Publications
Jill Landsberg, Craig D. James, Stephen R. Morton, Trevor J. Hobbs, Jacqui Stol, Alex Drew and Helen Tongway
CSIRO Division of Wildlife and Ecology
January 1997
Published by Environment Australia and CSIRO
© Commonwealth of Australia, 1997
ISBN 0 642 27010 4
The species diversity (numbers of species) in each plant group or animal taxon varied substantially between gradients (Table 2.1.1). However, some general trends were apparent. The most diverse of the groups and taxa we sampled were plants, particularly those in the understorey and in the seedbank, and ants. Birds were also moderately diverse. So too were overstorey plants, except at the chenopod gradients. Beetles were moderately diverse at some gradients, but not consistently so, while reptiles tended to be a little lower in diversity, but more consistent. Grasshopper diversity was variable, but rather low, while springtail diversity was consistently low at all gradients where it was assessed. The diversity of small mammals was consistently extremely low at all eight gradients.
The abundance of the different groups and taxa (measured as numbers of individual animals and plants), tended to mirror their diversity of species (Table 2.1.2). Ants were the most abundant taxon by far, with seedbank plants the next most abundant group. Numbers of individual birds were also moderately high, numbers of reptiles were lower, and numbers of small mammals were consistently very low at all gradients. Among the invertebrates however, the correlation between diversity and abundance was apparent for ants only. Despite their low diversity, springtails were the next most abundant invertebrate group after ants, while beetles were frequently the least abundant invertebrate group, despite their moderately high diversity.
The abundance of other invertebrate taxa from pit-traps is shown in Table 2.1.3. Ants were clearly the most numerically dominant taxon across all gradients (Table 2.1.2), but flies (Diptera) were also moderately abundant and spiders (Arachnida) were a consistent presence across all gradients (Table 2.1.3). In addition to springtails, beetles and grasshoppers (Table 2.1.2), other taxa that were abundant on some gradients included wasps (Hymenoptera), ticks and mites (Acarina) and bugs (Hemiptera) (Table 2.1.3). Despite their abundance, the pit-trap samples of winged invertebrates such as flies and wasps are unlikely to be very representative of these groups, because the trapping technique was not very appropriate for animals that can fly.
The diversity and abundance of invertebrates caught in sweep nets (Table 2.1.4) differed substantially from those caught in pit-traps. The sweep netting technique was successful in sampling moderately high numbers of some groups of invertebrates that live on herbage (particularly hemipterans; ie. bugs) but was disappointing in its catch of grasshoppers (Orthoptera), which were the main target taxon. The relatively low numbers of grasshoppers in the samples probably reflects low abundances, rather than any inherent problem with the techniques, since very few grasshoppers were seen during the surveys.
Rainfall in the months preceding the surveys had a marked effect on some, but not all plant groups and animal taxa. Rainfall preceding the 1994 surveys was below average, but average rainfall preceded the 1995 surveys (Table 1.3.5.1). In the gradients dominated by acacia woodland we recorded nearly twice as many plant species growing in the understorey for the two gradients sampled after an average season, compared with the two sampled after a season of below-average rainfall (Table 2.2.1.1). The diversity and abundance of birds showed similar trends. The abundance of both springtails and beetles seemed to show a similar correspondence to season, but diversity of beetles was more variable. The diversity and abundance of ants seemed to relate more to specific gradients than to season: abundance (but not diversity) was highest at the NT mulga gradient, which was sampled after a below average season, and both diversity and abundance were lowest at the Qld gidgee/chenopod gradient, although seasonal conditions there had been average. Table 2.2.1.1 Seasonal variation in species numbers and abundance for selected plant groups and animal taxa at the four gradients dominated by acacia woodland.
Group or taxon Season Gradient No. No.
species individuals
understorey below average NT mulga 54 -
plants
NSW mulga 55 -
average Qld mulga 127 -
Qld gidg/ chen 113 -
birds below average NT mulga 28 633
NSW mulga 28 701
average Qld mulga 50 1,377
Qld gidg/ chen 48 1,810
ants below average NT mulga 80 26,254
NSW mulga 99 9,396
average Qld mulga 92 18,882
Qld gidg/ chen 69 2,096
beetles below average NT mulga 11 16
NSW mulga 28 72
average Qld mulga 40 759
Qld gidg/ chen 24 399
springtails below average NT mulga 7 786
NSW mulga 8 152
average Qld mulga - 6,051
Qld gidg/ chen - 8,330
Diversity and abundance of birds and understorey plants appeared to be affected by vegetation type, particularly for the four gradients surveyed after the average seasonal conditions of 1995 (Table 2.2.1.2). The trend was most apparent for birds, where numbers of species and individuals declined from highest in the mulga gradients, intermediate in the mixed acacia/ chenopod gradients to lowest in the structurally simple chenopod gradient. This trend is not surprising; the relationship between diversity of bird species and vertical habitat diversity has been demonstrated on many occasions since it was first described by MacArthur (1958). Table 2.2.1.2 Variation with vegetation type in numbers of species of understorey plants and birds and numbers of individual birds, for the four gradients sampled after average seasons.
Group or taxon Vegetation type Gradient No. No.
species individuals
understorey closed woodland Qld mulga 127 -
plants
open woodland Qld gidg/ 113 -
chen
open shrubland WA chen/ acac 120 -
shrub steppe WA chenopod 63 -
birds closed woodland Qld mulga 50 1,377
open woodland Qld gidg/ 48 1,810
chen
open shrubland WA chen/ acac 22 1,201
shrub steppe WA chenopod 17 352
There were few trends relating species diversity and vegetation structure for any other plant groups and animal taxa. The WA chenopod/acacias gradient had the highest recorded diversity of overstorey plants and, with the SA chenopod/myall gradient, also had a higher than average diversity of reptiles (Table 2.1.1). In general, however, species richness of reptiles was rather similar across all gradients, irrespective of vegetation type and year of sampling. This is somewhat surprising because in past studies we have found that rain stimulates more species of reptiles to be active (James 1994) so we expected to catch more species in similar habitats in 1995 than 1994. Lower than expected species richness of reptiles in 1995 may have been due to a change in trapping technique: we began using reflective foil covers over pit-traps to protect the captured animals from sun and heat, and this may have deterred some animals.
Many rangeland plant species are short-lived herbs (grasses and forbs). They are usually an inconspicuous part of the flora, hidden in the soil as seeds or present as a very few plants growing just large enough to produce a few flowers and seeds before dying. In very dry seasons they may not be represented by growing plants at all. In very wet seasons they become widespread and abundant; under such conditions they are frequently the showiest plants in desert lands (Inouye 1991; Mott 1973). However, they are not the only plants represented in the soil seedbank: many perennial plant species also maintain stores of viable seeds in the soil. The ability of these species to "come-back" following the death of adult plants during a period of adversity is largely dependent on their ability to maintain stores of viable seed in the soil. We undertook seedbank assessment for two reasons: to gain an indication of the contribution of short-lived herbs to the total species richness of plant communities; and to gain some indication of the potential for species no longer present as adults to "come-back" under more favourable conditions.
Identification of all the plant species represented in the seedbank is made difficult by the requirements of some species for very specific combinations of temperature, quantity and timing of rainfall, and/or preceding seasonal conditions (including fire) (Inouye 1991; Mott 1973). Rather than attempt a complete elucidation of the seedbank, we focussed on the "readily germinable" portion of it: those species that germinated when soils were kept continuously moist under "average" temperature regimes representing warm and cool seasons in the regions from which the soils were collected. We were not able to assess the seedbanks from three of the gradients because of problems with sampling and storing soils (see Section 1.4.2.5).
Seedbanks were assessed as completely as possible for the remaining five gradients, by combining the counts of germinants identified in the samples watered at warm-season temperatures with the counts from samples watered at cool-season temperatures. The seedbanks revealed a rich array of species that tended to complement, rather than overlap, the species detected in the field (Appendix 4;Table 2.2.2.1). The identities of species could be confirmed only if they were grown through to their reproductive stages, and this was not always achieved. The only "morphospecies" that were included in counts, however, were those that grew to a stage sufficient to differentiate them from all others; germinants that died before they could be identified to this level were excluded from analyses. We kept voucher specimens for most species so we could compare field and seedbank "morphospecies" and be reasonably confident that each was in fact a unique species. However, vouchers were not always collected for field species (note the "identity status" field in Appendix 4). Thus there exists the possibility that some unconfirmed "morphospecies" may be the same as some uncollected species detected in the field; if so, the degree of overlap between seedbank and field species may be a little higher than calculated in Table 2.2.2.1.
Between 23% and 56% of the species at each gradient were detected in the seedbank but not in the field. Thus without the seedbank assessments our knowledge of each gradient's flora would have been deficient by at least a quarter to half of its species. In addition, because we assessed only the readily germinable seedbank there may be other species we have not yet detected.
Table 2.2.2.1 Numbers of plant morphospecies detected in the seedbank and in the field.
All morphospecies were judged to be separate species, but because reproductive material was not always available, not all identities could be confirmed. Field plants include both understorey and overstorey species.
Number of Identity NT NSW WA chen/ SA WA chen
species status mulga mulga acac chen/
detected: myall
Only in confirmed 28 56 18 28 14
seedbank
not 24 30 23 24 19
confirmed
total 52 86 41 52 33
% all 46.0 55.8 22.7 30.1 31.7
species
Only in field confirmed 38 40 85 88 30
not 5 12 13 3 23
confirmed
total 43 52 98 91 53
% all 38.1 33.8 54.1 52.6 51.0
species
Both seedbank confirmed 18 16 42 30 18
and field not 0 0 0 0 0
confirmed
total 18 16 42 30 18
% all 15.9 10.4 23.2 17.3 17.3
species
All seedbank conf. + not 70 102 83 82 51
species conf.
% all 61.9 66.2 45.9 47.4 49.0
species
All field conf. + not 61 68 140 121 71
species conf.
% all 54.0 44.2 77.3 69.9 68.3
species
All species conf. + not 113 154 181 173 104
conf.
The proportions of total species detected only in the seedbank were highest for the two mulga gradients assessed. For these gradients, the proportion of total species detected only in the seedbank exceeded the proportion detected only in the field (NT mulga and NSW mulga in Table 2.2.2.1). In part, this was probably because the field surveys for these gradients were undertaken after below-average seasons, when relatively low numbers of understorey plant species were detected in the field (Section 2.2.1). Although few short-lived species were apparent in the field when we surveyed these gradients, many were apparently still present as seeds in the soil.
There may also be an interaction between vegetation type and seedbank, however, independent of season. We assessed seedbanks at three chenopod gradients, two after average seasons (WA chenopod/ acacias and WA chenopod) and one after a below-average season (SA chenopod/ myall) (Table 1.3.5.1). However, the relationships between species detected in the seedbank and field were rather similar for all three gradients, with 23-32% of species detected only in the seedbank, 51-54% of species detected only in the field and 17-23% detected both in the seedbank and in the field (Table 2.2.2.1). Apparently for these chenopod gradients the proportion of plant species likely to be detected in a once-only field survey is rather constant (around half the total flora), regardless of preceding seasonal conditions.
Exotic species of mammals were present on every gradient, but only the House Mouse (Mus musculus) was actually sampled. The other exotics seen, in addition to livestock, were feral goats (Capra hircus), rabbits (Oryctolagus cuniculus) and foxes (Vulpes vulpes). No exotic bird species were detected, but many of the bird species recorded at the gradients have probably extended their range into areas where they previously would not have been able to persist because there was no drinking water (Appendix 1).
Exotic plant species were present and sampled on every gradient (Table 2.1.1). They were most diverse and abundant in the understorey and seedbanks of the chenopod gradients where they constituted between 3% and 11% of the flora (Table 2.2.3.1; Appendix 4). They were a less conspicuous component of the flora of the mulga gradients (range 1-3%) and overall comprised only 4% of the total number of plant species detected (Table 2.1.1). These proportions are at the low end of the national spectrum: naturalised exotic species in the Australian flora range from about 5% in the Northern Territory to 31% in Tasmania, with an overall average of around 15% (Humphries et al. 1992).
Table 2.2.3.1 Contribution of exotic plant species to the total flora detected at the gradients.
Values in italics are underestimates of the total flora, because seedbank assessments were not undertaken for these gradients. Totals for other gradients include understorey, overstorey and seedbank species.
Gradient Total number of plant Number of % exotics
species detected exotics
NT mulga 112 2 1.8
NSW mulga 154 5 3.2
Qld mulga 145 3 2.1
Qld gidgee/ chenopod 129 1 0.8%
WA chenopod/ acacias 181 8 4.4
SA chenopod/ myall 173 11 6.4
SA chenopod 74 2 2.7
WA chenopod 104 11 10.6
The most abundant exotics detected in the field were at the WA chenopod/ acacias gradient where the most abundant grass was the exotic buffel grass (Cenchrus ciliaris) and several other weeds were also moderately abundant (Table 2.2.3.2; Appendix 4). The presence of abundant buffel grass is of particular concern. It has been identified as one of the top environmental weeds in Australia, requiring very high urgency of action (Humphries et al. 1992). It is considered as a critical invasive species, because of its potential to displace native herbaceous species, affect the availability of food for native herbivores, and alter fire regimes. Originally introduced for erosion control and pasture, it has become an aggressive coloniser of mesic habitats (e.g. alluvial pans) in the arid zone. It has also been reported as capable of spreading into adjacent drier habitats, an observation supported by its occurrence along the generally dry WA chenopod/ acacias gradient.
Table 2.2.3.2 The exotic plant species detected along the gradients.
A plus indicates whether the species was detected in the field (Fld) and/or the seedbank (SB). ND indicates seedbanks for which no data are available.
Gradient Species name Common name Family Fld SB
NT Mu *?Brassica sp1362 a forb Brassicaceae +
NT Mu *Schismus barbatus Arabian Grass Poaceae +
NSW Mu *?Brassica sp1362 a forb Brassicaceae +
NSW Mu *Bidens sp+15 a cobbler's pegs Asteraceae +
NSW Mu *Oxalis corniculata Yellow Wood Sorrel Oxalidaceae +
NSW Mu *Rostraria pumila Tiny Bristle Grass Poaceae +
NSW Mu *Schismus barbatus Arabian Grass Poaceae +
Qld Mu *Bidens bipinnata Cobbler's Pegs Asteraceae + nd
Qld Mu *Malvastrum Spiked Malvastrum Malvaceae + nd
americanum
Qld Mu *Oxalis corniculata Yellow Wood Sorrel Oxalidaceae + nd
Qld *Malvastrum Spiked Malvastrum Malvaceae + nd
Gi/Ch americanum
WA Ch/Ac *Asphodelus Onion Weed Asphodelaceae + +
fistulosus
WA Ch/Ac *Cenchrus ciliaris Buffel Grass Poaceae + +
WA Ch/Ac *Cenchrus White Buffel Grass Poaceae +
pennisetiformis
WA Ch/Ac *Chloris virgata Feathertop Rhodes Poaceae +
Grass
WA Ch/Ac *Malvastrum Spiked Malvastrum Malvaceae +
americanum
WA Ch/Ac *Rostraria pumila Tiny Bristle Grass Poaceae + +
WA Ch/Ac *Setaria verticillata Whorled Pigeon Poaceae + +
Grass
WA Ch/Ac *Sonchus oleraceus Common Sowthistle Asteraceae +
SA Ch/My *?Brassica sp1362 a forb Brassicaceae +
SA Ch/My *Bidens bipinnata Cobbler's Pegs Asteraceae +
SA Ch/My *Carrichtera annua Ward's Weed Brassicaceae + +
SA Ch/My *Carthamus lanatus Saffron Thistle Asteraceae +
SA Ch/My *Emex australis Spiny Emex Polygonaceae +
SA Ch/My *Malva parviflora Small-flower Mallow Malvaceae +
SA Ch/My *Rostraria pumila Tiny Bristle Grass Poaceae +
SA Ch/My *Schismus barbatus Arabian Grass Poaceae + +
SA Ch/My *Sisymbrium Smooth Mustard Brassicaceae +
erysimoides
SA Ch/My *Solanum ?nigrum Black-berry Solanaceae +
Nightshade
SA Ch/My *Sonchus oleraceus Common Sowthistle Asteraceae +
SA Ch *Schismus barbatus Arabian Grass Poaceae + nd
SA Ch *Sonchus oleraceus Common Sowthistle Asteraceae + nd
WA Ch *Carrichtera annua Ward's Weed Brassicaceae + +
WA Ch *Centaurea St. Barnaby's Asteraceae +
?solstitialis Thistle
WA Ch *Centaurea melitensis Maltese Cockspur Asteraceae +
WA Ch *Centaurea sp63 a cockspur daisy Asteraceae +
WA Ch *Critesion murinum a grass Poaceae +
WA Ch *Medicago sp66 a medic Fabaceae +
WA Ch *Mesembryanthemum an iceplant Aizoaceae +
?aitonis
WA Ch *Oxalis corniculata Yellow Wood Sorrel Oxalidaceae + +
WA Ch *Sisymbrium irio London Rocket Brassicaceae +
WA Ch *Sonchus ?oleraceus Common Sowthistle Asteraceae +
WA Ch *Sonchus oleraceus Common Sowthistle Asteraceae +
Most of the exotic species detected were grasses or forbs, many of them short-lived members of the families Asteraceae, Brassicaceae and Poaceae (Table 2.2.3.2). The most widespread exotic was the Common Sowthistle (Sonchus oleraceus) which was found at five gradients; however it was generally locally uncommon to rare and showed no detectable response to distance from water (Appendix 4). Of more concern is Arabian Grass (Schismus barbatus) which, as well as being widespread (four gradients), was also frequently abundant and was frequently identified as an increaser species. It was a prominent increaser in the field at both chenopod gradients where it was detected and was the most abundant increaser in the seedbank at one of them. At the two mulga gradients where Arabian Grass was detected it was found only in the seedbank; this is not surprising since it is a short-lived grass (Cunningham et al. 1992) and the gradients were experiencing dry seasons at the time of their surveys (Section 1.3.5). It is worrying, however, since it was moderately abundant in the seedbank at both mulga gradients and was identified as an increaser in one of them, the NSW mulga gradient (Appendix 4). Seedbanks were not assessed at the other two mulga gradients, so we cannot determine whether it is generally widespread in mulga regions.
Also significant was Ward's weed (Carrichtera annua) which was moderately abundant increaser in the field and seedbank at both the WA chenopod and SA chenopod/ myall gradients. It is a common weed of roadsides and depleted pastures throughout semi-arid southern Australia and is prevalent on many parts of the Nullarbor Plain (Cunningham et al. 1992). It has also been identified as significant environmental weed in Western Australia and South Australia (Humphries et al. 1992). Spiked Malvastrum, Malvastrum americanum, was another very widespread weed (it was detected at both Queensland gradients, and also the WA chenopod/ acacias gradient) but it was never abundant and is not generally considered a significant environmental weed (Humphries et al. 1992).
At the chenopod gradients more exotic species were detected in the field than the seedbank, although a number of species were found only in the seedbank (e.g. Cenchrus pennisetiformis, Rostraria pumila, Sisymbrium irio; Table 2.2.3.2). For the two mulga gradients where seedbanks were assessed, most of the exotics were detected in the seedbank. This might partly relate to seasons (the two mulga gradients were dry at the time of the field survey) but not entirely, since a large number of exotic species were detected in both field and seedbank of the SA chenopod/myall gradient, which was also surveyed after a dry season.
Our surveys were very successful in the detection and identification of previously un-recorded species. We identified among our plant samples:
Five new species:
Four species that may be new but require more taxonomic work to determine their status:
Five species collected from well outside their previously known range:
Four species collected so rarely that little is known of their range:
Most of the invertebrate species that we collected have not been formally described in the scientific literature. Only 4-7% of the ants in our samples could be formally identified by an expert taxonomist (Dr Alan Andersen, CSIRO Division of Wildlife and Ecology) reflecting the generally poor species-level taxonomy of Australian ants. It was estimated that at least half of the species recorded in a collection of similar size from semi-arid South Australia were new to science (Andersen and Clay 1996). The most species-rich genera in the gradient samples were Melophorus, Iridomyrmex, Camponotus, Monomorium, Rhytidoponera, Tetramorium, Meranoplus, Pheidole and Stigmacros, in that order (Appendix 4). This is similar to the order of generic richness reported from the arid southern Australian study (Andersen and Clay 1996), except that Rhytidoponera and Pheidole were more prominent in the gradients' samples.
The number of ant species sampled is not known precisely, because the genera Melophorus, Monomorium and Pheidole were not integrated across gradients. That is, the specimens from these genera were not compared across gradients to determine how many species were common to more than one gradient. This was because of the taxonomic intractability, large numbers of species (e.g. 162 species of Melophorus) and limited taxonomic knowledge for these genera (only two of the sampled species have been described in the scientific literature). More than 270 species of ants were identified from the genera that were integrated across all gradients and nearly 250 additional species were identified but not integrated; thus the total number of ant species in the samples was probably somewhat less than 500 (allowing that there may have been some overlap of species in the non-integrated samples; see also Table 2.1.1.).
Of the 148 species of beetles that we sampled, only 13 (9%) could be formally named by an expert taxonomist (Dr John Lawrence, CSIRO Division of Entomology). In addition, approximately 20% of the genera to which the beetles belonged (16 out of 84) could also not been named and may be new to science. Five of the unnamed genera belonged to the Curculionidae (Weevil) family, five to the Melyridae family, three to the Elateridae (Click beetle) family, two to the Staphylinidae (Rove beetle) family and one each to several other families.
Of the 17 species of springtails (Collembola) that we sampled from four gradients, only one has been formally named to species level, four showed affinities with previously described species and 12 (70%) could not be formally named by an expert taxonomist (Dr Penelope Greenslade, CSIRO Division of Entomology).
The gradients proved to be rich in wildlife, particularly in numbers of species of understorey plants, birds, ants and beetles, as well as total numbers of animals (Tables 2.1.1-4). Reptiles were also moderately abundant and diverse, but very few species or individual small mammals were detected. Although this could be an artefact of the sampling (the pit-trap design has been shown to be efficient for reptiles but may under-sample small mammals) it is more likely to reflect the poor conservation status of this group. The disproportionate loss of species of small and medium sized mammals from the arid regions of Australia has been well documented (Appendix 1) and many of the still extant species are now considered rare, very scarce or declining throughout their range (Appendix 1; State of the Environment Advisory Council 1996; Dickman 1994).
Nevertheless, the gradients represent a rich and diverse array of mostly native species. The diversity of understorey plants at some gradients was particularly high. Because of differences in methods of collecting species richness data it is not possible to make direct comparisons with other plant communities, but it is possible to make inferences. The herbaceous understoreys of temperate grassy woodlands in western Victoria are considered to be among the richest plant communities in southern Australia, surpassing mallee and sclerophyll shrublands and comparable to the kwongan of south-west Western Australia (Lunt 1990). Data describing the grassy woodlands were compiled from species-area curves constructed after repeated visits to sites where species were identified in nested quadrats that increased geometrically in size from 0.25 m² to 128 m² (Lunt 1990). The species-area curves flattened out at around 80-90 species in an area of 30-60 m².
The total numbers of understorey species growing in the field at sites along the Queensland gradients were very similar, at between 60 and 90 species (Tables 2.1.1; 3.2.1), identified in 80 x 1m² quadrats (Section 1.4.2.3). The studies are not directly comparable, because gradient "sites" did not consist of nested quadrats but instead were stratified in patches of vegetation scattered over a total area of around 5 ha. But since the species-area curves at the Victorian sites had flattened out (Lunt 1990, described above), it is probably reasonable to assume that most of the species in these communities had already been detected. Similarly, we were confident that our methods successfully detected most plant species growing within a site area (Section 1.4.2.3). We recognised that this was an underestimate of total site richness, however, because our surveys were conducted once-only rather than as repeated visits. Seedbank assessment provided a useful surrogate for repeated site visits, but seedbanks were not assessed at the Queensland gradients. For those gradients where seedbanks were assessed, the number of additional species detected was between 23% and 56% of the gradients' total flora (Table 2.2.2.1). Thus the species richness of the Queensland gradients might be considerably higher than the 60-90 species per site indicated by the once-only field visits. If so, their richness probably exceeds that recorded for the temperate grassy woodlands and other renowned vegetation types described by Lunt (1990).
Exotic species were a minor component of the flora of the acacia woodland gradients but constituted from 3-11% of the species at gradients where chenopod shrubs were a major component of the vegetation. While this is at the lower end of the range recorded for other Australian ecosystems, two of the more abundant exotics have been identified as significant environmental weeds (Humphries et al. 1992).
Birds are a taxonomic group whose occurrence has been relatively well documented in the arid zone of Australia (Appendix 1) and the richness we recorded was within the ranges reported in other studies. For example, Smith and Smith (1994) estimated that there were 10 bird species confined to acacia woodlands in western NSW and 105 species that used this vegetation type regularly. For chenopod low shrublands the estimates were 19 (confined) and 84 (regular). Our counts for comparable environments ranged from 28-50 for acacia woodlands and 16-46 for chenopod shrublands (Table 2.1.1).
The diversity of the ant fauna at our gradients was also very high, but not unexpectedly so, since the semi-arid zone of Australia is known to support an exceptionally diverse ant fauna. Species numbers and the richest genera in our samples are similar to those reported from comparable environments in other parts of Australia (Andersen and Clay 1996; Andersen and Spain 1996). The species-richness of ants in these environments has been reported as rivalling that of lowland rainforests (Andersen and Clay 1996).
None of the species we detected appears to have been formally listed as nationally rare or endangered (State of the Environment Advisory Council 1996) but such listings are inevitably biased toward taxa and regions where sufficient information is available for due consideration. Lack of knowledge of the taxa present in an area or habitat, and therefore, of the influences of external factors on their diversity and abundance, are already recognised as major impediments to facilitating conservation of invertebrates (New 1995). The new species of plants we collected (Section 2.2.4) indicates that there is also a knowledge impediment to documenting the conservation status of plants in these rangelands.
Sampling efficiency is the trade-off between the success of sampling (ie. number, consistency and representativeness of organisms sampled) and the costs of sample collection, processing and identification (New 1996). The less processing required of samples after collection, the lower the costs of identifying them. Similarly, the more easily samples can be identified, the lower those costs.
Numbers of species and individuals surveyed give some indication of the success of sampling of the different plant groups and animal taxa (Tables 2.1.1-2). On these grounds, understorey plants were the most successful group or taxon for sampling, followed by seedbank plants, ants, birds, overstorey plants and reptiles. Springtails were very abundant but not very diverse at the times of our field visits, and beetles, though relatively diverse, were represented by very few individuals of each species sampled (usually less than five individuals per gradient). Small mammals were neither diverse nor abundant.
Arid and semi-arid environments are, by definition, often dry, so many organisms are often dormant or inactive, and are therefore inherently difficult to sample successfully in once-only field surveys. Because we did not encounter above average rainfall seasons during our surveys we undoubtedly missed many animal species, particularly invertebrates, that have a relatively "ephemeral" life-history. For example, springtails have little control over water loss and so tend to become inactive in dry conditions but rapidly resume activity after rain (Greenslade and Greenslade 1985). It is likely, therefore, that our catches of these animals did not reflect their potential diversity and abundance after rain. For plants, we partly overcame dormancy problems by determining the composition of their seedbanks in the glasshouse germination studies. This markedly improved the success of our sampling of plant species, particularly for surveys conducted during below-average seasons (Section 2.2.2).
The success was not without cost, however. Plants in the seedbank and invertebrates were the most expensive biota we sampled, because of the large amount of processing required for these groups (Sections 1.4.2.5 & 1.4.3.3). Their initial assessments took many months of glasshouse and laboratory work per gradient. Subsequently, formal identification of both insects and plants required comparison of voucher specimens with verified specimens catalogued in specialist collections, or identification by expert taxonomists; the latter was relatively efficient once the specimens had been processed.
The efficiency of our sampling of orthopterans was low. In addition to the low numbers of animals caught (Section 2.1), this taxon presented an additional problem in terms of identification of species. It is a difficult group to separate into morphospecies, because of the amount of morphological variation between nymphs and adults. This difficulty was exacerbated in our case because the alcohol in which we stored specimens removed most of their colour, and colour is an important characteristic for identification of orthopterans. Expert taxonomists usually work with pinned specimens. Although we pinned all our voucher specimens this was not done until after the samples had been transported and stored for some time in ethanol. The very large numbers of invertebrates we sampled made it logistically impossible for us to do otherwise.
Sweep netting did not prove to be as effective as expected for sampling invertebrates in the understorey vegetation. Low numbers of captures for target taxa such as orthopterans were probably a consequence of low population sizes rather than a failure of the technique. Low population sizes of orthopterans were probably due to the generally dry conditions when sampling. More successful sampling of orthopterans would probably require either careful timing to coincide with periods of high abundance, or an increase in sampling effort (ie., more sweeps of the vegetation) to increase sample sizes. Increasing the sampling effort raises other problems such as the additional time that would then be needed to extract animals in extractor boxes, and the number of days with low wind conditions that would be needed to conduct sweep netting. (Windy weather reduces the effectiveness of sweep nets, but windy weather frequently develops during the day in arid regions.) The problem of identification is an additional impediment to efficient sampling of orthopterans.
Sample costs were least for plant groups and animal taxa that could be fully processed and identified in the field, but were still substantial in the case of small mammals and reptiles because of the costs associated with purchase and installation of pit-fall traps (Section 1.4.3.2). Birds were the least costly organisms to sample, but required the full-time involvement of an expert to identify them during field surveys.
Weighing sampling success against costs, understorey plants and birds were arguably the most efficient to sample of the plant groups and animal taxa we surveyed. Seedbank plants and ants were similarly successful to sample but much more expensive to process. Decisions about which groups and taxa to target during surveys should not be based solely on efficiency, however. The seedbank plants provided valuable information about site floristic potential that was not apparent from the understorey species present at the time of our surveys. Ants were the most abundant animal group we sampled and represent a very significant component of faunal diversity along the gradients.
Several other invertebrate groups were also abundant in our samples and may provide useful information about faunal diversity. Spiders, for example, were consistently abundant in pit-trap samples from most gradients (Table 2.1.3) and are potentially very efficient to process. Recent work in Tasmanian heathlands showed that spatial patterns in spider communities detected at the species level were equally defined by family level data at larger spatial scales (Churchill 1995). Since differences in vegetation structure were important correlates of the community level spatial patterns, it is possible that similar trends may be apparent among the spiders sampled along the grazing gradients. Family-level characteristics have previously been identified among spider assemblages of heavily grazed sites in England (Gibson et al. 1992), but have not been investigated in Australia.
Regardless of which taxa are investigated in detail, careful documentation and storage of archival collections of all taxa collected during biodiversity surveys should always be a high priority, because responsible archiving retains options for future analysis and interpretation. It is particularly important in arid and semi-arid Australia: the number of new species we identified during our surveys shows how limited is the current state of knowledge about biodiversity in these regions. Taxonomic sufficiency is the appropriate level of identification that balances the need to know the biology (including diversity) of organisms, with accuracy in making identifications and the costs of doing so (New 1996). Unfortunately there is no universal taxonomic level that is sufficient for detecting response to change, because this depends on which attributes of the biology and ecology of the organisms being considered are most responsive to the disturbance regime being investigated.