Habitat structure is a key driver of animal diversity (Tews et al., 2004) because it enhances resource and niche availability and promotes species coexistence (Costanza et al., 2011; Pérez-Sánchez et al., 2023). The effect of habitat structure on species diversity, however, can vary across spatial scales (Cushman & McGarigal, 2002; Ross et al., 2017; Stoner & Joern, 2004), and comparing habitat–diversity relationships at local or landscape scales is key to improving our understanding of what features of ecosystems are important in promoting diversity in different regions and for different taxa (Barton et al., 2009; Campos et al., 2011; Cushman & McGarigal, 2002).
Ants are a conspicuous and ecologically important component of the insect fauna of many terrestrial ecosystems (Folgarait, 1998; Griffiths et al., 2018; Kass et al., 2022; Lach et al., 2010). Yet, despite their importance, the knowledge of ant communities and their diversity is still lacking for many environments and habitats around the world (Kass et al., 2022). One way to close this knowledge gap is by examining ant diversity in different habitat types and assessing the role of different habitat attributes in shaping ant community structure (Campos et al., 2011; Spiesman & Cumming, 2008). Vegetation cover created by trees and shrubs is a major component of habitat structure and a key driver of local ant diversity (Barton et al., 2016; Oliver et al., 2016; Pérez-Sánchez et al., 2023). It affects, among other factors, light penetration, microclimate, and litter or plant cover at ground level (Cerda et al., 1997; Dattilo & Izzo, 2012) and exerts an influence on ants through their tolerance to temperature and humidity, preferences for litter depth and composition, or availability of food resources such as honeydew, prey insects, and seeds (Gibb & Cunningham, 2013; Holldobler & Wilson, 1990; Lach et al., 2010).
Studies on animal communities, including ant communities (Cushman & McGarigal, 2002; Whittaker et al., 2001), have identified a hierarchical suite of factors affecting diversity and community composition, including large-scale biogeographic factors like primary productivity down to small-scale environmental heterogeneity and soil characteristics (Maravalhas & Vasconcelos, 2020; Pérez-Sánchez et al., 2023; Spiesman & Cumming, 2008). Within this range of scales, vegetation or land use type can vary across landscapes, whereas vegetation cover or complexity can vary at much smaller, local scales (Oliver et al., 2016; Spiesman & Cumming, 2008). A key characteristic of vegetation or habitat cover is the concept of “openness,” which has been noted as a key driver of ant community structure (Andersen, 2019). Vegetation cover and openness are important features of many woodland and savannah ecosystems characterized by a heterogeneous cover of trees and grassland, and distributed across the Australian, African, and South American continents (Asner et al., 1998; Pilon et al., 2021; Tongway & Ludwig, 1994). In Australia, temperate grassy woodlands once covered a significant area of the southeast of the continent. They have been cleared, fragmented, and degraded significantly over the last 150 years of pastoral and urban activity and now occupy approximately 5% of their pre-European range (McIntyre et al., 2015; Yates & Hobbs, 1997). As a result, there are very few areas of high-quality grassy woodland remaining, and conservation of biodiversity in these areas is a priority for land managers (Lindenmayer et al., 2012; McIntyre et al., 2015; Yates & Hobbs, 1997). Previous studies have identified important effects of sampling methodology (Ward et al., 2001), grazing (Barton et al., 2016), trophic structure (Gibb & Cunningham, 2011), farmland revegetation (Ng et al., 2021), and climate (Yates et al., 2011) on ant communities in the grassy woodlands of southeastern Australia. Quantifying the role of vegetation cover as a driver of ant community structure can improve the understanding of invertebrate biodiversity and its management in these threatened ecosystems.
In this study, we investigated the effects of habitat structure on ants in a critically endangered grassy woodland located in southeastern Australia. We took the approach of classifying habitat structure at two spatial scales: (1) microhabitat defined by sites in open ground (no tree or shrub cover), sites adjacent to logs (in open ground), or sites under trees, each separated by a few meters (<10 m), and (2) a macrohabitat scale characterized as four vegetation classes defined by combinations of low or high shrub and tree cover and separated by >100 m. Our aim was to examine ant assemblages sampled from within and among these micro- and macrohabitats to determine where species richness was highest or lowest, how ant community composition differed, and which species were associated with which habitat types. We made the following predictions:
- Ant species richness would be greatest in habitats with high cover and lowest in habitats with low cover. That is, species richness would increase as habitats increased in cover provided by trees and shrubs, driven by the concomitant moderated microclimate and plant resource attributes. This prediction stems, in part, from the well-documented habitat complexity–species diversity theory (Pérez-Sánchez et al., 2023; Tews et al., 2004), whereby greater vegetation cover adds vertical complexity and additional niche space that supports higher numbers of coexisting species. Although some thermophilic species may avoid habitats with higher cover, we expected more cryptic and litter-dwelling species as well as nectar-feeding species to occur under trees.
- Ant community composition would differ most clearly among microhabitat types but may be less clear among macrohabitat types. This prediction is based on other studies showing the importance of small-scale habitat for beetles (Ross et al., 2017) and spiders (Barton et al., 2017), as well as the perception of “landscapes” by small-bodied organisms (Barton et al., 2013; Manning et al., 2004; Wiens & Milne, 1989).
- Ant species within key genera will show preferences for either open habitat (e.g., arid-adapted Melophorus or seed-feeding Pheidole species) or treed habitat (e.g., honeydew feeding Camponotus or Iridomyrmex species) (Shattuck, 2000) at either micro or macro scales. Broad predictions of the ecology of ants at the genus level are possible (e.g., Andersen, 1995a, 1997), but it remains unclear whether species within key genera show similar or different habitat preferences in grassy woodlands.
We discuss our findings in light of potential land management practices that might promote ant biodiversity and conservation in landscapes characterized by heterogeneous vegetation cover.
METHODS Study area and designWe conducted our study within the Mulligans Flat and Goorooyarroo Nature Reserves along the northern edge of the Australian Capital Territory in the southeast of the Australian continent (centered at approximately 35.177538° S, 149.176265° E). These reserves protect an area of approximately 1400 ha of Yellow Box-Red Gum (Eucalyptus melliodora, E. blakelyi) grassy woodland, a critically endangered ecological community in Australia (Australian Government, 2006). The study area has an elevation of about 650–700 m above sea level and mean annual rainfall of 650 mm (Jenkins, 2000). Broad vegetation types recognized in the region include (1) open forest on the shallower soils of the ridges and slopes dominated by Eucalyptus macrorhyncha, E. rossii, and E. mannifera; (2) woodlands on the deeper soils of the lower slopes and flats (Eucalyptus blakelyi and E. melliodora); and (3) patches of grassland dominated by Panicum effusum, Themeda australis, and Rytidosperma (previously Austrodanthonia) spp. (Lepschi, 1993; McIntyre et al., 2010).
We examined the ant community and habitat openness at two spatial scales, which we called “microhabitat” and “macrohabitat.” Microhabitat scale: We conducted our sampling at the microhabitat scale by placing pairs of pitfall traps, separated by 1 m, either under a tree, adjacent to a log, or in open ground (Figure 1a,b). Samples from open ground and under trees provided two extremes of the small-scale habitat openness spectrum (openness defined by the degree of tree and shrub cover), but we were also interested in samples adjacent to logs. This was because these small-scale habitat structures have previously been shown to be important for arthropods and microbial communities (Barton et al., 2009, 2017; Hamonts et al., 2017). We sampled each of these three microhabitat types within plots of 25-m radius at each end of a 1-ha site (50 × 200 m) and replicated this across 96 sites (Figure 1a). Within each site, we separated microhabitat samples on average by a distance of 71 m (min = 6 m, max = 142 m). We placed pitfall traps under Yellow Box (Eucalyptus melliodora) or Blakely's Red Gum (Eucalyptus blakelyi) trees with a dbh of greater than 0.25 m. We placed traps in open ground beyond the perimeter of any tree canopy. We selected logs only if they were greater than 0.10 m in diameter and 1 m in length. In some cases, we could not locate a suitable tree or log within 25 m of the center of a plot, so the microhabitat was not sampled. Macrohabitat scale: The macrohabitat scale was defined by groups of four adjacent sites that shared one of four different vegetation structures: (1) high tree and high shrub cover (low habitat openness), (2) high tree and low shrub cover (medium habitat openness), (3) low tree and high shrub cover (medium habitat openness), and (4) low tree and low shrub cover (high habitat openness). The mean distance between sites from different macrohabitat types was 1266 m (min = 85 m, max = 2998 m).
FIGURE 1. (a) Ninety-six 1-ha sites (50 × 200 m) were surveyed for ants, with pitfall traps placed in open ground, adjacent to a log and under a tree in a 25-m radius plot located at each end of every 1-ha site. All microhabitats were >10 m apart. (b) Image of a 1-ha site showing the heterogeneous distribution of tree cover and grassland. Photo credit: Philip Barton.
We opened pitfall traps for three weeks during Autumn (March–April) of 2007. Our traps consisted of 200-mL plastic jars dug in flush with the soil surface, each with 100 mL of polypropylene glycol as a preservative. We installed the traps a week before opening them to reduce “digging-in” bias in collections of ground active arthropods (Digweed et al., 1995; Greenslade, 1973). Some traps were disturbed by inquisitive mammals or birds and so were excluded from the analysis. If a single trap was missing, then the pair was excluded, removing potential sampling bias linked to a single trap. After collection, we removed ants from the pitfall samples, counted and identified individuals using keys to Australian ant fauna (Shattuck, 2000) and with reference to the online database
We pooled each pair of traps to give one sample per microhabitat structure, resulting in a dataset of 533 samples of ants. We first examined the taxonomic profile of the whole ant community and constructed rank-richness and rank-abundance curves to identify species-rich and abundant genera, which is useful for comparison with other bioregions in Australia (e.g., Andersen, 1995b, 2016; Andersen et al., 2018) and globally (Andersen, 1997; Gotelli et al., 2011). We then conducted statistical analysis to address our three predictions relating to species richness, community composition, and abundant ant species.
Prediction 1: Ant species richnessWe wanted to know whether species richness of ants differed among microhabitat or macrohabitat types. We first used species accumulation curves using the “vegan” package (Oksanen et al., 2022) in R (R Core Team, 2022) to assess sampling completeness and among-sample heterogeneity, and by comparing the rate of accumulation and whether they approached an asymptote, respectively (Colwell et al., 2004; Magurran & McGill, 2011). For accumulation curves comparing microhabitat types, we pooled data across macrohabitat vegetation classes. For curves comparing macrohabitat vegetation classes, we pooled data across microhabitat types. We also tested for differences in mean species richness among micro- and macrohabitats by fitting a set of Bayesian generalized linear mixed models, using the “brms” package (Bürkner, 2021). We assumed a Poisson error distribution and used default priors and ran four Markov chains for 2000 iterations with a warm-up/burn-in of 1000. We specified site and polygon as random effects to account for potential spatial correlation. We fitted five models, with and without the microhabitat and macrohabitat predictors and their interaction. We compared models using the leave-one-out information criterion (LOOIC), considering the simplest model <2 ∆LOOIC as the best-fit model. We used the Gelman–Rubin (Gelman & Rubin, 1992) statistic, and examined trace plots to assess whether or not the chains showed adequate mixing. We assessed model fit using posterior predictive checks via the pp check in the bayesplot package (Gabry et al., 2019; Gabry & Mahr, 2022).
Prediction 2: Ant community compositionWe wanted to know whether ant communities were nonrandomly structured among our sites, and so conducted permutation-based multivariate analysis of variance (PERMANOVA, Anderson, 2001) on Bray–Curtis dissimilarities of presence/absence transformed data of all species in the community. We used site ID as a grouping variable and tested for the effects of microhabitat, macrohabitat, and their interaction. We also wanted to inspect ant community composition visually to determine whether there was any clear separation of habitat groups, and so conducted a principal coordinates analysis (PCoA) on the transformed data of species within sites. We plotted the PCoA axes using shapes and colors to denote microhabitat sites within the macrohabitat types, and fitted convex hulls and centroids to aid in group differentiation.
Prediction 3: Common ant speciesAny differences in communities among habitats will be due to the occurrences of different species, and so we examined the effect of microhabitat and macrohabitat on the occurrence of common ant species in our dataset. We defined “common” species as those that were caught in 5% or more of our samples. We fitted a Bayesian ordination and regression analysis of multivariate data model using the “boral” package (Hui & Blanchard, 2021) in R (Hui, 2016; R Core Team, 2022). We fitted a correlated response model, which combines separate species models with latent variable models to account for any residual correlation stemming from potential biotic interactions or unknown covariates. We fitted the model using binomial data (i.e., presence or absence) assuming a binomial error distribution. We used binomial instead of abundance data because of the highly uneven distribution of abundances among ant species, and because they are social insects where an individual of one species is likely to be accompanied by several individuals of that species (Gotelli et al., 2011). We used default priors, with 30,000 Markov Monte Carlo Chain (MCMC) samples using 3000 as burn-in and thinning every 15th sample. We specified site and polygon as random effects to account for potential spatial correlation. We inspected trace plots of the MCMC samples to check for convergence. We used the “ggplot2” package (Wickham, 2016) in R for plotting.
RESULTS Overview of the ant faunaWe identified 117 species from 41 genera from a total of 155,004 individuals collected in our traps (see full species list in Appendix S1: Table S1). The most species-rich genera were Iridomyrmex and Camponotus, both represented by 11 species, followed by Stigmacros (9 species), Melophorus (8 species), and Monomorium (7 species) (Figure 2a). Iridomyrmex was the most abundant genus accounting for 66% of all individuals collected (Figure 2b), and Iridomyrmex rufoniger was the single most abundant species (49% of all individuals) (Appendix S1: Table S1). We also recorded range extensions for Melophorus gibbosus and M. hexidens, and an undescribed species of Syllophopsis.
FIGURE 2. (a) Relative richness and (b) abundance of the 41 genera of ants collected in this study.
Our best-fit model for ant species richness included a significant effect of microhabitat only, and the second best model included both a significant effect of microhabitat and nonsignificant effect of macrohabitat (Appendix S1: Tables S2 and S3). Mean species richness was highest under trees and lowest next to logs (Figure 3). Our species accumulation curves also showed that observed species richness of ants was highest under trees, and fewer species occurred next to logs or in open ground (Figure 4a). At the macrohabitat scale, we found ant assemblages had similar numbers of species relative to sampling effort for each vegetation type (Figure 4b).
FIGURE 3. Predicted mean (±SE) species richness of ants sampled from microhabitat in open ground, under trees, and adjacent to logs.
FIGURE 4. Species accumulation curves showing species richness of ants relative to the number of samples from (a) three microhabitat types and (b) four macrohabitat types. Curves represent estimated mean species richness (±1 SD) calculated using the Mau Tau estimator (Colwell et al., 2012). Microhabitat was defined by open ground, next to logs, and under trees. Macrohabitat was defined by vegetation cover: HTHS, high tree and high shrub cover; HTLS, high tree and low shrub cover; LTHS, low tree and high shrub cover; LTLS, low tree and low shrub cover.
Our analysis of the ant community using PERMANOVA indicated there was a significant effect of microhabitat on ant in assemblages (F = 14.18, p < 0.001) and a significant but smaller effect of macrohabitat (F = 7.17, p < 0.001), and there was no interaction between these scales (F = 1.08, p = 0.07). Visual depiction of these patterns with principal coordinates ordination plots showed some separation of samples grouped by microhabitat along axes 3 and 4, but samples grouped by macrohabitat were not distinct (Appendix S1: Figure S1).
Ant species (prediction 3)Investigation of species-level differences provided a clearer picture. We found several effects of microhabitat on a range of ant species, but only few effects of macrohabitat (Figure 5). At the microhabitat scale, some genera displayed strong preferences for particular microhabitats, with all species of Camponotus having higher occurrences under trees compared with open sites, and several species of Melophorus (but not all) having lower occurrences under trees compared with open sites. We found a spectrum of negative and positive effects of tree microhabitat on species in the genera Iridomyrmex, Melophorus, Monomorium, and Pheidole, indicating different preferences among species within the same genus. At the macrohabitat scale, there were some species-rich genera (e.g., Camponotus, Iridomyrmex, Stigmacros) and many other individual species that showed little or no effects at this scale (Figure 5). For example, Crematogaster pallipes showed a positive response to high shrub cover, whereas Rhytidoponera metallica showed a negative response to high shrub cover, yet neither of these species showed responses to microhabitat (Figure 5).
FIGURE 5. Mean effect sizes of microhabitat types and macrohabitat vegetation classes on common ant species. Effects of tree and log microhabitats are relative to samples from open ground, and effects of macrohabitat vegetation classes are relative to samples from sites with low tree and low shrub cover. Species are grouped by genus to highlight where species within the same genus show contrasting responses to habitat “openness.” Large dots depict significant effects (95% credible intervals not overlapping zero).
Multiscale studies of habitat–diversity relationships are useful for identifying how variation in habitat drives patterns of diversity and species distributions (Barton et al., 2009; Campos et al., 2011; Costanza et al., 2011). In this study, we found that vegetation cover had its largest effects on ant species richness and individual ant species at the “microhabitat” scale, where variation in canopy cover and open ground occurred over distances of only a few meters. Our results also supported, in part, our predictions that (1) richness would increase with vegetation cover, (2) composition would vary most strongly at microhabitat scales, and (3) individual ant species would display clear preferences for either open or treed habitat. A key finding, however, was that some species within the same genus displayed contrasting habitat preferences, indicating genus-level functional classifications based on habitat may not apply at small scales. Further, we found that the ant fauna was both species rich and genus rich, was numerically dominated by a few species of Iridomyrmex, and contained a mix of species with xeric- and mesic-centered biogeographic distributions. Our study suggests that land management practices that maintain spatial heterogeneity in habitat openness at microhabitat scales will likely promote species coexistence within and among genera, and benefit ant diversity at larger scales. Below we discuss the key effects of habitat on ant species as well as new insights into the ant fauna found in temperate grassy woodlands.
Habitat effects were strongest at the microhabitat scaleWe found that variation in microhabitat defined by sites in open ground, adjacent to logs or under trees, was important in shaping the local richness of ant assemblages and the occurrence of a range of different ant species. Vegetation and habitat cover therefore appears to shape the distribution of ant species in Australia's temperate grassy woodlands over scales of only a few meters. This finding is congruent with other multiscale studies of ant communities that have identified important small-scale drivers of ant community structure (Maravalhas & Vasconcelos, 2020; Pérez-Sánchez et al., 2023). Similar findings have been reported for arthropods and soil microbes in grassy woodlands (Barton et al., 2010, 2017; Hamonts et al., 2017).
A key finding was that individual ant species displayed a spectrum of responses to microhabitat, which varied within and among different genera. Some genera displayed strong preferences for one particular microhabitat. For example, all species of Camponotus and most Stigmacros had higher occurrences under trees, whereas most Melophorus preferred open ground. However, some species within the same genus displayed contrasting responses to microhabitat. For example, species of Iridomyrmex and Monomorium showed a range of preferences for open or tree microhabitats. These genera are both very widespread, each with a large number of species, and are able to thrive in almost all areas of Australia (Heterick, 2001; Heterick & Shattuck, 2011; Sparks et al., 2014). Genera are sometimes viewed as having broad preferences at a whole-of-clade level, and this can have practical use in understanding community organization (e.g., Andersen, 1995a). Yet, our results demonstrate the plasticity and fine-scale preferences of species within the same genus and within the same ecosystem. This finding is significant insofar as it highlights the potential limitations of broader functional classifications of ants based on habitat associations and the need to build further knowledge of the ecology of individual species.
A majority of common ant species displayed a preference for trees. Trees are key structural features of grassy woodlands and savannas worldwide (Manning et al., 2006), and their canopies create more closed environments and produce leaf litter that is important for foraging and nesting for many ant species (Holldobler & Wilson, 1990; Lach et al., 2010). Leaf litter under the eucalypt trees in grassy woodlands provides a protective matrix for nesting, aids the retention of moisture, and supports abundant food in the form of collembola and other invertebrates. Litter is a critical resource for several cryptic species, such as Heteroponera, although these species were probably underrepresented in our pitfalls (Bestelmeyer et al., 2000). Shade and more moderate temperatures also occur under trees, which are unfavorable to ants preferring open areas, such as thermophilic species of Melophorus, Meranoplus and some species of Iridomyrmex. Cracks and hollows also provide nesting opportunities for some genera including Ochetellus, Colobopsis, some Crematogaster, and Anonychomyrma (Shattuck, 2000). Tree-feeding insects like caterpillars (Lepidoptera), scale, and lerps (Hemiptera) are also important producers of honeydew for species of Camponotus, Iridomyrmex, and Anonychomyrma, which have been observed feeding on sugars in eucalypts in grassy woodlands near our study area (Gibb & Cunningham, 2009). In contrast, the forbs and grasses found in open habitat provide resources for seed eaters like Pheidole, Chelaner, Melophorus, some Tetramorium, and some Monomorium. Further research aimed at discerning the use of different woodland strata by ant species would assist with identifying which genera and species are using resources distributed across ground level or in the canopy (e.g., Wilkie et al., 2010).
Logs generally had similar effects on the structure of the ant community as trees, but not in all cases. Like trees, logs are likely to offer a complex range of environmental conditions that make them suitable habitat for a range of ant species. For example, logs collect litter and provide cooler, moister conditions (Goldin & Hutchinson, 2014), and this may provide a range of suitable nesting niches. Ants including Stigmacros cf. clivispina and Iridomyrmex brunneus occurred at both trees and logs. Our results, however, show that logs do not provide a substitute for tree-loving genera like Camponotus, Colobopsis, and Myrmecia, or for species like Pheidole sp. 1 or Monomorium sydneyense (Shattuck, 2000). The reason for this is unclear but may be due to simple features missing from logs such as broad shading or suitable foraging opportunities—logs do not provide ready access to honeydew produced by sap-sucking insects. The role of logs in structuring ant communities in grassy woodlands therefore remains ambiguous, and they appear to be less important in driving variation than the level of openness. Nevertheless, logs provide at least some of the benefits associated with trees, and, in their absence, could provide useful habitat for some species.
We found only a few effects of macrohabitat on individual ant species, with many species and some whole genera (e.g., Camponotus, Iridomyrmex, Stigmacros) showing little or no response. This likely explains the lack of differences in species richness and smaller differences in composition we found in our analyses. Some macrohabitat effects on ant species also appeared to be absent at the microhabitat (e.g., Crematogaster pallipes, Rhytidoponera metallica). These mixed responses are difficult to interpret, and we conclude that mechanisms of habitat structure effects on ants appear most strongly at finer scales linked to niche requirements and foraging ecology (e.g., Radnan et al., 2018). Our scale of macrohabitat may also not reflect broader ecosystem variation such as strong edaphic or floristic gradients, or biogeographic factors such as climate or elevation, which may be more important in affecting ant communities (Dunn et al., 2009; Spiesman & Cumming, 2008; Yates et al., 2011).
New insights into grassy woodland ant faunaOur study was one of the most spatially intensive surveys of ants from the southeast of Australia and adds substantially to the taxonomic data gained from ant communities in grasslands and grassy woodlands of this region (Barton et al., 2016; Greenslade, 1994; Melbourne, 1999; Ng et al., 2021). Key features of the ant fauna in our grassy eucalypt woodland study area suggest some aspects of overlap with community organization patterns seen in both arid and mesic ant fauna. For example, we identified a large number of species (117), which were associated with a high number of genera (41), a pattern consistent with those known for mesic ant fauna (Andersen, 1995b). The dominance of Iridomyrmex found in this study also matches the descriptions of ant assemblages from other grassy woodlands (e.g., Barton et al., 2016; Ng et al., 2021). The dominance of this genus also suggests that the ant fauna of box-gum grassy woodlands in southeast Australia shares some features with arid zone ant fauna where Iridomyrmex also dominates (Andersen, 1986). The presence of several species of Melophorus, including new distribution records for mostly xeric species Melophorus hexidens and M. gibbosus, also indicates that grassy woodlands in our study region provide suitable conditions for a number of both mesic and xeric species. Melophorus gibbosus appears to be uncommon and previously known from only 11 collections, mainly in arid regions of Australia (Heterick et al., 2017). Similarly, the range of M. hexidens was previously known only from three collections in arid northwest NSW but is now extended into the cooler more humid region of our study area (Heterick et al., 2017).
Habitat openness as a driver of ant community structureOur study shows that habitat openness is important for generating differences in species richness and occurrence of ant species, and supports the hypothesis that habitat openness is a key driver of ant community structure (Andersen, 2019). There are many studies reporting on the effects of vegetation cover on ants (e.g., Neves et al., 2023; Oliver et al., 2016; Retana & Cerda, 2000; Schmidt et al., 2013), and as a result, there are many different interpretations of the concept of openness. Two recent definitions of openness are “level of vegetation cover” (Andersen, 2019) and, in response to Andersen (2019), “the amount of sun exposure on the ground” (Lessard, 2019). These definitions imply that shading is a key mechanism by which openness acts on ants. However, a suite of vegetation attributes could be implicated in openness, or in generating a gradient of vegetation density. For example, Spiesman and Cumming (2008) use “open structured habitats with a sparse understory,” whereas Berman et al. (2013) explicitly use tree canopy cover to define openness. It is clear that context, scale, vegetation type, and resulting microclimate all influence whether a site might be perceived as “open” by ant species and communities, making definition of openness difficult. Broadly, openness represents a complex syndrome of habitat features that mediates the penetration of sunlight and air, and litter and soil attributes, thereby acting on a number of ecological traits of ants (Gibb et al., 2023). We suggest that the features of openness be clearly defined by authors and (where relevant) linked to ecological or management questions being addressed to assist with interpretation of findings. Further development of the habitat openness concept should aim for a standardized approach to allow meaningful comparisons among biomes and land uses.
Implications and conclusionsSmall-scale habitat structure and their effects on animal assemblages are important from a biodiversity conservation perspective. Manipulation of key structural features of habitat, such as trees (Manning et al., 2006) or woody debris (Barton et al., 2011; Harmon et al., 1986), can provide a practical way to manage habitat and benefit biodiversity. For example, it is well established that individual trees provide localized hotspots of ecological function in landscapes by retaining soil moisture and nutrient content (McElhinny et al., 2010), and this can result in increased invertebrate diversity (Oliver et al., 2006) as we found in this study. Trees also produce clear gradients in the diversity and composition of invertebrates away from their canopy and into surrounding grassland (Oliver et al., 2006). These patterns are largely driven by inputs of leaf litter and associated soil nutrients under the tree crowns (Barton et al., 2010; Eldridge & Wong, 2005; McElhinny et al., 2010). Yet the converse—open spaces between trees—is also important for many ant species. Just as land management practices that remove large paddock trees may be detrimental to tree-dependent species of ants, plantings or natural regeneration that reduce open areas of grassland may similarly be detrimental to open habitat specialists and thermophilic species (e.g., Melophorus sp.). We suggest that land management in grassy woodlands and other similarly structured ecological communities like savannas and Cerrado should consider the maintenance of spatial heterogeneity in canopy gaps and open grassland at scales of tens of meters to promote species coexistence and benefit ant diversity at larger scales. Key ecological disturbances, such as fire and grazing, could play an important role in contributing to small-scale heterogeneity as they can shift the distribution of tree cover in woodlands and savanna ecosystems (Harrison et al., 2003; Price et al., 2019), and therefore alter the availability of treed or open habitat to ants. In our study system, eastern gray kangaroos (Macropus giganteus) are the dominant large herbivore and can reach high population densities with adverse effects on ground-dwelling fauna (Barton et al., 2011; Howland et al., 2014). The management of kangaroo populations at appropriate densities could be a useful tool to maintain a mix of open grassland and tree cover. Further research on the spatial arrangements of trees and ratio of tree cover to open ground would also assist with identifying the levels of cover that maximize local richness while also providing sufficient habitat for compositionally distinct assemblages. Our study also indicates that the concept of habitat openness is an important lens through which to examine ant community structure. Further refining of this concept using standardized measures and designing studies that consider smaller-scale perspectives could provide valuable new insights into ant communities and conservation and land management practices in heterogeneous landscapes.
AUTHOR CONTRIBUTIONSPhilip S. Barton designed the study. Philip S. Barton and Jon Lewis contributed to data collection, and Jon Lewis led the taxonomic work and data collation. Analyses were conducted by Philip S. Barton and Maldwyn J. Evans. Philip S. Barton wrote the manuscript with input from Jon Lewis and Maldwyn J. Evans. All authors edited and approved the manuscript for submission.
ACKNOWLEDGMENTSWe thank Adrian Manning, David Lindenmayer, David Shorthouse, Jeff Wood, and Ross Cunningham for their inputs to the design of the study. Heloise Gibb, Saul Cunningham, Adrian Manning, and Steve Holiday assisted with the pitfall trapping. We have benefited from thoughtful discussions with Sue McIntyre about grassy woodland structure and floristics. We are grateful to Brian Heterick and Steve Shattuck for valuable taxonomic advice. Open access publishing facilitated by Deakin University, as part of the Wiley - Deakin University agreement via the Council of Australian University Librarians.
CONFLICT OF INTEREST STATEMENTThe authors declare no conflicts of interest.
DATA AVAILABILITY STATEMENTThe data underpinning this study (Barton, 2023) are available from Figshare:
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Habitat structure is a key determinant of local animal diversity, with attributes of vegetation such as cover or complexity generating key resources for different species. However, habitat–diversity relationships can vary across spatial scales, and among different taxa and ecosystem types. Here we report on a study of habitat structure and its effects on ant communities at two spatial scales in a temperate grassy woodland characterized by heterogenous tree and grassland cover. We examined species richness and the occurrence of ground-dwelling ant species at (1) microhabitat scales defined by a triplet of sites comprising open ground, adjacent to a log, and under a tree, each separated by a few meters, and (2) at macrohabitat scales defined by sites grouped into broader vegetation types defined by low or high levels of shrub and tree cover and separated by 100s of meters. We identified 117 species of ant from 41 genera, from a total of 155,004 individuals collected. Ant community composition differed significantly among microhabitats and macrohabitats, but mean species richness only differed at the microhabitat scale where it was the highest under trees and lower adjacent to logs and in open ground. Notably, ant species within the genera
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details



1 School of Life and Environmental Sciences, Deakin University, Geelong, Victoria, Australia; Future Regions Research Centre, Federation University, Mount Helen, Victoria, Australia
2 Fenner School of Environment and Society, Australian National University, Canberra, ACT, Australia
3 Fenner School of Environment and Society, Australian National University, Canberra, ACT, Australia; Australian National Insect Collection, CSIRO, Canberra, ACT, Australia