Multi-taxa fauna survey (birds, reptiles, ants) in arid central Australia for impacts of buffel grass (Cenchrus ciliaris) on fauna communities

The data set accompanies the paper in submitted for review in Ecology and Evolution. The data set includes bird, reptile and ant surveys that were conducted concurrenlty in 2018-2019 in the Aṉangu Pitjantjatjara Yankunytjatjara (APY) Lands of arid central Australia to test whether the invasion of buffel grass (Cenchrus ciliaris) had consistent impacts on the functional structure of fauna communities across major taxons. Furthermore, the experiment compared how detecting impacts using the functional approach (aggregating species into functional groups of fauna) compared to measures that used total taxon abundance or diversity, or taxonomic composition. The experimental design was a spatially paired-plot (randomised-block) experiment at 16 sites (8 pairs) for birds and, within these sites, 12 sites (6 pairs) for reptiles and ants, with each pair representing native and invaded status of the same vegetation community. The experiments spanned two geographic regions (~ 300 km apart) and multiple vegetation communities amongst flat plains and rocky hills landforms. The data set also contains a plant species data file and site variables file including percentage cover of habitat components that was used to test the key habitat and vegetation differences between the native and invaded paired sites. Data collection methodology is described in the accompanying paper, and summarised here. This survey was conducted at a subset of sites used in Ryan-Colton (2024) DOI: 10.1002/ecs2.70033 and dataset 10.5281/zenodo.13627594. All coordinates provided in MGA 52 J Eastings and Northings (UTM, Australian National Grid). The plant and fauna data was collected in accordance with standard biological survey methods in South Australia (Heard and Channon 1997, Owens 2003). Birds, reptiles, ants and vegetation were surveyed concurrently at each site over four days in September 2018 (West region) and four days in October 2019 (East region). Within each uniform habitat patch (> 0.1 km2), vegetation was sampled within a one-hectare plot (100 m x 100 m) at the centre of each site. Reptiles and ants were surveyed along two 60 m traplines, one within and one outside the one-hectare plot within the extended habitat patch. Each pitfall line had 6 traps each (vertebrate pitfall traps for reptiles and ants, micropitfall traps for ants). Traps were left open for 4 days and checked daily. Reptiles were marked and released onsite and recaptures were excluded from the dataset. Ants were captured in ethanol and ant specimens were identified and lodged with the South Australian Museum. Abundance per trap lines (for reptile and ants) were pooled per site to form the taxon matrix. Ant abundance and ant diversity used micro-pit data only, whilst composition used ant captures from micro-pits and reptile pitfalls combined. Ant abundance in each vial was capped at 100 individuals per genus to reduce the influence of trap placement, in case this was near a nest or foraging trail. Bird surveys were conducted over the one-hectare plot and extended habitat patch, covering between 0.1 - 0.15 km2 in a timed survey. Each of the 16 bird sites was surveyed twice over a four-day period, once in the morning (between 0600 - 1000) and once in the afternoon (1400 - 1900) totalling 32 surveys. Each survey was a one-hour area search. Bird abundance was mean abundance per site averaged across the morning and afternoon surveys. Functional groups for birds, reptiles and ants were defined to best represent three functional axes of 1) diet, 2) habitat use, and 3) the scale of resource use in a way that was meaningful for each taxon. For example, the scale of resource use was defined by proxies of movement patterns (birds), body size (reptiles) or hierarchy-aggregation size (ants) to ensure taxa within each group could be represented along this axis, and we acknowledge these three taxa operate at different scales to begin with. Functional groups were defined based on literature searches, and full functional group assignment is available in the dats files, and literature sources are available in the paper’s Appendix. Given that fauna can respond to invasion via multiple traits, that may be interdependent, for birds and reptiles we coalesced the three functional axes (diet, habitat use and scale) into two-way combinations, such that final functional groups for each taxa represented their diet x habitat use, diet x scale and habitat use x scale. These two-way groups were used in the multivariate analysis. Ant functional groups have been well described in the context of predicting their response to environmental change (Andersen 1997, Santos et al 2021). Multivariate analysis was performed on these groups, and subsequently interpreted post-hoc along three functional axes. Total abundance, diversity and taxonomic composition of birds, reptiles and ants were also analysed separately to comapre to the functional approach. Vegetation sampling methods are reported in more detail in Ryan-Colton et al (2024). At each one-hectare plot we recorded plant species cover-abundance estimated in the field using a modified Braun-Blanquet scale and later converted to a raw continuous variable based on the mid-point of the cover class: 1% (1-10 plants, <5% cover); 2% (sparsely present, <5% cover; 3% (plentiful but <5% cover); 15% (5 to 25% cover class); 37% (25 to 50% cover class); 63% (50 to 75% cover class). Buffel grass was recorded on the same scale. The raw mid-point percent cover was converted to relative percent cover by dividing each species’ raw cover by the summed cover of all species at that site (including buffel grass + understorey + overstorey species). Plant species were vouchered and identification checked post-field by the South Australian Hebarium. Plant taxonomy reflects current names (as of 2015) in the Biological Databases of South Australia. Sensitive species (one species in this dataset) has had the coordinates denatured by 10km due according to the requirements of the Biological Database of South Australia for sensitive species. We also visually estimated percentage bare ground and rock, and litter, over the one-hectare plot. We then summed the relative visual cover of different plant groups to produce five habitat variables in total: buffel grass, bare ground, litter, native understorey plants and overstorey shrubs/trees (dead or alive). On the flatter plains (12 all taxa sites) percent cover of these same five habitat variables was estimated using point-sampling image analysis (SamplePoint, Booth et al 2006) from top-down drone images (Mavic Pro, e.g. Appendix S1: Figure S1). Steep inclines on rocky sites led to oblique drone images and thus image analysis was not suitable for these sites (which were only surveyed for birds), so for consistency for bird sites, the ground-based visual estimates of percent cover for the habitat variables were used for all bird sites. The authors wish to acknowledge Traditional Owners and Aṉangu Pitjantjatjara Yankunytjatjara (APY) Lands Organisation who gave permission for collaboration, data collection, photographs and reporting on and about their Traditional Lands. Data is jointly the Intellectual Property of Aṉangu as the Traditional Owners and the author team, and approval has been granted for research and publication use of the data with acknowledgment of Aṉangu and the author team as IP holders. Many people assisted in the field during the fauna surveys and with fauna identification and are wholly acknowledged in the paper. APY Land Management, Alinytjara Wilurara Landscape Board, Central Land Council, Ten Deserts Project, Charles Darwin University, South Australian Department for Environment and Water, State Herbarium of South Australia, Holsworth Wildlife Research Endowment, Jill Landsberg Trust and Ecological Society of Australia all provided either funding and/or in-kind support of the project. Study conducted with APY Executive Board approval, South Australian Scientific Permit Q26782 and Northern Territory Wildlife Permit 63104.