the airboat over a distance of 2 m at net depth (i.e. 30 cm) and then back in the ... m of the edge of a patch and in areas that had been disturbed by the airboat.
Weed management and the biodiversity and ecological processes of tropical wetlands DRAFT FINAL REPORT
National Wetlands R & D Program Environment Australia &
Land and Water Australia
2001 Douglas, M. M.1, Bunn, S.E.,2 Pidgeon, R. J.W.3, Davies, P.M.4, Barrow, P.5, O’Connor, R. A.1, Winning, M. 2
1. Centre for Tropical Wetlands Management, Northern Territory University. 2. Centre for Catchment and In Stream Research, Griffith University. 3. Environmental Research institute of the Supervising Scientist, Environment Australia. 4. Department of Zoology, University of Western Australia. 5. Parks Australia North, Environment Australia.
Executive summary •
An experimental study was carried out on the Magela Creek Floodplain in Kakadu National Park between 1997 and 2000. The aims of the experiment study were to: (1) assess the impact of Para grass (Urochloa mutica) on faunal biodiversity and ecosystem processes (community metabolism and trophic pathways) of tropical wetlands; (2) assess the effectiveness of herbicide treatments in the control Para grass, and; (3) determine the impact of chemical control of Para grass on wetland biodiversity and ecosystem processes, and to monitor recovery in treated areas.
•
Biodiversity was examined by sampling floodplain vegetation, aquatic and terrestrial invertebrates, and fish. Ecosystem processes were examined by using stable isotope analysis to determine aquatic food web structure and by measuring community metabolism during the wet seasons. Comparisons were made between samples collected in replicate areas of Para grass, Hymenachne acutigluma (Hymenachne), Oryza meridionalis (Rice), and Para grass that been treated with herbicide. An experimental trial was done to assess the effectiveness and impacts of herbicide control of Para grass.
•
Para grass invasion had significant negative effects on native vegetation. Compared with native Rice and Hymenachne grass communities, Para grass habitats had markedly different plant composition and significantly lower plant biodiversity, particularly during the dry season. Para grass was found across a wide range of water depths and therefore has the potential to spread widely, greatly reducing habitat diversity.
•
Para grass invasion had significant negative effects on wet season terrestrial invertebrate biodiversity, but few negative impacts on aquatic invertebrate or fish communities. In some cases, Para grass invasion resulted in the reduction of aquatic invertebrate biodiversity, and fish community composition differed between Para grass and the two native grasses. However, fish and aquatic invertebrate communities appear to be largely insensitive to changes in grass composition.
•
Para grass invasion had little effect on the aquatic food web of the floodplain. Epiphytic algae were the main energy source for the aquatic food web with minimal contribution from Para grass or native macrophytes. Assuming it provides suitable
substrate for epiphytic algae, Para grass invasion will have limited impact on the aquatic food web. This contrast markedly with the terrestrial food web of the floodplain, as important bird and mammal species rely heavily on native grass seeds and will be negatively impacted by Para grass invasion. •
Para grass had higher rates of autotrophy than other habitats. Whilst this shows that Para grass was a more productive habitat, the greater oxygen demand overnight created anoxic conditions. Floodplain fish must either be tolerant of these anoxic conditions or must rely on moving to areas of open water in less dense native vegetation types. Para grass invasion reduces these refugia.
•
Para grass has the potential to alter floodplain fire regimes, with negative consequences for native flora and fauna. Despite having higher rates of decomposition, Para grass produced approximately twice the dry season fuel load of Rice. This is likely to result in hotter fires, which would be detrimental for floodplain fauna and native vegetation, and would facilitate the spread of Para grass.
•
Roundup Biactive reduced the cover of Para grass by over 90% and there was no evidence of adverse effects of herbicide on invertebrates or fish communities. Native vegetation quickly recovered following the removal of Para grass, however, follow-up control would be required as Para grass readily reinvades treated areas.
•
Although this study represents one of the most comprehensive investigations of the ecological impacts of an aquatic weed, the results of this study must be viewed in light of two limitations. Only a subset of the floodplain flora and fauna were sampled, with groups such as amphibians, reptiles and birds remaining unstudied. Secondly, the conclusions are drawn from sites where Para grass invasion was still patchy; these conclusions may not apply to larger scale infestations of Para grass where potential refuges of native vegetation may be absent.
•
Despite the limited effects on aquatic food webs, Para grass should be viewed as a significant threat to tropical wetlands as it has the potential to spread widely and to adversely impact biodiversity (primarily native vegetation and terrestrial fauna reliant on it) and ecosystem processes such as fire regimes.
i
Table of Contents Acknowledgements iii Introduction 1 1 General methods 4 2 2.1 Study site desription 4 2.2 Experimental design 5 2.3 Data analysis 6 3 Effects of Para grass on floodplain vegetation communities 9 3.1 Aims 9 3.2 Methods 9 3.2.1 Richness and cover 9 3.2.2 Surface area and wet season biomass 10 Breakdown rates 11 3.2.3 3.2.4 Fuel load 12 3.3 Results 13 3.3.1 Richness and cover - Dry season 13 Richness and cover – Wet season 16 3.3.2 3.3.3 Surface area and wet season biomass 22 3.3.4 Breakdown rates 25 3.3.5 Dry season biomass 25 Discussion 26 3.4 3.4.1 Vegetation communities. 26 3.4.2 Surface area and wet season biomass 28 3.4.3 Dry season biomass 28 Vegetation breakdown 29 3.4.4 4 Effects of Para grass on aquatic macroinvertebrate and terrestrial invertebrate communities30 4.1 Aims 30 4.2 Methods 30 Aquatic invertebrates 30 4.2.1 4.2.2 Terrestrial invertebrates 32 4.3 Results 33 4.3.1 Aquatic macroinvertebrates 33 Terrestrial invertebrates 40 4.3.2 4.4 Discussion 45 5 Effects of Para grass on floodplain fish communities 47 5.1 Aims 47 Methods 47 5.2 5.2.1 Sampling methods 47 5.2.2 Fish measurement 49 5.2.3 Habitat structure 49 Results 50 5.3 5.3.1 Sampling methods 50 5.3.2 Species richness and fish abundance 51
ii
5.3.3 Mulivariate Analysis Discussion 5.4 6 Effects of U. mutica on aquatic food webs 6.1 Aims 6.2 Methods Sample collection 6.2.1 6.2.2 Stable isotope analysis 6.2.3 6.3 Results Variation in isotope ratios of primary sources 6.3.1 6.3.2 Major sources of organic carbon supporting invertebrates 6.3.3 Major sources of organic carbon supporting fish 6.4 Discussion 7 Effects of U. mutica on community metabolism 7.1 Aims: 7.2 Methods 7.3 Results and Discussion 8 Aspects of Herbicide control of U. mutica 8.1 Aims 8.2 Methods 8.2.1 Herbicide control trial Rate of spread measurements 8.2.2 8.3 Results 8.3.1 Herbicide control trials 8.3.2 Chemical control and rate of spread measurements Discussion 8.4 8.4.1 Effectiveness of herbicide control 9 Conclusions 9.1 Summary of effects of U. mutica on biodiversity and ecosystem processes Summary of herbicide control of U. mutica 9.2 9.3 Limitations of this study 9.3.1 Choice of study organisms 9.3.2 Spatial scale 10 References
55 61 63 63 63 63 63 64 65 65 65 67 68 78 78 78 79 83 83 83 83 84 85 85 85 86 86 90 90 91 92 92 93 96
iii
1 Acknowledgements This research was funded by Environment Australia and Land and Water Australia, under the National Wetlands R & D Program. Additional funds were provided by the Northern Territory University. The traditional owners of Kakadu National Park and Parks Australia North generously provided access to the study sites. Michael Welch (NTU), Ben Bayliss and James Boyden (eriss) and Buck Salau, Michelle Hat, Fred Baird, Kelvin Murikami and Guy [MMD1]McSkimming,the
staff from the Natural Resource Management section, Parks Australia
North, provided a great deal of field support and technical assistance.
1
2 Introduction The spread of weeds is one of the greatest threats to Australia’s biodiversity (Humphries et al 1991, Low 1999). Wetlands are particularly susceptible to weed invasion, with 14 of the top 18 environmental weeds in Australia occurring in wetlands (Humphries et al., 1991). Eleven of these occur in tropical wetlands and consequently, tropical wetlands are recognised as ecosystems which are in "critical danger" of impacts from weed invasion (Humphries et al., 1991). Despite these concerns, effective weed management in tropical wetlands is hampered by a lack of information on the ecological impacts of weeds and on suitable control methods (Douglas et al., 1998). This lack of knowledge makes it difficult to assess the need and priority for management action, and difficult to prescribe effective control measures for species that warrant such action (Douglas et al., 1998). Even for species where a great deal of effort has been directed towards developing both chemical and biological control, such as mimosa (Mimosa pigra) and salvinia (Salvinia molesta), relatively little effort has been directed towards assessing the extent of ecological change caused by these species (Finlayson et al., 1994). Such information is becoming increasingly important to justify the increasing costs of weed management and to influence policy regarding the introduction and spread of potential new weeds (Douglas et al., 1998). A weed of increasing concern in tropical wetlands throughout northern Australia is Para grass (Urochloa mutica Forssk. Stapf). Para grass is a perennial stoloniferous grass with stout culms that root at the nodes and can grow up to 10 m long and 2 m tall. It was introduced to Queensland from North Africa in the 1880’s and has been actively is promoted as an improved pasture species, particularly in Queensland and the Northern Territory (Lonsdale 1994). The first record of Para grass in the Northern Territory was its establishment in the Darwin Botanical Gardens in the late 1800’s (Wesley-Smith 1973). It is currently promoted as an improved pasture species for cattle in seasonally inundated areas of northern Australia and it is estimated that Para grass now covers some 40 000 ha in the Northern Territory (Low 1997).
2 The attributes of Para grass that are valued by the pastoral industry include rapid spread, high yield, tolerance of waterlogging and drought, and recovery from heavy grazing (Anning and Hyde 1987). However, outside pastoral land, its aggressive colonising habit in wetland environments has lead to Para grass being listed among the 18 taxa identified as having “potential to cause serious impact on a nationally significant scale” (Humphries, Groves et al., 1991). Para grass is also listed as a serious agricultural weed of the Commonwealth Caribbean (Hammerton 1981) and in Australia is a serious pest in supply and drainage channels, an aggressive weed of sugar cane and an unsightly weed in urban parks (Low 1997). Manual methods are used to control Para grass in developing countries (Hammerton 1981) while elsewhere a range of herbicides are generally used in control (e.g. (Horng and Leu 1979)). In northern Australia, the possible benefits of programs to control important wetland weeds such as mimosa may be not be fully realised because Para grass has the potential to invade areas formerly occupied by mimosa and may prove even more intractable (Cook and Setterfield, 1996). Although Humphries et al (1991) recommend that the first urgent step in weed management is to provide formal documentation of the impact of weeds as a conservation threat, little research has been done on the possible effects of Para grass on wetland ecology. One notable exception is in Queensland where Para grass from ponded pastures has become a weed in sugar cane and has also invaded the riparian zone of streams. In this situation, Para grass was found to make little contribution to stream food webs (Bunn, Davies et al., 1997) but was responsible for increased flooding frequency by changing channel morphology (Bunn, Davies et al., 1998). Little work has been done to assess the ecological impacts of Para grass in floodplain environments. The high conservation value of northern Australia’s wetlands (Whitehead, Wilson et al., 1990) means that impacts of Para grass on biodiversity are of particular interest and this study investigates the effect of Para grass invasion on wetland biodiversity or ecosystem processes. A large-scale experiment was conducted on a floodplain in Kakadu National Park, NT.
3 The aims or this research project were: (1) To assess the impact of Para grass on faunal biodiversity and ecosystem processes
(community metabolism and trophic pathways) of tropical wetlands in Kakadu National Park, Northern Territory. (2) To assess the effectiveness of herbicide treatments in the control Para grass. (3) To determine the impact of chemical control of Para grass on wetland biodiversity and
ecosystem processes, and to monitor recovery in treated areas.
4
3 General methods 3.1
Study site desription
This study was undertaken in Kakadu National Park in the wet-dry tropics of northern Australia. Kakadu covers 19,804 km2 and its natural and cultural heritage is recognised by its inscription on the World Heritage List. The wetlands of Kakadu have been designated as internationally important under criteria established by the Ramsar Convention (Finlayson and Woodroffe 1996). The Park was designated in a series of stages, stage one in 1979, stage 2 added in 1984 and stage 3 in 1987. Prior to its declaration, non-Aboriginal economic activity in the region was small scale and diverse, including hunting of feral buffalo and various attempts at cattle grazing (Levitus 1995). In the early 1900’s Para grass was planted on the eastern bank of the East Alligator River in Arnhem Land as a pasture grass and by the 1920’s it had spread to the western side of the river, which is now Kakadu National Park (Wesley-Smith 1973). In the early 1990’s Para grass was identified as a species with fairly limited distribution in Kakadu but with the capacity to dominate large areas of relatively undisturbed plant communities (Cowie and Werner 1993). Study sites were located on the floodplain of Magela Creek (Figure 3-1). Magela Creek is a tributary of the East Alligator River, which arises from sandstone plateau country in the western part of Arnhem Land before it enters the low sandy plains which cover most of Kakadu. There it divides and distributes water in an expansive floodplain system that covers about 150 km2 of the 1600 km2 catchment (Williams 1979). At the onset of the dry season Magela Creek dries up and ceases to flow and the floodplain starts to dry. The amount of water and the number of waterbodies persisting vary markedly from year to year, depending on the length and intensity of the wet season (Russell-Smith, Needham et al., 1995). Vegetation types are a direct indicator of water depth on the floodplain and consist of paperbark forests, open perennial and annual swamps, billabongs and grass/sedge herbfields (Williams 1979). The most widespread grassland communities in the late 1980’s were dominated by three grasses: Hymenachne acutigluma (Steud.) Gilliland (15% of floodplain), Pseudoraphis spinescens (R.Br.) Vick. (14%), and Oryza meridionalis N. Ng. (12%) (Finlayson 1993). A comparison of aerial photography from 1991 and
5 1996 showed large areas of O. meridionalis (wild rice) had been replaced by Para grass over that period (Knerr 1996). 3.2
Experimental design
Six sites were sampled on the Magela Creek floodplain (Figure 3-2). Three sites were located between Nankeen Billabong and the western floodplain margin and the other three further north between a palaeochannel and the western floodplain margin. For statistical purposes, sites were considered to be in two distinct zones (north and south). Each site consisted of four different ‘vegetation types’ occurring in patches ranging in size from 20 x 20 m to approximately 2 ha. Three of the vegetation types were defined by the dominance of the one of the following grass species: Hymenachne acutigluma (Hymenachne), Oryza meridionalis (wild rice), or U. mutica. The fourth vegetation type consisted of U. mutica that had been sprayed with herbicide (details below). These four vegetation types will be referred to as ‘Hymenachne’, ‘Rice’, ‘Para grass’ or ‘Sprayed’. In addition to the Munmarlary trial, Para grass was aerially sprayed at each of the six sites on the Magela floodplain that were used for all other components of the study. The vegetation cover of sites before spraying did not differ significantly from Para grass sites that were left unmanipulated (F1,1 = 0.17, p = 0.75). On December 20 1997, the herbicide Roundup biactive (Monsanto Co. MON 77920) was applied to the six sites at a rate of 12 L ha-1. The area sprayed at each site varied from approximately 20 x 20 m to 80 x 80 m. The time of spraying corresponded to the first rains of the wet season when plant growth is initiated, but before plants are submerged. By January 23 1998, areas of herbicide application were clearly evident as bare patches on the floodplain. Rice was chosen as a study species because it is the native plant species most frequently replaced by Para grass on the Magela floodplain (Knerr 1996, Cowie, 1993 #136). Hymenachne was studied as it can be considered a “native analogue” of Para grass because it is a perennial grass that can form dense monospecific stands. The primary constraint on site selection was that representative patches of each vegetation treatment were closer to each other than patches from other sites. The other constraints were access considerations and fire history; areas burned in the 1997 dry season were avoided. Each
6 site was marked during the first dry season with a 2 m long PVC pole attached to a steel post to aid location during the wet season when sites were inundated. Rice tended to occur at shallow areas on the floodplain margin (except for site 4 on a billabong levee –Figure 3-2) while Hymenachne occurred at deeper sites next to permanent waterbodies. Para grass occurred across this range of depths. 3.3
Data analysis
Effects of vegetation, spatial and temporal variation on univariate community indices, such as species richness and total abundance, were examined by multifactor ANOVA. Homogeneity of variances was assessed by Cochran’s test and plots of variance against means were examined. Data were transformed when assumptions of ANOVA were not met. Multivariate pattern analysis was carried out using PATN software (Belbin 1995) on appropriate biological community data. Bray & Curtis dissimilarity values amongst all samples were calculated using log (X+1) transformed abundance data. For fish data, dissimilarity values were used in a one-way ANOSIM analysis (Belbin 1995) to evaluate the influence of vegetation on fish community structure by testing for evidence of differences among the groups of different vegetation types. Semi-strong HMDS ordination was used to display the patterns of relationships of community structure at different sites. Principal Axis Correlation (PCC) was used to calculate the correlation of environmental variables and community data with the ordination space. The significance of these values was determined using a Monte Carlo randomisation procedure involving 100 iterations.
7
N
Figure 3-1
.
Escarpment & plateau
Sample sites
Major roads
Lowlands
Uranium Mine
Minor roads
Seasonally inundated
Uranium mine development
Settlement or site
Mining lease boundary
Location of the study area in relation to the East Alligators River and the township of Jabiru.
8
Woodland
Melaleuca swamp
Palaeochannels - open water
Seasonally inundated
Palaeochannels - vegetated
Figure 3-2
Location of study sites within the study area. Sites are numbered 1 – 6 and the lettering denotes the vegetation type: P = Para grass, R = Rice, H = Hymenachne, S = Sprayed Para grass.
9
4 Effects of Para grass on floodplain vegetation communities
4.1
Aims
The aims of this component of the study were to examine the effects of Para grass on:
4.2
•
Plant species richness and cover during the wet and dry seasons,
•
Plant biomass and plant surface area during the wet season,
•
Rates of vegetation breakdown, and
•
Dry season fuel loads.
Methods
4.2.1 Richness and cover Vegetation sampling was done during both the wet and dry seasons to assess the effect of Para grass invasion on native plant communities. Wet season samples were collected twice each year in February and April of 1998 and in January and March of 1999. Dry season sampling was done in November 1997 and again in November 1998. In the dry season of 1997 only two of the four vegetation types were sampled: Para grass and Hymenachne. It was not possible to locate areas of Rice during November as it dies off at the end of the wet season. Sprayed sites were not sampled in November, as the herbicide was not applied until late December (see below). In the dry season, vegetation cover and richness were measured at three randomly selected points within each patch. At each point, vegetation height was measured and a nested quadrat (2 x 2 m) was used to assess the percentage cover of each species. In the wet season, four sites were randomly selected within each patch. Depth was measured and the cover of both emergent and floating
10 plant species was estimated over a 1 x 2 m area. The presence of submerged plant species was also noted.
4.2.2 Surface area and wet season biomass To assess possible changes in habitat structure caused by invasion of Para grass, sampling was done to determine aboveground plant biomass and surface area during the wet season. Vegetation was harvested in May 1999 at four sites; two randomly selected from both the northern and southern zones. Patches of Sprayed Para grass were not included. Vegetation was harvested from a 1 x 2 m quadrat placed randomly within each patch, giving a total of four biomass estimates per vegetation type. Water depth was measured at each quadrat and vegetation was harvested using a Recipro submersible grass cutter fitted to an Echo brushcutter and hand shears. Cut material (including emergent sections) was pulled out using rakes. In quadrats where the free-floating aquatic fern salvinia was abundant, it was removed using hand nets prior to the cutting of attached vegetation. Vegetation was sorted into species in the boat and allowed to drain for approximately 1 min before being weighed with scales of 100 g accuracy. Specimens of Para grass, Hymenachne and Wild Rice were collected for surface area measurements adjacent to the plots harvested for biomass. Ten individual plants of each species were pulled from the water by hand in an effort to collect the entire stem although it is likely that some snapped above the roots. Sections of plant that were above the water surface were cut off and discarded and the submerged portions folded and put on ice in plastic bags. Leaves were removed from stems and electronically scanned in the laboratory. Leaf area was determined by importing scanned images into the program DPS-Delta scan with 0.5% accuracy. To determine the surface area of the stems, width was measured at 25 cm intervals using vernier calipers (accuracy to 0.05 mm). Surface area of the 25 cm lengths was calculated using the top diameter measurement in the equation SA = πDh, where D = diameter and h = 25 cm or length of the terminal section of stem. The use of 25 cm segments rather than one measurement for the whole stem was an attempt to take into account tapering of stem width from base to tip. Leaves and stems were dried in an oven set to 700 C and weighed using an electronic balance accurate to two decimal places. The dry mass to surface area ratio was then calculated. Wet weight to dry weight ratios were not calculated with these samples given the loss of moisture from the plant
11 material in storage and during measurement of surface area. Plant material was collected in the same manner from the same sites in the following wet season and dried in the same manner to obtain these ratios.
4.2.3 Breakdown rates Changes in the rate of vegetation breakdown (sensu (Hanlon 1982)) in areas invaded by Para grass was studied by measuring weight loss in bundles of grass that were placed in the field at the start of the dry season. Leaf and stem material harvested during biomass sampling from the three key grass species (Para grass, Hymenachne and Wild Rice) was used. Plant material was ovendried at 700C then formed into bundles of roughly equivalent sizes (30 cm x 10 cm). Use of green leaf material, and drying of leaves before immersion are both likely to have an effect on results (Boulton and Boon 1991). Decomposing plant material often comprises dead leaves which have a different chemical composition to green leaves. Litter of this kind was not available for the three grass species in the dry season and was inaccessible in the wet season. Drying of leaves can increase the rate of decomposition by rendering the leaf more susceptible to attack by microbes and invertebrates, particularly with submerged aquatic plants (Boulton and Boon 1991). Given that we were interested in relative breakdown rates among species rather than accurate decay estimates, drying was considered an appropriate pre-treatment to standardise for water content. Bundles were wrapped in nylon onion bag material with mesh size 1 cm. The ends of each bundle were secured with plastic cable ties and labelled with an aluminium tag. Each bundle was weighed with an electronic balance. Hymenachne bundles ranged in weight from 16 – 21 g, Rice 13 – 18 g and Para grass 16 – 27 g. An a priori assumption was made that water depth and the suite of factors related to it (e.g. period of inundation) were likely to influence vegetation breakdown. Sites for vegetation breakdown trials were thus chosen according to water depth. Under this scheme Hymenachne sites were always “deep” and Rice sites were always “shallow”. Para grass sites were selected in both deep areas for comparison to Hymenachne and in shallow areas for comparison to Rice. This meant that the overall design was similar to that used in other components of the study except Para grass was represented twice at a given site and the specific location of sites sometimes differed by up to approximately 100 m.
12 In the field, bundles were attached to plastic or steel posts using nylon fishing line. At shallow sites, it was possible to place bundles on the ground at the base of the post while in deep sites greater lengths of fishing line were used for attachment and bundles were pushed through small gaps made in the surrounding vegetation to a depth of approximately 70 cm. Three replicate bundles of each grass species were deployed in the corresponding vegetation type e.g. bundles of Hymenachne were placed in patches of Hymenachne. Three bundles of Para grass were deployed at all sites and vegetation types. This enabled us to determine whether breakdown rates of Para grass were affected by conditions particular to a vegetation type. Samples were left on the floodplain for 10 weeks from May – July 1999. This period was a compromise between leaving samples out for as long as possible, or access problems in the early dry season and the risk of samples being burned by floodplain fires later in the dry season. At the time of deployment, water depth in the deep sites was around 100 cm while in the shallow sites it was around 25 cm. At the time of retrieval the shallow sites were dry and the deep sites were less than 50 cm deep. Bundles were placed in zip-lock plastic bags for transport. Mesh and cable ties were cut off and any new growth from the bundles or surrounding vegetation was removed in the lab. Samples were then dried and weighed in the same manner as pre-deployment. The percentage weight lost was then calculated.
4.2.4 Fuel load To determine changes in fuel load caused by Para grass invasion, biomass was sampled at monthly intervals on three occasions in the 1999 dry season. The same sampling design was adopted as for the decomposition work. Three points were randomly selected within each vegetation patch and aboveground vegetation was harvested from 1 m2 quadrats using hand shears or collecting dead material by hand. The minimum area required to obtain an estimate of above-ground biomass for Hymenachne and Rice is 0.25 m2 (Finlayson 1991). Sampling was not conducted in areas that had been burned, disturbed by feral pigs, or was still inundated. Sampling of Rice and Para grass from shallow areas was only done after 10:30 am to avoid possible overestimation of moisture content due to adsorption of dew. Because of these constraints not all sites and vegetation types were sampled on the same day. Harvested material was placed in nylon mesh (mesh size approximately 2 mm) and weighed in the field using scales to 100 g accuracy. Representative sub-samples (< 1 kg wet weight) were
13 then selected for calculation of wet weight to dry weight ratios. Samples were placed in zip-lock plastic bags in the field to retain moisture and weighed on return to the lab with an electronic balance, then oven-dried at 700C and re-weighed. Above-ground biomass was then calculated from these data. 4.3
Results
4.3.1 Richness and cover - Dry season
A total of 17 plant species were recorded during the Dry season, but there were marked differences in the plant communities found in the four vegetation types (Table 4-1). Para grass always occurred in monospecific stands while all other vegetation types occurred as mixed communities. A total of five species occurred in the native perennial grass, Hymenachne, while the Rice and Sprayed patches each had a total of 11 species (Table 4-1). Mean species richness ranged from 1 to 4 species per quadrat and was a significantly difference between the vegetation types, but this difference was not consistent between sites (F12, 48 = 7.7, p < 0.001). At four of the six sites, richness was higher in Hymenachne than in Para grass (Figure 4-1). At five of the six sites there was a trend of lower richness in the Para grass than in the Rice but this difference was only significant at Site 2 (Figure 4-1). There was a trend of lower richness in Para grass than in sprayed Para grass at five sites, and this difference was significant at three of these (Figure 4-1).
14
Table 4-1
Percentage occurrence of each species in the 18 quadrats from each of the four vegetation types.
Vegetation Type Species
Para
Hymenachne
U. mutica
100
17
H. acutigluma
Rice
Sprayed 28
100
O. meridionalis Ludwigia adscendens
72
Persicaria sp.
33
Cyperus platystylis
22
Pseudoraphis spinescens
72
33
6
22
33
17
Glinus oppositifolius
11
Phyla nodiflora
17
Ipomoea aquatica
6
6
Euphorbia vachellii
6
6
Goodenia purpurascens
17
Heliotropium indicum
6
39
Coldenia procumbens
6
44
Passiflora foetida
6
Eclipta prostrata
6
Eleocharis sp. Total species in veg. type
1
5
15
23
11
11
15
0.6
0.4
log 10 Species richness
0.2
0.0
hymen
para rice Site 1
spray
hymen
para rice Site 2
spray
hymen
para rice Site 3
spray
hymen
para rice Site 4
spray
hymen
para rice Site 5
spray
hymen
para rice Site 6
spray
0.6
0.4
0.2
0.0
Figure 4-1
Mean (± s.e. ) dry season plant species richness per quadrat in the four vegetation types at the six sites. The horizontal lines indicate significant differences between vegetation types at each site.
Mean vegetation cover in the dry season was highly variable and ranged from less than 5% to 100%. There was a significant difference in cover between vegetation types, but this difference was not consistent between sites (F12, 48 = 60.39, p < 0.001). Cover was 100% in all Hymenachne patches, and greater than 95% in all Para grass patches ( Figure 4-2), indicating that both perennial grasses formed dense stands that persist throughout the dry season. Rice and Sprayed patches had significantly lower cover than Para grass at five of the six sites ( Figure 4-2).
16
1.6 1.2 0.8
Vegetation cover (%)
0.4 0.0
hymen
para rice Site 1
spray
hymen
para rice Site 2
spray
hymen
para rice Site 3
spray
hymen
para rice Site 4
spray
hymen
para rice Site 5
spray
hymen
para rice Site 6
spray
1.6 1.2 0.8 0.4 0.0
Figure 4-2 Mean (± s.e. ) vegetation cover per quadrat in the four vegetation types at the six sites. Cover values are arc-sin transformed.
Richness in the sprayed Para treatment was not consistently higher or lower than richness in other vegetation types (Figure 4-1). In the 1998 dry season (11 months after herbicide application) vegetation cover in the sprayed Para treatment was significantly lower than in perennial grass types but variable in relation to Rice ( Figure 4-2). Physical disturbance (rooting) by feral pigs (Sus scrofa) was observed in half the sprayed Para treatments and contributed to low vegetation cover. At one site, grazing by wild horses (Equus caballus) also reduced the cover of Para grass regrowth.
4.3.2 Richness and cover – Wet season A total of 30 plant taxa were recorded in the wet season samples (Table 4-2). As in the dry season, total richness was highest in the Rice and Sprayed habitats and lower in both the perennial grasses, with Para grass having the lowest total richness (Table 4-2). Mean richness per quadrat ranged from 1 to 6 taxa and was significantly different between vegetation types for nine
17 of the 24 Vegetation by Site by Time comparisons (F36, 288 = 3.14; p < 0.001; Figure 4-3). In five of these cases, Para grass contained significantly fewer taxa than all other vegetation types. In three cases, both Para grass and Hymenachne, and in one case both Para grass and Rice, contained significantly fewer species than other habitats.
Table 4-2
List of aquatic plants and the number of records for each over the four sampling occasions within each vegetation type (maximum = 96). Vegetation type
Species/growth form
Hymenachne
Rice
Para grass
Sprayed
Hymenachne
96
1
Rice
4
92
Para grass
5
6
96
17
60
28
52
24
Emergent Eleocharis sp. Ipomoea aquatica
1
Limnophila sp.
1
2
4
11
Panicum palludosum Pseudoraphis spinescens
4 2
39
2
11
Emergent/free floating Leersia hexandra
1
Ludwigia adscendens
65
10
3
7
Persicaria sp.
11
1
2
1
Floating Azolla sp.
10
18 Hygrochloa sp. Nymphaea sp.
4 19
Nymphoides sp. Salvinia molesta
32
26
13 50
1
76 2
15
9
Submerged Chara sp.
2
Maidenia sp.
42
Najas sp.
8
Utricularia sp
10
17
5
31 19
20
Vallisneria sp.
23 6
Terrestrial Commelina sp.
1
Cyanotis axillaris
15
Cyperus platystylis
2
1
1
Echinochloa elliptica
6
Eriocaulon sp.
12
Euphorbia vachelli
5
Passiflora foetida
1
Sesbania sp.
3
Sida sp.
2
4
1
1
19
8
Site 1
6 4 2 0 8
Site 2
6 4 2 0 8
4 2 0 8 6
Site 4
Floral richness
Site 3
6
4 2 0 8
Site 5
6 4 2 0 8
Site 6
6 4 2 0 hymen
rice para
hymen sppara
February 98
Figure 4-3
rice para April 98
hymen sppara
rice para
hymen sppara
January 99
rice para
sppara
March 99
Mean (± s.e. ) vegetation richness per quadrat in the four vegetation types over the four sampling times.
20
Site 1
1.6 1.2 0.8 0.4 0.0
Site 2
1.6 1.2 0.8 0.4 0.0
Site 3
1.2 0.8 0.4 0.0 1.6
Site 4
(arcsine transformed)
Vegetation cover (%)
1.6
1.2 0.8 0.4 0.0
Site 5
1.6 1.2 0.8 0.4 0.0
Site 6
1.6 1.2 0.8 0.4 0.0 hymen rice hymen rice hymen rice hymen rice para sppara para sppara para sppara para sppara February 98
April 98
January 99
March 99
Figure 4-4 Mean (± s.e. ) wet season plant cover per quadrat in the four vegetation types over the four sampling times.
21 Mean vegetation cover was extremely variable and showed interactions between all experimental factors. Overall, the two perennial grasses, Para grass and Hymenache, had higher mean cover (~70%) than Rice (~20%), and the Sprayed patches were the lowest (~20%). However, this general pattern was not consistent across all sites or times (F36, 288 = 2.92; p < 0.001). A relatively high number of species with a terrestrial growth form were recorded and these were most common in Rice, which may be related to the shallower depths typically associated with this vegetation type (Table 4-2). Compared with Rice habitats, there was a lower incidence of emergent growth forms in Para grass (particularly Eleocharis sp and Pseudoraphis spinescens) and a lower incidence of Maidenia sp (a rooted submerged genus) (Table 4-2). Several aquatic perennial species (Eleocharis sp., Ludwigia adscendens, and Pseudoraphis spinescens) occurred in Para grass in the wet season but not during the dry season. These species were also present during the dry season, but only in the Rice habitat. In the wet season after spraying, no Para grass was recorded in sites sprayed with herbicide and vegetation communities in sprayed sites were dominated by Nymphaea spp. and Eleocharis spp.. In February 1998, vegetation cover was significantly lower in sprayed Para grass patches than in the other three vegetation types (Figure 4-4). By the second wet season after herbicide application, however, mean cover had increased, with only one-third of sprayed sites now having significantly lower cover than other vegetation types (Figure 4-4). In this second season after herbicide application, both Para grass and Rice were recorded in sprayed sites. Rice occurred in shallower areas than the other vegetation types, and was therefore exposed to shorter periods of inundation (Figure 4-3). Hymenachne generally occurred in deeper water than the other grasses (Figure 4-3). Para grass sites were at intermediate depths but it should be noted that Para grass occurred in water as deep as 2.2 m (Figure 4-3).
22
Table 4-3
Summary of water depths recorded during wet season vegetation sampling in the four vegetation types.
Vegetation type
Mean (s.e.)
Min.
Max.
Range
Hymenachne
181 (5)
100
273
173
Rice
96 (5)
40
193
153
Para grass
126 (5)
48
220
172
Sprayed Para grass
123 (5)
39
222
183
4.3.3 Surface area and wet season biomass In the late wet season, leaves accounted for a greater proportion of total surface area than stems for Para grass. In contrast, the two native grasses had a higher proportion of total surface area as stems. Total plant surface area per m2 of floodplain was significantly greater for Hymenachne than the other two grasses (F2, 9 = 8.7, p = 0.007; Figure 4-5a). However, when differences in depth were taken into account (expressed as plant surface area per m3) there were no differences between vegetation types (F2, 9 = 0.9, p = 0.5; Figure 4-5b). Wet season biomass was significantly different between vegetation types when expressed as either per m2 (F2, 9 = 15.7, p = 0.001) or per m3 (F2, 9 = 14.5, p = 0.002). Para grass had a similar biomass per m2 to Hymenachne (Figure 4-6 a) but occurred in shallower depths, so biomass per m3 was greater for Para grass (Figure 4-6 b). Rice had significantly lower biomass than Para grass for both measures of biomass (Figure 4-6).
23 Figure 4-5 Mean (+/- se) wet season surface area estimates for Para grass and the two native grasess, Hymenachne and Wild Rice. Biomass estimates are expressed as (a) cm2 m2 , and (b) cm2 m-3. Horizontal lines indicate significant differences.
(a) 15
2
-2
Surface area (cm m )
12
9
6
3
0 Hymenachne
Para grass
Rice
Hyemachne
Para grass
Rice
(b)
2
-3
Surface area (cm m )
15
10
5
0
24 Figure 4-6 Mean (+/- se) wet season biomass estimates for Para grass and the two native grasess, Hymenachne and Wild Rice. Biomass estimates are expressed as (a) kg dry weight m-2 and (b) kg dry weight m-3. Horizontal lines indicate significant differences.
(a) 2.0
-2
Biomass (kg m )
1.5
1.0
0.5
0.0
(b)
Hymenachne
Para grass
Rice
Hyemachne
Para grass
Rice
2.5
-3
Biomass (kg m )
2.0
1.5
1.0
0.5
0.0
25
4.3.4 Breakdown rates Breakdown rates significantly different between the four vegetation types (F3, 3 = 405; p < 0.001). Para grass broke down significantly slower than Hymenachne in deep water, but in shallow water, Para grass broke down significantly faster than Rice (Table 4-4). There was a significant difference in the breakdwon of Para grass across the four vegetation types but this was not consistent across all sites (F12, 48 = 2.4; p = 0.01). Breakdown rates for Para grass were the same in of three vegetation types, but were significantly lower Rice, due to the influence of a single site (Table 4-4). This indicates that differences in environmental characteristics between vegetation types (such as depth) did not strongly influence breakdown rates of Para grass in the early dry season. Table 4-4 Mean (± se) percentage weight loss of the three key plant species in four floodplain vegetation types.
Vegetation type
Average weight lost (%) Hymenachne
Hymenachne
Rice
56.98 ± 1.48
Rice
Para grass 43.56 ± 1.61
20.83 ± 2.09
28.54 ± 1.47
Para grass (deep)
42.87 ± 1.10
Para grass (shallow)
32.58 ± 0.94
4.3.5 Dry season biomass Maximum above-ground biomass was 5.36 kg m-2 for Para grass in deep areas and 2.89 kg m-2 in shallow areas, 4.43 kg m-2 for Hymenachne and 1.17 kg m-2 for Rice. Low-lying areas of the floodplain that remain wetter for longer (Hymenachne and deep Para grass sites) had significantly
26 higher dry season biomass than areas with a shorter period of inundation (Rice and shallow Para grass sites) in 7/9 comparisons (F12, 71 = 2.56, p = 0.007; Figure 4-7). In the shallow areas, Para grass had significantly higher biomass than Rice in 11/15 site by time comparisons (F4, 60 = 5.7, p < 0.001).
Site 1
0.8 0.6 0.4 0.0
2
log10 Dry Biomass (kg m- )
0.2
Site 5
0.8 0.6 0.4 0.2 0.0
Site 6
0.8 0.6 0.4 0.2 0.0
H
DP
R
SP
August
H
DP
R
September
SP
H
DP
R
SP
October
Figure 4-7 Mean (± s.e) dry season biomass in four vegetation types (H = Hymenachne, DP = Deep Para grass, R = Rice, SP = Shallow Para grass) during the dry season.
4.4
Discussion
4.4.1 Vegetation communities. On the Magela Creek floodplain, Para grass occupied a broad range of habitats spanning from Melaleuca woodland fringing the floodplain to the edge of permanent billabongs within the floodplain. The maximum water depths in which Para grass grew on the Magela floodplain
27 (frequently > 2m) were deeper than previously reported for this species (e.g. (Anning and Hyde 1987; Clarkson 1991; Low 1997). It is generally considered that Para grass does not grow in water depths greater than 50 – 60 cm and this has been used as a rationale for introducing another exotic grass for ponded pastures (Hymenachne amplexicaulis), which can grow at greater depths (Pittaway and Chapman 1996). Underestimation of the maximum depth for Para grass occurrence may have arisen because previous observations were based on Para grass subject to cattle grazing or in ungrazed creek-side locations that do not reach the depths seen on the Magela floodplain. Para grass clearly has a wider ecological tolerances than previously reported and can potentially invade a greater range of habitats than previously thought possible. Predictions of areas at risk of Para grass invasion need to be revised in light of these results, as most of the native floodplain vegetation communities can potentially be displaced by Para grass. In the dry season, Para grass occurred in monospecifc stands, so displacement of either Rice or Hymenachne by Para grass represents a marked loss of plant species richness. In the wet season, the lower species richness in Para grass was not as marked at the quadrat scale, but total richness (summed across all quadrats) was much lower in Para grass, indicating a large reduction in plant species richness at this larger spatial scale. Others have suggested that the importance of floodplains for biodiversity lies in their highly heterogeneous nature (Whitehead, Wilson et al., 1990). Our results support the contention that Para grass, with its wide ecological tolerances, can reduce plant species richness in both the wet and dry season and across a range of spatial scales. Reductions in the distribution and abundance of important plant species such as Rice will result in reduced food resources and habitat diversity for floodplain fauna. For example, O. meridionalis is a prolific producer of high protein seeds which form an important component of the diets of the Dusky Plains rat (Rattus colletti) and the Magpie goose (Anseranus semipalmata) (Whitehead 1998; Wurm 1998). Furthermore, bulbs of the sedge Eleocharis spp, which were more common in Rice than Para grass (Table 4-1, Table 4-2), are another important food source for Magpie geese (Corbett and Hertog 1996). Para grass cannot support the growth rates obtained from a diet of native grasses and this can be fatal for fledging goslings (Whitehead and Dawson 2000). Magpie geese may also be detrimentally affected by Para grass invasion as they preferentially nest in Eleocharis/Oryza (Corbett and Hertog 1996).
28
4.4.2 Surface area and wet season biomass As well as changes species richness, Para grass invasion results in structural changes in habitat. The higher wet season biomass of Para grass represents a greater density of vegetation in the water column. This may impede the movement of larger aquatic animals such as turtles and this possibility warrants further investigation, given the importance of turtle harvesting for Aboriginal people. [MMD2]The higher surface area may represent an increase in habitat area to very small organisms such as macroinvertebrates or epiphytic algae. The consequences of this are discussed in section 5. In the wet season, Para grass had a different architecture to the native grasses, with a higher leaf to stem surface area ratio. The greater leaf surface area may also represent a significant physiological difference between Para grass and native grasses because leaves, rather than leaf sheaths (or stems) are the primary organs of photosynthesis. The higher proportion of leaf area in Para grass means that it has the potential to convert solar energy to biomass more efficiently than the native grasses and this may be another factor contributing to its invasive success.
4.4.3 Dry season biomass The higher biomass of Para grass in the wet season persisted throughout the dry season. Increased plant biomass in the dry season associated following Para grass invasion has implications for fire regimes on the floodplain. Para grass invasion leads to greater fuel loads in both the Rice and Hymenachne habitats. Greater fuel biomass could increase the intensity as well extent of fire on floodplains particularly, in the late dry season, when the grasses have had a chance to dry out. Hotter fires pose a threat to adjacent lowland communities such as Melaleuca forests and rainforests. On the Magela floodplain, there are several stands of dead Melalucas which have an understorey of Para grass (Douuglas, pers. obs.). Many of the trees in these stands have deep fire scars around the base. Observations by staff at Parks Australia North indicate that fires from Para grass have been responsible for the reduction in size of monsoon vine forest patches (like those near Ubirr) (Barrow, pers. obs.). More intense fires may also pose a threat to animals such as turtles, which aestivate in the floodplain soil during the dry season. Changes in fuel dynamics due to Para grass may even facilitate its spread. Work in the Mary River floodplains has indicated that Hymenachne is acutely sensitive to fire, possibly due to low seed viability (Whitehead and McGuffog 1997). The burning of Hymenachne may represent a
29 disturbance that allows the invasion of Para grass into deeper areas of the floodplain. It is likely that this has already taken place on Magela Creek, where Hymenachne was only present adjacent to permanent water, whereas further south on the floodplain where Para grass is less prevalent, Hymenachne occurs in large stands that dry out seasonally.
4.4.4 Vegetation breakdown Para grass invasion also has the potential to change the rates of vegetation breakdown on the floodplains. In shallow areas, Para grass decomposes more rapidly than the Rice is displaces. The high silica content of Rice probably contributes to its slower rate of breakdown. In contrast, Para grass is more refractory than Hymenachne and consequently breaks down more slowly than the native perennial in the deeper water habitats. The slower breakdown of Para grass than Hymenachne is somewhat surprising, given the higher leaf:stem surface area ratio for Para grass and the fact that leaves generally breakdown more rapidly than stems. The slower rate of breakdown would contribute to the higher biomass of Para grass in these deeper habitats. In contrast, the higher biomass of Para grass in the shallower habitats accrues despite the fact that the Para grass breaks down more rapidly than Rice, highlighting the greater productivity of Para grass. The consequences of changes in grass breakdown rates on the floodplain are difficult to predict, but may include changes in decomposer communities, differences in water quality (specifically dissolved oxygen concentrations), and altered rates of ecosystem processes such as nutrient cycling.
30
5 Effects of Para grass on aquatic macroinvertebrate and terrestrial invertebrate communities 5.1
Aims
The aim of this component of the study was to examine the effects of Para grass on: •
The taxonomic richness, abundance and community structure of benthic and epiphytic aquatic macroinvertebrates.
•
The taxonomic richness and abundance of terrestrial invertebrate communities during the dry season and the wet season.
5.2
Methods
5.2.1 Aquatic invertebrates Aquatic invertebrate samples were collected at four sites (two sites in each zone) from the four vegetation types described in section 2. High densities of aquatic vegetation, deep water (on some occasions over 2 m) and the presence of saltwater crocodiles (Crocodylus porosus) limited the choice of macroinvertebrate sampling strategies. Within these constraints, two sampling strategies were adopted that could be performed from airboats and were semi-quantitative. Two habitats were sampled in the first year: epiphytic and benthic. The epiphytic habitat consisted of open water, floating and submerged portions of vegetation in the top 60 cm of the water column. The benthic habitat consisted of silt, clay and detritus on the floodplain bed. Sampling occurred twice (early- and late-wet season) each year in February and April of 1998 and January and March of 1999. Benthic samples were not collected in the second year due to long processing times and relatively low invertebrate richness in samples collected during the first year. Epiphytic samples were collected using a D-shaped dip net measuring 30 cm across the bottom and 25 cm at the sides and fitted with a net of 250 µm mesh. The net was swept along the side of the airboat over a distance of 2 m at net depth (i.e. 30 cm) and then back in the opposite direction to collect any dislodged invertebrates. A sweep sample consisted of 8 x 2 m sweeps. The contents
31 of the net were preserved in 70% ethanol in the field. Two replicate samples were taken from within each vegetation type (at least 10 m apart) with attempts made to avoid sampling within 5 m of the edge of a patch and in areas that had been disturbed by the airboat. Benthic samples were collected on the opposite side of the airboat to the epiphytic samples with an electric suction sampler of the type described by Brooks [/1994 #21]. A suction sampler was used because it was not possible to use a dip bet in such dense vegetation and depths sometimes exceeding 2 m. The only modification to the sampler was the attachment of an aluminium pole to the intake hose so that approximately 2 cm of pole protruded beyond the end of the hose. The pole was used to disturb the substratum, which was then sucked up through the hose. Samples were standardized to 1 min based on the results of a pilot study and were collected over an area of approximately 0.5 x 1 m. The contents of the collecting net and jar were emptied into a sieve of 300 µm and agitated in water to rinse out excess fine silt and mud. The samples were then transferred to plastic sample jars (500 ml) and preserved in 70% ethanol. Water depth at each sampling location was measured with a marked PVC pole. Water quality measurements were made at the water surface and immediately above the substratum at two locations used for epiphytic sampling within each vegetation type. Dissolved oxygen, pH, electrical conductivity, turbidity and water temperature were measured with a Horiba U-10 water quality meter. The cover and diversity of aquatic vegetation were visually assessed at each epiphytic sampling location. For a full description of vegetation assessment methods, see section 4.2. Two replicate samples were processed from each habitat in each vegetation type. Samples were emptied into a developing tray, rinsed with tap water and poured through stacked 2.8 mm, 1 mm and 300 µm sieves to separate invertebrates from the coarse organic detritus and to split the remaining sample into two fractions for sorting. The contents of the 2.8 mm sieve was put back in the developing tray with water and scanned for remaining invertebrates using an illuminated magnifying glass. The contents of the 1 mm and 300 µm sieves were sorted and invertebrates identified using a Leica MZ-8 dissecting microscope. The 300 µm fraction was sub-sampled when it contained large numbers of invertebrates. Subsampling was done using a 1 L Nalgene (Nalge Nunc International) imhoff cone. The cone was
32 modified by gluing an air stone into the base, which allowed aeration of the sample via an aquarium air pump. Use of bubbling air to homogenise samples prior to sub-sampling has been described elsewhere (Hickley 1975) and was considered suitable for these samples given that large and heavy animals (such as gastropods) had been caught in the 1 mm fraction and filamentous algae, which can cause clumping, was rare. Samples were made up to 1 L and aerated for at least 2 min. A 50 mL ladle was then used to remove at least two sub-samples (10% of the 300 µm fraction). The sub-sampling rationale was to retrieve at least 200 animals from the 300 µm fraction within a 2 hr sorting time. Identification was to the family level for samples collected in 1998 except for microcrustacea, Oligochaeta, Acarina and Chironomidae (sub-family). In 1999 specimens were identified to genus where possible, with the above exceptions. Identification was done using published keys.
5.2.2 Terrestrial invertebrates At the end of the 1998 dry season, terrestrial invertebrates were sampled from the six sites used for all other components of the project. Sampling was done using a standard insect net. The operator paced through the vegetation, sweeping the net across the tips of the vegetation at every step. A total of 25 paces (or 25 sweeps) constituted a sample, and two samples were taken in each stand. Invertebrates were then transferred to sealed plastic bags and refrigerated until they were preserved in 100% ethanol, within 48 hrs of collection.
During the wet season, a standard insect net was used to collect invertebrates living above the water surface. The net was swept repeatedly through emergent vegetation (or above the water surface if no emergent vegetation was present) along one side of the airboat (covering an area of 0.5 x 2 m) until no more invertebrates were captured. Specimens were transferred to labeled plastic bags and put on ice until the end of the day when they were preserved in 100% ethanol. Terrestrial invertebrates were collected twice during each wet season, on the same days that aquatic invertebrates were sampled. Two replicate samples were collected and sorted from four sites (two sites in each zone).
33 5.3
Results
5.3.1 Aquatic macroinvertebrates From all the samples processed, a total of 72 invertebrate taxa (primarily families) were recorded. The richest insect order (at family level) was Coleoptera (beetles) followed by Diptera (true flies) and Hemiptera (true bugs) (Table 5-1). Other non-insects included Crustacea, Platyhelminthes, Nematoda, Hirudinea, Oligochaeta and Acarina (Table 5-1). Coleoptera and Hemiptera included groups with life stages considered to be semi-aquatic in habit and associated with still waters or vegetation (Williams 1980; Hawking and Smith 1997).
Table 5-1
Breakdown of the number of invertebrate groups recorded in epiphytic and benthic habitats.
Benthic habitat Taxonomic group Feb 98 Apr 98 Total 7 9 9 (6*) Coleoptera Diptera 5 8 8 2 2 2 Ephemeroptera 1 5 5 Hemiptera 1 1 1 Lepidoptera Odonata 2 3 3 2 2 3 Trichoptera 2 2 2 Gastropoda Other non-insects 9 9 10 * total with adult and larval stages pooled
Epiphytic habitat Feb 98 Apr 98 Jan 99 11 16 17 5 9 11 2 2 2 10 11 9 1 1 1 3 5 3 3 3 2 4 3 2 12 11 9
Mar 99 Total 13 19 (12*) 7 12 2 2 11 11 1 1 4 5 3 3 2 4 9 13
The epiphytic habitat supported a greater richness of invertebrates than the benthic habitat (Table 5-1). The epiphytic habitat supported 65 taxa in 1998 compared with 43 in the benthos. Taxa found in the benthos also occurred at the surface with the exception of two rare taxa. In contrast, 25 of the taxa collected at the surface in 1998 were exclusive to that habitat. Many of the taxa found exclusively in the epiphytic habitat were from the orders Coleoptera and Hemiptera (Table 5-1). Total invertebrate richness in the two habitats differed between the vegetation types. For epiphytic invertebrates, richness was highest in Hymenachne and lowest rich in Sprayed Para grass (Table 5-2). In contrast, benthic invertebrate richness was lowest in Hymenachne and highest in Rice (Table 5-2). Richness in the benthos of the sprayed Para treatment was higher
34 than both Para grass and Hyemachne, whereas epiphytic richness in sprayed Para grass was lower than all other vegetation types (Table 5-2).
Table 5-2
Total invertebrate taxa richness (primarily families) for benthic and epiphytic habitats in the four vegetation types.
Benthic
Epiphytic
Feb 98
Apr 98
Feb 98
Apr 98
Jan 99
Mar 99
Hymenachne
22
15
43
53
47
46
Rice
31
39
39
45
41
42
Para grass
27
31
46
46
39
41
Sprayed
28
37
32
42
38
44
n= 8-10 for each vegetation type per occasion Within the benthic habitat, mean richness was consistently lowest in Hymenachne and this difference was significant in three of the four Zone by Time comparisons (F3, 6 = 9.3, p = 0.01; Figure 5-1). Benthic abundance varied among vegetation types, but not consistently between sites and time (F6,34 = 5.9, p < 0.001)
35 Figure 5-1
Mean (+/- se) benthic invertebrate richness for the four vegetation types at the two zones and times.
30 25
north
15 10 5 0 30 25 20
south
Invertebrate richness (taxa sample
-1
)
20
15 10 5 0
hymen
para
rice
sp para
hymen
para
February 98
Figure 5-2
rice
sp para
April 98
Mean (+/- se) benthic invertebrate abundance for the four vegetation types (H = Hymenachne, P = Para grass, R = Rice, and S = Sprayed Para grass) at the four sites during February and April 1998.
February 98
9000 6000 3000 0 12000
April 98
Abundance (animals sample
-1
)
12000
9000 6000 3000 0
H
P
R
Site 1
S
H
P
R
Site 3
S
H
P
R
Site 5
S
H
P
R
Site 6
S
36 For epiphytic invertebrates, mean richness and abundance differed significantly between vegetation types but these differences were not consistent between sampling occasions or sites (Richness: F18, 66 = 3.1, p < 0.001, Figure 5-3; Abundance: F18, 66 = 7.7, p < 0.001, Figure 5-4). This result is unlikely to be an artefact of low taxonomic resolution, because analysis of samples from 1999, mostly at genus level, gave no significant differences in richness among vegetation types.
Figure 5-3
Mean (+/- se) epiphytic invertebrate Richness for the four vegetation types at the four sites during the four sampling times.
Site 1
40 20
40 Site 3
-1
Richness (taxa sample )
0
20 0
Site 5
40 20 0 Site 6
40 20 0
H
R
P
February 98
S
H
R
P
April 98
S
H
R
P
January 99
S
H
R
P
March 99
S
37
Site 6
Site 5
Site 3
-1
Abundance (animals sample )
Site 1
Figure 5-4
6 5 4 3 2 1 0 6 5 4 3 2 1 0 6 5 4 3 2 1 0 6 5 4 3 2 1 0
Mean (+/- se) epiphytic invertebrate Abundance for the four vegetation types at the four sites during the four sampling times
H
P
R
February 98
S
H
P
R
April 98
S
H
P
R
January 99
S
H
P
R
S
March 99
Multivariate analysis of benthic invertebrate data showed few differences between the vegetation types (Figure 5-5). Invertebrates from Para grass were generally interspersed with those from other vegetation types. Late in the wet season, hymenachne had distinctive invertebrate communities that were characterised by low richness and abundance. This vegetatio type was also characterised by low dissolved oxygen concentrations. Most vegetation types showed some separation between the two sampling times, with the second collection characterised by lower depths and temperatures.
38 Figure 5-5
1.6
SSH ordination of benthic invertebrate communities in the four vegetation types.
stress=0.17
1.2
0.8
Axis 2
Time
0.4 Depth
0.0
-0.4 Temperature
-0.8
-1.2
Dissolved Oxygen
-1.6
-1.2
-0.8
Invertebrate Abundance
-0.4
0.0
Invertebrate Richness
0.4
0.8
1.2
Hymenachne (Feb) Hymenachne (Apr) Rice (Feb) Rice (Apr) Para (Feb) Para (Apr) Sprayed para (Feb) Sprayed para (Apr)
Axis 1
Multivariate analysis of epiphytic invertebrate data also showed very little separation between vegetation types (Figure 5-6). The invertebrate communities in Para grass were interspersed with other vegetation types, indicating similar invertebrate communities. A subset of samples from the Sprayed Para grass were the only samples that were distinct. These samples were primarily collected during the first year and were charatcerised by relatively low invertebrate richness and low vegetation cover.
39 Figure 5-6
(a) SSH ordination of epiphytic invertebrate communities in the four vegetation types and (b) vector diagram showing highly correlated environmental and community variables. The four vegetation types are abbreviated to: H, Hymenachne; P, Para grass; R, wild rice; and S, Sprayed Para grass.
(a) 2.0
H P R S
1.5
Axis 2
1.0 0.5 0.0 -0.5 -1.0 -2.0
-1.4
(b)
-0.8
-0.2
0.4
1.0
Axis 1 1.0 Hymenachne Abundance
0.5 Temperature
Veg. Cover
Axis 2
Turbidity
0.0 Invertebrate Richness
-0.5 Maidenia
stress=0.20
-1.0 -1.0
-0.5
0.0 Axis 1
0.5
1.0
1.6
40
5.3.2 Terrestrial invertebrates Over 2,600 invertebrates from eight Orders were collected during the dry season (Table 5-3). Over half of these were grasshoppers (Orthoptera); spiders (Aranea), true bugs (Hemiptera) and beetles (Coleoptera) were also numerically important (Table 5-3). Total richness of terrestrial invertebrates was highest in Para grass and Hymenachne, and lowest in Rice and Sprayed Para grass (Table 5-4). Total abundance was much higher in Hymenachne and much lower in Rice than in the other habitats (Table 5-4). Both mean richness and abundance of terrestrial invertebrates per sample showed significant differences between habitats but these were not consistent between the six sites (Richness: F12, 47 = 5.7, p < 0.001. Abundance: F12, 47 = 9.6, p < 0.001). At five of the six sites there was no significant difference in invertebrate richness or abundance between Para grass and Hymenachne, whereas richness and abundance in Rice was significantly lower than in Para grass at four of the six sites (Figure 5-7). Richness and abundance in Sprayed Para grass were more variable (Figure 5-7). Over 5,000 invertebrates from 12 Orders were collected during the two wet seasons (Table 5-3). Three times as many invertebrates were collected during 1998 than in 1999. In the fisrt year, flies (Diptera), wasps and ants (Hymenoptera), and spiders (Aranae) were the most abundant groups (Table 5-3). In the second year, flies were again the most abundant group, but beetles (Coleoptera), Dragonflies and Damselflies (Odonata), true bugs (Hemiptera) and caterpillars (Lepidopetra) were also abundant.
41
Table 5-3
Summary of % abundance of terrestrial invertebrates collected during the dry season and the wet season over two years.
Dry season
Wet season
Order
1998
1998
1999
Total
Aranae
21
16
7
16
Blattodea
