(shrubs
R E L A T I O N S H I P S B E T W E E N S H R E W S (SOREX SPP.) A N D D O W N E D W O O D IN T H E V A N C O U V E R W A T E R S H E D S , B.C.
by
V A N E S S A JOY CRAIG B.Sc. (Hon.), Simon Fraser University, 1990
A T H E S I S SUBMITTED IN PARTIAL F U L F I L L M E N T O F THE REQUIREMENTS FOR THE D E G R E E OF MASTER OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Forestry)
W e a c c e p t this thesis as conforming to the requifBcl,standard
T H E UNIVERSITY O F BRITISH C O L U M B I A September 1995 © Vanessa Joy Craig, 1995
In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department
or by his
or
her
representatives.
It
is
understood
that
copying or
publication of this thesis for financial gain shall not be allowed without my written permission.
Department of
V~-r\r4.?z¥c\J
The University of British Columbia Vancouver, Canada
•ate
DE-6 (2/88)
CMoher JO. 1446*
II Abstract
I studied relationships between shrew populations, downed wood, and vegetation in 1991 and 1992 on study areas in the Vancouver watersheds, B.C. I conducted a mark-recapture study on grids with varying amounts of downed wood ('low', 'medium', and 'high'). Three species of shrews, Sorex monticolus, S. vagrans, and S. cinereus were captured during the study. Shrew populations were small, and appeared to increase between 1991 and 1992, but trends were influenced by a change in trapping regime between years. Size of shrew populations did not vary directly with amount of downed wood but showed a more complex relationship with habitat. In general, larger shrew populations were associated with abundant, moderate-sized pieces of wood, and increased vegetative and litter cover. Larger populations also had a disproportionately greater number of reproductively active females, suggesting that the areas they lived in were better shrew habitat. Relation of shrew captures to the microhabitat surrounding trap sites indicated that shrews associated with larger pieces of downed wood and potential foraging sites. Shrews inhabiting areas with diversity in characteristics of downed wood, both in size and in abundance (on 'high' grids), were caught more frequently on sites with larger pieces of wood and on sites where downed wood was closer together. The three species of shrew captured were similar in terms of life history and microhabitat use, and appeared to be dispersed in the environment through territoriality/dominance relationships. Tracking individual shrews confirmed that downed wood was an important habitat component. Shrew travel routes had greater continuity of cover than did areas farther away from their trails. Shrew trails were negatively associated with downed wood 12 cm in diameter. To encourage larger, reproductively active shrew populations in managed forest stands, I recommend that a range of tree species and lengths of wood pieces be left on a cutblock. Emphasis should be placed on providing pieces >6 cm in diameter. Pieces should be dispersed to provide a continuous log network. Patches of trees should be left standing to provide a future source of downed wood on the site. Areas should not be burned, because it destroys litter, moss, associated insect communities, and hardens logs.
iv Table of Contents Abstract
ii
Table of Contents
iv
List of Tables
vi
List of Figures
vii
Acknowledgements
x
Chapter 1. General Introduction
1
Chapter 2. Study Area
3
Introduction Methods Analyses Results Discussion Chapter 3. Shrew Populations and Habitat Associations Introduction Methods Analyses Results (1) Species (2) Estimating shrew populations (3) Trap success in 1992 (4) Population dynamics (5) Relation of shrew population size to habitat attributes (6) Habitat attributes of successful trap sites (7) Shrew range Discussion (1) Species (2) Estimating shrew populations (3) Trap success in 1992 (4) Population dynamics (5) Relation of shrew population size to habitat attributes (6) Habitat attributes of successful trap sites (7) Shrew range Chapter 4. Shrew Travel Routes Introduction Methods Analyses Results
3 5 9 10 17 20 20 21 23 26 26 26 34 37 39 40 44 46 46 56 58 59 61 63 64 67 67 68 70 71
V
Discussion
74
Chapter 5. Conclusions and Recommendations
78
Literature Cited
80
List of Tables Table 1. Description of habitat variables measured on study grids in the Vancouver watersheds, B.C., and their abbreviations used in the text Table 2. Description of study grids in the Vancouver watersheds, B.C
1
Table 3. Mean (and SE) of habitat variables measured on nine study grids in the Vancouver watersheds, B.C
1
Table 4. Mean (and SE) of Jolly trappability in 1991 and 1992 for each grid in the Vancouver watersheds, B.C Table 5. Mean Jolly-Seber shrew population estimates for each study grid in the Vancouver watersheds, B.C Table 6. Mean 1992 Jolly-Seber shrew population estimates for each study grid in the Vancouver watersheds, B.C. using only capture data from livetraps Table 7. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'low' grids in the Vancouver watersheds, B.C Table 8. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'medium' grids in the Vancouver watersheds, B.C Table 9. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'high' grids in the Vancouver watersheds, B.C Table 10. Number of captures (N), species identification, weight, mean distance travelled, and 90% range area for each shrew caught > 5 times on study grids in the Vancouver watersheds, B.C. Sm = Sorex monticolus, Sc = S. cinereus, Sv = S. vagrans Table 11. Cumulative frequency distributions of number of wood pieces by diameter class within corridor intervals. An asterisk (*) indicates a 'greatest difference' (P < 0.05) determined through a Kolmogorov-Smirnov test; (+) or (-) indicates whether the association is positive or negative
List of Figures Figure 1. Study areas in the three Vancouver watersheds, B.C., Canada (adapted from EES Inc., 1991)
4
Figure 2. Layout of four, 100-m downed wood transects within 1-ha small mammal trapping grids in the Vancouver watersheds, B.C. Numbers indicate transects one through four. Dots indicate trap stations
6
Figure 3. A) Mean volume of downed wood (and SE) on each grid in the Capilano, Seymour, and Coquitlam watersheds. Wood volumes were different across wood volume classes (P 0.1). B) Mean volume of downed wood on 'low', 'medium', and 'high' grids. 'High' grids had greater volumes of wood than 'medium' and 'low* grids (P < 0.0005). Wood volumes on 'medium' and 'low' grids did not differ significantly (P = 0.064)
12
Figure 4. Volume of downed wood within each diameter class on each study grid in the Vancouver watersheds, B.C. 'High' grids had greater volumes of wood in the large diameter classes (48-100 cm and >100 cm, P < 0.01) than 'medium' or 'low' grids. 'Medium' and 'high' grids had greater volumes of wood in the 6-12-cm diameter class (P < 0.05) than 'low' grids.
13
Figure 5. Proportions and total number (n) of shrews captured per grid of three species of shrews caught in 1992 in the Vancouver watersheds, B.C. Grids were classified as having 'low' (L), 'medium' (M) and 'high' (H) amounts of downed wood. Sm = Sorex monticolus, Sc = S. cinereus, Sv = S. vagrans
27
Figure 6. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Capilano watershed, B.C
30
Figure 7. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Seymour watershed, B.C
31
Figure 8. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Coquitlam watershed, B.C
32
Figure 9. Proportions and total number (n) of shrews caught by both trap types, or only by pitfalls or livetraps in 1992 of A) residents (shrews caught > once), and B) transients (shrews caught only once). Grids were classified as having 'low' (L), 'medium' (M), and 'high' (H) wood volumes
36
Figure 10. Proportion of reproductively active females (and total number of reproductive females captured, n) relative to 1991 and 1992 shrew populations on each grid with 'low' (L), 'medium' (M) and 'high' (H) downed wood volumes in the Vancouver watersheds, B.C 38 Figure 11. Resident shrew (captured >once) ranges on Capilano 'low'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
47
Figure 12. Resident shrew (captured >once) ranges on Capilano 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
48
Figure 13. Resident shrew (captured >once) ranges on Capilano 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
49
Figure 14. Resident shrew (captured >once) ranges on Seymour 'low'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
50
Figure 15. Resident shrew (captured >once) ranges on Seymour 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
51
Figure 16. Resident shrew (captured >once) ranges on Seymour 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
52
Figure 17. Resident shrew (captured >once) ranges on Coquitlam ' low'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
53
Figure 18. Resident shrew (captured >once) ranges on Coquitlam 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
54
Figure 19. Resident shrew (captured > once) ranges on Coquitlam 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals
55
Figure 20. Diagram of a shrew trail, illustrating the placement of transects and corridor intervals. Thick, dark lines represent logs, thin lines represent transects, placed every 1-m along the shrew trail
69
Figure 21. Cumulative frequency distribution of downed wood pieces >3 cm in diameter at increasing distances from shrew trails within wood volume classes on study grids in the Vancouver watersheds, B.C. An asterisk (*) denotes the corridor interval with the greatest difference in number of pieces vs. expected (P < 0.05) by Kolmogorov-Smirnov test
X
Acknowledgements I thank Dr. Fred Bunnell, my supervisor, for providing me with the opportunity to conduct applied research. He tried to keep me on the right path, and sometimes succeeded; I could not have done this project without his support and patience. Dr. Alton Harestad inspired me to pursue this area of research, and has served as a role model. Drs. Tom Sullivan and Walt Klenner have been extremely supportive throughout. Dr. Val Lemay provided statistical advice. I could not have conducted this research project without my field assistants: Don Demarchi, Sandi Lee, Bob Moody, Peter Opie, Brian Poole, and Kristine Webber. Special thanks to Peter for doing site classifications of my study grids, and Kristine for fun and enthusiasm. Thanks also to Jackie Johnson, without who's advice I would probably still be attempting to disentangle the mysteries of graduate life at U.B.C. The Greater Vancouver Regional District permitted me to conduct this study in the Vancouver Watersheds. Ken Juvik and Derek Bonin from the Greater Vancouver Water District were very helpful in ensuring that I had wide access to facilities within the Vancouver watersheds. Dr. Harvey-Clarke from the U.B.C. South Campus farm assisted me in designing a safe method for anaesthetizing shrews. Thanks to the many graduate colleagues who've made grad school interesting, stimulating, and fun. I especially thank Barry Booth, a great friend. I was supported by a G.R.E.A.T. Scholarship from the Science Council of B.C., and by the Greater Vancouver Water District and Ministry of Forests. My partner, Steve Wilson, provided limitless and invaluable support, statistical advice, computer time, and editing power. I dedicate this thesis to my family, who have supported me throughout the many phases of this thesis.
1 Chapter 1. General Introduction
Shrews (Sorex spp.) are important forest inhabitants. They are both prey and predator. Shrews are preyed upon both by avian and terrestrial predators, and prey upon both injurious and beneficial invertebrates. Considerable research has concentrated on determining the habitat features that are important to small mammals (primarily mice and voles), but little research has investigated a particular habitat feature in detail. Because of their high metabolic rate and small size, shrews have a high mortality rate under conventional trapping regimes (e.g. Sullivan and Sullivan 1982, Churchfield 1990). Therefore, very few researchers in North America have examined a shrew population through mark-recapture techniques in its natural environment (Buckner 1966, Hawes 1975, Cawthorn 1994). Downed wood is thought to be important to small mammals for cover, nesting, and foraging sites (Cowan and Guiguet 1975, Thomas 1979, Maser et al. 1981, Van Home 1981, Kaufman et al. 1983, Maser and Trappe 1984, Harestad and Shackleton 1990, Barnum et al. 1992, Planz and Kirkland 1992, Carter 1993, Amaranthus era/. 1994, Tallmon and Mills 1994). However, research on the relationship between small mammal abundance and downed wood has failed to find a clear relationship (Nowotny et al. unpubl., Corn et al. 1989). Shrews may be more strongly associated with downed wood than are deer mice. Because of their high metabolic rate, shrews must forage almost continuously and may demonstrate a strong relationship with microhabitat features, such as downed wood, that are associated with invertebrates. In addition, the poor vision and hearing of shrews may increase their vulnerability to predators and so cause them to associate more closely with cover than do other small mammal species. This study had three objectives: 1) Determine if abundance of downed wood influences shrew species presence or their relative abundance in second-growth forest; 2) Determine if abundance of downed wood influences shrew population sizes or
2 demography in second-growth forest; 3) Determine habitat components important to shrew populations and individuals. In Chapter 2,1 examine the habitat structure on my study grids in the Vancouver Watersheds, B.C. In Chapter 3,1 examine relationships between shrews and their habitat, and downed wood, in particular. I also examine relationships among individual shrews. In Chapter 4,1 describe movements of shrews tracked with fluorescent powder and how they related to downed wood in the environment. In Chapter 5,1 summarize the results from my study and offer recommendations for managing second-growth forest stands for shrews.
3 Chapter 2. Study Area Introduction The Vancouver watersheds comprise three large areas: the Capilano, Seymour, and Coquitlam watersheds. They total 57,591 ha in size, and are located north of (Capilano and Seymour watersheds), and east of (Coquitlam watershed) the city of Vancouver, B.C., Canada (Economic and Engineering Services Inc.; hereafter EES Inc. 1991, Fig. 1). In the late 1920's, the watersheds were developed as areas managed for drinking water quality and were closed to the public. By 1936, all commercial logging operations had ceased and access to the areas was controlled to prevent water contamination. Some logging still occurs to prevent fire and insect outbreaks. The watersheds include three biogeoclimatic zones (Meidinger and Pojar 1991): Coastal Western Hemlock (CWH, 63%), Mountain Hemlock (MH, 31%) and Alpine Tundra (AT, 6%) (EES Inc., 1991). All grids for my study are in the CWH biogeoclimatic zone. Study grids in the Capilano and Coquitlam watersheds were in the CWH
v m 1
(very moist) subzone; the Seymour watershed study grids were on an alluvial flood plain, in the CWH
d m
(dry maritime) subzone. Dominant tree species present were
western hemlock (Tsuga heterophylla), Douglas-fir (Pseudotsuga menziesii), and western redcedar {Thuja plicata). Dominant understory species were salal (Gaultheria shallon), red huckleberry (Vaccinium parvifolium), blue huckleberry {Vaccinium ovalifolium), and foam flower (Tiarella unifoliata). All of my study sites had been logged 60-80 years ago and had regenerated naturally. Some areas had been high-graded, leaving some old-growth trees. All of the Vancouver watersheds have a history of severe periodic natural wildfires, and some broadcast burning (EES, Inc. 1991). All of my study areas bore evidence of fire history in the form of charring of the outside of some stumps and snags.
Figure 1. Study areas in the Vancouver watersheds, B.C., Canada (adapted from EES Inc., 1991).
5
In this chapter I describe the vegetation and downed wood components on my study grids. I determine whether the study grids differ only in amount of downed wood present, or if some other habitat component also varies among treatments. Methods I examined many second-growth forest stands within each watershed during June and July 1991. Within each potential study site, I conducted a preliminary assessment of the volume of downed wood on the area using three, 100-m transects, placed in an equilateral triangle. I measured the diameter of every piece of wood that crossed the transects and calculated a downed wood volume for each site following the method and formula described by Van Wagner (1968): V (m3 / ha) = (n
2
x E(D2)) / 8 X L
where D = diameter of the piece of wood (in centimetres) L = length of the transect (in metres) A total of 9 study sites in 60 to 80-year-old, second-growth forest stands were selected. Three study sites were selected within each of the three watersheds, corresponding to three apparent volume classes of downed wood: 'low', 'medium', and 'high'. I chose areas that were as similar as possible; however, I was constrained by the limited number of forest stands that were at least 300 x 300 m in size (to allow a 100 x 100-m grid with a 100-m forest buffer on each side), had minimal understory vegetation, and a suitable volume of downed wood. Downed wood was patchily distributed within grids, therefore I used a more intensive sampling scheme to more accurately estimate the volume of downed wood on each study grid. Four 100-m transects were laid out in a star-pattern (Fig. 2), and I calculated the volume of wood on the grid as before. I did not measure downed wood volumes in 1992. Although a few new trees had blown down as a result of strong winds in the late fall of 1991,1 felt the difference in wood volume would not be appreciable.
Figure 2 . Layout of four, 100-m downed wood transects within 1-ha small mammal trapping grids in the Vancouver watersheds, B.C. Numbers indicate transects one through four. Dots indicate trap stations.
7 I measured various habitat variables on each grid in August, 1992. The methods described below were modified from Walmsley et al. (1980). Understory vegetation measurements were taken at each trap station. I compiled data on four vegetative strata around each station: B1 (tall shrub >2 m), B2 (shrubs 6 cm in diameter that crossed the transect. I used three decay classes: (1) log with few signs of decay, bark and limbs present; (2) log fairly soft but still retaining shape, little or no bark or limbs remaining; and (3) log soft, no bark or limbs remaining. Each 1 m along the transect the canopy was defined as open or closed using a gimbal scope. I converted this to a proportion of canopy present (CANOPY). I sampled trees in ten, 10-m2 circular, fixed-area plots distributed systematically throughout each 1-ha grid. Within each tree plot, I recorded species, DBH (diameter at breast height), crown, and tree class of each tree >2 m in height. I measured the distance from the station to the four closest pieces of downed wood >6 cm in diameter (hereafter four closest pieces) to provide an estimate of dispersion of downed wood. Each piece was identified to species (if possible), and length (LENGTH), width (DIAMETER), and decay class (DC) recorded. Habitat variables and their abbreviations are listed in Table 1.
8
Table 1. Description of habitat variables measured on study grids in the Vancouver watersheds, B.C., and their abbreviations used in the text.
VARIABLE
DESCRIPTION
B1
Percent cover of vegetation in the tall shrub (>2 m in height) layer
B2
Percent cover of vegetation in the small shrub ( 0.1). Wood volumes were different among volume classes (F = 12.262, P < 0.0005, Fig. 3a). Within wood volume classes, volumes were similar across watersheds (F = 0.323, P > 0.1). The wood volume varied between 249 m3 / ha on Capilano 'low', and 1077.34 m^ / ha on Seymour 'high'. 'High' grids had significantly higher volumes of downed wood than 'low' or 'medium' grids (P < 0.0005). Volume of wood on 'low' grids did not differ significantly from that on 'medium' grids, primarily because of the relatively low volume on Seymour 'medium' (P = 0.064, Fig. 3b). Volume of downed wood by diameter class was not consistent across 'low', 'medium', and 'high' grids (Fig. 4). The 'high' grids had greater wood volumes (n= 3 watersheds, n = 3 volume classes, F = 8.748, F = 6.814, P < 0.005) in the large diameter classes (48-100 cm and >100 cm diameter, respectively) than either the 'medium' or 'low' grids (P < 0.005). The 'medium' and 'high' grids also had significantly greater wood volumes (F = 20.449, P < 0.005) than 'low' grids in the 6-12-cm diameter class (P < 0.005). Volumes in the 3-6 cm diameter class and 12-24 cm diameter class varied differently with wood class across watersheds (F = 9.685, F = 4.096 respectively, P < 0.05). In the 3-6 cm diameter class, Capilano 'high' and Coquitlam 'medium' had high volumes, which were different from the grids with the lowest volumes (Capilano 'low' and 'medium', and Coquitlam 'low'). In the 12-24 cm diameter class, the grids with the highest volumes (Capilano 'medium' and Seymour 'medium') were different from the grids with the lowest volumes (Capilano 'low' and Coquitlam 'medium'). All other comparisons were not significant (P > 0.1).
Table 2. Description of study grids in the Vancouver watersheds, B.C.
Watershed Capilano
Seymour
Description
Wood volume class Low
Topography/Aspect
Site classification
Gentle/SW
Poor
Very open, minimal understory vegetation. Old railway bed transected the grid.
Medium
Gently rolling/S
Poor
Locally dense pockets of western hemlock regeneration on the lower half of the grid.
High
Steep/W
Poor
Little vegetation, but contained dense patches of small, dead western redcedar trees.
Low
. Flat
Poor-medium
1
Small stream associated with vegetation; the rest of the grid had patchily-distributed vegetation, and vine maple (Acer circinatum) patches.
Coquitlam
Medium
Flat
Rich
Near Rice Lake in the Seymour Demonstration Forest. Fairly dense vegetation consisting primarily of blue huckleberry and western hemlock regeneration.
High
Steep/SE
Medium(rich)
Patchily distributed vegetation which tended to be associated with vine maple gaps.
Low
Steep/S
Poor
Medium
Gentle/SE
Medium
High
Steep/W
Poor
Following Green era/. (1984)
Rocky area with brushy patchily-dense trees and little understory vegetation. Patchily distributed vegetation and some vine maple patches. Very rocky with an extensive underground rock tunnel network. Little understory vegetation.
12
Figure 3. A) Mean volume of downed wood on each grid in the Capilano, Seymour, and Coquitlam watersheds. Wood volumes were different across wood volume classes (P < 0.005), and similar across watersheds within wood volume classes (P > 0.1). B) Mean volume (and SE) of downed wood on 'low', 'medium', and 'high* grids. 'High' grids had greater volumes of wood than 'medium' and 'low' grids (P < 0.0005). Wood volumes on 'medium' and 'low' grids did not differ significantly (P = 0.064).
13 A) Capilano 600 |
Diameter classes • 3-6 cm 500. 0 6-12 cm 12-24 cm H 24-48 cm 83 48-100 cm • >100cm
m
Low
Medium
High
Medium
High
Medium
High
B) Seymour 600
Diameter classes • 3-6 cm 500. E3 6-12 cm 12-24 cm B 24-48 cm E3 48-100 cm -g 400• >100 cm
I f I
o 300
?
200
E • i 100
Low
B) Coquitlam 600_, Diameter classes • 3-6 cm 500 • 6-12 cm E3 12-24 cm E H 24-48 cm 400 E3 48-100 cm • >100 cm 300
Low
Figure 4. Volume of downed wood within each diameter class on each study grid in the Vancouver watersheds, B.C. 'High' grids had greater volumes of wood in the large diameter classes (48-100 cm and >100 cm, P < 0.01) than 'medium' or 'low' grids. 'Medium' and 'high' grids had greater volumes of wood in the 6-12-cm diameter class (P < 0.05) than 'low' grids.
14 The F a x -tesf indicated that, even when transformed, some of the microhabitat m
variables were heteroscedastic. The normal probability plot for all of the vegetation variables (B1, B2, C, D, CANOPY) showed severe deviations from normality. Accordingly, variables were ranked and two-way non-parametric ANOVAs were used to identify differences within strata among watersheds and wood volume classes. Percent cover in the tall shrub (B1) layer did not vary differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 0.852, P > 0.1; Table 3). Percent cover in the 'B1' layer did not vary among wood volume classes (F = 2.276, P > 0.1), but did vary among watersheds (F = 22.367, P < 0.005). The Seymour watershed had significantly more tall shrub cover than the Capilano or Coquitlam watersheds (P < 0.005). Percent cover in the small shrub (B2) layer varied differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 7.050, P < 0.005). Although the 'B2' percent cover was highest on 'medium' grids in each watershed, the Coquitlam grid had much less cover, causing a watershed x volume class interaction. The herb layer (C) varied differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 4.416, P < 0.005). Herb cover was higher on grids in the Seymour watershed, but did not consistently vary with volume class, causing a watershed x wood volume class interaction. Mean moss cover (D) varied differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 18.222, P < 0.005). Moss cover was higher on 'medium' grids and higher on grids in the Seymour watershed causing a watershed x wood volume class interaction. Canopy presence (CANOPY) varied differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 2.836, P < 0.005). Percent canopy present was lowest on 'medium' grids in each
15
Table 3. Mean (and SE) of habitat variables measured on nine study grids in the Vancouver watersheds, B.C. CAPILANO
SEYMOUR
COQUITLAM
WOOD VOLUME CLASS
WOOD VOLUME CLASS
WOOD VOLUME CLASS
LOW n = 49
MEDIUM n = 38
HIGH n = 49
LOW n = 49
MEDIUM n = 49
HIGH n = 49
LOW n = 49
MEDIUM n = 49
HIGH n = 49
B1
0.0 (0.0)
0.0 (0.0)
0.0 (0.0)
1.02 (1.020)
8.86 (2.931)
5.18 (2.373)
2.69 (1.037)
1.88 (1.190)
0.0 (0.0)
B2
0.20 (0.130)
11.55 (2.360)
0.65 (0.428)
1.19 (0.345)
15.39 (12.90)
5.45 (1.752)
3.35 (0.958)
3.96 (1.369)
0.45 (0.179)
C
1.69 (1.035)
0.62 (0.360)
0.16 (0.110)
3.63 (1.112)
4.58 (1.726)
3.71 (1.405)
0.63 (0.272)
1.20 (0.528)
1.26 (0.935)
D
0.15 (0.105)
1.63 (1.334)
0.02 (0.020)
3.68 (0.937)
6.18 (1.633)
0.78 (0.315)
0.14 (0.109)
5.56 (2.443)
0.03 (0.023)
CANOPY
85.49 (1.976)
67.25 (4.018)
84.58 (2.145)
77.24 (3.286)
69.06 (3.258)
75.30 (2.802)
82.99 (2.918)
74.94 (3.100)
83.90 (1.716)
MOSS (cm)
0.02 (0.01)
0.18 (0.054)
0.01 (0.007)
0.46 (0.096)
0.55 (0.082)
0.25 (0.037)
0.04 (0.023)
0.34 (0.054)
0.01 (0.006)
LITTER (cm)
1.27 (0.118)
1.63 (0.110)
1.73 (0.143)
1.16 (0.119)
1.56 (0.139)
1.57 (0.157)
1.40 (0.088)
1.78 (0.138)
1.34 (0.108)
TREE DENSITY
29.80 (2.260)
9.20 (1.172)
58.90 (9.579)
17.30 (1.106)
20.40 (2.349)
16.80 (1.467)
32.80 (3.980)
20.10 (2.100)
22.10 (1.703)
DBH (cm)
21.02 (0.773)
34.41 (2.520)
14.55 (0.669)
23.90 (1.022)
26.98 (2.023)
29.10 (3.075)
12.50 (0.726)
24.61 (2.162)
22.25 (1.464)
DISTANCE (m)
1.03 (0.091)
0.85 (0.091)
0.64 (0.051)
1.12 (0.089)
0.73 (0.051)
0.67 (0.054)
1.17 (0.111)
0.75 (0.065)
0.61 (0.050)
DIAMETER (cm)
21.24 (1.792)
21.38 (1.491)
22.34 (1.396)
25.88 (1.823)
18.81 (1.214)
25.48 (2.737)
23.99 (1.780)
21.12 (2.005)
22.37 (1.393)
LENGTH (m)
2.92 (0.286)
5.22 (0.751)
5.00 (0.547)
5.14 (0.484)
4.29 (0.359)
5.08 (0.480)
4.99 (0.362)
4.57 (0.447)
6.12 (0.531)
DC
2.41 (0.061)
2.69 (0.050)
2.41 (0.075)
2.37 (0.046)
2.46 (0.059)
2.46 (0.051)
2.49 (0.065)
2.35 (0.046)
2.22 (0.043)
VARIABLE
16 watershed, however the relatively low canopy presence on all grids in the Seymour watershed caused a watershed x wood volume class interaction. Mean moss depth (MOSS) showed both severe deviations from normality and heteroscedasticity, and was analyzed with a two-way non-parametric ANOVA on ranked data. Moss depth varied differently with wood volume class among watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 18.830, P < 0.005). Although moss depth was greatest on 'medium' grids in each watershed, the Seymour watershed had a relatively deep moss layer on all grids causing a watershed x wood volume class interaction. Mean litter depth (LITTER) was transformed with a logarithmic transformation. Mean litter depth did not vary differently with wood volume class across watersheds (n = 430 vegetation plots, n = 3 watersheds, n = 3 volume classes, F = 1.936, P > 0.1). Litter depth did not vary among watersheds (F = 0.976, P > 0.1), but did vary among wood volume classes (F = 7.604, P < 0.005). Mean litter depth was lower on the 'low' grids (P < 0.005) than on the 'medium' grids. 'Medium' and 'high' grids had similar litter depths (P> 0.1). Mean plot density of trees (TREE DENSITY) varied differently with wood volume class across watersheds (n = 10 plots, n = 3 watersheds, n = 3 volume classes, F = 25.027, P < 0.005). Tree density tended to be lower overall on the grids in the Seymour watershed, and low on 'medium' grids which contributed to the watershed x volume class interaction. Mean tree DBH on each grid varied differently with wood volume class across watersheds (n = 10 plots per grid, n = 3 watersheds, n = 3 volume classes, F = 28.830, P < 0.005). The 'medium' grids tended to have larger DBH trees than 'low' and 'high' grids, and Seymour watershed tended to have larger DBH trees than Capilano or Coquitlam watersheds which contributed to the watershed x volume class interaction. Distance to the four closest pieces of wood >6 cm in diameter (DISTANCE) was transformed with a logarithmic transformation. DISTANCE did not vary differently with
17 wood volume class across watersheds (n = 430 plots, n = 3 watersheds, n = 3 volume classes, F = 0.690, P > 0.1). Wood dispersion did not vary across watersheds (F = 0.041, P > 0.1) but differed across wood volume classes (F = 32.664, P < 0.0005). Neighbouring pieces of downed wood on 'low' grids were further apart (P < 0.0005) than on 'medium' or 'high' grids. Dispersion of downed wood on 'high' grids was not significantly different from that on 'medium' grids (P = 0.06). Mean length of the four closest pieces (LENGTH) was also transformed with a logarithmic transformation. LENGTH varied differently with wood volume class across watersheds, causing a watershed x volume class interaction (n = 430 plots, n = 3 watersheds, n = 3 volume classes, F = 3.462, P < 0.05). Mean diameter of the four closest pieces (DIAMETER) was transformed with a logarithmic transformation. DIAMETER did not vary differently with wood volume class across watersheds (n = 430 plots, n = 3 watersheds, n = 3 volume classes, F = 1.644, P > 0.1). DIAMETER did not differ significantly among watersheds (F = 0.646, P > 0.1), but did vary among wood volume classes (F = 3.048, P < 0.05). Multiple comparisons showed that the four closest pieces on 'low' grids tended to have a larger log-mean diameter than 'medium' grids (P = 0.064). Decay class of the four closest pieces of wood (DC) varied differently with volume class across watersheds, causing a watershed x volume class interaction (n = 430 plots, n = 3 watersheds, n = 3 volume classes, F = 5.113, P < 0.005). Discussion Downed wood volumes were different on 'low', 'medium', and 'high' grids, and this difference was consistent across watersheds. The range of wood volume across volume classes was high (250-358
/ ha on 'low', 443-486 m^ / ha on 'medium', and
833-1077 nr)3 / ha on 'high; Fig. 3a). Recent logging techniques were estimated to leave an average wood volume of 280 m^ / ha on clear-cuts (Howard 1981). Many of the large logs on the 'high' grids appeared to be left during logging operations prior to
18 1936; they did not appear to have broken off or blown over. Douglas-fir logs are heavily decayed after approximately 80 years on the forest floor (Means et al. 1985). The two most likely contributors to decayed downed wood on these study sites, Douglas-fir and western hemlock, show a similar rate of decay (Erickson et al. 1985, Edmonds 1987). Many of these logs, although hollow, still retained their shape, which is unexpected for logs that have been on site for 60-80 years (Grier 1978). Therefore, these large, relatively undecayed logs were most likely western redcedar, which decay more slowly than other species (Daniels et al. in press). Some large logs showed evidence of windthrow; selective logging of some of the sites left residuals which have since fallen. Most of the heavily decayed wood of moderate size classes was probably added during logging or shortly thereafter; windthrow appeared to be the primary contributor of less decayed logs. The disproportionate presence of large pieces of downed wood on 'high' grids heavily influenced the volume estimate on those areas. Because the amount of ground covered by logs is not proportional to the volume of wood on an area, the volume classes may not represent an increase in the amount of ground covered by downed wood. I obtained a better measure of the amount of ground covered by wood by calculating the volume of wood in relatively narrow diameter classes. This measure confirmed that the amount of ground covered by wood increased from 'low' to 'high' grids, because 'high' grids on average had greater wood volume in every diameter class than 'medium' grids, which had greater volumes than 'low' grids (Fig. 4). Amount of vegetative cover varied among watersheds and wood volume classes. In general, among wood volume classes, 'medium' grids had the greatest amount of understory vegetative cover and trees with the largest DBH. Among watersheds, the Seymour watershed had the most understory vegetative cover and trees with the largest DBH. All Seymour grids had relatively sparse canopy cover. Site classification, when interpreted in conjunction with these other variables merits interpretation (Table 2). Grids classified as medium or rich growing sites tended to
19
have more vegetation, although the heavy cover in the B2 layer on Capilano 'medium' remains an outlier. No evidence of tree competition in the form of dead or stunted trees was seen on the Capilano 'medium' grid. Thus, the relatively open canopy is most likely a function of regeneration patterns on the site. Downed wood attributes followed those expected within different wood volume classes. No pattern in log size differences (e.g. length or diameter) among watersheds was found, indicating that logging impacts on piece size and dispersion were similar across watersheds within wood volume classes. Original pre-logging area composition may have been different on 'high' grids than on 'medium' or 'low' grids. The presence of large, old, relatively solid logs on 'high' grids suggests that western redcedar was present on these sites. A few similar logs were seen on 'medium' grids but none on 'low' grids indicating that western redcedar was absent or uncommon on these sites. The dispersion of pieces around trap stations as measured by distance to the four closest pieces corroborated my preliminary assessment that 'high' grids had greater woody ground cover than 'medium' grids which had more than 'low'. Wood pieces were more dispersed on 'low' grids than 'medium' and 'high' grids and mean dispersion of pieces on 'medium' grids was greater than that on 'high' grids (P = 0.06).
20 Chapter 3. Shrew Populations and Habitat Associations Introduction Vegetation and downed wood are important habitat components for small mammals. Vegetation provides food (for herbivorous species) and cover (shelter, and escape, e.g. Morris 1979, Carter 1993). Downed wood also provides cover, as well as nesting and foraging sites (Cowan and Guiguet 1975, Thomas 1979, Maser etal. 1981, Van Home 1981, Kaufman etal. 1983, Maser and Trappe 1984, Harestad and Shackleton 1990, Tallmon and Mills 1994). A few studies have investigated the importance of downed wood for travel by deer mice (Barnum et al. 1992, Planz and Kirkland 1992, Carter 1993). Shrews are voracious insectivorous mammals. They have an extremely high mass-specific metabolic rate, and so require constant access to adequate food sources. As a result, they are continuously active (Churchfield 1990). Members of the genus Sorex require between 80 and 125 percent of their body weight in food daily, and at times of nutritional stress (i.e. pregnancy or lactation) may require up to 3 times their body weight in food (Churchfield 1990). Individuals of the smaller shrew species have a maximum life span of approximately 16 months, reproducing in their second calendar year; larger species have a slightly longer life expectancy (Churchfield 1990). Because of their high metabolism and narrow prey-base, shrews may be more heavily influenced by changes in their habitat than larger species that are able to move relatively long distances and take advantage of a variety of habitats and food sources (Teferi and Millar 1993). Most studies on habitat use and the impact of habitat changes on small mammals have concentrated on larger species such as mice (Peromyscus spp.) and voles (Clethrionomys and Microtus spp.), which are more numerous and can be studied more easily than shrews. Few studies in North America have examined a shrew population in its natural habitat by livetrapping (but see Buckner 1966, Hawes 1975, Cawthorn 1994); most
21 have relied on shrews inadvertently or purposefully killed during other studies for analysis of habitat associations (Getz 1961, Spencer and Pettus 1966, Brown 1967, Choate and Fleharty 1973, Richens 1974, Wrigley et al. 1979, Porter and Dueser 1982, Sullivan and Sullivan 1982, Yahner 1982, Dickman and Doncaster 1987, Snyder and Best 1988, Doyle 1990, Innes era/. 1990, Pagels era/. 1994). I studied shrew populations on areas with differing amounts of downed wood to address my three study objectives: 1) determine if areas in second-growth forest with greater amounts of downed wood had different shrew species than areas with low amounts of downed wood, 2) determine if areas in second-growth forest with greater amounts of downed wood had larger or more reproductively active shrew populations, 3) determine if other habitat components are associated with larger and more reproductively active shrew populations. Methods I established a 1.0-ha small mammal-trapping grid within each of three study areas described in Chapter 2. Each grid consisted of a 7 x 7 array of trap stations spaced 14.3-m apart. At each trap station I baited a single Longworth livetrap with oats, raw beef or pork, and carrot. Coarse brown cotton was provided for insulation. Traps were placed within 1-m of the station beside or under a log or rock. Grids were pre-baited approximately two weeks prior to the beginning of trapping. In 1991, trapping occurred from mid-July until the end of November. During this period, each grid was trapped for three consecutive days, every three weeks. In 1992, in addition to Longworth traps, I placed pitfall traps (hereafter "pitfalls") flush with the ground at each trapping station. Pitfall traps consisted of two coffee cans duct-taped together resulting in a trap approximately 12-cm across and 24-cm deep. A 1-litre plastic tub with its bottom cut out set into the trap provided a funnel into the trap, and prevented animals from jumping out. I used the tub's lid to close pitfalls when not in use. I placed a bit of raw beef or pork in each pitfall trap during each trapping
22 session to serve as a food source. The bottom of the pitfall was cool, thereby limiting the odour that the raw meat produced. Pitfall traps were placed within 1 m of the station beside a different log or rock than were the livetraps. Livetraps were pre-baited approximately two weeks prior to the beginning of trapping. Using both types of traps, I trapped each grid one day per week from the third week of June until the end of August in 1992. Shrews have a very high mortality rate under normal trapping procedures because they starve within 2-3 hours without food (Sullivan and Sullivan 1982, Churchfield 1990). To minimize shrew mortality, I opened traps for approximately 8 hours each day and checked traps every two hours. I found the addition of a food source to pitfall traps to be essential. After two mortalities in pitfall traps at the beginning of the field season in 1992,1 had the shrews necropsied by Dr. HarveyClarke from the U.B.C. South Campus farm to ensure that bodily trauma from the fall into the pitfall was not a cause of death. He found that, even though I had been checking traps every 11/2 hours, shrews were dying from starvation. After the addition of food to the pitfall traps, the mortality rate was < 0.5% during trapping. Traps were closed (pitfall) or locked open (Longworth) over night. Although shrews can survive up to six hours in a trap when provided with food, stress associated with the experience causes a higher mortality rate after release than those animals kept in a trap for a shorter time (W.J. McShea, pers. comm.). I considered this trapping regime to be the most practical and safe way to capture shrews. All individuals caught were identified to species, sexed (if possible), weighed with a Pesola® spring balance and ear-tagged with an individually numbered fish fingerling tag. I also placed spots of hair-dye on each shrew in a unique pattern to identify individuals. Shrews caught in 1991 were identified to species by examining front incisors, third and fourth unicuspids, and toe pads (van Zyll de Jong 1983). I used a headband magnifier during this procedure, which freed my hands to better handle the shrew. In 1992,1 used Metofane® to briefly anaesthetize shrews, allowing me to more
23 accurately identify species. Shrews were placed in a jar with a small amount (approximately 0.5 cc) of the anaesthetic and watched closely as they lost consciousness, which took approximately 30 seconds. Shrews were then placed on a board with loose Velcro straps to hold them in place during inspection. Most shrews regained consciousness within a minute of being removed from the anaesthetic. If a shrew started to regain consciousness before its examination was complete, I placed a nose cone containing a small amount (approximately 0.1 cc) of Metofane® over the shrew's head to keep it unconscious. I did not ear-tag shrews in 1992 because 1991 results indicated that shrews quickly lost their tags, and that tagging damaged their ears. Shrews were kept warm and released upon full recovery from the effect of the anaesthetic; all other animals were released at point of capture immediately following data collection. Analyses Differences in species presence in watersheds and volume classes were tested with two, one-way ANOVAs. Alpha was adjusted to 0.025 to ensure that the experiment-wise type-1 error rate did not exceed 0.05 (Beal and Khamis 1991). I used paired t-tests on arcsine squareroot-transformed trappability data to determine whether average trappability was different between years. I tested for differences in trappability among watersheds and wood volume classes by year using two, one-way ANOVA's. Alpha was again adjusted to 0.025. Mark-recapture data were analyzed using Small Mammal Programs for MarkRecapture Data Analysis (Dr. C.J. Krebs, Department of Zoology, University of British Columbia, 1995). I used the Jolly-Seber (hereafter "JS", Jolly 1965, Seber 1982) estimate of population size and trappability (the probability of capturing an individual known to be in the population). The JS estimate is more robust to low trappability and unequal catchability than the Minimum Number Alive (hereafter "MNA") estimator
24 (Efford 1992, Hilborn and Krebs 1992), and provides a less biased population estimate than MNA under most conditions (Nichols and Pollock 1983). Variation in population size with watershed or wood volume class was analyzed using a split-plot ANOVA, with watershed as the block, grid as the plot, and time (population estimates) as the sub-plots (Sokal and Rohlf 1981). This is similar to a repeated-measures ANOVA, except that the within-subject factor (time) is considered a random variable (von Ende 1993). I compared the ranking of shrew populations with respect to size between years with the Spearman rank correlation (Conover 1980). I classified shrews caught in 1992 as either transient (caught only once) or resident (caught >once). To determine if the more frequent trapping rate in 1992 influenced population estimates, I calculated the proportion of residents including only animals with captures at least 2 weeks apart, and again including only individuals with captures at least 3 weeks apart. Within each shrew class (resident or transient), I classified each shrew as being in one of three groups: those caught solely in pitfalls, solely in livetraps, or in both types of traps (residents only). I tested whether pitfall traps or livetraps were more efficient at capturing shrews with Wilcoxon signed-ranks tests. I counted the number of reproductive females captured on each study grid in 1991 and 1992. I performed a linear regression analysis to determine if the number of reproductively active females was related to mean estimated population size of shrews. I counted the number of juvenile shrews captured on each grid. Juvenile shrews have been previously defined as those 5 times was calculated with the computer program Calhome (Kie et al. 1994). I examined the relationship between number of captures and range size with linear regression analysis to determine if five capture locations were adequate to calculate ranges (Batzli and Henttonen 1993). I used linear regression analysis to determine relationships between range size and mean distance travelled within wood volume class, watershed, and shrew size (estimated by body weight). Alpha was adjusted to 0.025 to compensate for experiment-wise error. Two-sample t-tests on ranked data were used to test for differences in range size and mean distance travelled among species; a was again adjusted to 0.025 to compensate for multiple tests. I drew ranges of all shrews caught at least twice to identify shrews with overlapping ranges.
26 Results (1) Species I caught three species of shrew in the Vancouver watersheds: the vagrant shrew (Sorex vagrans), common shrew (S. cinereus), and dusky shrew (S. monticolus, nomenclature follows Cannings and Harcombe 1990). I also caught red-backed voles {Clethrionomys gapperi; 23 in 1991, 9 in 1992), mice (Peromyscus spp.; 12 in 1992, 17 in 1992), and shrew-moles (Neurotrichus gibbsii; 1 in 1991, 4 in 1992). These latter species were not caught in sufficient numbers to permit analysis. Numbers of shrews captured by species were calculated for 1992, when the species identification procedure was improved. The dusky shrew was the most frequently captured species, followed by the common shrew, and vagrant shrew, respectively (Fig. 5). Number of shrews caught of each species was not related to wood volume class (n = 3 wood classes, F = 1.01, F = 1.46, F = 0.66; P> 0.1 for S. monticolus, S. vagrans, and S. cinereus, respectively). Although S. vagrans appeared to comprise a greater proportion of the shrew populations in the Coquitlam watershed than in the Capilano or Seymour watersheds, numbers caught of each species did not vary among watersheds (n = 3 watersheds, F = 0.41, F = 2.79, F = 5.37 respectively; P > 0.05). (2) Estimating shrew populations a) Trappability Mean trappability differed between years (f = 2.45, P < 0.05), but did not vary with watershed (F = 0.51, F = 0.03, P > 0.1) or wood class (F = 0.24, F = 3.89, P> 0.05) in 1991 or 1992 respectively. In general, trappabilities were lower in 1992 than in 1991. Mean (and SE) trappabilities are presented in Table 4 for descriptive purposes. Mean trappabilities for the watersheds were: Capilano 54.6%, Seymour 55.4%, and Coquitlam 56.5%.
27
Figure 5. Proportions and total number (n) of shrews captured per grid of three species of shrews caught in 1992 in the Vancouver watersheds, B.C. Grids were classified as having 'low" (L), 'medium' (M) and 'high' (H) amounts of downed wood. Sm = Sorex monticolus, Sc = S. cinereus, Sv - S. vagrans.
28
Table 4. Mean (and SE) of Jolly trappability in 1991 and 1992 for each grid in the Vancouver watersheds, B.C.
1991 WATERSHED
CAPILANO
SEYMOUR
COQUITLAM
WOOD CLASS
1992
% TRAPPABILITY
SE
TRAPPABILITY
SE
LOW
80.0
27.76
61.9
10.52
MEDIUM
74.4
24.18
61.4
9.79
HIGH
100.0
0.00
40.7
14.53
LOW
60.0
33.99
48.5
9.95
MEDIUM
100.0
0.00
60.4
9.58
HIGH
20.0
27.8
57.2
11.98
LOW
100.0
0.00
55.6
20.26
MEDIUM
84.5
24.66
67.4
10.07
HIGH
76.4
21.11
46.6
11.20
%
29 b) Shrew populations Capture data for individual species (except S. monticolus) were insufficient to estimate population sizes separately. Also, the sex of shrews was very difficult to identify; only pregnant/lactating females and very scrotal male shrews could be accurately identified. Therefore, shrew data were combined with respect to species and sex within each year to obtain shrew population estimates (Figs. 6-8, Table 5). Because the trapping regime changed between 1991 and 1992,1 analyzed recapture data separately for each year. The relatively low trappabilities ( 0.1). Populations did not differ among wood volume classes ( F = 1.17, P > 0.1), but tended to differ among watersheds (F = 2.98, P = 0.089). The Coquitlam watershed tended to have larger populations (x = 7.10 shrews / ha) than did the Capilano and Seymour watersheds (x = 3.07, 2.91 shrews / ha, respectively). The high average for the Coquitlam watershed was due primarily to the 'high' grid; during one trapping session I captured nine new animals which resulted in a large estimated population size for that week (Fig. 8c), thus increasing mean population size. Data for 1992 showed a wood class x watershed interaction (n = 3 watersheds, n = 3 wood classes, n = 9 population estimates, F = 2.596, P < 0.05). 'Medium' and 'high' grids had similar population sizes (x = 10.9, 9.0, respectively). 'Low' grids had the smallest population in each watershed (x = 5.6). Grids in the Seymour watershed had large population sizes within all wood classes relative to the other watersheds. However, population sizes varied within wood class across watersheds, causing the wood x watershed interaction. Overall, mean estimated 1992 populations were larger than 1991 populations (Z = 1.955, P= 0.05). A Spearman rank correlation performed on 1991 and 1992 data indicated that mean population size did not follow the same trend across grids between years (p = -0.361, P > 0.05).
30 A) 'Low' 35. 30
S
20.
1991
B) 'Medium' 35 30
S •a
25
Ia S
15 20.
g
1991
1992
C) 'High' 35
S
30
•a
25
§
1991
1992
Figure 6. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Capilano watershed, B.C.
31 A) 'Low' 35 30
8 "» g
25
3
20
1991
1992
B) 'Medium' 35 30
a TX
25
C) 'High'
1991
199Z
Figure 7. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Seymour watershed, B.C.
32 A) 'Low' 35.
8
* c
0
1
Ia
30 25 20. 15
1991
1992
C) 'High' 35
1991
1992
Figure 8. Jolly-Seber shrew population estimates for grids on 'low', 'medium' and 'high' downed wood volume classes in the Coquitlam watershed, B.C.
33
Table 5. Mean Jolly-Seber shrew population estimates for each study grid in the Vancouver watersheds, B.C.
A)
1991 CAPILANO
WOOD CLASS
SEYMOUR
COQUITLAM
MEAN
SE
MEAN
SE
MEAN
SE
LOW
1.4
0.51
1.4
0.93
4.4
1.21
MEDIUM
4.8
0.97
4.4
2.73
4.1
2.84
HIGH
3.0
0.89
0.2
0.20
12.7
5.96
B)
1992 CAPILANO
WOOD CLASS
SEYMOUR
COQUITLAM
MEAN
SE
MEAN
SE
MEAN
SE
LOW
6.4
0.93
9.0
1.62
1.3
0.36
MEDIUM
8.0
1.40
16.2
3.00
8.6
1.57
HIGH
8.9
2.33
10.8
1.92
7.4
1.34
I calculated population estimates for each grid in 1992 including only data from livetrap captures to identify whether pitfall traps influenced 1992 population estimates (Table 6). This is a conservative estimate, because animals caught in pitfalls cannot be caught in livetraps simultaneously. A comparison of the ranking of grids using only livetrap capture data from 1992, with all 1992 capture data, showed that the rankings were similar (p = 0.736, P < 0.05). Comparing 1992 mean estimated population sizes calculated with livetrap data, with 1991 population estimates showed that populations did not change significantly between years (Z = 0.89, P > OA), however, they were ranked differently (p = -0.215, P > 0.05).
c) Effect of trapping rate on 1992 population estimates To illustrate the influence of the increased trapping rate (every week) on 1992 population estimates, I calculated the proportion of residents (those shrews caught more than once), including only shrews caught two weeks or more after their initial capture. The percentage of residents fell from 64.3% (for the 1-week trapping interval), to 54.8%. When I used a three week interval, the percentage fell further to 45.2%. The lower number of individuals classified as 'resident' would result in lower population size estimates. This estimate is only slightly higher than the 34.1% of shrews classified as resident in 1991. (3) Trap success in 1992 I caught and marked 126 shrews in 1992 (Fig. 9). Eighty-one (64.3%) of these shrews I considered resident, and 45 shrews (35.7%) transients. Of the 81 resident shrews, 58 (71.6%) were caught by both livetraps and pitfall traps. Of the 23 resident shrews caught in only one type of trap, more were caught in pitfall traps (16, 69.5 %) than in livetraps (7, 8.6%; n = 9 grids, Z = -2.25, P < 0.05). More transient shrews were caught in pitfalls (29, 64.4%) than in livetraps (16, 35.6 %; n = 9 grids, Z = -2.588, P < 0.05). Overall, shrews were caught more frequently in pitfall traps than livetraps:
35
Table 6. Mean 1992 Jolly-Seber shrew population estimates for each study grid in the Vancouver watersheds, B.C. using only capture data from livetraps.
CAPILANO
WOOD CLASS
SEYMOUR
COQUITLAM
MEAN
SE
MEAN
SE
MEAN
SE
LOW
4.5
0.9
5.9
1.33
0.6
0.06
MEDIUM
4.9
1.23
14.2
3.31
5.9
1.31
HIGH
2.8
0.95
5.2
1.17
3.5
0.98
36
A) Resident shrews • Both ffl Pitfalls • Livetraps
L M H
n
(8)
(8)
(9)
Capilano
L
M
(9) (15)
H
(11)
Seymour
L
M
(3) (11)
H
(7)
Coquitlam
B) Transient shrews
L M H n
(1)
(4)
L M H
(7)
Capilano
(9) (10)
(7)
Seymour
L M H
(1)
(3)
(3)
Coquitlam
Figure 9. Proportions and total number (n) of shrews caught by both trap types, or only by pitfalls or livetraps in 1992 of A) residents (shrews caught > once), and B) transients (shrews caught only once). Grids were classified as having 'low' (L), 'medium' (M), and 'high' (H) wood volumes.
37 103 (81.7 %) shrews were caught in pitfall traps at least once, and 81 (64.3%) were caught in livetraps at least once. (4) Population dynamics a) Reproductive females Number of reproductive females in 1991 was low, ranging from 0 to 2 per grid, representing between 0% to 71.4% of the population (Fig. 10). There was no relationship between the average estimated population size and the number of reproductive females (n = 9 grids, r = 0.006, F = 0.040, P > 0.1), primarily because 2
there were no reproductively active females caught on the grid with the largest population. A regression solely on the other 8 grids (excluding Coquitlam 'high') showed a strong positive relationship (n = 8 grids, r = 0.729, F = 16.15, P < 0.05) 2
between population size and number of reproductive females. Number of reproductive females on study grids in 1992 varied from 0 to 6 (0% to 37% of the population). Again, grids with larger population sizes had disproportionately more reproductive females than grids with smaller populations (n = 9 grids, r = 0.707, F = 16.88, P < 0.05). 2
b) Juveniles I caught few juvenile shrews in both years of the study. Seven of 108 shrews caught in 1991, and four of 126 caught in 1992 were juveniles. In 1991,1 caught one 3g shrew on Seymour 'medium' which I did not recapture. I caught six other shrews 0.1). a) Within volume classes I distinguished between microhabitat surrounding successful and unsuccessful trap sites on 'low' grids using discriminant function analysis (n = 147 plots, n = 3 grids, Wilks' X = 0.83, F = 1.92, P < 0.05). Successful and unsuccessful sites were more precisely discriminated in the second analysis (Wilks' X = 0.86, F = 1.84, P < 0.05, Table 7). Overall success in classification was 70.7%, greater than the classification success expected by chance (54.3%). Microhabitats associated with successful and unsuccessful sites were identified on 'medium' grids (Table 8). The first analysis using all of the variables was unable to discriminate between sites (n = 136 plots, n = 3 grids, Wilks' X =0.85, F = 1.63, P = 0.086). By using the variables that had a canonical coefficient >0.2 in the first analysis, I was able to discriminate between sites (Wilks' X = 0.88, F = 3.64, P < 0.005). Overall success in classification was 65.4%, higher than that expected by chance (56.8%). I was also able to discriminate between successful and unsuccessful sites on 'high' grids (Table 9). The first analysis using all variables was unable to discriminate between cases (n = 147 plots, n = 3 grids, Wilks' X = 0.884, F = 1.23, P > 0.1). The second analysis, using the highest loading variables (>0.2) from the first analysis was able to discriminate trap success (Wilks' X = 0.91, F = 2.98, P < 0.05). Overall success in classification (65.3%) exceeded that expected by chance (53.9%).
41
Table 7. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'low' grids in the Vancouver watersheds, B.C.
DISCRIMINANT FUNCTION
VARIABLE
CANONICAL CORRELATION
CLASSIFICATION
TRAP SITES
% CORRECTLY CLASSIFIED
MOSS
0.70
UNSUCCESSFUL
67.3
DC
0.42
SUCCESSFUL
72.6
D
0.41
TOTAL
70.7
CANOPY
-0.34
B1
0.29
LITTER
-0.26
VOLUME
0.24
42
Table 8. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'medium' grids in the Vancouver watersheds, B.C.
DISCRIMINANT FUNCTION
VARIABLE
CANONICAL CORRELATION
CLASSIFICATION
TRAP SITES
% CORRECTLY CLASSIFIED
DC
0.61
UNSUCCESSFUL
65.1
LITTER
0.49
SUCCESSFUL
65.6
C
-0.34
TOTAL
65.4
MOSS
0.29
LENGTH
0.24
43
Table 9. Variable names and canonical correlations for discriminant function analysis performed to identify differences between successful and unsuccessful trap sites on 'high' grids in the Vancouver watersheds, B.C.
DISCRIMINANT FUNCTION
VARIABLE
CANONICAL CORRELATION
CLASSIFICATION
TRAP SITES
% CORRECTLY CLASSIFIED
VOLUME
0.76
UNSUCCESSFUL
64.2
B1
0.49
SUCCESSFUL
66.0
DIAMETER
0.45
TOTAL
65.3
D
-0.28
44 b) Among species Using discriminant function analysis, I could find no difference in microhabitat surrounding trap sites where S. cinereus and S. vagrans were captured in 1992 (n = 75, n = 41; Wilks' X = 0.91, F = 0.85, P > 0.1). I also could not differentiate between habitat around traps where S. cinereus and S. monticolus (n = 75, n = 146; Wilks' X = 0.97, F = 0.62, P > 0.1); or S. vagrans and S. monticolus (n = 41, n = 146; Wilks' X = 0.96, F = 0.58, P > 0.1) were captured. (7) Shrew ranges Thirty-two shrews were captured five or more times in 1991 and 1992. Number of captures varied from 5 to 12, with a mean of 6.6 (± 0.29) captures per shrew. There was a great deal of variability in shrew range size, from 102.2 m2 to 3170
(Table
10). Mean distance travelled varied from 12.2 m to 56.8 m. Range size did not vary with number of captures (fi = 0.00, F = 0.11, P > 0.1). Mean distance travelled between captures also did not vary with number of captures (r? = 0.03, F = 2.19, P > 0.1). Thus, five captures appeared adequate to provide estimates of range size and mean distance travelled. Mean size of shrew ranges did not differ with wood class (F = 0.25, P > 0.1) or watershed (F = 0.77, P > 0.1). Mean distance travelled also did not vary with wood volume class (F = 0.61, P> 0.1) or watershed (F = 0.41, P> 0.1). Maximum shrew weight was used as an index of shrew size. Neither range area (adjusted r2 = 0.29, F = 1.17, P> 0.1) nor mean distance travelled (adjusted r2 = 0.00, F = 1.14,P>0.1) varied with shrew size. Mean range size and mean distance travelled did not differ among shrew species in 1992 at a = 0.025 (S. monticolus and S. vagrans; n = 21, n = 3; range t = 2.75, P = 0.03; distance t = -1.03, P > 0.1. S. cinereus and S. monticolus, n = 5, n = 21; range t = 0.68, P > 0.1; distance; t = 0.07, P > 0.1. S. cinereus and S. vagrans, n = 5, n = 3; range f = -0.88, P > 0.1; distance t = -0.46, P> 0.1).
Table 10. Number of captures (N), species identification, weight, mean distance travelled, and 90% range area for each shrew caught > 5 times on study grids in the Vancouver watersheds, B.C. Sm = Sorex monticolus, Sc = S. cinereus, Sv = S. vagrans.
SHREW I.D.
CAPILANO
SEYMOUR
COQUITLAM
N
SPECIES
(9)
MEAN DISTANCE (m)
90% RANGE AREA (m )
WEIGHT
2
LOW
11 12 14
5 6 9
Sc Sm Sc
5 6 5
23.5 29.3 35.3
409.0 306.7 1840.0
MEDIUM
3 20 22 23
5 12 7 6
Sv Sc Sm
6 6 8.5 5
24.4 27.1 10.5 30.7
409.0 818.0 102.2 204.5
HIGH
18
5
Sm
8
34.4
511.2
LOW
10 12 15 20
5 7 8 7
Sm Sm Sm Sm
6 6.5 6 6
27.3 19.8 28.6 28.3
204.5 306.7 818.0 613.5
MEDIUM
8 20 22 23 24 26 30 32
7 7 5 6 6 5 9 7
Sc Sm Sv Sm Sm Sm Sm
5 6 8 6 8.5 6 7.5 8
50.5 13.5 56.8 29.5 30.6 17.3 11.5 15.8
3170.0 306.7 2658.0 409.0 818.0 204.5 204.5 306.7
HIGH
8 9 10 11 13
7 5 9 8 5
Sm Sc Sm Sm Sm
6 5 6.5 7 7.5
12.1 38.7 16.6 23.3 52.3
204.5 613.5 204.5 306.7 1022.0
LOW
5
5
6
48.7
715.7
MEDIUM
14 15 17 20
5 6 9 5
Sm Sm Sm Sm
6 6 6 6.5
12.2 18.5 19.3 20.2
101.2 204.5 409.0 204.5
HIGH
26 30
6 6
Sv Sm
5.5 7
21.9 18.5
511.2 204.5
46 To determine whether shrews had overlapping ranges, I drew ranges by connecting each individual's outermost points of capture (Figs. 11 -19). Most shrew ranges did not overlap. Four grids; Capilano 'low', and 'high', and Seymour 'low' and 'medium' grids, had shrew ranges with substantial (>25 % of area) overlap. Discussion (1) Species The dusky shrew was consistently the most numerous shrew caught in 1992, followed by the common shrew, and vagrant shrew. Although population sizes differed among grids, the rank of each species did not vary, suggesting that the species were consistent in their response to their environments. A simultaneous killtrapping study carried out in different second-growth (40-80-year-old) forested stands within the Vancouver watersheds ranked the abundance of these three shrew species in the same order (Seip and Savard 1993). All three species compete for the same general prey base: invertebrates. Among sympatric species of shrews, resource use is often partitioned by differences in size, anatomical specialization, abundance, microhabitat use, or direct competition (Whitaker and French 1984). Populations of up to nine shrew species co-exist where shrew body size enables partitioning of the food resource (Whitaker and Maser 1976, Whitaker and French 1984, Sheftel 1994). However, shrew species captured in my study have similar mean body sizes, and are considered 'small' relative to shrew species in general. S. vagrans and S. obscurus (which has since been taxonomically reclassified as S. monticolus, van Zyll de Jong 1983), normally attain weights >6 g in their second year, and only reach a mean weight of 4.5 - 5 g during their first year (Hawes 1975). Both of these species are slightly larger than S. cinereus, which has a mean adult body weight of 4.6 g (Innes 1994).
47 .100 m
Key
•
• •• • X
PI
Shrew I.D. 10 11 12 13 14 15 16 18
Species Sc Sc Sm Sc Sc Sm Sm Sm
Sex ? ? ? ? ? ? M ?
Max. Wt. (g) 5 5 6 5 5 6 7.5 5
Weeks captured 51-59 51-56 51-59 52-59 51-56 55-58 55-58 57-59
Figure 11. Resident shrew (captured >once) ranges on Capilano 'low'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
100
Key X
Shrew I.D.
*
16 17 18 19
+
20 22 23 24
•
Species Sm Sm Sm Sm Sv Sc Sm Sc
Sex ? F ? ? ? F ? ?
m
Max. Wt. (g) 4.5 7.5 5.5 5 6
8.5 5 5
Weeks captured 50-52 51-59 51-58 51-53 53-59 53-59 54-58 54-55
Figure 12. Resident shrew (captured >once) ranges on Capilano 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
49
.100 m
Key X
Shrew I.D.
+
20 22 24 25 26 29
m *
•
t
•t
15 17 18
Species Sm Sm Sm Sc Sm Sm Sv Sm Sm
Sex ? F F ? M ? M ? ?
Max. Wt. (g) 5 8 8 5 6.5 5 8 6 5
Weeks captured 51-52 52-59 52-59 53-58 54-56 54-57 57-59 57-59 57-59
Figure 13. Resident shrew (captured >once) ranges on Capilano 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
100 m
Key X
m *
• • +
Shrew I.D. 10 11 12 13 14 15 18 20 23
Species Sm Sm Sm Sm Sm Sm Sv Sm Sm
Sex ? ? F M F ? F ? M
Max. Wt. (g) 5 5 6.5 5 6.5 5.5 6 6 7
Weeks captured 51-59 51-53 51-58 51-58 51-59 52-59 54-57 55-59 57-59
Figure 14. Resident shrew (captured >once) ranges on Seymour 'low*. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
Key
• • • • •
•x H
H
+
*
Shrew I.D. 20 22 23 24 25 26 28 30 32 33 35 37 38 39
Species Sc Sm Sv Sm Sm Sm Sm Sm Sm Sv Sm Sc Sv Sm
Sex ? F ? F ? ? F ? F ? F ? ? ?
Max. Wt. (g) 6 8 6 7 6 6 9 7.5 8 6 9 5.5 6 6.5
Weeks captured 50-59 50-54 50-59 51-55 51-55 52-58 52-55 52-59 53-56 53-59 54-56 54-59 55-59 55-56
Figure 15. Resident shrew (captured >once) ranges on Seymour 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
Figure 16. Resident shrew (captured >once) ranges on Seymour 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
Key X *
+
Shrew I.D. 10 11 13
Species Sm Sm Sm
Sex F F M
Max. Wt. (g) 6.5 7 7.5
Weeks captured 52-59 53-59 53-55
Figure 17. Resident shrew (captured >once) ranges on Coquitlam 'low". Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
Figure 18. Resident shrew (captured >once) ranges on Coquitlam 'medium'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
Key X
EH *
+
H
t
Shrew I.D. 25 26 27 29 30 34
Species Sc Sv Sm Sv Sm Sm
Sex ? ? ? ? ? ?
Max. Wt. (g) 5 5 5.5 6 6.5 6
Weeks captured 50-59 50-59 51-57 53-59 53-59 57-59
Figure 19. Resident shrew (captured > once) ranges on Coquitlam 'high'. Dots indicate trap stations. Numbers beside trap stations indicate multiple captures of individuals.
56 Small shrews have smaller, less powerful jaws than larger species; they are usually less robust physically, and are not strong diggers (Terry 1981). Thus, small shrews (4-6 g), are most often surface foragers, while larger species (>8 g) are able to use subterranean microhabitats (Churchfield and Sheftel 1994). The diet of small shrews is comprised primarily of surface-dwelling (84-86%) invertebrates, with soildwelling invertebrates present in much lower proportions (8-12%, Churchfield 1994, Churchfield and Sheftel 1994). Small shrews also differ from large shrews in the size of prey caught. Churchfield (1994) found that the majority of prey consumed by 'small' species varied between 3 and 10 mm in length. Because the 3 shrews in my study area are similar in body size, they likely prey on the same types of insects, and their consistent differences in abundance must be due to some factor other than partitioning through body size. (2) Estimating shrew populations a) Trappability Although I caught more shrews in 1992 (126 vs 108), the mean trappability in 1992 tended to be lower. Krebs and Boonstra (1984) suggested two scenarios where reduced trappability might occur: increased competition for trap sites, or reduced trapping intensity. Competition for a trap in an area of local shrew abundance was not likely to occur because I released animals from livetraps every two hours, and used pitfall traps which can capture more than one animal at a time. Trapping rate likely affected the trappability estimate to some degree. Mean Jolly trappability (%, Krebs and Boonstra 1984) is defined as: (Z (Total number of marked individuals caught at time i / estimated population size at time i) x 100) / Number of sampling periods With the same number of captures, an increase in the trapping rate causes a decrease in the trappability estimate (Krebs and Boonstra 1984). Shrews captured in 1992 were often not recaptured every week, depressing the trappability estimate. Although the
57 frequent trapping rate provided a more detailed record of changes in the shrew populations; one trapping-day per trap-session may not have been enough to ensure a high probability of capturing animals each trapping session.
b) Shrew populations Shrew populations change in composition annually (Hawes 1975, Innes et al. 1990, Sheftel 1994, Cawthorn 1994). The estimates of shrew population size on the grids that I studied differed between 1991 and 1992. Population estimates on all grids except Coquitlam 'low' and 'high' appeared higher in 1992 than 1991. All of the populations studied had large changes in estimated populations within years (Figs. 6-8). The size of the variation is likely exaggerated by the JS estimate. The JS estimate reflects the proportion of new shrews relative to the known population of shrews. Population estimates can vary dramatically, especially in small populations such as the ones in this study where the capture of a new individual results in a large increase in estimated population size. Even with these JS estimates, shrew populations in the Vancouver watersheds, ranging from approximately 1 shrew / ha to a maximum of 16 shrews / ha, tended to be slightly smaller than reported in other live-trapping studies. Hawes (1975) reported MNA estimates of 12 S. vagrans I ha and 5-12 S. monticolus I ha (17-24 shrews / ha in total) in a young second-growth forest in southwestern B.C. Sorex cinereus densities ranged from 1-23 / ha (1-28 shrews / ha in total) in a bog in Manitoba (Buckner 1966) and 3-13 / ha (MNA estimates) in young second-growth deciduous forest (4-19 shrews / ha in total, Cawthorn 1994). On very productive sites, shrew populations can reach much higher densities. Sorex araneus can reach densities of 200 / ha on insect-rich grassland areas (Churchfield 1990). Unfortunately, populations of the dusky, common, or vagrant shrew have not been estimated in areas of greater shrew abundance (e.g. old-growth, Seip and Savard 1993).
58 Second-growth forests have relatively closed canopies which reduces the amount of understory vegetation present. Vegetation on my study grids that could support insect communities was patchy and low in diversity. However, logs on most grids were present in a range of decay classes and sizes, which should increase insect diversity (Maser and Trappe 1994). Shrew abundance was higher in old-growth forests, which usually have greater microhabitat diversity than second-growth forest, in the Vancouver watersheds (Seip and Savard 1993). Shrew populations respond numerically to increased food resources (Holling 1959, Stewart et al. 1989, Churchfield 1990). Therefore, the relatively small populations that I studied may reflect relatively poor food resources on these sites.
c) Effect of trapping rate on 1992 population estimates The proportion of resident shrews (those caught > once) increased between 1991 and 1992. The difference in proportion of residents between years was influenced by the change in trapping rate. By increasing the trapping rate and decreasing the length of individual trapping periods, I increased the chance of capturing an individual in different trapping periods, thus increasing the potential number of captures per individual per unit time. This change increased my 1992 population estimates, making it difficult to ascertain whether shrew numbers actually varied in size between years. (3) Trap success in 1992 I included pitfall traps in 1992 because they are more efficient than livetraps at capturing shrews (e.g. Williams and Braun 1983). Although pitfalls are normally run as 'kill traps', I virtually eliminated trap mortality by restricting the amount of time they were open. A potential bias in small mammal population analysis is differential catchability of individuals or species (Hilborn et al. 1976, Jolly and Dickson 1983). By using two types
59 of traps I was more likely to capture all sex/age classes of the population than with livetraps alone (Innes and Bendell 1988, Szaro et al. 1989). Resident shrews were caught by both livetraps and pitfall traps. However, 20% of residents were caught solely in pitfalls, and 8.5% were caught solely in livetraps, which suggested that although pitfalls were more efficient at capturing shrews, more than one type of trap should be used to reduce bias in the types of individuals captured. Transients were much more likely to be caught in pitfall traps than in livetraps. Shrews, like most small mammals, avoid strange new objects in their environments. Thus pitfall traps, which are placed flush to the ground and do not require an animal to make a conscious decision to enter the trap, are likely to be more successful at capturing animals. The success of pitfall traps compared to livetraps indicated that shrew populations in 1991 may not have been adequately sampled. Population estimates for 1992 were significantly lower overall when population sizes were recalculated using data only from livetraps. Thus, although shrew populations may have been larger in 1992 than 1991,1 can not determine whether shrew population sizes actually changed because of the change in trapping regime. (4) Population dynamics a) Reproductive females I used the ratio of reproductive females to the mean estimated population size as an indicator of the reproductive health of the populations. Larger populations had a greater proportion of reproducing females, suggesting that areas with larger populations were more productive, and possibly had better shrew habitat than those sites with smaller population sizes. This trend was evident in 1991 (except for Coquitlam 'high'), and in 1992. Grids with the two largest populations in 1992 (x = 16.2,T= 10.8), had the same proportion of reproductive females (37%), while the two grids with a mean estimated population size < 7 had no reproductive females.
60 Populations in 1991 had a much greater range in the proportion of breeding females than in 1992. The apparent large number of breeding females in small populations in 1991 may be related to the inadequately censused populations in 1991. All reproductively active females caught in 1992 (except one female on Capilano 'low') were captured at least once in a livetrap. When only livetrap captures were included for 1992 data, proportions of reproductive females ranged up to 76.9% of the population (4 reproductive females on Seymour 'high' with an estimated population size of 5 shrews), and had similar numbers of reproductive females in relatively small populations as in 1991. Hawes (1975), using Longworth livetraps, captured similar proportions of reproductive females: 59-66% (% of breeding females to estimated populations size (MNA)). Thus, Longworth livetraps appear to be quite effective at capturing reproductive females, relative to the rest of the population. Many reproductive females were captured repeatedly within a single trapping session, eating all of the food in the trap; two sat on my hand as I fed them more food. Reproductively active females have extremely high metabolic demands, and may be more likely to be captured in livetraps because of the large available food source they provide, thus overestimating their numbers in the population.
b) Juveniles Breeding of most shrew species begins in March and continues through August (Churchfield 1990). I caught breeding shrews until the end of September in 1991, and until the end of the trapping season (August 30) in 1992. I caught few juvenile shrews in both years of the study. Although livetraps may not be sensitive enough to capture very small shrews, they can be caught in pitfall traps (Cawthorn 1994). However, the ratio of juveniles in the population decreased in 1992 even with the addition of these less selective traps. Juveniles caught only once were most likely dispersing from the population (Moraleva and Telitzina 1994). Some small shrews caught repeatedly did show a weight increase over the trapping period, from an initial weight of 6 cm in diameter were classified as: (1) log with few signs of decay, bark and limbs present; (2) log fairly soft but still retaining shape, little or no bark or limbs remaining; and (3) log soft, no bark or limbs remaining.
69
Figure 20. Diagram of a shrew trail, illustrating the placement of transects and corridor intervals. Thick, dark lines represent logs, thin lines represent transects, placed every 1 -m along the shrew trail.
70 Analyses I used two, one-way ANOVAs to determine whether trail length varied with wood volume class or watershed. Alpha was adjusted to 0.025 to ensure that the experiment-wise type I error rate did not exceed 0.05 (Beal and Khamis 1991). Within 'low', 'medium', and 'high' wood volume classes I compared the number of pieces of wood >3.1 cm in diameter among corridor intervals with a KolmogorovSmirnov (hereafter "KS") test to determine if corridor intervals close to the shrew trail (00.5 m, 0.5-1 m) had more wood than intervals further away (1-2 m, 2-3 m). I calculated expected values with the assumption that each corridor interval would have an equal number of pieces of wood. Thus, given x number of pieces within the transect, I calculated expected values of number of pieces of wood for each interval as x/4. To compensate for the discrepancy in size of corridor intervals (e.g., 0-0.5 m is 50 cm, 1-2 m is 100 cm), I divided number of pieces of wood counted in the 100 cm corridor intervals in half. To determine if shrew trails were associated with larger pieces of downed wood, I examined differences in distribution of wood pieces within diameter classes by corridor interval, within volume class. Data were divided into 6 diameter classes: 3.1-6 cm, 6.112 cm, 12.1-24 cm, 24.1-48 cm, and >48 cm. I used a KS test to identify greatest differences in number of pieces by diameter class across corridor intervals (described above). Distributions of pieces within diameter classes across corridor intervals were compared to expected values (calculated as above). To estimate continuity of cover along the shrew trail, I converted each corridor interval along the trail to a presence/absence measure of downed wood. I calculated the proportion of trail length that had cover within each subsection, and the variates were transformed with arcsine squareroot. I then used a two-sample t-test to compare continuity of cover within three corridor intervals: 0-1 m, 1-2 m, and 2-3 m from the shrew trail. Alpha was adjusted to 0.0167 to compensate for multiple comparisons.
71 Within the 0-1 m subsection, I compared continuity of cover 0-0.5 m from the trail from that 0.5-1 m from the trail in the same manner. I used a two-way ANOVA to identify differences in the decay class of logs >6 cm in diameter among trail subsections within volume classes. For significant relationships, Bonferroni multiple comparison tests were used to distinguish differences (Sokal and Rohlf 1981). Results Nineteen shrews were tracked. Trail lengths ranged from 4 m to 31 m. Trail length did not vary with wood volume level (F = 2.141, wood levels = 3, P > 0.1), or watershed (F = 0.037, watersheds = 3, P> 0.1). The KS test revealed that the greatest difference between expected and observed numbers of pieces in each wood volume class was in the first corridor interval (0-0.5 m) on 'low' and 'high* wood volume classes, and the first two subsections (0-1 m) on 'medium' grids (Fig. 21). Thus, shrews appeared to choose travel routes associated with downed wood. Vegetation was sparse along the shrew trails; a KS test including all cover data (vegetation and rocks) provided the same results as the downed woodonly test. Number of pieces of wood within each diameter class by volume class revealed that shrew trails were associated with different wood diameter classes in different volume classes (Table 11). On areas with 'low' amounts of downed wood, shrew trails were not associated with pieces 3.1-6 cm in diameter, and were associated with pieces > 6 cm, for travelling. On 'medium' grids, shrew trails also were not associated with small logs (.13-6 cm in diameter), and showed strong association with downed wood >6 cm in diameter. Shrews on 'high' grids were not associated with small to moderatesized logs (3.1-12 cm in diameter), but were associated with logs >12 cm in diameter. When cover data were included in each diameter class (vegetation and rocks), the same results were obtained.
72 A) 'Low'
0-0.5m
0.5-1 m
1-2m
2-3m
Distance from shrew trail
B) 'Medium'
0-O.Sm
0.5-1 m
1-2m
2-3m
Distance from shrew trail
C) 'High'
0-0.5m
0.5-1 m
1-2m
2-3m
Distance from shrew trail
Figure 21. Cumulative frequency distribution of downed wood pieces >3.1 cm in diameter at increasing distances from shrew trails within wood volume classes on study grids in the Vancouver watersheds, B.C. An asterisk (*) denotes the corridor interval with the greatest difference in number of pieces vs. expected (P < 0.05) by Kolmogorov-Smirnov test.
73 Table 11. Cumulative frequency distributions of number of wood pieces by diameter class within corridor intervals. An asterisk (*) indicates a 'greatest difference' (P < 0.05) determined through a Kolmogorov-Smirnov test; (+) or (-) indicates whether the association is positive or negative.
Volume Class Corridor Interval (m)
Low
Medium
High
Diameter class 3.1-6 cm
6.1-12 cm
12.1-24 cm
24.1-48 cm
>48 cm
0-0.5
12.9
28
38.1* (+)
24.4
36.5
0.5-1
45.2
54
55.6
50.9
67.4* (+)
1-2
58.9*(-)
71
72.2
78
88.8
2-3
100
100
100
100
100
0-0.5
20*(-)
29.9
34.5*(+)
35.9
36.9*(+)
0.5-1
51
57.5*(+)
55.5
58.6
56.9
1-2
77
78.5
77.3
87.6*(+)
78.4
2-3
100
100
100
100
100
0-0.5
26
21.6*(-)
30.6
38.0*(+)
43.1*(+)
0.5-1
38.1*(-)
47.1
60.4*(+)
55.1
69.4
1-2
66.9
74.7
77.8
78.7
86.5
2-3
100
100
100
100
100
74 On 'low' grids, where diversity in size of downed wood is relatively low, shrew trails were related to larger pieces of wood (>6 cm) and very strongly related to number of pieces of wood. On 'medium' grids, shrews were more strongly related to larger pieces of wood (>6 cm), as well as number of pieces of wood. On 'high' grids, where there is high diversity in size of downed wood, shrews were strongly associated with downed wood >12 cm in diameter, as well as number of pieces. The corridor within 1 m of the shrew trails had significantly greater continuity of cover than the corridor 1-2 m away from the trail (f = 3.612, df = 33, P < 0.005). The continuity of cover close to (within 1 m of) the shrew trail did not differ significantly from the continuity 2-3 m away (n = 34, f = 2.092, P < 0.05, a = 0.0167); continuity of cover 1-2 m away did not differ significantly from that 2-3 m away (n = 34, t = 0.864, df = 33, P > 0.1). Within the corridor 0-1 m from the shrew trail, the corridor within 50 cm of the trail (0-0.5 m) had significantly greater continuity of cover than that 0.5-1 m away (n = 34, t = 2.539, P < 0.05, a = 0.05). Thus, shrews appeared to choose travel routes with high continuity of cover. Decay class of logs did not vary differently with distance from shrew trail across wood volume classes (n = 1363 pieces of wood, F = 0.63, P > 0.1). Decay class did not differ among trail subsections (F = 0.15, P > 0.1), but did vary among wood volume classes (F = 18.19, P < 0.005). 'Medium' grids (x = 2.49) had more decayed wood than 'low' (x = 2.38) or 'high' (x = 2.28) grids (P < 0.05). Discussion Use of fluorescent pigments allows fine-scale discrimination of microhabitat used by small mammals. As the animal moves, the pigment falls off or is brushed against downed wood, vegetation, and other animals (Kaufman 1989) resulting in a trail that can be followed with an ultraviolet light. This method provides an exact record of an individual's movements and therefore is an accurate reflection of microhabitats used
75 (Graves etal. 1988, Barnum etal. 1992, McShea and Gilles 1992, Planz and Kirkland 1992, Carter 1993). Although the first section of the trails were likely escape routes, shrews did not appear adversely affected by the experience. I captured three shrews within the same trapping day that they were fluoresced. Two shrews were captured two hours after release; they still had a light coating of powder on their body. One shrew was captured four hours after release; it had very little powder left on it. All shrews trailed were recaptured at least once after they were trailed. Shrew movements are constrained by various factors: 1) they have a high metabolic demand and must eat almost continually; 2) unlike deer mice, shrews have very poor vision, and are likely capable only of light/dark discrimination (Branis and Burda 1994). These two factors may increase the association of shrews with downed wood on my study areas. Insects may be more numerous on richer sites (Shvarts and Demin 1994). My study sites were in second-growth forest, and tended to be dry and had little vegetation. Therefore, on my sites, downed wood, which is associated with insects (Maser and Trappe 1984), may represent localized areas of insect abundance. Shrews' poor vision may cause them to rely more heavily on objects such as logs and rocks in their environment for navigation, such as those described for the well-sighted deer mouse (Barry and Francq 1980). Shrews of the Sorex species may have a crude form of echolocation (Buchler 1976, Branis and Burda 1994) which is too insensitive to use while foraging, but may be used while exploring; Blarina spp., have been able to distinguish open and closed tunnels at short distances (Tomasi 1979). Because of their extremely poor vision, it is unlikely that shrews could see far enough to choose a trail with a lot of cover. Three shrews released from the same trap site on Coquitlam 'high , 1
followed the same trail for the first 5 m. All of the shrews moved very quickly and showed no hesitation in their movements. Thus, I believe that shrews demonstrated evidence of a spatial map (Pierce 1987, Churchfield 1990), assisted by light/dark discrimination that allowed them to identify horizontal logs that could provide cover.
76 Shrews appeared to use downed wood differently than do deer mice. Downed wood was used exclusively for travelling beside or under. Unlike deer mice, which often use downed wood as a highway (Barnum etal. 1992, Planz and Kirkland 1992, Carter 1993), shrews did not travel on top of logs. One shrew travelled on top of a log, but only to cross it, at which point the shrew resumed travelling alongside it. Shrews used tunnels quite frequently; 11 of 19 shrews trailed used tunnels while they were covered with fluorescent powder, travelling up to 7 m in a straight-line distance before re-emerging. No mention of tunnel use by deer mice was made in previous studies (Barnum etal. 1992, Planz and Kirkland 1992, Carter 1993). Use of downed wood for travelling by deer mice may be a predator-avoidance tactic; they make less noise than when travelling on the ground, and can move more quickly. Use of tunnels by shrews would serve a similar function. Decay class of logs, which may reflect use of tunnels in heavily decayed logs, did not differ with distance from shrew trails. When shrews were travelling underground in tunnels, I did not collect habitat measurements as I was uncertain of the exact trail. Thus, I may have underestimated shrew use of heavily decayed logs. Two attributes of cover: effectiveness (the degree to which the animal is protected) and continuity, determine how useful cover is to the animal (Harestad and Shackleton 1990). Shrews selected travel routes that had much more cover, as measured by the amount and continuity of downed wood along their travel path, than available in areas farther away from the trail. Other forms of cover (vegetation and rocks) were very patchy; when these data were included it did not influence my results. Because other forms of cover were very sparse, I believe that the primary determinant of shrew travel routes on my study areas was the amount and continuity of cover in the form of downed wood. Shrew trails were associated with larger pieces of cover (>6 cm); small pieces of wood may be ineffective as cover. With increasing diversity and abundance of downed wood, shrew trails were more strongly associated with larger pieces. Thus, it appears
77 that, where shrews have a choice in size of downed wood (in areas of high diversity in number and size of downed wood), they prefer larger pieces.
78 Chapter 5. Conclusions and Recommendations Amounts of downed wood varied as expected across wood volume classes, and were consistent across watersheds. Downed wood attributes were similar within wood classes across watersheds, and indicated that 'high' grids had greater woody ground cover, and a different dispersion pattern of downed wood than 'medium' grids, which were different from 'low' grids. 'High' and 'medium' grids differed primarily in the disproportionate presence of large logs (> 48-cm in diameter) on 'high' grids, but varied also in the overall greater amount of downed wood on 'high' grids, which was reflected by the lower dispersion of downed wood. 'Medium' grids had greater volumes of downed wood than 'low' grids (P =0.064), but differed primarily in the disproportionate presence of medium sized pieces (6-12 -cm in diameter) on 'medium' grids, and the greater dispersion of wood pieces on 'low' grids. Logging practices such as selection logging, and forest attributes such as tree size and species present prior to logging, likely contributed to the variability in size and decay profile of downed wood. I was unable to select only sites that had little or no understory vegetation; small stand sizes and variability in downed wood volumes limited my options. There appeared to be a consistent difference among watersheds and wood classes in site productivity as indicated by vegetative cover. This is a result of more than one factor; general site productivity and regeneration patterns appeared to influence current vegetation patterns. Increased vegetative cover and continuity of downed wood was related both to increased shrew population size, and reproductive status of the population. Shrews were caught more frequently in areas with greater cover and access to foraging sites. In areas with large amounts of downed wood in a range of size classes, shrews became more selective in terms of size of wood pieces used. All three shrew species were caught in similar habitats. Resource partitioning appeared to occur primarily through maintenance of ranges and dominance interactions. In larger
79
populations, range overlap occurred more frequently and was most likely moderated through differential use of microhabitat at a very fine scale. Tracking shrews corroborated the finding that shrews were likely to associate with cover. Shrews chose travel routes that were closely associated with downed wood. Areas immediately surrounding the shrew trails had both greater numbers of pieces of downed wood, and higher continuity of cover, than areas farther away from the trail. Bigger logs provided more effective cover for shrews. On grids with low diversity in size of logs, shrew trails were less strongly associated with size of log, but were very strongly associated with number of pieces. On grids with greater diversity in size of logs, shrew trails were more strongly associated with bigger logs. Within second-growth forests, shrew populations benefited from increased amount and continuity of cover and increased microhabitat diversity. To maximize these elements in managed stands, a range of log piece size in length, diameter (with emphasis on pieces >6 cm in diameter), and tree species should be left on a cutblock. Downed wood pieces should be distributed evenly over the cut-block to provide adequate cover; the amount of wood should be enough to provide a network of logs across the area, without covering the ground to the extent that it seriously impedes vegetation growth. Patches of trees including snags should be left standing; these will fall over time, providing a source of downed wood to the stand. Broadcast burning of cut-blocks should be avoided because it reduces the litter layer, may reduce the amount of downed wood, kills insects already present on the site, and chars wood pieces, which could decrease their usefulness to shrews. Some thinning of dense second-growth forest stands may encourage vegetation growth, and increase microhabitat diversity.
Literature Cited Amaranthus, M., J. M. Trappe, L. Bednar, and D. Arthur. 1994. Hypogeous fungal production in mature Douglas-fir forest fragments and surrounding plantations and its relation to coarse woody debris and animal mycophagy. Canadian Journal of Forest Research 24:2157-2165. Barnum, S. A., C. J. Manville, J. R. Tester, and W. J. Carmen. 1992. Path selection by Peromyscus leucopus in the presence and absence of vegetative cover. Journal of Mammalogy 73:797-801. Barry, R. E. J., and E. N. Francq. 1980. Orientation to landmarks within the preferred habitat by Peromyscus leucopus. Journal of Mammalogy 61:292-303. Barry, R. E., and E. N. Francq. 1982. Illumination preference and visual orientation of wild-reared mice, Peromyscus leucopus. Animal Behaviour 30:339-344. Batzli, G. O., and H. Henttonen. 1993. Home range and social organization of the singing vole Microtus miurus. Journal of Mammalogy 74:868-878. Beal, K. G., and H. J. Khamis. 1991. A problem in statistical analysis: simultaneous inference. Condor 93:1023-1025. Boonstra, R., and I. T. M. Craine. 1986. Natal nest location and small mammal tracking with a spool and line technique. Canadian Journal of Zoology 64:10341036. Branis, M., and H. Burda. 1994. Visual and hearing biology of shrews. Pages 189-200 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania PA. Brown, L. N. 1967. Ecological distribution of six species of shrews and comparison of sampling methods in the central Rocky Mountains. Journal of Mammalogy 48:617-623. Buchler, E.R. 1976. The use of echolocation by the wandering shrew (Sorex vagrans). Animal Behaviour 24:858-873. Buckner, C. H. 1966. Populations and ecological relationships of shrews in tamarack bogs of southeastern Manitoba. Journal of Mammalogy 47:181-194. Cannings, R. A., and A. P. Harcombe, editors. 1990. The vertebrates of British Columbia: scientific and English names. Volume Royal British Columbia Museum heritage Record No. 20; Wildlife Report No. R24. Ministry of Municipal Affairs, Recreation and Culture and Ministry of Environment, Victoria, B.C.
81 Carter, D. W. 1993. The importance of serai stage and coarse woody debris to the abundance and distribution of deer mice on Vancouver Island, British Columbia. M. Sc. Thesis. Simon Fraser University, Burnaby, BC. Cawthorn, J. M. 1994. A live-trapping study of two syntopic species of Sorex, S. cinereus and S. fumeus, in southwestern Pennsylvania. Pages 39-43 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of Carnegie Museum of Natural History No. 18, Pittsburgh, PA. Choate, J. R., and E. D. Fleharty. 1973. Habitat preference and spatial relations of shrews in a mixed grassland in Kansas. The Southwestern Naturalist 18:93-114. Churchfield, S., editor. 1990. The natural history of shrews. Christopher Helm Ltd., London, England. .
. 1994. Foraging strategies of shrews, and the evidence from field studies. Pages 77-88 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania PA.
Churchfield, S, and B. I. Sheftel. 1994. Food niche overlap and ecological separation in a multi-species community of shrews in the Siberian taiga. Journal of Zoology, Lond. 234:105-124. Comrey, A. L. 1973. A first course in factor analysis. Academic Press, New York. Conover, W. J. 1980. Practical nonparametric statistics, 2nd Edition. John Wiley & Sons, Inc., Toronto. Corn, P. S., R. B. Bury, and T. A. Spies. 1989. Douglas-fir forests in the Cascade mountains of Oregon and Washington: is the abundance of small mammals related to stand age and moisture? (from Symposium: Management of amphibians, reptiles, and small mammals in North America (Flagstaff, AZ, July 19-21,1988)) Cowan, I.., and C. J. Guiguet. 1975. The mammals of British Columbia. British Columbia Provincial Museum, Handbook 11. Victoria, B.C. Daniels, L. D., J. Dobry, K. Klinka, and M. C. Feller. In press. Death and decay of Thuja plicata. Canadian Journal of Botany. Dickman, C. R. 1988. Body size, prey size, and community structure in insectivorous mammals. Ecology 69:569-580.
82 Dickman, C. R., and C. P. Doncaster. 1987. The ecology of small mammals in urban habitats 1. Populations in a patchy environment. Journal of Animal Ecology 56:629-640. Doyle, A. T. 1990. Use of riparian and upland habitats by small mammals. Journal of Mammalogy 71:14-23. Economic and Engineering Services Inc. 1991. Watershed management evaluation and policy review. Final summary report. Greater Vancouver Water District. Edmonds, R. L. 1987. Decomposition rates and nutrient dynamics in small-diameter woody litter in four forest ecosystems in Washington, U.S.A. Canadian Journal of Forest Research 17:499-509. Efford, M. 1992. Comment - Revised estimates of the bias in the 'minimum number alive' estimator. Canadian Journal of Zoology 70:628-631. Erickson, H. E., R. L. Edmonds, and C. E. Peterson. 1985. Decomposition of logging residues in Douglas-fir, western hemlock, Pacific silver fir, and Ponderosa pine ecosystems. Canadian Journal of Forest Research 15:914-921. Getz, L. L. 1961. Factors influencing the local distribution of shrews. American Midland Naturalist 65:67-88. Graves, S., J. Maldonado, and J. O. Wolff. 1988. Use of ground and arboreal microhabitats by Peromyscus leucopus and Peromyscus maniculatus. Canadian Journal of Zoology 66:277-278. Green, R.N., P.J. Courtin, K. Klinka, R.J. Slaco, and C A . Ray. 1984. Site diagnosis, tree species selection, and slashburning guidelines for the Vancouver Forest Region. B.C. Min. For., Land Manage. Handbook No. 8, Victoria, B.C. Grier, C. C. 1978. A Tsuga heterophylla - Picea sitchensis ecosystem of coastal Oregon: decomposition and nutrient balances of fallen logs. Canadian Journal of Forest Research 8:198-206. Halfpenny, J. C. 1992. Environmental impacts of powdertracking using fluorescent pigments. Journal of Mammalogy 73:680-682. Harestad, A. S., and D. M. Shackleton. 1990. Cover and use of travel routes by female Townsend's voles in a laboratory arena. Biology of Behaviour 15:196204. Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, G. W. Lienkaemper, J. Cromack, and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Adv. Ecol. Res. 15:133-301.
83 Hawes, M. L. 1975. Ecological adaptations in two species of shrews. Ph.D. Dissertation. University of British Columbia, Vancouver. . 1977. Home range, territoriality and ecological separation in sympatric shrews, Sorex vagrans and Sorex obscurus. Journal of Mammalogy 58:354-367. Hayes, J. P., and M. P. Cross. 1987. Characteristics of logs used by western redbacked voles, Clethrionomys califomicus, and deer mice, Peromyscus maniculatus. Canadian Field-Naturalist 101:543-546. Henttonen, H. 1985. Predation causing extended low densities in microtine cycles: further evidence from shrew dynamics. Oikos 45:156-157. Hilborn, R., and C. J. Krebs. 1992. Bias in the 'minimum number alive' estimator: a reply. Canadian Journal of Zoology 70:632. Hilborn, R., J. A. Redfield, and C. J. Krebs. 1976. On the reliability of enumeration for mark and recapture census of voles. Canadian Journal of Zoology 54:10191024. Holling, C. S. 1959. The components of predation as revealed by a study of small mammal predation of the European pine sawfly. The Canadian Entomologist 91:293-320. Howard, J. O. 1981. Ratios for estimating logging residue in the Northwest. U.S. For. Serv. Res. Pap. PNW-288 (cit. in Erickson et. al. 1985). Innes, D. G. L , and J. F. Bendell. 1988. Sampling of small mammals by different types of traps in Northern Ontario, Canada. Acta Theriologica 33:443-450. Innes, D. G. L , J. F. Bendell, B. J. Naylor, and B. A. Smith. 1990. High densities of the masked shrew Sorex cinereus, in Jack Pine plantations in northern Ontario. American Midland Naturalist 124:330-341. Innes, D.G.L 1994. Life histories of the Soricidae: A review. Pages 111-136 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania PA. Jolly, G. M. 1965. Explicit estimates from capture recapture data with both death and immigration - stochastic model. Biometrika 52:225-247. Jolly, G. M., and J. M. Dickson. 1983. The problem of unequal catchability in markrecapture estimation of small mammal populations. Canadian Journal of Zoology 61:922-927.
84 Kaufman, G. A. 1989. Use of fluorescent pigments to study social interactions in a small nocturnal rodent, Peromyscus maniculatus. Journal of Mammalogy 70:171-174. Kaufman, D. W., S. K. Peterson, R. Fristik, and G. A. Kaufman. 1983. Effect of microhabitat features on habitat use by Peromyscus leucopus. American Midland Naturalist 110:177-185. Kie, J. G., J. A. Baldwin, and C. J. Evans. 1994. Calhome: a home range analysis program. U.S. Forest Service, Pacific Southwest Research Station, CA. July, 1994. Krebs, C. J., and R. Boonstra. 1984. Trappability estimates for mark-recapture data. Canadian Journal of Zoology 62:2440-2444. Lemen, C. A., and P. W. Freeman. 1985. Tracking mammals with fluorescent pigments: a new technique. Journal of Mammalogy 66:134-136. Maser, C , and J. M. Trappe, editors. 1984. The seen and unseen world of the fallen tree. Volume Gen. Tech. Rep. PNW-164. US Dept. Agriculture, For. Serv. PNW Forest and Range Experiment Station, Portland, Oregon. Maser, C , B. R. Mate, J. F. Franklin, and C. T. Cyrness, editors. 1981. Natural history of Oregon Coast mammals. USDA For. Serv. Gen. Tech. Rep. PNW-133, Portland, Oregon. McLeod, J. M. 1966. The spatial distribution of cocoons of Neodiprion swainei Middleton in a Jack pine stand I. A cartographic analysis of cocoon distribution, with special reference to predation by small mammals. The Canadian Entomologist 98:430-447. McShea, W. J., and A. B. Gilles. 1992. A comparison of traps and fluorescent powder to describe foraging for mast by Peromyscus leucopus. Journal of Mammalogy 73:218-222. Means, J. E., K. J. Cromack, and P. C. MacMillan. 1985. Comparison of decomposition models using wood density of Douglas-fir logs. Canadian Journal of Forest Research 15:1092-1098. Meidinger, D.V., and J. Pojar, editors. 1991. Ecosystems of British Columbia. B.C. Special Rep. Series No. 6, B.C. Min. For., Victoria, B.C. Merkens, M., A. S. Harestad, and T. P. Sullivan. 1991. Cover and efficacy of predatorbased repellents for Townsend's vole, Microtus townsendii. Journal of Chemical Ecology 17:401-412.
85 Moraleva, N. 1989. Intraspecific interactions in the common shrew Sorex araneus in central Siberia. Annates Zoologici Fennici 26:425-432. cit. in Moraleva and Telitzina 1994. Moraleva, N., and A. Telitzina. 1994. Territoriality in juveniles of the common shrew Sorex araneus in prepeak and peak years of population density. Pages 67-76 in K. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania PA. Morris, D. W. 1979. Microhabitat utilization and species distribution of sympatric small mammals in southwestern Ontario. American Midland Naturalist 101:373-384. Nichols, J. D., and K. H. Pollock. 1983. Estimation methodology in contemporary small mammal capture-recapture studies. Journal of Mammalogy 64:253-260. Nowotny, C , T. P. Sullivan, and A. S. Harestad. 1990. Woody debris and abundance of rodents on clearcuts in sub-boreal spruce forest. Unpubl. Pagels, J. F., K. L. Uthus, and H. E. Duval. 1994. The masked shrew, Sorex cinereus, in a relictual habitat of the southern Appalachian Mountains. Pages 103-110 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania PA. Pierce, G. J. 1987. Search paths of foraging common shrews Sorex araneus. Animal Behaviour 35:1215-1224. Planz, J. V., and G. L. J. Kirkland. 1992. Use of woody ground litter as a substrate for travel by the white-footed mouse Peromyscus leucopus. Canadian FieldNaturalist 106:118-121. Porter, J. H., and R. D. Dueser. 1982. Niche overlap and competition in an insular small mammal fauna: a test of the niche overlap hypothesis. Oikos 39:228-236. Richens, V. B. 1974. Numbers and habitat affinities of small mammals in Northwestern Maine. Canadian Field-Naturalist 88:191-196. Seber, G.A.F. 1982. The Estimation of Animal abundance and Related Parameters. 2nd edition. Charles Griffin & Company, London. Seip, D., and J. P. Savard. 1993. Wildlife diversity in old growth forests and managed stands. B.C. Ministry of Forests unpubl. report. 44pp.
86 Sheftel, B. I. 1994. Spatial distribution of nine species of shrews in the central Siberian taiga. Pages 45-56 in J. F. Merritt, G. K. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania, PA. Shvarts, E. A., and D. V. Demin. 1994. Community organization of shrews in temperate zone forests of northwestern Russia. Pages 57-66 in J. F. Merritt, G. L. Jr. Kirkland and R. K. Rose, editors. Advances in the biology of shrews. Special publication of the Carnegie Museum of Natural History No. 18, Pennsylvania, PA. Snyder, E. J., and L. B. Best. 1988. Dynamics of habitat use by small mammals in prairie communities. American Midland Naturalist 119:128-136. Sokal, R. R., and F. J. Rohlf. 1981. Biometry: The principles and practice of statistics in biological research, 2nd Edition. W.H. Freeman and Co., New York. Spencer, A. W., and D. Pettus. 1966. Habitat preferences of five sympatric species of long-tailed shrews. Ecology 47:677-683. Stapp, P., J. K. Young, S. VandeWoude, and B. Van Home. 1994. An evaluation of the pathological effects of fluorescent powder on deer mice Peromyscus maniculatus. Journal of Mammalogy 75:704-709. Stewart, D. T., T. B. Herman, and T. Teferi. 1989. Littoral feeding in a high-density insular population of Sorex cinereus. Canadian Journal of Zoology 67:20742077. Sullivan, D. S., and T. P. Sullivan. 1982. Effects of logging practices and Douglas-fir, Pseudotsuga menziesii seeding on shrew, Sorex spp., populations in coastal coniferous forest in British Columbia. Canadian Field-Naturalist 96:455-461. Systat. 1992. SYSTAT for Windows: Statistics, Version 5 Edition. Systat, Inc., Evanston, Illinois. Szaro, R. C , L. H. Simons, and S. C. Belfit. 1989. Comparative effectiveness of pitfalls and live-traps in measuring small mammal community structure, from Symposium: Management of amphibians, reptiles, and small mammals in North America (Flagstaff, AZ, July 19-21,1988). Tallmon, D., and L. S. Mills. 1994. Use of logs within home ranges of California Redbacked Voles on a remnant of forest. Journal of Mammalogy 75:97-101. Teferi, T., and J. S. Millar. 1993. Long Distance Homing by the Deer Mouse, Peromyscus maniculatus. Canadian Field-Naturalist 107:109-111.
87 Terry, C. J. 1981. Habitat differentiation among 3 species of Sorex and Neurotrichusgibbsii'm Washington USA. American Midland Naturalist 106:119-125. Thomas, J. W., editor. 1979. Wildlife habitats in managed forests. USDA Forest Service (Agricultural Handbook No. 533), Washington,DC. Tomasi, T. E. 1979. Echolocation by the short-tailed shrew Blarina brevicauda. Journal of Mammalogy 60:751-759. Van Home, B. 1981. Demography of Peromyscus maniculatus populations in serai stages of coastal coniferous forest in southeast Alaska. Canadian Journal of Zoology 59:1045-1061. . 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901. Van Wagner, C. E. 1968. The line intersect method in forest fuel sampling. Forest Science 14:20-26. van Zyll de Jong, C. J. 1983. Handbook of Canadian Mammals I. Marsupials and Insectivores. National Museums of Canada, Ottawa, Canada. von Ende, C. N. 1993. Repeated-measures analysis. Pages 113-137 in S. M. Scheiner and J. Gurevitch, editors. Design and analysis of ecological experiments. Chapman and Hall, New York, NY. Walmsley, M., G. Utzig, T. Void, D. Moon, and J. van Barneveld. 1980. Describing ecosystems in the field. B.C. Min. Environ, and Min. For., RAB Tech. Paper 2/Land Manage. Rep. No. 7. 224 pp. Victoria, B.C. Whitaker, J. O. J., and T. W. French. 1984. Food of six species of sympatric shrews from New Brunswick. Canadian Journal of Zoology 62:622-626. Whitaker, J. O. J., and C. Maser. 1976. Food habits of five western Oregon shrews. Northwest Science 50:102-107. Williams, D. F., and S. E. Braun. 1983. Comparison of pitfall and conventional traps for sampling small mammal populations. Journal of Wildlife Management 47:841-845. Wrigley, R. E., J. E. Dubois, and H. W. R. Copland. 1979. Habitat abundance and distribution of six species of shrews in Manitoba. Journal of Mammalogy 60:505520. Yahner, R. H. 1982. Microhabitat use by small mammals in farmstead shelterbelts. Journal of Mammalogy 63:440-445.
88 Yoshino, H., and H. Abe. 1984. Comparative study on the foraging habits of two species of Soricine shrews. Acta Theriologica 29:35-43.