Jill Landsberg, Craig D. James, Stephen R. Morton, Trevor J. Hobbs, Jacqui Stol, Alex Drew and Helen Tongway
CSIRO Division of Wildlife and Ecology
Biodiversity Convention and Strategy Section of the Biodiversity Group, Environment Australia, January 1997
ISBN 0 6422 7010 4
3. Analysis of biodiversity change along gradients (continued)
Correspondence analyses of the more diverse and abundant plant groups and animal taxa revealed consistent patterns of variation in abundance of individual species with distance from water, for most of the groups and taxa at most of the gradients. On the correspondence graphs (Appendix 3), the distance between sites is a measure of their similarity. Site 1 was often plotted a long way from the other sites, indicating that the species profile for that site often differed from the species profiles for the other sites. Site 6, also, was often plotted separately from all other sites. Sites 1 and 6 were not always isolated, but when they were not, they nearly always grouped with their neighbours. For example, sites 1 and 2 were sometimes grouped together, and occasionally grouped with site 3. Site 6 was often grouped with site 5, particularly for the NT mulga gradient, where both sites were more than 9 km from water. The middle sites (site 3 particularly, but often with site 4 and sometimes with sites 2 and/or 5) were not grouped so consistently; sometimes they clumped together indicating a similarity in their species profiles and sometimes they were moderately distant from each other, as well as being distant from the sites at either end of the gradient. This shows that the middle sites shared similar species profiles for some groups and taxa at some gradients, but for others each of the middle sites had its own distinctive profile.
The consistent grouping of sites among eight gradients indicates two things: (1) that the biotic assemblages do vary along gradients away from water; and (2) that this variation is a consistent result across the range of environments sampled.
For the species corresponding to these site groups, there were consistent trends in the ways their abundance varied with distance from water (Figure 188.8.131.52; Tables 184.108.40.206-6). Regression analyses revealed consistent, significant, log-linear relationships between species abundance and distance from water for the site groups at both ends of the gradients. The species associated with sites closest to water (usually site 1, sometimes sites 1 and 2) consistently showed an "Increaser" pattern of response to the disturbance associated with water, in that their abundance significantly increased with increasing proximity to water (Figure 220.127.116.11). The species associated with sites remote from water consistently showed a "Decreaser" pattern of response (ie opposite to the "Increaser" pattern).
The general linear regressions showing these trends were invariably highly significant (P usually < 0.001) and generally explained high proportions of the variation in species abundance (%Var in Tables 18.104.22.168-6). Occasionally, all the species in a response group were present at similar levels of abundance and showed the same trend with water (ie. only the distance term was significant; Tables 22.214.171.124-6). More frequently, however, the species in each response group had different average levels of abundance (from locally abundant to locally rare; Appendix 4) but shared the same log-linear trend of changing abundance with distance from water (Tables 126.96.36.199-6).
The species corresponding to the middle sites were more variable in their responses. They were frequently present at different average levels of abundance (ie. the "species" term was often significant in Tables 188.8.131.52-6) but their relationship with distance from water was more variable. Some species probably showed flat linear trends comparable to the "neutral species" in Figure 184.108.40.206. Others, however, may have shown non-linear trends (e.g. quadratic relationships with distance from water). Species in this more variable response group have been designated "Not Determined" in the analysis summary (Tables 220.127.116.11-6)
Increaser and decreaser response groups were generally most apparent and most consistently associated with opposite ends of the gradients for plants growing in the understorey (Table 18.104.22.168) and detected in the seedbank (Table 22.214.171.124). There was a moderate degree of overlap in the species represented in these two plant groups (Section 2.2.2) and some of the increaser and decreaser species identified growing in the understorey were also identified as increasers or decreasers in the seedbank (Section 3.4.1). This indicates that the increaser/decreaser patterns apparent among plants growing in the understorey are likely to be perpetuated in future generations of understorey plants, via the seedbank.
Increaser and decreaser response groups were also consistently apparent for birds along most gradients, with the exception of WA chenopod, where no response groups were detected (Table 126.96.36.199). Abundance and diversity of birds were low at both chenopod gradients (Section 2.2.1) and livestock grazing is a much more recent development on the WA chenopod gradient than on any of the others (31 years, compared with at least 100 for the other gradients; Section 1.2.1). This combination may explain why trends in the composition of bird species with distance from water were less apparent here. Response groups at the NSW mulga gradient were only marginally significant (Table 188.8.131.52; 0.05<P<0.10) but birds had access to drinking water from the reference at this gradient, although large grazing animals did not (Sections 184.108.40.206 and 1.3.4). Thus at this gradient there were two potential confounding influences on birds: changes in habitat caused by grazing, and access to drinking water independently of any habitat change.
Increaser and decreaser response groups were also readily apparent among ant species at most gradients (Table 220.127.116.11). Trends were weakest at the chenopod gradients, with no decreaser response group apparent at the SA chenopod gradient and no increaser response group apparent at the WA chenopod gradient.
Trends were generally weakest for overstorey plants (Table 18.104.22.168) and for reptiles (Table 22.214.171.124) which were the least diverse and/or abundant groups and taxa analysed in this way. Increaser and decreaser response groups were evident among reptiles at three of the gradients. Only at one gradient (Qld mulga) were there no response groups detected. The abundance and diversity of reptiles were low at this gradient, but not exceptionally so (Section 2.1).
No increaser or decreaser groups were detected among overstorey plant species at four of the gradients (Table 126.96.36.199). Two of these were chenopod gradients where there were very few species of overstorey plants (Section 2.2.1). Generally trends were most apparent at the gradients that had highest diversity of overstorey plants, with the exception of the Qld gidgee/chenopod gradient where no response groups were apparent despite a moderate diversity of species. In any case, overstorey plants are all perennial shrubs or trees which have much longer life-spans than most species detected in the understorey. For some overstorey plant species (e.g. western myall; Lange and Purdie 1976) the time since the advent of pastoralism is little more than one generation (where generation time is the average age of reproductive adults). It is not surprising that these long-lived plants show fewer signs of change than species that have passed through many generations since pastoralism was introduced.
Tables 188.8.131.52-6 Correspondence and regression relationships for species response groups in diverse, abundant plant groups and animal taxa
Summary of results of correspondence and regression analyses used to identify species response groups for the more diverse and abundant groups and taxa at each gradient. Correspondence graphs (Appendix 3) were used to identify associations between groups of species and sites. The resulting groups of species were analysed by regressions of the form log(species abundance) = ai + bi (distance) where ai represents the intercept and bi represents the slope of the regression for the ith species. For plants growing in the field, abundance was measured as frequency of occurrence of species in quadrats. For seedbank plants and animals it represents the total number of individuals counted.
Explanatory notes for headings in summary tables:
- Terms that were significant in the best-fit model. D=Distance of site from water; S=species, D.S = the interaction between them. D was tested first, to determine whether there was a significant trend in abundance with distance, regardless of any differences among species. If D was significant, it was then tested in combination with S, to test whether each of the species had significantly different abundances (different intercepts) but the same trend with distance (same slope). If D was not significant, S was fitted on its own, to see whether the abundances of the different species were significantly different from each other. If D and S were both significant, they were tested in combination with D.S, to test whether each of the species had significantly different abundances (different intercepts) and significantly different trends with distance from water (different slopes).
- Regression slope (± standard error) of the best-fit model, for the relationship between (log) abundance of species and distance from water. Slopes are only recorded where there was a common slope for all species (ie. where the best-fit model terms were D, or D and S).
- Summary of best-fit model: Prob. = probability that the slope is significant; d.f. = degrees of freedom (model & residual) and %Var = percentage of variance accounted for by the model.
- Response groups are determined from the slope of the best-fit model: INCREASERS have negative regression slopes (indicating increasing abundance with increasing proximity to water) and decreasers have positive regression slopes (indicating the reverse). Species response groups are shown as "nd" (NOT DETERMINED) if none of the regressions was significant, or if S was the only significant term, or if the interaction D.S was significant and indicated a mixture of positive and negative trends with distance.
The probability of the regression slope differing from zero is considered marginal at 0.10>P>0.05; where this occurs the response group is shown in lower case.
The proportions of each gradient's species in the different response groups were moderately consistent for most of the plant groups and animal taxa surveyed at most of the gradients (Tables 184.108.40.206-6). From 36-75% of species did not show any linear trend in abundance with distance from water (those in the "not determined" response group). Overstorey plants were the most prominent group in this category (Table 220.127.116.11). The presence of mature individuals may not be a very good indication of trend in overstorey plants, however, because they tend to be long-lived (see section 3.3.1). Therefore they may not be doing as well as the high proportions of apparently unaffected species indicate.
The abundance of the remaining species in all groups and taxa was strongly influenced by proximity to water. Many were decreasers; species whose abundance decreases with proximity to water. Across all gradients this averaged around 15% of the overstorey plant species at each gradient, around 22% of reptiles, 23% of plants in the seedbank, 23% of birds, 26% of ants and 38% of understorey plants (Tables 18.104.22.168-6). It is worth noting that the trend in abundance shown by species in these groups is log-linear (Section 22.214.171.124). That is, their abundance drops logarithmically with proximity to water. Conversely, as distance from water increases the abundance of decreaser species rises logarithmically. There is no evidence that their abundance has begun to stabilise even at the reference sites, which are 8-15 km from water (Table 126.96.36.199). Presumably their optimal habitat lies even further away from water.
Other species were increasers, whose abundance is favoured by proximity to water. On average, the proportions of increaser species in the different groups and taxa were very similar to the proportions of decreasers, ranging from 10% for overstorey plants to 33% for plants in the seedbank. (Paired two-tailed t-tests comparing increasers and decreasers in each plant group and animal taxon showed no significant differences at P<0.10). Although there was considerable variation between individual gradients, the highest average proportion of increaser species was found in the seedbank (Table 188.8.131.52). The prominence of increaser species in this plant group may be partly an artefact of the way it was assessed. The glasshouse trials detected only the readily germinable component of the soil seedbank; that is, only those species that germinated when water and temperature were not limiting and did not have any complex dormancy-breaking requirements (Section 184.108.40.206). Since large seedbanks and relative ease of germination are common attributes of colonising species (Thompson 1992) it is likely that our estimation of seedbank species was biased toward those species most likely to be favoured by disturbance. These are also the species most likely to be increasers.
Whatever the reason, both increaser species and those for which we could determine no response to distance from water, appear able to cope with the changes associated with water provision. Therefore, around 75% of species at each gradient (from 62% for understorey plants to 85% for overstorey plants) appear able to cope. Even if overstorey plants are not included, an average of around 73% of species at each gradient appear not to be disadvantaged by the provision of water for pastoralism.
These are not the species of greatest conservation concern, however. Water points are now so widespread across the pastoral rangelands that most of the chenopod and acacia rangelands now lie within 10 km of an artificial source of water (Landsberg and Gillieson 1996; Appendix 2). Therefore most of the pastoral rangelands now provide habitat that is potentially suitable for species that are advantaged or unaffected by the provision of water. In contrast, habitat likely to be suitable for the persistence of decreaser species has been reduced to a very small fraction of its former extent. Perhaps as little as 3-8% of pastoral rangelands are now potentially remote from water (calculated from Appendix 2; assuming that only one third of water points are named, and the remainder have similar patterns of distribution). Therefore the area of potentially suitable habitat remaining for the 15-38% of species showing decreaser trends may be as little as 3-8% of its original extent.
Tables 220.127.116.11-6 Proportion of each gradient's total species in each response group, for the more diverse and abundant plant groups and animal taxa.
See Tables 18.104.22.168-6 for an explanation of column headings. Only response groups where the trend is significant at P<0.05 are included.
Gradient Total No. Spp. %Increasers %nd %Decreasers NT mulga 54 16.7 53.7 29.6 NSW mulga 55 18.2 49.1 32.7 Qld mulga 127 29.9 16.5 53.5 Qld gidgee/chenopod 113 15.9 0.0 84.1 WA chenopod/acacias 120 33.3 48.3 18.3 SA chenopod/myall 121 18.2 43.0 38.8 SA chenopod 71 50.7 19.7 29.6 WA chenopod 63 22.2 57.1 20.6 mean ± std error 90 ± 11 26 ± 4 36 ± 7 38 ± 8
Gradient Total No. %Increasers %nd %Decreasers Spp. NT mulga 19 0.0 100.0 0.0 NSW mulga 19 42.1 36.8 21.1 Qld mulga 24 0.0 41.7 58.3 Qld gidgee/chenopod 28 0.0 100.0 0.0 WA chenopod/acacias 50 18.0 56.0 26.0 SA chenopod/myall 25 16.0 68.0 16.0 SA chenopod 6 0.0 100.0 0.0 WA chenopod 13 0.0 100.0 0.0 mean ± std error 23 ± 5 10 ± 5 75 ± 10 15 ± 7
Gradient Total No. Spp. %Increasers %nd %Decreasers NT mulga 69 34.8 65.2 0.0 NSW mulga 102 36.3 41.2 22.5 WA chenopod/acacias 83 43.4 21.7 34.9 SA chenopod/myall 82 19.5 42.7 37.8 WA chenopod 51 33.3 49.0 17.6 mean ± std error 77 ± 8 33 ± 4 44 ± 7 23 ± 7
Gradient Total No. %Increasers %nd %Decreasers Spp. NT mulga 26 15.4 26.9 57.7 NSW mulga 28 0.0 100.0 0.0 Qld mulga 44 59.1 25.0 15.9 Qld gidgee/chenopod 46 0.0 43.5 56.5 WA chenopod/acacias 18 11.1 77.8 11.1 SA chenopod/myall 47 21.3 57.4 21.3 SA chenopod 16 31.2 43.8 25.0 WA chenopod 15 0.0 100.0 0.0 mean ± std error 30 ± 5 17 ± 7 59 ± 11 23 ± 8
Gradient Total No. %Increasers %nd %Decreasers Spp. NT mulga 14 14.3 28.6 57.1 NSW mulga 14 0.0 85.7 14.3 Qld mulga 15 0.0 100.0 0.0 Qld gidgee/chenopod 14 0.0 85.7 14.3 WA chenopod/acacias 20 45.0 35.0 20.0 SA chenopod/myall 23 0.0 56.5 43.5 SA chenopod 13 61.5 15.4 23.1 WA chenopod 13 23.1 76.9 0.0 mean ± std error 16 ± 1 18 ± 8 60 ± 11 22 ± 7
Gradient Total No. Spp. %Increasers %nd %Decreasers NT mulga 79 51.9 19.0 29.1 NSW mulga 100 0.0 40.0 60.0 Qld mulga 92 21.7 28.3 50.0 Qld gidgee/chenopod 69 0.0 89.9 10.1 WA chenopod/acacias 96 44.8 32.3 22.9 SA chenopod/myall 87 28.7 58.6 12.6 SA chenopod 34 47.1 52.9 0.0 WA chenopod 50 0.0 76.0 24.0 mean ± std error 76 ± 8 24 ± 8 50 ± 9 26 ± 7
The statistical determination of response groups did not differentiate between exotic and indigenous species, since both contribute to patterns of changing abundance. Exotics were only apparent in response groups for understorey plants and seedbank plants, where they were most prominent as increasers at some chenopod gradients (Tables 22.214.171.124-2). The highest proportion of increaser exotics was found growing in the understorey at the WA chenopod gradient. Exotics comprised half of the understorey plant species in the increaser group at this gradient (11% of all species; Table 126.96.36.199). Increaser exotics were a smaller proportion of the seedbank flora at this gradient (4%) but were still apparent (Table 188.8.131.52). Exotics were also moderately prominent among the increaser plants at the SA chenopod/myall gradient, where increaser exotics constituted 5% of both understorey and seedbank plants. Exotic species were also found in other response groups, but generally as a minor component. The highest proportions of exotic decreasers were identified in the seedbank at the NSW mulga gradient and the WA chenopod/acacias gradient, where they constituted 1-2% of species detected (Table 184.108.40.206). No exotic decreasers were detected growing in the understorey during the field survey of these gradients (Table 220.127.116.11) but their presence in the seedbank is of concern, since it suggests that exotics have the potential to establish at these gradients in little disturbed sites a long way from water. The detection of exotics among the decreasers growing in the understoreys of four of the gradients supports these concerns, although the exotics currently contribute only minor proportions of the flora (Table 18.104.22.168).
Gradient Total No. % Increaser % Not %Decreaser Determined of species Native Exotic Native Exotic Native Exotic NT mulga 54 16.7 0.0 53.7 0.0 29.6 0.0 NSW mulga 55 18.2 0.0 47.3 1.8 32.7 0.0 Qld mulga 127 29.1 0.8 16.5 0.0 52.0 1.6 Qld gidg/ 113 15.9 0.0 0.0 0.0 83.2 0.9 chen WA chen/ 120 30.8 2.5 14.2 2.5 18.3 0.0 acac SA chen/ 121 13.2 5.0 42.1 0.8 38.0 0.8 myall SA chenopod 71 47.9 2.8 19.7 0.0 29.6 0.0 WA chenopod 63 11.1 11.1 54.0 3.2 19.0 1.6 Mean 90.5 22.9 2.8 30.9 1.0 37.8 0.6 Standard 10.7 4.1 1.3 6.9 0.4 7.0 0.2 Error
Gradient Total No. % Increaser % Not %Decreaser Determined of species Native Exotic Native Exotic Native Exotic NT mulga 70 34.3 0.0 15.7 0.0 0.0 0.0 NSW mulga 102 34.3 2.0 16.7 0.0 20.6 2.0 WA chen/ 83 41.0 2.4 19.3 2.4 33.7 1.2 acac SA chen/ 82 14.6 4.9 41.5 1.2 37.8 0.0 myall WA chenopod 51 29.4 3.9 47.1 2.0 17.6 0.0 Mean 77.6 30.7 2.6 28.1 1.1 21.9 0.6 Standard 7.5 4.0 0.8 6.0 0.4 6.0 0.4 Error
Fortuitously, the four gradients sampled to represent two main vegetation types (acacia woodland and chenopod shrubland) also represented contrasting seasonal conditions in terms of rainfall: for each vegetation type two of the gradients were sampled after "below average" seasonal conditions while the other two were surveyed after "average" seasonal conditions (Section 1.3.5). This allowed us to investigate whether there were any consistent trends suggesting potential seasonal differences in the proportion of species in different response groups. We were particularly interested to see whether the proportion of decreaser species appeared to drop when seasonal conditions were relatively good ("average"), compared with relatively bad ("below average"). If so, it could indicate that populations of some decreaser species may be capable of recovering to their full potential abundance in very favourable seasons.
Unfortunately, our data provided no clear indication of consistent seasonal trends in proportions of decreaser species at the gradients. This may have been because of the very small number of replicates (only two vegetation types per season) and partly because seasonal effects were only readily apparent among species of understorey plants and birds at the acacia woodland gradients (Section 2.2.1). For birds there was no evidence of a consistent seasonal trend at these gradients: both high and low proportions of decreaser species being apparent on different gradients, regardless of season (Table 22.214.171.124). For understorey plants, however, the proportion of decreaser species was much higher on the two woodland gradients surveyed after average seasons than on the two woodland gradients surveyed after below average seasons. Thus for understorey plants, rather than there being any indication of a seasonal recovery in decreaser trends, higher proportions of decreasers were apparent in better seasons.
(See Table 126.96.36.199 for derivation of seasonal variation data, but note that some bird species were excluded from statistical analyses and are not included below.)
Group or taxon Season Gradient No. % decreasers species understorey below average NT mulga 54 29.6 plants NSW mulga 55 32.7 average Qld mulga 127 53.5 Qld gidg/ 113 84.1 chen birds below average NT mulga 26 57.7 NSW mulga 28 0.0 average Qld mulga 44 15.9 Qld gidg/ 46 56.5 chen
The numbers of species of small mammals and springtails detected were so low (Table 2.1.1) that correspondence analysis was not appropriate for these taxa. Nor was it appropriate for any other of the invertebrate taxa identified to species, apart from ants. Species numbers for beetles and grasshoppers tended to be higher than for springtails, but the numbers of animals per species were still very low (Table 2.1.2), and high counts were restricted to very few species. Since there were so few species with moderately high counts among these taxa, regression analyses were undertaken individually for each species with a total count of > 5 animals per gradient (Section 188.8.131.52).
However, few of the regressions were significant, and these did not tend to be consistent. For small mammals the two different response types that were significant were only marginally so, and of doubtful biological significance (Table 184.108.40.206). Except for unidentified immature springtails, only three species of springtails showed significant trends: one showed a decreaser trend at one gradient, another an intermediate (hump-shaped) trend at another gradient, and a third showed an increaser trend at a third gradient (Table 220.127.116.11). Drepanura cinquilineata was one of the most abundant and widespread springtails (occurring at three of the four gradients assessed) but its abundance did not generally vary significantly with distance from water, apart from a weak decreaser trend at one gradient.
Similarly, trends shown by individual species of beetles were not significant or consistent (Table 18.104.22.168). Of the eleven species for which regressions were tested only three showed any significant trends and they were all different. The most abundant and widespread beetle, Corticaria subtilissima, did not show any significant trends with distance from water. Twelve species of grasshoppers and crickets were sufficiently abundant to analyse (Table 22.214.171.124). Two showed significant and consistent trends of increasing in abundance close to water, but ten, including the most widespread and abundant species, Endocusta spA, showed no significant relationships with distance from water.
Summary of results of regression analyses for individual species in animal taxa with low species diversity and/or low abundance. The regressions tested were of the form log(count) = a + b1 (distance) + b2 (distance)². Regressions were only tested for those species that had a total count of > 5 animals across a gradient.
Explanatory notes for headings in summary tables:
- Species with total counts > 5 animals in the gradient
- Significant terms and their coefficients (regression slope ± standard error) for the most significant relationship between (log) count and distance from water.
- Summary of best-fit model: Prob. = probability that the slope is significant; d.f. = degrees of freedom (model & residual) and %Var = percentage of variance accounted for by the model.
- Response types are recorded as INC (increaser) when log (count) = a - b1 (distance); DEC (decreaser) when log (count) = a + b1 (distance); INT (intermediate) when log (count) = a + b1 (distance) - b2 (distance)² and INV (inverted) when log (count) = a - b1 (distance) + b2 (distance)².
When regressions were significant at P<0.05 response types are shown in upper case; when regressions were only marginally significant (0.10>P>0.05) response types are shown in lower case.
Notes: See above
Notes: See above
Notes: See above
Notes: See above
Many of the species identified as decreasers were locally rare (Appendix 4) and moderately high proportions of species were found only at the reference sites (Tables 126.96.36.199-6) which were 8-15 km from water (Table 188.8.131.52). If the distance from water is the primary reason for the local rarity of these species it is a matter of great conservation concern, because there are very few areas left this far from water in the more productive rangelands (Appendix 2).
Gradient Total number of Number found only % found only at species at reference reference NT mulga 54 5 9.3 NSW mulga 55 6 10.9 Qld mulga 127 9 7.1 Qld gidg/chen 113 10 8.8 WA chen/acacias 120 9 7.5 SA chen/myall 121 16 13.2 SA chenopod 71 5 7.0 WA chenopod 63 3 4.8 mean ± std error 91±11 8±1 9±1
Gradient Total number of Number found only % found only at species at reference reference NT mulga 19 0 0.0 NSW mulga 19 4 21.1 Qld mulga 24 4 16.7 Qld gidg/chen 28 0 0.0 WA chen/acacias 50 3 6.0 SA chen/myall 25 1 4.0 SA chenopod 6 0 0.0 WA chenopod 13 1 7.7 mean ± std error 23±5 2±1 7±3
Gradient Total number of Number found only % found only at species at reference reference NT mulga 71 3 4.2 NSW mulga 103 7 6.8 WA chen/acacias 89 6 6.7 SA chen/myall 83 2 2.4 WA chenopod 56 1 1.8 mean ± std error 80±8 4±1 4±1
Gradient Total number of Number found only % found only at species at reference reference NT mulga 21 3 14.3 NSW mulga 28 3 10.7 Qld mulga 44 0 0.0 Qld gidg/chen 44 4 9.1 WA chen/acacias 18 1 5.6 SA chen/myall 47 4 8.5 SA chenopod 16 1 6.2 WA chenopod 15 0 0.0 mean ± std error 29±5 2±1 4±1
Gradient Total number of Number found only % found only at species at reference reference NT mulga 14 1 7.1 NSW mulga 14 2 14.3 Qld mulga 15 2 13.3 Qld gidg/chen 14 0 0.0 WA chen/acacias 20 1 5.0 SA chen/myall 23 3 13.0 SA chenopod 13 3 23.1 WA chenopod 13 0 0.0 mean ± std error 16±1 2±0.5 9±3
Gradient Total number of Number found only % found only at species at reference reference NT mulga 80 4 5.0 NSW mulga 99 8 8.1 Qld mulga 92 5 5.4 Qld gidg/chen 69 3 4.3 WA chen/acacias 96 2 2.1 SA chen/myall 87 4 4.6 SA chenopod 34 2 5.9 WA chenopod 50 2 4.0 mean ± std error 76±8 4±1 5±1
Unfortunately, it is not possible to determine from the gradient surveys the reasons why some species were found only at the reference sites. One possibility is that the number of species was not related to grazing intensity, but was merely the result of natural variation in the distribution and abundances of some species. Study sites were not arrayed at uniform distances along the gradients; instead the spacing between the reference sites and their nearest neighbour was greater than the spacing between the sites closest to water (Table 184.108.40.206). Thus it is possible that the reference sites had high numbers of species not found at other sites because they were relatively isolated, while the sites close to water may have had more species in common because they were closer together.
One way to test this is to consider those species found only at a subset of equidistant sites. For all the gradients except NT mulga, sites 2, 4, 5 and 6 were approximately 2.5 km apart. For these gradients, a subset was formed by excluding the other two sites (sites 1 and 3) and the species found only at them. For most plant groups and animal taxa (the exception was understorey plants) there were no significant differences among any of these four sites in the numbers of species found at only one of them (Tables 220.127.116.11 -13). That is, for most groups and taxa the number of species found only at a reference site was comparable to the number of species found only at any one of the other three equidistant sites along a gradient, regardless of distance from water. For the understorey plants, however, the number of species found only at site 6 was significantly higher than the number of species found only at site 2 or site 4 or site 5 (Table 18.104.22.168 and 22.214.171.124; 0.01<P<0.05).
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 3 6 4 7 Qld mulga 4 5 10 13 Qld gidgee/chenopod 5 8 7 14 WA chenopod/acacias 10 8 8 10 SA chenopod/myall 9 11 6 20 SA chenopod 3 11 4 5 WA chenopod 6 4 1 3
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 2 1 0 4 Qld mulga 3 0 1 5 Qld gidgee/chenopod 0 6 3 3 WA chenopod/acacias 7 3 4 3 SA chenopod/myall 4 0 1 2 SA chenopod 0 1 2 0 WA chenopod 1 0 3 2
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 9 13 5 11 WA chenopod/acacias 9 7 7 13 SA chenopod/myall 6 8 10 2 WA chenopod 4 2 4 3
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 5 1 3 3 Qld mulga 7 1 5 3 Qld gidgee/chenopod 1 3 2 4 WA chenopod/acacias 2 2 0 1 SA chenopod myall 1 5 5 6 SA chenopod 4 1 0 1 WA chenopod 0 1 2 0
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 0 0 2 3 Qld mulga 3 1 1 2 Qld gidgee/chenopod 0 4 3 1 WA chenopo/acacia 2 1 4 1 SA chenopod myal l 2 4 2 5 SA chenopod 1 0 1 3 WA chenopod 0 0 2 1
Gradient Site 2 Site 4 Site 5 Site 6 NSW mulga 8 7 9 12 Qld mulga 5 6 11 7 Qld gidgee/chenopod 6 4 14 4 WA chenopod/acacia 13 4 9 4 SA chenopod myall 8 12 4 6 SA chenopod 8 5 1 2 WA chenopod 3 5 2 2
(See Section 1.5.4 for analytical details.)
Group or taxon d.f. deviance ratio Prob. Understoreyplants 3, 18 3.76 <0.05 Overstorey plants 3, 18 0.46 >0.10 Seedbank plants 3,9 0.07 >0.10 Birds 3, 18 0.24 >0.10 Reptiles 3, 18 0.69 >0.10 Ants 3, 18 0.54 >0.10
Apparently for most plant groups and animal taxa, some species were so uncommon along the gradients that they were detected at only one of the four equidistant sites. They may have been locally rare along a gradient because the gradient's environment was only marginally suitable for them. If this was the case they were probably more common in other parts of the landscape (e.g. upslope or downslope of our sites), or other landscape types (e.g. dunefields or floodplains), or other climatic zones (e.g. more humid or more arid). For species for which a gradient's environment was only marginal, occurrence and detection at any one site might reflect nothing more than chance.
This possibility does not negate the concern about species found only at the reference sites, however. The total number of species found at each of the sites along a gradient was usually rather constant (Section 3.2), but this was a composite result masking considerable variation in underlying composition. The total number of species at each site represented a changing balance between increaser species predominating near water and decreaser species predominating far from water (Figure 126.96.36.199). It is possible that some of the species found at only one site reflect this underlying compositional shift. For understorey plants, the number of species occurring only at one site was significantly higher at the reference sites than at any of the other equidistant sites. Understorey plants constituted the most species-rich group or taxon sampled and are also most directly affected by the elevated grazing that occurs around water points. Thus it is possible that this group might be the most sensitive indicator of species in decline.
A more regional picture of the distribution and abundance of species is needed before it will be possible to determine whether the species found only at the reference sites are more abundant elsewhere in the region, independently of proximity to water. Alternatively, systematic regional surveys may identify species that are everywhere restricted to sites remote from water; for most of the productive rangelands water is now so widespread that any such species may be at risk of regional extinction.