Studies using climatic gradients play a key role in our understanding of the importance of rainfall and temperature as factors regulating species diversity and distribution, and thus of likely responses to climate change. However, such studies currently consider above-ground species only, ignoring the diverse hypogaeic (subterranean) invertebrate fauna. Here, we assess the extent to which the responses of hypogaeic ant assemblages to a rainfall gradient in the Australian seasonal tropics are congruent with those of epigaeic (ground-dwelling) assemblages. We sampled at 20 sites covering a gradient of over 1000 km and 1100 mm in mean annual rainfall (MAR). We predicted that the different assemblages would display contrasting patterns because the epigaeic fauna is highly arid-adapted whereas the hypogaeic fauna is mesic-adapted in Australian savannas. We collected 392 epigaeic and 17 hypogaeic species. Epigaeic ant abundance and species diversity tended to increase with decreasing rainfall, whereas the reverse was shown by hypogaeic species. In contrast, the phylogenetic diversity (PD) of both epigaeic and hypogaeic assemblages showed only weak or no variation with rainfall. The low response of PD reflected the fact that additional species tended to come from the same rather than additional genera: genus-level diversity did not change with MAR. Epigaeic species composition formed three distinct clusters, corresponding to low, moderate and high MAR, whereas hypogaeic ants did not. Our findings show that ant assemblages can have stratum-specific responses to climatic gradients, reflecting different evolutionary histories of their constituent taxa. This may be true for other taxa, and has important implications for predicting responses to climate change. Lastly, we also show that different levels and facets of diversity have contrasting responses to climatic gradients. We thus advocate for ecological research simultaneously assessing assemblages of several strata, as well as varying levels and facets of their diversity. This dataset consists of ant records collected along the Northen Australian Tropical Transect (NATT) in 2023. Ants were sampled every 50 km along a 1000 km transect. Two methods were used at each site: pitfall and subterranean baited traps. In total, 20 sites were sampled with 400 pitfall and 800 subterranean traps. These traps were left open for 4 days to collect ants. No ethics or sampling permits were required to collect this data. The dataset contains site name, location and environmental attributes (collected from Worldclim). It also contains ant species abundance for each trap used in the study. 

Study sites

Our study was conducted along an extended North Australian Tropical Transect (NATT) in the Northern Territory, running from the northern coast inland toward central Australia. In September 2023, we sampled 20 equidistant sites over a transect of 1000 km, 383 – 1563 mm MAR (Figure 1, Table S1). All sites supported eucalypt-dominated tropical savanna on well-drained sandy or loamy soils.

Sampling

At each site, we arranged a 2 x 10 grid with 10 m spacing, where each point represents the location of one ground pitfall trap (diameter = 4.2 cm) and two subterranean baited pitfall traps (with 4 side entrance holes with diameter = 2.5 mm) placed at a depth of 10 cm. The baits in subterranean traps consisted of an equal mixture of tuna, peanut butter, and honey, which we spread as an inside coating near the aperture of the traps. We chose this mixture to attract a maximum number of ant species, which would find the baits attractive whether they were foraging for carbohydrates, proteins, or lipids. All traps were left operating for 96 hours (4 days). In total, we deployed 400 pitfall and 800 subterranean traps.

We sorted ants to species. The great majority of species are undescribed; we identified these to species group and assigned species codes that apply to this study only. The fauna includes many highly diverse and taxonomically difficult species groups, and for these we used CO1 barcoding to inform species delimitation. We assigned each species a ‘hypogaeic’ or ‘epigaeic’ status according to their morphology and microhabitat. All voucher specimens were deposited in the Darwin collection.

Environmental variables

For each site, we extracted mean annual rainfall, mean annual temperature, mean annual maximum temperature, mean annual minimum temperature, mean annual solar radiation and mean annual water vapor pressure from Worldclim 2.