupants may have reduced the densities to such an extent that local deer hunting was no longer feasible. As Thomas argues (1996d:166), population numbers of such a valued resource, with high mobility and low reproductive rates, would be the first to suffer. Indeed, as was noted in Chapter 1, ethnohistoric sources suggest the contact period Huron appear to have obtained their deer over 100 km away, from areas just north of Lake Ontario (Heidenreich 1971:207). Subsistence change in the Kempenfelt Bay area Faunal abundance Is it possible to relate the differences in the zooarchaeological assemblages of these three sites to the cultural and physical environment? Do these differences relate at all to precontact human population movements in the area? It appears that these assemblages do, indeed, offer support for changes through time and space, especially when key interpretive taxa like Salmonidae, dog and deer are considered individually.
Diversity Perhaps the variety of species within these categories can tell us about changes in time and space. Bearing in mind that exploitation does not necessarily have to be strongly correlated with availability, zooarchaeological data may reflect only a portion of the fauna available in the vicinity of the site. Taxa may be absent from the assemblage for a variety of reasons: because they were not in the area at the time; because their remains did not preserve; because they were not recovered; or because the sample size is too small (Lyman 1995:374-375). A direct comparison of the number of mutually exclusive taxa represented is likely to reflect sample size. A measure of species richness that compensates for sample size circumvents some of these problems. Menhinick’s richness index (Magurran 1988) is only one of many possible calculations of richness.
Taphonomic and recovery bias implies that interpretations derived from class distributions are, of necessity, somewhat problematic. For example, Lennox, Dodd and Murphy (1986:166-169) have suggested that fish were consumed extensively at the Wiacek site because few mammals were available, and that the earliest Iroquoian settlement in the area may have impacted the local mammal population so drastically that later villages show a shift to fish within a short time. While this is an interesting hypothesis, it cannot be tested without a much better understanding of taphonomic differences between sites. Even within-class comparisons between fish species are problematic, since some elements are more robust and can be more confidently identified to an analytically useful taxonomic level.
DMn = S / N where S is the number of species recorded, and N is the total number of individuals summed over all S species. In this zooarchaeological application, the number of species in each assemblage was based on the lowest unique taxonomic level. If some material was identified to family level and other material in that same family was identified to species level, only the lowest taxonomic level, that of species, was counted, so as to avoid inflating the number of taxa represented (Horwitz 1996). N here equals the NISP. Only the three most numerous classes are included (fish, birds, mammals) (Figure 27).
It is almost certain that a comparison of relative class contribution under-represents the contribution of fish. The exact extent of this bias is unknown, and this likely varies by assemblage. Depending on feature sample sizes (the unit of aggregation in this particular case), fragment sizes, and taxonomic distribution, BW may present quite a different picture from NISP (Figure 24, 25). Large- and heavy-boned taxa, such as deer and black bear (Ursus americanus), appear relatively more important, resulting in a change of rank-ordering of the taxonomic categories. Most notable are the drastic increases in mammal contribution at Barrie and Carson. Quantification by BW really emphasises the contribution of whitetailed deer to the Barrie diet (and bone tool/ornament technology). Recovery methods at Carson should favour fish bones. The reduction in fish contribution at Carson may relate to the increased emphasis on dogs, who may have eaten the butchering and post-consumption refuse of fish.
Species richness is relatively low at Barrie (3.2), much higher at Dunsmore (4.2), and then slightly lower again at Carson (3.9) (Figure 27). The low values for Barrie may relate to the more limited range of environments exploited – a plausible explanation for a “pioneer” site. Species richness indexes for fish, however, are very similar. Indeed, so are the cumulative percentages of taxonomic abundance based on NISP (Figure 28). While the data points are technically non-continuous, they are presented here as a line graph rather than a bar chart to emphasize the similarity in 58
the shapes of the curves (note that Dunsmore has more data points because it has more species). Subtle differences in the curves notwithstanding, all three sites contain small numbers of a variety of species, more substantial numbers of a few other species, and much larger numbers of only two or three species. The real difference, of course is in which species occupy the upper end of the curve.
for a poor harvest or a shortage of venison.” Indeed, Thomas (1996a; 1996d) has argued that fishing for Salmonidae may have been a more predictable and reliable autumn activity than hunting deer. Both deer hunting and autumn lake fishing might take groups of people some distance from the village. However, while autumn fishing involved a predictable location and a fairly reliable return, deer hunters might have had to roam very far to find their prey, especially given the local limitations on deer productivity. This may have discouraged later occupants of the area from actively pursuing this taxon. This crucial difference between these two autumn resources may explain why Salmonidae fishing persists through time, albeit at a low level, and deer hunting is a minor activity, even at the Barrie site.
Species richness of all three classes is greater at Dunsmore than at Carson. It has been suggested by the excavator (Ron Williamson, personal communication 1999) that this may reflect the nature of the evolving occupation at Dunsmore between seasonal and permanent occupation. The restriction in the range of species exploited at the Carson site may relate to the increased emphasis on dog as a source of food.
Rather than the expected inverse relationship between the proportion of Salmonidae and the proportion of deer at Barrie area sites, the pattern is quite the opposite, especially when we include the vertebrae. Barrie and Molson have slightly higher number of both Salmonidae and deer (Figure 26). This appears to suggest that there was no scheduling conflict relating to these two autumn resources, and that the shift to Salmonidae was not the result of de-emphasis of deer or vice versa. The occupants of Dunsmore, Carson, Wiacek and Hubbert appear not to have exploited either autumn resource intensively (Figure 26). It is interesting to note in this context that the unpublished Holly site contained no deer (Ron Williamson, personal communication 1999). The numbers of deer bones identified on sites in the area between Lake Simcoe and Nottawasaga Bay are extremely low compared to adjacent areas (Robertson et al. 1995: Table 14).
It is interesting to note that the same temporal pattern pertains to the Lover’s Creek sites. The number of species identified at Wiacek is low compared to Hubbert and Molson. This is in part because analysis of the 1983 fish bones consisted of many family and genus level identifications (Lennox et al. 1986); analysis of the much smaller 1990 fish sample (Thomas 1993) was restricted to an inventory of species present and approximate NISP. If this collection had been treated in the same way as the other five collections, the species richness index would be somewhat higher, but the intensive flotation programme likely accounts for some of the non-specific identifications. Given the relatively short interval between occupation of the Wiacek and Hubbert sites, and the small sample size at Hubbert, it is interesting to note the dramatic increase in number of species exploited. The decrease in number of species exploited at Molson may relate to an increased focus on the autumn Lake Fishery.
The autumn Lake Fishery in Lake Ontario appears to have its origins at least in the late fifteenth to early sixteenth century at the Keffer (Stewart 1991a) and Over sites (Thomas 1996a; Thomas 1996c). In the Lake Simcoe area, a full-scale autumn Lake Fishery appears to have started a bit later. Considering the ubiquity of lake herring, lake whitefish, and especially lake trout vertebrae at Barrie, Dunsmore, Carson and Hubbert, it seems probable that the autumn component of the Lake Fishery was more important than the limited NISP suggests, but a comparison with Molson suggests that this fishery was not yet fully developed until the sixteenth or seventeenth century. Rather, these earlier sites may represent an early stage in the development of the autumn Lake Fishery. It could even be argued that all Salmonidae remains are the result of shallow water fishing, either in autumn or spring, or ice fishing. There is no need to evoke canoe journeys to the lake shoals to explain such small numbers. By the time the Molson site was occupied that all had changed. The inhabitants of the Molson site were exploiting Salmonidae on a larger scale during the autumn spawning season. While Salmonidae do come near the shore in autumn, mass exploitation during that season would have involved setting gill-nets from canoes, often in bad weather, as described so vividly by Sagard in 1623 (Kinietz 1965:28).
Changes in seasonal emphasis Cleland (1982) argues that the autumn gill-net fishery for Salmonidae (the autumn component of the Lake Fishery described above), with its restricted time frame and increased labour requirement for processing and preservation, resulted in even greater cooperation than that required for net fishing during the spring spawning-run. As Thomas (1996d) emphasises, fishing for spawning Salmonidae also entails a substantial social commitment in terms of the tools of production: lake-going canoes, cordage production for large nets, and mass food processing techniques. It also implies extensive exploitation of a different kind of habitat – the shoals of deep lakes. Certainly the autumn Lake Fishery differs from the other kinds of fishing activity, both qualitatively and quantitatively. The fall Lake Fishery would have been extremely productive, both in terms of volume of fish that could be obtained in a short period of time, their ability to be preserved by freezing and their nutritional value (Cleland 1982), especially lake trout (see Table 30). Chapdelaine (1993:189) has suggested that communal hunting of deer may have facilitated annual sedentariness, and that “fishing was probably the main strategy to make up 59
The increase in Salmonidae exploitation at the protohistoric Molson site fits well with what descriptions of historic period fishing from the ethnohistoric sources for the area. Fish exploitation at Molson was beginning to concentrate on an intensive autumn Lake Fishery, which became one of the most important economic institutions of the Huron some thirty-five years later (Sagard 1939; Thwaites 1896-1901; Trigger 1976).
element identifications. As was noted previously, the preliminary list of 8 diagnostic elements discussed in Chapter 3 should be compared in a systematic manner to other Great Lakes assemblages to establish to what extent the findings at the Barrie, Dunsmore and Carson sites hold true at other sites. It should be noted that the presence of freshwater salmon family fishes in an assemblage can go largely unnoticed unless fish vertebrae are analysed at least to family (see also Molnar 1997). While this is a somewhat time-consuming task, it does provide important clues to a fishery that differs both in kind and in scale from either the Spring Spawningrun Fishery, or the Generalized Warm Weather Fishery.
Concluding remarks and recommendations for future research To conclude, I offer some further thoughts on methodology. Throughout this work it has been stressed that taphonomic issues must be understood before an attempt is made to interpret changes in time and space. Useful insights in taphonomy are gained from a consideration of BW, MNI, diagnostic elements, scales, vertebrae, and element size. Details on the “unidentified” component (NUSP) are important.
Age-at-death interpretations from fish vertebrae have been used to establish MNI at the McKeown site, and Iroquoian village near the St. Lawrence River (Stewart 1992). It has been shown here with walleye scales and brown bullhead spines that season of capture data can be used to further refine MNI estimates. A thorough understanding of the principles of fish growth, combined with CSAGES age and growth interpretation, greatly increases the information potential of several kinds of archaeological fish elements that are often under-utilised. While the use of incremental structures was recommended by Thomas several years ago (1993), the current research presents the first systematic attempt at age and growth studies on an Ontario site. An understanding of fish growth indicates that there are some limitations to the procedure, especially with older fish that were killed during the period of growth cessation.
It must be acknowledged that bias is introduced by the actual method of recovery (dry screening vs. water screening vs. floatation), screen sizes, and how consistently the material is recovered in the field. It would be useful for future research to rigorously compare fish recovery between different methods at Iroquoian sites, in particular fine mesh dry screening vs. wet screening, so that future excavations can take account of optimal recovery techniques and sampling design, not only for ceramic, lithic and palaeobotanical remains, but also for zooarchaeological remains. Larger sample sizes and increased fish bone and scale recovery would allow for more robust findings. While optimal recovery was not investigated in detail in this thesis, it seems likely that both the quality and quantity of data could have been improved by large-scale flotation and/or fine mesh water screening. Recommendations in favour of floatation in an Ontario context were made previously by Stewart (1991b) and Prevec (1985). While it is beyond our power to control for taphonomic bias resulting from food preparation, and dog scavenging, additional, avoidable bias resulting from recovery process must be avoided.
The success of the scale analysis at Dunsmore and Carson was due in part to the fact that there were so many scales from a single taxon. The key questions of when and where these fish were caught – during the spawning-run, or during other times of the year, perhaps more opportunistically or incidentally – could be answered, even though interpretation of early spring catches is considered more difficult (NeedsHowarth and Brown 1998). CSAGES resulted in a confident sub-species identification. It also indicated that the dates of capture were tightly clustered, and that the age distribution was highly skewed. The CSAGES data also confirmed a general trend towards smaller fish at Carson. Numerous lines of evidence suggest this trend is not simply a reflection of the smaller dry screen mesh size used at Carson. If this trend is a result of extended fishing in the same locations over a period over several decades, it supports the relative chronology of site sequences. Comprehensive age and growth studies by fisheries biologists can help to elucidate some of these issues. Prerequisites are optimal recovery of fish scales and a substantial financial commitment.
An important component to understanding recovery bias is soil volumes (screened as well as floated), as was demonstrated 15 years ago by Junker-Andersen (JunkerAndersen 1984). An exploration of more comprehensive MNI estimates that take into account element size as well as L/R matches and element duplication show that only three or four cranial bones of each yellow perch survived to be identified. The number of vertebrae recovered was even lower. This issue deserves further attention. Perhaps MNI to NISP ratios for other taxa are much lower, but the large yellow perch deposit at the Dunsmore site seem to have suffered an astounding amount of taphonomic loss.
An understanding of fishing strategies was facilitated by a consideration at the feature level and site level of detailed species identifications, fish biology, scales, vertebrae, element size, age and season of capture. As was noted in Chapter 5, some components of this approach have been
An investigation of diagnostic elements indicates a lot of intra- and inter-species, as well as inter-site variation in fish 60
employed by other researchers working with freshwater fish remains in a North American context. The multidisciplinary, comparative approach advocated here offers some different and additional ways to understand fish subsistence strategies. Differences in fish faunal assemblage composition that may reflect differences in timing of procurement events and refuse disposal were examined at the feature level. Proportional element size and osteometrics proved helpful not only in elucidating taphonomy, but also in establishing the nature of fish catches. Earlier discussions on parts of this method resulted in some preliminary conclusions (Needs-Howarth and Thomas 1998), primarily that the spatial and seasonal patterning of precontact fish exploitation cannot be explained only in terms of habitat preferences and spawning dates of the various species involved. The current research has confirmed that it is not possible to identify in detail these different kinds of fishing events by restricting analysis of the fish assemblage to the site as a whole. Fish procurement activity must also be investigated at the level of the single feature or feature stratum, wherever possible (Needs-Howarth and Thomas 1998).
economy outlined by Cleland for the Late Woodland (Cleland 1976), which relies on high quality, abundant resources that are consistently available. Resource reliability is key, and was provided in an Iroquoian context by a combination of hunting, fishing and plant cultivation. Temporal and spatial variation within this broad pattern is to be expected (Cleland 1976); fish subsistence evidenced at these six precontact sites, ranging in time from the end of the thirteenth century to the beginning of the seventeenth century, seems to show such variations. The Dunsmore, Carson and Hubbert assemblages appear somewhat less focal than the Barrie, Wiacek and Molson assemblages. Long-distance deer hunting, and an emphasis on mass exploitation of deep water fishes suggest subsistence strategies of the early contact period Huron were yet more focal, involving more specialized fishing gear and perhaps a better knowledge of animal movements and behaviour. It appears, therefore, that subsistence patterns documented by explorers and missionaries in the early contact period shows clear similarities with the general precontact subsistence pattern. What has become very obvious through this research is that many variations exist within this general subsistence precontact pattern. It will always be a challenge to separate out taphonomy, environment and human preference, but the multidisciplinary approach used in this thesis goes some way towards providing answers. New insights can be gained from even modest samples of fish bones.
More can be done. This research has shown the potential of fish element size data on Great Lakes sites. Accurate reconstructions of fish size would lend allow for more convincing arguments to be made regarding size distributions and method of capture. While building up a database of regression equations derived on modern specimens takes a lot of commitment, it is something Great Lakes zooarchaeologists should strive for in the immediate future. Obtaining DNA signatures also merits further consideration. It would be an accurate way of establishing which body of water the fish stock originally derived from. Lastly, it is important to recognize that fish remains at the villages may be only part of the fish procurement pattern. Zooarchaeological samples from contemporaneous fishing camps are a vital missing link. Given the location of the sites, and the wide variety of fish exploited, it appears that a lot of fish were captured locally. Some of the fish represented at the Barrie site may have been caught at a temporary camp on Nottawasaga Bay, although the large number of vertebrae recovered from the village, suggest most fish were brought back to the village whole. Despite shortcomings in the datasets and the fish size methodology, the different approaches presented here allow us to come to new insights on the nature of precontact subsistence. The Three Fisheries Model points to inter-site differences in precontact fishing strategies, especially between the Barrie site, and the Dunsmore and Carson sites. The assemblages discussed above provided some details on the origins of a large-scale autumn Lake Fishery. Among the Huron, fish were a staple, while fowl and game meat were a welcome seasonal supplement (Heidenreich 1972:73).Considering the biassing effects of taphonomy, the class distributions at Barrie, Dunsmore and Carson, and the Lover’s Creek sites suggest that fish were important also in the precontact period. These sites conform to the focal 61
SUMMARY
This thesis presents several lines of investigation, which, in combination, offer a detailed way to investigate precontact fishing strategies in the Great Lakes area. Questions to be answered include when, where, and how fish remains originated. Palaeoenvironmental, biogeographical and fisheries science data are used to understand the ancient landscape, fish habitat and distribution, and fish behaviour. Descriptions in the ethnohistoric literature, written by missionaries and explorers in the seventeenth century, are used to understand the techniques and social customs surrounding fishing practices in the early contact period. The main body of information, however, is zooarchaeological. This thesis develops ways of getting more information from small collections of fish bones by looking at species distribution; co-occurrence of fish taxa; fish bone size and state of sexual maturity; and age and season of death. This information is considered at the level of individual archaeological deposits, as well as at the site level.
and the Carson site, dating to ca. A. D. 1475-1525, were part of a proliferation of permanent villages in the area. These two later sites are located close together, about 4 km north of the Barrie site, near Little Lake and Willow Creek, which are part of the Nottawasaga River drainage. Both historic reports and more recent ethnographies emphasize the importance of fishing to the Iroquoian people in the area. Until now, this has not really resulted in the necessary refocusing of (zoo)archaeological research aims and methods to include ways of dealing with fish remains in detail, using additional and different approaches to those used on other classes, such as birds and mammals. Fish remains are not necessarily a passive reflection of local availability or ease of capture. In order to understand the nature of fish subsistence strategies, we have to examine collections of fish bones in more detail, going beyond traditional bone fragment counts. The methodological potential and limitations of the fish component of the collections are discussed in detail because many of these issues have not been previously explored in an Ontario context. The discussion focusses on fish remains; discussion other classes of animals is limited to those aspects that may help understand taphonomic or subsistence issues relating to the fish component of the assemblages.
The multidisciplinary approach developed in this thesis allows an understanding how people in the past scheduled their time, energy, material and labour resources. Inter- and intra-site differences in fishing strategies are investigated at three communities of Iroquoian-speaking people who lived between Lake Simcoe, and Georgian Bay of Lake Huron, Ontario. These sites range in date from the end of the thirteenth century to the beginning of the sixteenth century A. D. The occupants are the cultural antecedents of the Huron, who were living in the area at European contact.
The Barrie, Dunsmore and Carson sites were salvage excavated in the last 15 years. The zooarchaeological samples derive from refuse deposits in houses and from external middens. These assemblages are small (number of fish bones identified below class, excluding scales and vertebrae, is 380, 665, 558, respectively), with a great diversity of species. Small sample sizes, especially at the feature level, are a problem. In addition, the assemblages were recovered with differing intensities of flotation and dry screening, and differing sieve mesh sizes. It was necessary, therefore, to develop ways of sorting out taphonomic differences from real differences in resource exploitation. Prior to discussing fishing strategies, therefore, this thesis presents a detailed comparative analysis of site-specific taphonomic issues, with an emphasis on fragment sizes, fish element sizes, fish vertebrae and scales.
The first two chapters place the research in its environmental and cultural context. The Iroquoian food economy was based on slash-and-burn maize horticulture, gathering, hunting and fishing. Settlement in the contact period consisted of permanently occupied villages with many multi-family longhouses and large external refuse deposits; satellite villages; hamlets and special purpose camps for activities such as corn horticulture and fishing. Most of this settlement pattern probably already existed in the precontact period. In the precontact period, sites were probably occupied for a period of about 20-30 years, after which time the community moved to a new location. Corn horticulture and village life in Ontario have their origin in the first millennium. Unlike in other areas of Ontario, there is no in-situ development of Iroquoian villages in the area between Lake Simcoe and Georgian Bay. The Barrie site represents a community that migrated to the area between Lake Simcoe and Georgian Bay from the north shore of Lake Ontario around A. D. 1280. This site, occupied between A. D. 1280-1330, is the earliest longhouse village known in the area. It was located on Dyment's Creek, at the head of Kempenfelt Bay of Lake Simcoe, and close to Minesing Swamp and the Nottawasaga River, which drains into Georgian Bay of Lake Huron. The Dunsmore site, dating to the period ca. A. D. 1430-1510,
The third chapter discusses laboratory identification and computer-based quantification of the zooarchaeological assemblages. The samples are quantified as BW and bone counts. These abundance measures are qualified by a discussion of eight so-called “diagnostic elements” that are readily identified to species even when broken: articular, ceratohyal, cleithrum, dentary, hyomandibular, operculum, preoperculum and quadrate. It is argued that other cranial bones are unevenly represented because of differences in osteology, mechanical strength and associated susceptibility to fragmentation and other taphonomic factors, rather than small sample size. This preliminary discussion on diagnostic 62
elements highlights the over-representation of, for example, pectoral spines and vertebrae surrounding the air bladder, and problems of quantifying lake sturgeon remains.
Reliance on current biogeographical data for species presence or absence in the past results in some circular arguments. In order to understand what kinds of habitats were exploited, it is useful to employ habitat preference studies. A tabulation of habitat preferences at each site displayed considerable differences. The taxonomic distribution at the Barrie site suggests an emphasis on open water, large bays and estuaries, while those at Dunsmore and Carson show an emphasis on large bays, estuaries and coastal marshes.
The fourth chapter deals with various aspects of taphonomy: fish butchering, processsing, consumption, discard, burial and recovery. Vertebrae were identified to the level of family in order to identify fish processing. Expected ratios of cranial bone to vertebrae must take into account the range of cranial elements that can be identified, as well as the range of vertebrae in different fish species, and their sizes in relation to the sieve mesh aperture. In general, vertebrae were under-represented with respect to cranial bones, suggesting that catch site butchering was limited. Iroquoian cooking methods relied heavily on stews. Fish bones were likely discarded with the head, or included in the cooking process. Many of the fish bones originally deposited would be unlikely to be preserve or recovered. Dogs were likely a major taphonomic agent at each village, but especially at Carson, which produced a lot of dog bones. Burial conditions at the Barrie, Dunsmore and Carson sites were favourable to the preservation of all bone, including, at Dunsmore and Carson, fish scales.
Since species were not equally available throughout the year, however, seasonal variation in fish behaviour must also be considered as well as habitat. Both active and passive fishing techniques can be expected to have left signatures in terms of quantity of remains, co-occurrence of taxa, and, especially, fish element size. For example, fish caught with techniques of mass capture would probably show somewhat restricted size range in each depositional context. The 25 or so fish species represented at each site exhibit great seasonal variation in habitat and behaviour, and for most species a spawning-run catch implies something very different in terms of fishing strategies than does a nonspawning-run catch. It was, therefore, necessary to somehow distinguish mature from immature fish in species that can be readily caught both during the spawning season and outside of it. The second application of fish element size, therefore, is as an approximation of state of maturity, in order to identify probable spawning-run catches. For example, if the reference specimen of species A is 30 cm long, and the size at maturity for both sexes in local waters is 15 cm, it can be argued that fish cranial bones less than half the size of the comparable measurement in the reference specimen probably belong to immature individuals. Macroscopic size observations and osteometrics can thus be used as a general guide for maturity and hence time, location and mode of capture. Mature fish may have been part of a spawning-run catch, whereas most immature fish were not. Immature fish could still have been caught at the time of the spawning-run, but for most taxa this represents a different intensity of fishing effort, in a different location.
BW was used to assess fragment sizes of the different taxonomic classes. Since individual fish bones often weigh very little, taphonomic issues within this class had to be assessed in a different manner, using fragmentation rates and fish element size. Fish element size was estimated using a proportional method, whereby the relative size of elements is expressed as a percentage of that same element in the reference specimen of known size. Wherever possible, these size observations are augmented by osteometrics. By back-calculating the osteometrics to a size percentage in relationship to the comparable metric in the reference specimen, it was possible to utilize all fish element size data and thus increase sample size. These data were used to document differences in fish element sizes between sites and between recovery methods. For example, fish cranial elements from floatation heavy fraction were larger at the Barrie site than at Dunsmore and Carson, despite similar heavy fraction sieve sizes. As expected, flotation has resulted in better fish recovery. At Dunsmore and Carson it has resulted in retrieval of smaller fish remains. At Carson this was unexpected, since the dry screen mesh size and heavy fraction sieve size were almost identical. There are surprisingly few differences between the screened components from Dunsmore and Carson, despite a 3.2 mm difference in mesh aperture.
Working at the feature level, we examined differences in fish faunal assemblage composition that may reflect differences in timing of procurement events and refuse disposal. This suggests that, while some taxa were probably exploited mostly during their spawning-run, other taxa were exploited throughout the warmer months, including, but certainly not restricted to, their spawning season.
Chapter 5 presents approaches from various biological, environmental, historical and archaeological subdisciplines that can help interpret archaeological fish remains. The ethnohistorical sources contain some useful details on technique of capture for fish, involving nets, canoes, and spears, although it is recognized that the described events may have been atypical, and that precontact fishing may have differed substantially from that of the early contact period.
Calcified structures, such as scales and pectoral spines, offer the potential of establishing age structure of the catch, as well as season of capture. The age of the fish can be used to infer whether the individual was sexually mature or not. This is a major benefit, since age is a more reliable indicator of maturity than size. Using these diverse sources of information, fish remains 63
from the Barrie, Dunsmore and Carson sites were assigned to one of three fisheries complexes: 1) Spring Spawning-run Fishery: a watercourse-oriented inland fishery that focuses on intensive exploitation of spring-spawning taxa such as the lake sturgeon, white sucker, longnose sucker, yellow perch and walleye; 2) Generalized Warm Weather Fishery: a generalized bay or inland fishery for opportunistic warm weather exploitation of resident taxa that do not aggregate in harvestable quantities during their spawning-runs, such as pikes, brown bullhead, members of the Sunfish family, and of immature and non-spawning yellow perch; and 3) Lake Fishery: a lakeoriented fishery on Kempenfelt Bay and Nottawasaga Bay that included inshore exploitation of autumn-spawning Salmonidae.
mature individuals, caught in (early) spring. This conforms with walleye behaviour and biology, which make it susceptible to mass catches during its spawning-run in midApril. Currently, this species comes up the Nottawasaga River to spawn in Willow Creek. The survival of so many walleye scales and almost none of any other taxon is hard to explain, although relative strength and size must play a role. What is even harder to explain is the paucity of walleye cranial bones. Since there are few larger perch family vertebrae, it seems likely that walleye were being processed at the catch site. The skin from the fillets, with adhering scales, may have been deposited in refuse dumps after return to the village. In general, fishing events at Dunsmore appear to have been less concentrated on the Spring Spawning-run Fishery than at Barrie. The cranial bones can all be interpreted in terms of the Generalized Warm Weather Fishery or the autumn component of the Lake Fishery. Even the spawning-run exploitation of walleye is congruent with a local fishing effort.
Fish may be found together in a deposit because they inhabit the same waters and/or they spawn together and/or they are amenable to the same techniques of capture. Chapter 6 provides details of the fish assemblages, and discusses how fish remains from individual features and from the site as a whole relate to the Three Fisheries Model. The Barrie site fish assemblage is dominated by lake sturgeon and yellow perch. The most productive and predictable place for a sturgeon fishing expedition would probably be at or close to the mouth of the Nottawasaga River during the spring spawning-run. The size distribution of yellow perch suggest that the majority were sexually mature when caught, and the contents of certain features appear to be the result of masscapture events. The ratio of cranial bone to vertebrae may indicate that more off-site processing was practised at Barrie than at the later two sites, and thus suggest a substantial exploitation of waters away from the site, which may have made catch site decapitation more desirable. Fishing at Barrie appears to make use of rivers and the lake shore, rather than the local stream.
The fish bones at Carson are dominated by brown bullhead, pumpkinseed and yellow perch. The yellow perch are slightly smaller than at Dunsmore. All major features with northern pike also contain brown bullhead and yellow perch. The peak in the size distribution of northern pike probably represents immature individuals. The consistent cooccurrence of northern pike with brown bullhead suggests that pike were exploited throughout the warm seasons as part of the Generalized Warm Weather Fishery, probably mostly in small fish-traps, rather than during their spawningrun. Only one feature contains unequivocal cranial bone evidence for the Spring Spawning-run Fishery, although selected fish scales from two further contexts indicate a substantial spring spawning-run exploitation of walleye. The first part of Chapter 7 compares the fish bone assemblages from Barrie, Dunsmore and Carson to suggest inter-site differences in fishing strategies and processing. Fish cranial bone sizes are on average smaller at Dunsmore and Carson. Since recovery at Barrie was more favourable than at Dunsmore, we may assume that differences in average fish element sizes between these two sites are not entirely the result of taphonomic factors; the larger fish element sizes at Barrie may be the result of a greater emphasis on Lake Simcoe. In this respect it is informative to compare Dunsmore and Carson, which are relatively close in time and space. Relative size observations and osteometrics on cranial bone, as well as age-at-death data from spines and scales suggest that differences in fish element size and age between Dunsmore and Carson are not solely a function of differing dry screen mesh sizes. Osteometrics of brown bullhead opercula and growth studies of pectoral spines show a trend towards smaller/younger fish at Carson, both in terms of size distribution and average age. If these fish were caught in Little Lake, it may suggest that fishing in this limiting environment was causing a decline in average size. A similar decline in cranial bone size through time is observed in yellow perch. A decline in mean size caused by fishing pressure is most readily acceptable if the
At the Dunsmore site yellow perch is the most important taxon, followed by brown bullhead, Pumpkinseed and northern pike. The average size of brown bullhead cranial bones suggests they belonged to mature individuals. In contrast, the northern pike are probably mostly not sexually mature. The yellow perch are on average somewhat smaller than those at Barrie, and represent a wider range in total length. The difference is most marked in the heavy fraction, suggesting that more intensive flotation would have weighted the assemblage further in favour of smaller individuals. Unequivocal evidence for spring exploitation comes from age and growth analysis on brown bullhead spines (although this does not necessarily represent spawning-run exploitation) and Percidae scales. Analysis of these scales was carried out in conjunction with a fisheries biologist. The substantial numbers of yellow perch cranial bones initially led to the assumption that the scales derived from this species. Absolute scale size, absolute age, age distribution, and growth pattern, however, suggested a related taxon, the walleye. All analysed scales from the Dunsmore and Carson site were of older, sexually 64
site occupants were exploiting the same population of yellow perch, probably in the lagoon at Kempenfelt Bay, or the tributary streams. The Three Fisheries Model, however, suggest an alternate explanation: the yellow perch at Barrie were mostly obtained during the Spring Spawning-run Fishery, whereas those at Dunsmore and Carson were mostly obtained during the Generalized Warm Weather Fishery, which is to hypothesized to have included larger numbers of smaller fish. There is also a decline in average age of walleye scales. While there are many large fish represented at either site, the decrease in average age between Dunsmore and Carson may indicate slightly increased pressure on the walleye populations.
increases substantially. Despite these taphonomic issues, however, it appears that only in the seventeenth century, with the occupation of the Molson site, do we see the start of the intensive autumn lake fishery that became one of the most important aspects of the economy during the early contact period. The autumn lake fishery differs from other kinds of fishing activity, both qualitatively and quantitatively. This fishery would have been extremely productive, both in terms of volume of fish that could be obtained in a short period of time and their nutritional value (especially lake trout). Analysis of non-fish taxa provides support for the trends identified above. The occupants of the Barrie site appear to have had a more seasonally focussed fishing strategy and a concurrent focus on birds and mammals preferring riverine environments and deciduous /coniferous forest. This may indicate that there were few clearings or secondary growth areas in the vicinity of the Barrie site. It was expected that continued human occupation of the area around Kempenfelt Bay would have resulted in more evidence of forest clearance and forest edge habitats at Dunsmore and Carson. Indeed, the emphasis on local, predictable, small-unit resources at Dunsmore suggests these resources were specifically targeted. Contrary to expectations, the Dunsmore and Carson sites contained fewer of these disturbed habitat species than the Barrie site. This finding is less contradictory than it seems. The pristine environment may have offered more subsistence options, however, the lack of forest edge and clearance habitats may have directed subsistence efforts towards a more limited range of habitats. Thus, while the Barrie people travelled further for their fish, they also small mammals that were attracted to the horticultural activity around the site. In this respect relative representation of deer bones is informative. The fact that this area of Ontario provides little suitable deer habitat is reflected in small numbers of deer bones at all three sites. However, there are substantially more at the Barrie site than at the later sites. While deer thrive in forest ecotones, they will also browse on stands of corn. The presence of deer is, therefore, compatible with the pioneer status of the Barrie village – deer would be attracted to the novelty of the newly created corn fields. They may also relate to the hypothesized greater mobility of the Barrie people, enabling them to encounter deer in adjacent areas that had larger deer populations. The abrupt decline in deer representation at Dunsmore and Carson could have been in part the result of the hunting efforts of the Barrie people; they may have reduced the already minimal deer populations to such an extent that hunting them was no longer feasible. Alternatively, the deer at the Barrie site may have derived from areas further south.
Decrease in net mesh apertures is an unlikely explanation for the size differences, since the nets were handmade from plant fibres, and probably displayed a lot of variation in mesh aperture within a single net. Local fisheries biologists have suggested that fish community structure could change quite rapidly, even with modest levels of exploitation, especially in a body of water such as Little Lake, or under heavy spawning-run exploitation. A decline in average fish size in the population can only be explained if the Dunsmore and Carson people were using the same location to fish. If the fish element size decline is a result of extended fishing in the same location over a period of maybe 50 years, it may support the relative chronology of site sequences based on the radiocarbon dates. When combined, the evidence appears to support differences in location of fish capture. Lake sturgeon bones at the Barrie site suggest exploitation of the Nottawasaga River. Yellow perch at all three sites were likely obtained from Kempenfelt Bay of Lake Simcoe and tributary streams. The large numbers and larger element size of yellow perch suggest a more intensive exploitation of Lake Simcoe and tributaries at the Barrie site. Both the species and size/age distributions of the remaining taxa indicate a growing emphasis on Willow Creek and Little Lake at the Dunsmore site and, especially, at the Carson site. For comparative purposes the three sites near Kempenfelt Bay were compared with three precontact sites located on Lover’s Creek, south of Kempenfelt Bay, the Wiacek, Hubbert and Molson sites. Large numbers of suckers at these sites appear to indicate a more consistent, stronger emphasis on spring time exploitation and/or lake fishing. Most striking, however, are the large numbers of lake trout, lake herring and lake whitefish at the Molson site, which was occupied in the last decades before contact. These freshwater Salmonidae were probably more important at all these sites than the limited cranial bone counts suggest. Cranial bones may have been left at the catch site, or they might not have preserved well because of their oily, fragile nature. Vertebrae, on the other hand, would be brought back to the site, either in a whole fish, or as part of a fillet. When vertebrae are added to the cranial bone counts, Salmonidae representation at the Barrie site, for example,
A major difference between these three sites is seen in the reliance on dog, which is much greater at Carson than at the other two sites. People can make up for a loss of animal protein from large mammal resources by more intensively exploiting the smaller mammals and the one domesticated resource, the dog. Dogs may thus have allowed and/or encouraged people to spend more time in and around the 65
village. This fits with the interpretation of the deer data, and the more local focus of fishing efforts. Species diversity also fits the fish and mammal data. Species richness indexes for fish are very similar at all three sites. Species richness of mammals and birds reflects the more limited range of environments exploited at the Barrie site, again confirming its “pioneer” status. Richness of fish, bird and mammal is greater at Dunsmore than at Carson. This may relate to the relative site chronology. The Dunsmore people were exploiting a more pristine environment than the Carson people in the same general catchment area and may, therefore, have been slightly less focussed on resources available in and around the village. Or the greater species richness may relate to functional differences within the occupation at Dunsmore. The decrease in number of species exploited at the Molson site, the latest of the Lover’s Creek sites, may relate to an increased focus on the autumn Lake Fishery. Among the contact period Huron, fish, especially Salmonidae, were a staple, while meat of mammals and birds was a welcome seasonal supplement. This in contrast to the more diffuse strategies of the precontact period, with an emphasis on shallow water fish and smaller mammals. The fishing economy of the Barrie site is more focal than that of Dunsmore and Carson. This is evident from the taxa exploited, their sizes, and the richness and evenness of the fish assemblage. It appears that the economy documented in the early contact period can be viewed as an extension of the general precontact subsistence pattern. What has become obvious through this research is that there are many variations within this general precontact pattern. It will always be a challenge to separate out taphonomy, environment and human preference, but this multidisciplinary approach has provided some new insights.
66
SAMENVATTING
"De inheemse visserij in het Grote Meren gebied: een multidisciplinaire benadering van zooarcheologische resten van pre-koloniale Irokeese dorpen nabij Lake Simcoe, Ontario"
Maar anders dan in andere delen van Ontario, is er in het gebied rond Lake Simcoe geen in-situ ontwikkeling van nederzettingen. De Barrie nederzetting, de vroegste van de drie die hier besproken worden, vertegenwoordigt mensen die, rond A. D. 1280, naar dit gebied migreerden vanuit het gebied net ten noorden van Lake Ontario. Deze nederzetting, die werd bewoond ergens tussen A. D. 1280-1330, is het eerste longhouse dorp in dit gebied. De nederzetting ligt aan een kreek, net ten oosten van Kempenfelt Bay van Lake Simcoe, en dichtbij Minesing Swamp en de Nottawasaga rivier, die uitmondt in Georgian Bay van Lake Huron, ten westen van de nederzettingen. De Dunsmore nederzetting, bewoond in de periode A. D. 1430-1510, en de Carson nederzetting, bewoond ergens tussen A. D. 1475-1525, maken deel uit van een verspreiding van dorpen in dit gebied. Deze twee latere nederzettingen liggen dicht bij elkaar, ongeveer 4 km ten noorden van de Barrie nederzetting, bij Little Lake en Willow Creek, die deel uitmaken van het stroomgebied van de Nottawasaga rivier.
In dit proefschrift worden een aantal verschillende benaderingen besproken die tezamen de mogelijkheid geven om, in detail, de visvangst-strategieën in de periode vòòr Europese kolonisatie te onderzoeken. De vragen die in deze context beantwoord moeten worden zijn: wanneer, waar en hoe de visresten op archeologische opgravingen hun origine hebben. Paleoecologische, biogeografische and visserijbiologische data worden in dit proefschrift gebruikt om het vroegere landschap te begrijpen, alsmede de natuurlijke leefomgeving, de verspreiding en het gedrag van de lokale vissoorten. Etnografische beschrijvingen van 17de eeuwse missionarissen en ontdekkingsreizigers worden gebruikt om de technologie en sociale context van de visvangst in de vroegste periode van Europese kolonisatie te begrijpen. Het grootste deel van dit proefschrift heeft betrekking op zooarcheologische vondsten. In dit proefschrift worden methoden ontwikkeld om meer informatie te verkrijgen over kleine collecties visbotten door een combinatie van soorten distributie; het tezamen voorkomen van soorten; visbot grootte; de leeftijdsgegevens van de vissen; en het seizoen waarin zij stierven. Deze informatie wordt vergaard zowel van individuele archeologische contexten, als van de nederzetting als geheel.
Het belang van de visvangst voor de Irokeese bevolking van dit gebied ten tijde van Europees contact wordt benadrukt in, zowel historische beschrijvingen als, meer recente etnografieen. Tot voor kort heeft dit nog niet geleid tot een nodige herwaardering van (zoo)archeologische doelstellingen en methoden zodat visresten in detail worden onderzocht met een andere benadering, dan die gewoonlijk van toepassing is op andere zoologische klassen, zoals vogels en zoogdieren. Om visvangst-strategieën te begrijpen moeten collecties van visresten worden onderzocht op meer dan alleen de aantallen botfragmenten. Zowel het methodologisch potentieel als de limitaties van visresten worden uitvoerig beschreven, omdat veel aspecten zo nog niet eerder zijn benaderd in Ontario. Dit proefschrift spitst zich toe op de visresten; de discussie van andere dieren is beperkt tot alleen die aspecten, die verheldering kunnen geven over taphonomie of proviandering, die te maken hebben met de visresten.
De multi-disciplinaire methode die in dit proefschrift ontwikkeld wordt maakt het mogelijk om beter te begrijpen hoe mensen in het verleden hun tijd, energie, materiaal en arbeid regelden. Verschillen in visvangst-strategieën worden besproken zowel binnen als tussen drie nederzettingen van Irokees-sprekende mensen die woonden in het gebied tussen Lake Simcoe en Georgian Bay van Lake Huron, in Ontario. Deze nederzettingen werden bewoond in de periode aan het eind van de 13de eeuw tot het begin van de 16de eeuw A. D. De bewoners zijn verwant aan de Huron, die in het gebied leefden ten tijde van het eerste Europese contact.
De Barrie, Dunsmore en Carson nederzettingen werden opgegraven als nood- opgravingen in de laatste 15 jaar. De zooarcheologische collecties komen uit (afval)putten in huizen en in de open lucht. Het aantal visbotten (minus schubben en wervels) dat tot het niveau van taxonomische familie of lager kon worden gedetermineerd is 380, 665 en 558, respectievelijk, terwijl het aantal vissoorten rond de 20-25 ligt. Het kleine aantal botten in individuele contexten is problematisch. Ook werden de botten verzameld met verschillende intensiteit van flotatie en zeven en met verschillende maaswijdten. Voor Barrie en Dunsmore gebeurde het zeven in het veld met 6.4 mm maaswijdte, flotatie met 2 mm; voor Carson was dit 3.2 mm, respectievelijk, 2.4 mm. Het was daarom nodig om taphonomische verschillen te proberen te onderscheiden van verschillen in vis-exploitatie, zowel binnen een nederzetting als tussen de drie nederzettingen. Dit gebeurt in Hoofdstuk
De eerste twee hoofdstukken plaatsen het onderzoek in een landschappelijke en culturele context. De voedseleconomie van de Irokezen was gebaseerd op het slash-and- burn verbouwen van maïs, verzamelen, jagen en vissen. In de periode van het eerste Europees contact waren er permanent bewoonde dorpen met veel longhouses waarin meerdere families samenleefden; kleinere satelliet-dorpen; gehuchten en speciale kampementen voor maïsverbouwing of visvangst. Een groot deel van dit patroon bestond waarschijnlijk al in de periode vòòr Europees contact. In die periode werden nederzettingen waarschijnlijk bewoond voor 20 of 30 jaar, waarna de bevolking naar een nieuwe lokatie verhuisde. In Ontario ligt de oorsprong van maïsverbouwing en nederzettingen in dorpen in het eerste millennium A. D. 67
4.
datzelfde element in het skelet in de referentiecollectie, waarvan de totale lengte bekend was. Waar mogelijk werden deze schattingen aangevuld met metingen. Door het terug rekenen van de metingen, in relatie tot dezelfde meting bij het skelet in de referentiecollectie, was het mogelijk om gebruik te maken van alle elementen per soort in een grafiek en op die manier de monsters te vergroten. Deze gegevens werden gebruikt om verschillen in visbot grootte op te sporen tussen de verschillende contexten, tussen de drie nederzettingen en tussen de verschillende botverzamelmethoden.
Hoofdstuk 3 geeft aan hoe het materiaal werd gedetermineerd en gekwantificeerd. Ook wordt beschreven hoe visbotten niet gelijkmatig vertegenwoordigd zijn vanwege verschillen in osteologie, mechanische sterkte en daarmee samenhangende fragmentatie en andere taphonomische factoren. Het gebruik van aantallen gedetermineerde botfragmenten (NISP) als methode van kwantificering wordt gekwalificeerd met een discussie over acht zogenaamde "diagnostische elementen" die gemakkelijk/betrouwbaar tot soort kunnen worden gedetermineerd, zelfs als ze gebroken zijn: articulare, ceratohyale, cleithrum, dentale, hyomandibulare, operculare, preoperculare en quadratum. Deze discussie over "diagnostische elementen" attendeert ons op de over-representatie van, bijvoorbeeld, de pectoral spine en de wervels rond de luchtblaas in de familie Ictaluridae en problemen in het kwantificeren van steur-resten. De validiteit van deze elementen moeten worden geverifieerd op andere zooarcheologische collecties in het grote meren gebied.
Zoals verwacht, resulteerde flotatie in betere werving van de visbotten. In Dunsmore en Carson resulteerde het in het winnen van botten van kleinere vissen, vergeleken met het zeven in het veld. Dit was onverwacht, gezien de maaswijdten op Carson minder dan 1 mm verschillen. Ook onverwacht was het feit dat er slechts kleine verschillen zijn tussen het materiaal dat op Dunsmore en Carson in het veld gewonnen werd, terwijl hier de maaswijdte juist sterk verschilde. Deze bevindingen duiden er misschien op dat de opgravers gefragmenteerde en heel kleine visbotten op Carson niet uit de zeef hebben herkend.
Hoofdstuk 4 beschrijft verschillende aspecten van taphonomie: butchering, processing, consumptie, afvalverwerking, burial en recovery. Viswervels werden gedetermineerd tot het taxonomisch niveau van familie om processing te beschrijven. De verwachte verhouding, tussen botten uit de viskop aan de ene kant en viswervels aan de andere kant, moet rekening houden met: het aantal kopbotten dat normalerwijs gedetermineerd kan worden en variaties in hoeveelheid van de wervels bij verschillende soorten, zowel als met: de botgrootte in relatie tot de maaswijdte. Wervels in de zooarcheologische collecties waren in het algemeen onder-vertegenwoordigd vergeleken met kopbotten. Dit geeft misschien aan dat bewerking van de vissen op de vangst-lokatie niet vaak plaats vond. De maaltijden van de Irokezen bestonden voornamelijk uit soep of stamppot-achtige gerechten. De vissenkop werd waarschijnlijk, ofwel weggegooid, ofwel in zijn geheel meegekookt. Veel van de visbotten die oorspronkelijk werden weggegooid hebben maar een kleine kans om bewaard te blijven in de grond, of om te worden teruggevonden in de zeef. Honden op deze nederzettingen aten waarschijnlijk veel van het visafval. De burial conditions op de Barrie, Dunsmore en Carson nederzettingen waren gunstig voor alle categorieën bot, inclusief, bij Dunsmore en Carson, van visschubben.
Visresten zijn niet noodzakelijkerwijs alleen een passieve weerspiegeling van hun locale verkrijgbaarheid, of hoe makkelijk ze te vangen zijn. Hoofdstuk 5 beschrijft methoden van verschillende biologische, landschappelijke, geschiedkundige en archeologische sub-disciplines die kunnen helpen bij het interpreteren van visresten. De etnografische geschriften bevatten nuttige informatie over verschillende vangsttechnieken, o.a. met weren, netten, speren, en kano’s. Het is belangrijk te erkennen dat de beschreven visvangst-expedities misschien niet typisch zijn en dat de visvangst, in de periode vòòr Europees contact, misschien heel anders was. Zowel actieve als passieve vangsttechnieken laten een signatuur achter van de hoeveelheid resten, het samengaan van bepaalde soorten en vooral visbot maten. Een vertrouwen in moderne biogeografische gegevens, over het wel of niet voorkomen van bepaalde vissen in bepaalde wateren in het verleden, kan lijden tot een cirkelredenering. Om de begrijpen wat voor soort wateren werden bevist is het nuttig om te kijken naar biologische studies omtrent de groei- en leef-omgeving voorkeur van verschillende soorten. De visresten van de drie nederzettingen verschilden in hun indeling. De taxonomische distributie op Barrie suggereert een voorkeur voor open water, grotere inhammen en baaien, terwijl de vissoorten die domineren op Dunsmore en Carson meer de voorkeur geven aan grotere inhammen en baaien, en oever moerassen.
Het botgewicht werd gebruikt om de gemiddelde fragment grootte in de verschillende taxonomische klassen te berekenen, en op die manier taphonomische effecten op te sporen. Daar individuele visbotten vaak maar heel weinig wegen, moesten taphonomische kwesties binnen deze klasse op een andere manier worden onderzocht, namelijk door middel van de hoeveelheid fragmentatie en de grootte van de visbotten. De grootte van de visbotten werd geschat door middel van een proportionele methode, waarbij de relatieve grootte van een bot werd uitgedrukt als een percentage van
Omdat deze vissoorten niet allemaal het hele jaar gemakkelijk te vangen waren, moet seizoengebonden variatie in visgedrag en leefomgeving ook in acht genomen worden. De vissoorten die op deze nederzettingen vertegenwoordigd zijn vertonen seizoengebonden variatie in leefomgeving en gedrag. Voor de meeste soorten vergt 68
vangst gedurende de paaitijd een andere vangst-strategie dan vangst buiten de paaitijd. Het was daarom nodig om, voor vissen die zowel binnen als buiten het paaiseizoen makkelijk te vangen zijn, een onderscheid te kunnen maken tussen geslachtsrijpe en onrijpe vissen. Een tweede toepassing van visbotmaatgegevens was daarom als ruwe indicatie van vis grootte en daarmee samenhangende geslachtsrijpheid, om op die manier paaitijd vangsten te kunnen onderscheiden.
visschub van deze nederzetting werd onderzocht, gebruik makende van de bovengenoemde CSAGES methode. De schub was afkomstig van een vijf jaar oude gele baars die in de lente werd gevangen. Daar de gele baars waarschijnlijk niet in het Nottawasaga stroomgebied voorkwam, werd deze soort waarschijnlijk bevist in Lake Simcoe en zijrivieren. Het merendeel van de andere visresten duidt ook op vangsten gedurende de paaitijd, het eerste soort visserijcomplex. Een klein aantal zalm-achtigen werd gevangen op het meer.
Elementen zoals schubben en spines kunnen gebruikt worden om de leeftijdsstructuur alsmede het vangstseizoen van vissen vast te stellen. De leeftijd van de vis kan ook aangeven of de vis geslachtsrijp was of niet. De CSAGES methode die recentelijk ontwikkeld werd door visserijbiologen werkzaam bij het Ontario ministerie van natural resources is veelbelovend voor archeologisch materiaal.
Op Dunsmore is de gele baars de meest talrijke soort, gevolgd door de kleine bruine dwergmeerval, een zonnevis (Lepomis gibbosus) en snoek. De snoekbotten waren veelal te klein om te behoren tot geslachtsrijpe individuen. De grootte van de dwergmeervalbotten geeft aan dat ze waarschijnlijk toebehoorden aan volwassen vissen. Leeftijds- en seizoensinterpretatie van zes dwergmeerval spines door middel van CSAGES wijst op volwassen vissen, waarvan twee mogelijk gedurende de paaitijd zijn gevangen. De baarsbotten zijn wat kleiner dan op Barrie en ze vertonen een grotere maatverspreiding, wat aangeeft dat minder van deze vissen werden gevangen gedurende de rijtijd. Het consequent samengaan van snoek, dwergmeerval en gele baars in de vijf contexten met de meeste visbotten geeft aan dat veel van de visresten op Dunsmore waarschijnlijk afkomstig zijn van een visserij gedurende de warme maanden (visserij 2). De zalm-achtigen en de sommige Catostomidae duiden op een visserij op het meer (visserij 3).
Gebruik makend van deze verschillende soorten informatie, werden de visresten van Barrie, Dunsmore en Carson onderverdeeld in drie visserijcomplexen: 1) een visserij in rivieren en kreken die zich concentreert op intensieve exploitatie van lente-paaiende soorten zoals zoetwater steur (Acipenser fulvescens), suckers (Catostomus catostomus, C. commersoni), gele baars (Perca flavescens) en snoekbaars (Stizostedion vitreum); 2) een minder gespecialiseerde visserij in rivieren, kreken en moerassen voor vissen die niet massaal samenkomen gedurende de paaitijd, zoals snoek (Esox lucius), bruine dwergmeerval (Ameiurus nebulosus) en zonnevissen (familie Centrarchidae); en 3) een visserij georiënteerd op Lake Simcoe of Georgian Bay, onder andere voor herfst-paaiende zoetwater zalm-achtigen. Vissoorten kunnen dus samen in een archeologische context voorkomen omdat ze dezelfde wateren bewonen en/of in dezelfde tijd paaien en/of gevangen kunnen worden met dezelfde technieken. Hoofdstuk 6 geeft meer informatie over de visresten verzamelingen en beschrijft hoe visresten, van individuele (afval)putten en van de nederzetting als geheel, kunnen worden gerelateerd aan het bovengenoemde visserijcomplexen-model. Op het niveau van individuele (afval)putten bestaan verschillen in het faunaspectrum die misschien het resultaat zijn van verschillen in het seizoen van vangst of afvalverwerking. Dit geeft aan dat, terwijl sommige soorten vrijwel exclusief gedurende hun paaitijd werden gevangen, andere soorten gedurende een veel langere periode in de late lente, zomer en vroege herfst werden geëxploiteerd, inclusief, maar niet alleen, gedurende hun paaitijd.
Onmiskenbaar bewijs voor exploitatie van een soort gedurende de paaitijd werd gevonden in 12 Percidae schubben van 4 contexten. Vanwege de grote aantallen gele baars botten in deze collectie werd eerst aangenomen dat de schubben ook van deze soort afkomstig waren. Maar zowel de absolute grote, absolute leeftijd, leeftijds-spreiding and groeipatroon, gaven een verwante soort aan, een snoekbaars, specifiek de ondersoort Stizostedion vitreum vitreum. De schubben waren allen afkomstig van oudere, volwassen individuen, die in de (vroege) lente werden gevangen. Tegenwoordig zwemt deze soort in grote getallen via de Nottawasagarivier naar Willow Creek om te paaien, en is daar zeer makkelijk te vangen. Met uitzondering van de snoekbaars-vangst, lijkt de visvangst op Dunsmore zich minder te hebben toegespits op de paaitijd in de lente dan op Barrie. De vangst van de snoekbaarzen, gedurende de paaitijd, duidt op een zeer lokale vangst. De rest van de visresten kunnen veelal geinterpreteerd worden als meer incidentele vangsten gedurende de warmere maanden van het jaar (visserij 2) of de visserij op het meer (visserij 3).
De visresten van Barrie bestaan voor een groot deel uit steur en gele baars. Slijpplaatjes van drie steur pectoral spines, geven aan dat sommige van de steuren volwassen waren. De meest productieve en voorspelbare plaats om steur te vangen zou zijn: dicht bij de monding van de Nottawasaga rivier, gedurende de paaitijd in mei. De botmaten van de baars geven aan dat de meesten volwassen waren en de inhoud van verschillende (afval)putten wijst op netvangsten. De enige
De visresten op Carson bestaan overwegend uit dwergmeerval, zonnevis en gele baars. De baars- en meervalbotten zijn over het algemeen net zo groot als die van Dunsmore. De snoekbotten zijn iets groter dan op Dunsmore, maar zijn waarschijnlijk ook voor het merendeel afkomstig van onvolwassen individuen. De CSAGES analyse van de spines van de dwergmeerval geeft aan dat sommige individuen nog niet geslachtsrijp waren. Het seizoen van 69
exploitatie varieert van lente tot herfst. Vier van de acht spines zouden kunnen duiden op een paaitijd vangst. Zoals op Dunsmore, werden snoek, dwergmeerval en kleinere baarzen waarschijnlijk geexploiteerd van lente tot herfst. En zoals op Dunsmore is de enige duidelijke indicatie van paaitijd exploitatie te vinden in 24 snoekbaarsschubben, afkomstig van twee grote (afval)putten.
Om een wat breder beeld te krijgen van de visvangst rondom Lake Simcoe werden Barrie, Dunsmore en Carson vergeleken met drie andere nederzettingen gelegen nabij een kreek die uitmondt ten zuiden van Kempenfelt Bay: Wiacek, Hubbert en Molson. Grote hoeveelheden "suckers" duiden op een grotere nadruk op visvangst gedurende de paaitijd in de lente (visserij 1), of een visserij op het meer (visserij 3). Wat meteen opvalt in deze vergelijking zijn de grote hoeveelheden zalm-achtigen op de meest recente van deze nederzettingen, Molson, die bewoond werd net vòòr Europese kolonisatie. Het is duidelijk, van een vergelijking van kopbotten en wervels tussen deze nederzettingen, dat er op de vroegere nederzettingen wel wat gevist werd op deze zalm-achtigen (in de lente of herfst), maar dat een grote nadruk op visvangst op het meer, in de late herfst, pas begint na de bewoning van Dunsmore en Carson. De visserij op Lake Simcoe en Georgian Bay in oktober en november, was één van de meest belangrijke aspecten van de voedselvoorziening in de periode van het eerste contact met de Europeanen. Daar deze zalm-achtigen in de paaitijd heel talrijk zijn en daar zij relatief veel calorieën bevatten (vooral Salvelinus namaycush), zal het herfst aandeel van deze derde visserij heel productief zijn geweest.
In het eerste deel van Hoofdstuk 7 worden de visresten van Barrie, Dunsmore en Carson vergeleken voor aanwijzingen over verschillen in visvangst-strategieën en visverwerking. De spreiding van de verschillende vissoorten duidt op verschillen in vangstlocatie. De verhouding tussen kopbotten enwervels geeft aan dat verwerking op de vangstlocatie belangrijker was voor de bewoners van Barrie dan van de twee latere nederzettingen. Dit duidt op exploitatie van wateren verder weg van de nederzetting, wat fileren op de vangstlocatie aantrekkelijker maakte. De vele gele baarzen op alle drie de nederzettingen kwamen waarschijnlijk van Lake Simcoe en zijrivieren, wateren die vrij dicht bij lagen. De vele steur op Barrie werden waarschijnlijk verder weg gevangen, in de Nottawasaga rivier, misschien in de monding, bij Georgian Bay. De vele dwergmeervallen op Dunsmore en Carson werden waarschijnlijk locaal gevangen, in Little Lake en Willow Creek. De gele baarzen zijn over het algemeen groter op Barrie dan op Dunsmore en Carson. Daar de maaswijdten van Barrie en Dunsmore hetzelfde waren is het mogelijk dat deze verschillen in botgrootte niet alleen voortkomen uit taphonomische verschillen. De grotere vissen op Barrie duiden misschien op een grotere nadruk op vissen met netten op Lake Simcoe of de Nottawasaga rivier gedurende de paaitijd. De viswervel determinaties duiden op een grotere nadruk op zalmachtigen op Barrie.
Informatie over vogels en zoogdieren geeft aanvulling op de visresten. De bewoners van Barrie hadden een seizoengebonden visvangst en aten meer vogels en zoogdieren afkomstig van de rivier of rivieroever en het bos. Zoals men zou verwachten van een dorp dat een soort pioniers status heeft, geeft dit aan dat er weinig ontgonnen gebied of nieuwe aanwas was in de omgeving. De vele kleine knaagdierbotten en een klein aantal hertenbotten lijken op het eerste gezicht niet in dit plaatje te passen. Maar zowel de knaagdieren als de herten zullen aangetrokken geweest zijn tot de maïsvelden, die een noviteit waren. Herten waren waarschijnlijk schaars in het gebied vanwege ongunstige landschappelijke omstandigheden. Op Dunsmore en Carson zijn hertenbotten nog veel zeldzamer. Misschien had de hertenjacht gedurende de 14de eeuw de lokale hertenpopulatie zover teruggedrongen dat er weinig meer te jagen viel tegen de tijd van bewoning van Dunsmore en Carson in de 15de eeuw. Het is ook mogelijk dat de herten op Barrie werden gevangen in het gebied ten zuiden van de nederzettingen, waar de Barrie bewoners oorspronkelijk vandaan kwamen, en waar herten veel talrijker waren.
Metingen van verschillende botten van dwergmeerval, alsmede de leeftijdsgegevens uit de dwergmeerval spines en snoekbaars schubben geven aan dat Carson jongere/kleinere vissen bevat, zowel in gemiddelde als in spreiding. Om verscheidene redenen wordt geargumenteerd dat verschillen in zeef maaswijdte of veranderende visaparatuur niet de enige oorzaak zijn van het aldaar voorkomen van kleinere/jongere vissen. Het visserijcomplex model geeft aan dat de twee nederzettingen het merendeel van hun visvangst uitvoerden in de kleine zijrivier van de Nottawasaga rivier, alsmede in een klein meertje dat daaraan vastzit. Het is mogelijk dat de afname in gemiddelde visbot grootte en/of leeftijd bij Carson gedeeltelijk een weerspiegeling is van verandering in de visgemeenschap. Het feit dat er zich onder de snoekbaarzen op zowel Dunsmore als Carson zeer oude individuen bevonden, duidt erop dat deze eventuele veranderingen in de visgemeenschap heel subtiel waren. Van overbevissing mag zeker niet gesproken worden. Als de afnemende leeftijd van vissen inderdaad het resultaat is van menselijke bevissing, en niet van taphonomie, sampling error, of natuurlijke veranderingen in de vispopulatie, zou dit de relative chronologische positie van de twee nederzettingen bevestigen.
De hoeveelheid hondenbotten op Carson is veel groter dan die op Barrie en Dunsmore. Sommige honden werden waarschijnlijk geofferd gedurende religieuze feesten (deze worden teruggevonden als aparte begravingen, zoals in één van de huizen op Carson). Vondsten van hondenbotten met snijsporen in de afvalhopen geeft aan dat anderen gewoon werden gegeten. De voedselwaarde van honden, alsmede het gemak maakte het misschien mogelijk en/of wenselijk voor de Carson bevolking om meer tijd door te brengen in en rondom het dorp. Dit past weer bij de lokale toon van de visvangst. De voedseleconomie die beschreven wordt door de eerste 70
missionarissen en ontdekkingsreizigers is duidelijk een voortzetting van de periode daarvoor. Wat dit vergelijkend onderzoek duidelijk maakt, is, dat er toch veel variatie was binnen dit patroon. Het zal altijd moeilijk blijven om te kiezen tussen taphonomie, landschappelijke omstandigheden en menselijke keuze als causale factoren, maar deze multidisciplinaire, vergelijkende aanpak heeft wat nieuwe inzichten gegeven.
71
TABLES Table 1. Archaeological cultural chronology of research area. Euro-Canadian historic period
A. D. 1820-present
limited permanent re-occupation of Simcoe County
A. D. 1750-1820
Late Woodland
French / Contact (Huron)
A. D. 1615-1650
Protohistoric Iroquoian
A. D. 1600-1615
Late Iroquoian Middle Iroquoian
A. D. 1400-1600 Middleport
A. D. 1330-1400
Uren
A. D. 1280-1330
Early Iroquoian
A. D. 900-1280
Transitional Woodland
A. D. 700-900
Middle Woodland
300 B. C.-A. D. 700
Early Woodland
800-300 B. C.
Late Archaic
2500-800 B. C.
Middle Archaic
6000-2500 B. C.
Early Archaic
7000-6000 B. C.
Palaeoindian
9000-7000 B. C.
Note: Nomenclature and dates mostly follow Ellis and Ferris (1990).
Table 2. Dates of occupation of the Barrie, Dunsmore and Carson sites advocated by respective excavators. Carson
ca. A. D. 1475-1525
14
Dunsmore
ca. A. D. 1430-1510
Based on ASI ceramic seriation and a recent AMS date on maize from House 8, Feature 206, of cal A. D. 1430-1510 at 1 sigma; 14C date of cal A. D. 1307-1411 at 1 sigma obtained on wood charcoal in 1977 is rejected by ASI (Dave Robertson, personal communication 1998) and Dodd, Poulton, Lennox, Smith and Warrick (1990:328). Sutton's ceramic seriation favours A.D. 1425-1475 (personal communication 1999 and Sutton 1996a).
Barrie
ca. A. D. 1280-1330
Based on ceramic seriation; 14C date on maize kernels of cal A. D. 1409 ± 40 rejected (Sutton 1996a).
C date of cal A. D. 1507 ± 27 is in agreement with ceramic seriation to late fifteenth-early sixteenth century (Varley 1993; Varley and Cannon 1994).
72
Table 3. Overview of contexts analysed and recovery methods at the Barrie site. Context
Screened only
Also floated
Total
House 1 features
5
3
8
House 2 features
14
7
21
Exterior features
4
-
4
23
10
33
Midden A squares
5
7
12
Midden B squares
2
3
5
Midden D squares
-
1
1
Number of 1m squares
7
11
18
Number of features
Table 4. Overview of contexts analysed and recovery methods at the Dunsmore site. Context
Screened only
Also floated
Total
Northeast section House 1 features
32
1
33
House 4 features
4
-
4
South-central section House 7 features
16
4
20
House 9 features
2
1
3
House 13 features
1
1
2
House 15 features
1
1
2
House 8 features
4
1
5
House 10 features
7
1
8
House 11 features
2
-
2
House 12 features
2
-
2
71
10
81
4
-
4
9
3
12
7
1
8
20
4
24
West section
Number of features Northeast section Midden A squares South-central section Midden B squares West section Midden C squares Number of 1m squares
Table 5. Overview of contexts analysed and recovery methods at the Carson site. Context
Screened only
Also floated
Total
House 1 features
12
14
26
House 2 features
1
1
2
House 3 features
64
5
69
House 5 features
35
6
41
House 6 features
1
-
1
House 7 features
1
-
1
House 8 features
1
-
1
115
26
141
Midden 4 squares
6
-
6
Number of 1m squares
6
0
6
Number of features
73
Table 6. Binomials and generic names of taxa identified at the three sites. Scientific name
Vernacular name / possibilities intended by scientific name
OSTEICHTHYES Acipenseriformes: Acipenseridae Acipenser fulvescens
lake sturgeon
Semionotiformes: Lepisosteidae Lepisosteus osseus
longnose gar
Salmoniformes: Salmonidae Salvelinus namaycush
lake trout
Coregonus artedi
lake herring
Coregonus clupeaformis
lake whitefish
Coregonus sp. (large)
lake whitefish or lake herring
Esociformes: Esocidae Esox americanus
grass pickerel (ssp. vermiculatus)
Esox lucius
northern pike
Esox masquinongy
muskellunge
Esox sp. (small)
grass pickerel, small northern pike or hybrid
Esox sp. (large)
larger northern pike, muskellunge or hybrid
Cypriniformes: Cyprinidae Semotilus sp.
creek chub or fallfish
Cypriniformes: Catostomidae Catostomus catostomus
longnose sucker
Catostomus commersoni
white sucker
Catostomus sp.
longnose or white sucker
Moxostoma sp.
redhorse sucker
Siluriformes: Ictaluridae Ameiurus melas
black bullhead
Ameiurus natalis
yellow bullhead
Ameiurus nebulosus
brown bullhead
Ameiurus sp. (large)
yellow or brown bullhead
Ictalurus punctatus
channel catfish
Ameiurus/Ictalurus sp.
yellow or brown bullhead or channel catfish
Gadiformes: Gadidae Lota lota
burbot
Perciformes: Centrarchidae Ambloplites rupestris
rock bass
Lepomis gibbosus
pumpkinseed
Lepomis macrochirus
bluegill
Lepomis sp.
green sunfish, pumpkinseed, bluegill or longear sunfish or hybrid
Micropterus dolomieu
smallmouth bass
Micropterus salmoides
largemouth bass
Micropterus sp.
smallmouth or largemouth bass
Pomoxis nigromaculatus
black crappie
Pomoxis sp.
black or white crappie
Perciformes: Percidae Perca flavescens
yellow perch
Stizostedion vitreum
walleye
Stizostedion sp.
sauger or walleye
74
Table 6 continued Scientific name
Vernacular name / possibilities intended by scientific name
AMPHIBIA Anuraa: Ranidae Rana catesbeiana
bullfrog
Rana sp.
frog
REPTILIA Chelonia: Chelydridae Chelydra serpentina
snapping turtle
Chelonia: Emydidaea Chrysemys picta
painted turtle
Emydoidea blandingi
blanding’s turtle
Graptemys geographica
map turtle
AVES Gaviiformes: Gaviidae Gavia immer
common loon
Podicipediformes: Podicepedidae Podiceps auritus
horned grebe
Podiceps grisegena
red-necked grebe
Podiceps sp.
red-necked or horned grebe
Anseriformes: Anatidae Branta canadensis
canada goose
Anas clypeata
northern shoveller
Aythya collaris
ring-necked duck
Aythya sp.
freshwater diving duck
Bucephala clangula
common goldeneye
Bucephala albeola
bufflehead
Melanita sp.
black, surf or white-winged scoter
Mergus merganser
common merganser
Anatinae
duck
Falconiformes: Accipitridae Accipiter gentillis
northern goshawk
Buteo jamaicensis
red-tailed hawk
Haliaeetus leucocephalus
bald eagle
Galliformes: Phasianidae Bonasa umbellus
ruffed grouse
Meleagris gallopavo
wild turkey
Gruiformes: Rallidae Fulica americana
american coot
Gruiformes: Gruidae Grus canadensis
sandhill crane
Columbiformes: Columbidae Ectopistes migratorius
passenger pigeon
Strigiformes: Strigidae Strix varia
barred owl
Aegolius acadicus
northern saw-whet owl
Piciformes: Picidae Sphyrapicus varius
yellow-bellied sapsucker
Passeriformes: Corvidae Corvus brachyrynchos
common crow
75
Table 6 continued Scientific name
Vernacular name / possibilities intended by scientific name
MAMMALIA Lagomorphaa: Leporidae Lepus americanus
snowshoe hare
Sylvilagus floridanus
eastern cottontail
Rodentia: Sciuridae Sciurus carolinensis
grey squirrel
Tamiasciurus hudsonicus
red squirrel
Marmota monax
woodchuck
Tamias striatus
chipmunk
Glaucomys sp.
southern or northern flying squirrel
Rodentia: Castoridae Castor canadensis
beaver
Rodentia: Cricetidae Peromyscus sp.
white-footed or deer mouse
Ondatra zibethicus
muskrat
Microtus pennsylvanicus
meadowvole
Microtus sp.
meadowvole or woodland vole
Rodentia: Erethizonidae Erethizon dorsatum
american porcupine
Carnivora: Canidae Canis familiaris
domestic dog
Vulpes vulpes
red fox
Urocyon cinereoargenteus
grey fox
Canidae
small dog or fox
Carnivora: Ursidae Ursus americanus
american black bear
Carnivora: Procyonidae Procyon lotor
raccoon
Carnivora: Mustelidae Mustela vison
mink
Martes americana
marten
Martes penanti
fisher
Lutra canadensis Artiodactyla: Cervidae
river otter a
Cervus elaphus
elk
Odocoileus virginianus
white-tailed deer
Note:Taxonomy follows Mandrack and Crossman (1992) for family, genus and species, and Nelson (1994) for orders of fish; Logier and Toner (1961) for amph bians and reptiles; Godfrey (1986) for birds; and Honacki et al. (1982) for mammals. Because of the GFAUNA2 database coding system, the position in the phylogenetic sequence of the family Esocidae and the family Salmonidae continues to follow an earlier source (Scott and Crossman 1973). a Higher phylogenetic level considered analytically useful
76
Table 7. Taxonomy, biology and ecology of the major fish families. Mature Max. Spawning seasona cm cm
Spawning habitat
Normal habitat
Behaviour
Comments
river or lake over rocks at up to 5 m and 14-15 C
deep, cool lake or river
bottom feeder; moves into deeper water in summer, active all winter; migratory; caught with spear or gill net
osteology extremely distinctive
Vert.
Acipenseridae (Sturgeons) Acipenser fulvescens
lake sturgeon
90
200
MMMM
Salmonidae (Trouts) OOOO
Salvelinus namaycush
lake trout
40
50
open shoreline or promontory of deep, cold lake lake over angular cobble and rubble at up to 36 m and 14-9 C
dispersed in winter; at lake margin at ice breakup; moves into deeper water in summer; caught by gill or seine net, spearing, angling; spawns at night during 9-16 day period
Coregonus artedi
lake herring
21
30
NN
Coregonus clupeaformis
lake whitefish
40?
59
NNNN
61-69
lake shallows over gravel or stone at up to 12 m and 5-3 C
deep, cold lake
schooling; moves into deeper water in summer; caught by gill or seine net; spawns at night
hybridizes with C. clupeaformis; great variability in morpho-metry
50-63
lake shallows over stone or sand at up to 8 m and 8-3 C
deep, cold lake
bottom feeder; schooling; moves into hybridizes with C. deep water in summer; caught by gill or artedi; dwarf form seine net, spearing and angling; spawns exists sympatrically at night
55-64
AAAAM
stream or marsh over vegetation at
