Mountains and their ecotones increase landscape heterogeneity and maintain a unique assemblage of grasshoppers in the southern Kalahari

South Africa is a megadiverse country. Here, natural communities are unevenly distributed across, and within, seven distinct biomes. In such heterogeneous landscapes, understanding spatial patterns of biodiversity is essential for planning and implementing efficient conservation measures. The southern Kalahari, forming part of South Africa's savanna biome, is an arid region of peculiarly high diversity and endemism. The responses of orthopteran assemblages to changing environmental conditions across the Kalahari were investigated by comparing alpha and beta diversity levels across discrete vegetation types in the Tswalu Kalahari Reserve. The degree of association between species and specific vegetation types were also studied and how a key life history trait ‐ dispersal ability – influences community composition was determined. This study identified 46 grasshopper species within the reserve, which compares well with richness levels in other more productive habitats of the country. Local (alpha) diversity was higher in mountain and mountain‐ecotone sites versus vegetation types on the plains, and species turnover was also exceptionally high – approaching 100% ‐ across these two groups. The few (3) dispersal limited species recovered were associated only with the mountain‐ecotone group, with emergent dominance patterns suggesting that competitive rather than dispersal abilities determine the species composition of unique assemblages in the landscape. Topology plays a key role in maintaining spatial diversity across the southern Kalahari landscape. Mountains, and their ecotones, promote not only species turnover, but also richness and functional diversity. These can be viewed as islands of diversity, and should be targeted priority areas for conservation beyond the boundaries of protected areas.


INTRODUCTION
South Africa is a megadiverse country. Three global biodiversity hotspots occur within the countrythe Cape Floristic Region, Succulent Karoo and Maputaland-Pongola-Albany (Myers et al., 2000Mittermeier et al., 2004. It also comprises seven distinct biomes, all with high levels of species diversity and endemism (Mucina & Rutherford, 2006).
In these conditions, understanding the spatial patterns of diversity is essential for the implementation of efficient conservation measures (Harvey et al., 2020;Meyer et al., 2015), and avoiding underperformance of interventions and the waste of valuable resources . In addition, much of the country is a semi-arid to arid landscape, and is therefore sensitive to changing climatic conditions, such that knowledge of the factors driving diversity patterns is necessary for adaptive conservation planning (Hannah et al., 2002).
Biodiversity can be partitioned into three levelsalpha, beta and gammaallowing for the comparison of diversity at site level, as well as community turnover between sites, and across regions, respectively (Colwell & Coddington, 1994;Crist et al., 2003). Alpha diversity indices range from species richness counts, to estimates of evenness such as the Simpson and Shannon indices which take into account how rare or abundant individual species are at a site. Beta diversity refers to the turnover of species between communities occurring at different sites, which can be quantified using measures of dissimilarity such as the Bray-Curtis and Jaccard's indices (Chao et al., 2005;Magurran, 2004). Beta diversity on its own is a diverse concept, and depending on the calculation method, interpretations of what beta diversity values are quantifying vary. Compositional changes can be measured in regard to similarity across the landscape, as a ratio of gamma to alpha diversity of a site, as well as the distance related decreases in similarity (Jurasinski et al., 2009;Tuomisto, 2010). In many cases beta diversity is used as a differential diversity measure (Jurasinski et al., 2009). Here, beta diversity is regarded as the proportionate turnover of species compositionand their respective abundancesfrom one locality to the next. Importantly, alpha and beta diversity may often vary independently. High levels of alpha diversity are usually associated with productive systems, but high levels of beta diversity only occur in heterogeneous habitats (although anthropogenic disturbance and invasions could contribute to heterogeneity, in which case high beta diversity may be undesirable; Socolar et al., 2016). Amongst orthopterans, for example, it has been found that species' distribution ranges expand due to climate and land-use changes, leading to increases in alpha diversity at local scales, but an overall homogenisation of assemblages and thus a decrease in beta diversity between sites (Ogan et al., 2022).
Grasshoppers are ubiquitous in most habitats (Cigliano et al., 2022). In South Africa, grasshopper diversity is known to respond to environmental parameters like rockiness (Crous et al., 2014), perturbations such as fire and grazing regimes, and grassland mowing (Joubert et al., 2016). In some sense, these responses may be considered generic indicators. For one, grasshopper species richness and guild diversity in grasslands is known to be congruent on that of butterfly communities (Bazelet & Samways, 2012). Some species have also been shown to be indicative of habitat quality (Bazelet & Samways, 2011), hence grasshopper communities have been used to identify areas of conservation priority within the Cape Floristic Region (Matenaar et al., 2015). Furthermore, grasshopper assemblages show responses to local disturbances that occurred up to 2 years prior (Theron et al., 2022a). The abundance of these and similar studies makes it possible to contextualise and compare relative diversity and the turnover of species between studies and sites.
The Kalahari is a semi-desert component of the southern African savanna biome, extending from Botswana and Namibia into the northern reaches of the Northern Cape and Northwest provinces of South Africa. Floral and faunal biodiversity in the Kalahari is exceptionally high, especially considering the area receives less than 400 mm annual precipitation on average, and lies on sandy soils with a high rate of water and nutrient leaching (Abraham et al., 2021;Tokura et al., 2018). For example, in the Tswalu Kalahari Reserve, a 111,000 ha protected area in the Northern Cape, some 70 species of beetle, and 136 species of spider, have been identified (Davids et al., 2010;Dippenaar-Scoeman et al., 2018), with additional work on hymenopteran species distributions and densities having been conducted (Moritz et al., 2008;van Noort et al., 2021). Little is known about orthopteran diversity in the region, aside from initial surveys conducted in the former Kalahari Gemsbok National Park, now the Kgalagadi Transfrontier Park (KTP), that indicated sparse populations of some 35 grasshopper species (Barker, 1983(Barker, , 1984). An attempt was also made to determine how common each species was (Barker, 1983), yet no method is provided as to how the estimates were obtained.
Regardless, high levels of biodiversity and endemism are expected in the Kalahari, given its biogeographical uniqueness within sub-Saharan Africa. Linder et al. (2012) found that many species likely have their evolutionary origins here. However, it is not known how habitat heterogeneity across the landscape shapes biodiversity patterns.
The overall aim of this study was to investigate how grasshopper diversity responds to environmental variation across the Tswalu Kalahari Reserve landscape. Comparison was made on the local patterns of species richness and evenness (alpha diversity), as well as community turnover (beta diversity) between distinct naturally occurring vegetation types and their ecotones. As grasshopper diversity and species composition have been shown to vary in accordance with rockiness (Crous et al., 2014), bare ground (Gebeyehu & Samways, 2003;Matenaar et al., 2015) and grass cover (Gebeyehu & Samways, 2003;Gebeyehu & Samways, 2006), this study hypothesise that levels of species richness and evenness will vary across the different vegetation types, and that communities in each would comprise unique assemblages of species. In particular, it was expected that both richness and uniqueness would be most evident at ecotonal sites. Ecotones are zones of transition in the landscape, where conditions are a mixture of two adjacent vegetation types (Turner & Gardner, 2015). Consequently, this study predict that grasshopper communities here will comprise subsets of species from both, resulting in high levels of alpha diversity and a unique species composition. This study predict that dominance patterns of species will also vary across the vegetation types as different species are able to utilise different micro habitats present in the different vegetation types (Rominger et al., 2009). Finally, the levels of association between each grasshopper species with the various vegetation types (and ecotones) sampled were determined. As grasshoppers can be indicative of habitat quality (Bazelet & Samways, 2011;Matenaar et al., 2015), species was expected to exhibit preferences towards specific vegetation types, or groupings of vegetation types, in the landscape as vegetation structure and composition varies between vegetation types. Measuring specieshabitat associations allows for identifying indicator species. Moreover, attempt to link these associations with the spatial distribution of species' functional traits, in particular species' dispersal ability was performed. In changing environments, dispersal limited species are at a higher extinction risk than more mobile species (Koot et al., 2022;Ogan et al., 2022), thus identifying regions in the landscape where such species occur will aid reserve management in identifying key conservation areas in the landscape.

Grasshopper sampling
Sampling was conducted over two separate seasons, first during April-May 2021 to coincide with the late Austral summer period (late summer), and then again in October-November 2021 during the early Austral summer period (early summer). Grasshoppers in the region can go through two generations in a summer period, and so this sampling protocol allowed us to sample the mature adult individuals in both sampling seasons. All sites were sampled between 9h00 and 16h30 on sunny, windless days. Sites were sampled twice, once in each season (late and early summer) and the same sampling methods were applied to both seasons. To randomise the stratified sampling protocol, sites were sampled in groups, with two replicates per vegetation type per group, equalling a total of 12 sites in each group ( Figure 1).
There were thus four sampling groups. All sites in a group were sampled within 3 days, and once complete, sites in the successive group were then sampled. Thus, 100 sweeps were conducted per transect, and 400 sweeps were conducted overall, at each site each season. After each transect was sampled, the contents of the nets were placed directly into sealable plastic bags, with samples from the four transects pooled and treated as one. After 24 h in the freezer, specimens were removed from the vegetation matter, and placed into plastic sample bottles, and kept frozen until identification.
Following sweep-net sampling, 15 min of active searching was conducted in the area covered by the transects (30 min in total per site per season). The collectors would randomly walk around and catch and collect any grasshoppers sighted. This was done to augment the sweep-net samples which are biased to less-mobile species. Individuals were removed immediately from the net upon being caught and placed into sealable plastic bags. Samples from the two collectors were pooled, frozen and placed into sample bottles and kept frozen until identification.
Adult specimens were identified to species level where possible.
Where species identification was not possible, the genus of the specimen was identified and assigned a species number. Only adult specimens were identified to avoid identification errors on nymphs.
Various taxonomic keys found in Dirsh (1965) were used to identify species, identifications were then verified with species records found in the Orthoptera Species File (Cigliano et al., 2022). Grasshopper species were then assigned to one of three dispersal classeswinged where both sexes have fully formed wings, dimorphicwhere only one sex is winged, or winglesswhere neither sex possess wings (Matenaar et al., 2015). The reference collection of grasshopper specimens is housed in the Department of Zoology and Entomology, University of the Free State, South Africa.

Environmental variables
Environmental variables were recorded at each site, each season. Four 25 m transects were laid out in an 'X' arrangement ( Figure 1 insert), with the transect beginning at the edge of a bush in the centre of the site (this layout was due to additional acoustic sampling that was conducted concurrently at each site; van der Mescht et al., 2022). At each meter point of the transect, a graduated pole (10 cm increments) was dropped and the dominant growth form or substrate type recorded, totalling 100 data points (Thompson et al., 2020). As grasshoppers respond more to vegetation structure than species composition (Bazelet & Samways, 2012;Hochkirch & Adorf, 2007;Joern, 2005), they were only distinguished between growth forms rather than plant species, that is, grass, forb, shrub or ground cover (creeper). When no plant cover was encountered, bare substrate was classified as bare soil or rock. These six categories were then represented as percentage cover per site. In addition, average vegetation height was calculated, as the mean of 30 randomly placed measures of plant height within the site. Elevation and the central co-ordinates of each site were recorded with a Garmin e-Trex 22x hand-held GPS. Collinearity was found to exist between the measured environmental variables and the different vegetation types, confounding any statistical model attempted. Thus, the environmental variables are presented as qualitative, rather than quantitative.

Statistical analysis
All analyses were conducted in R v 4.0.3 (R Core Team, 2020). A Mantle test was conducted on grasshopper species richness at each site to test for the presence of spatial autocorrelation in the data. The package ade4 (Dray & Dufour, 2007) was used to calculate the correlation between two distance-based matrices, one with the latitude and longitude of the central co-ordinates, and the other with the species F I G U R E 2 Changes in vegetation structural composition of each sampled vegetation types between the late and early summer sampling periods.
T A B L E 1 Results of the generalised linear mixed effects models describing the distribution of species richness and Shannon's entropy across all sites in the six different vegetation types    Figure 3b).

Community composition
Grasshopper species composition showed marked turnover between vegetation types, with two distinct groups emerging during analyses,  (Table 2).
Despite the clear separation between the mountain-ecotone and plain vegetation types, species turnover was also relatively high within these groupings: 0.74 between the two mountain-ecotone vegetation types, and across plains vegetation types dissimilarity ranged from 0.54 to 0.6 ( Table 2). Species (or sets of species) remained consistent within either of the mountain-ecotone and plains groups (see below).
At local scales, species were not evenly distributed within any of the six vegetation types sampled. Rank abundance curves show that few species were dominant within each vegetation type ( Figure 5).
Interestingly, and corroborating the turnover patterns described above, all four plains vegetation types had the same two dominant species (Acrotylus diana and Dnopherula sp. 9). Within the mountainecotone group, the Korannaberg-Langeberg Mountain Bushveld and Koranna-Olifants Ecotone sites shared the same first ranked species (Amesotropis sp. 1), although the second ranked species differed (Sphodroneaus gilli and A. diana, respectively).

Species-vegetation associations
Grasshoppers exhibit strong species associations with vegetation types and type groupings. Sixteen species were selected based on both the Indicator Value (IndVal) and phi coefficient of association, and virtually the same species were selected with both approaches (Table 3) Sphodroneaus gilli, the second ranked species in the Koranna-Olifants Ecotone, was also found to be indicative of the mountain-ecotone grouping (Table 3). Although ranked first in all the plains vegetation types, A. diana had IndVal scores only indicative of the Gordonia Dune Veld and Gordonia-Gordonia Ecotone sites. Two of the three wingless species -Thericles flavoangulatus and Kalaharicus elongatuswere indicative of the mountain-ecotone grouping, although a larger number of such species would need to be evaluated to substantiate this as a general pattern.

DISCUSSION
Grasshoppers were found to be considerably diverse within Tswalu Kalahari Reserve. Not only was a large number of 46 species recovered, but diversity varied across the landscape. However, contrary with expectations, diversity did not vary substantially between individual vegetation types, or even within ecotones. Instead, two distinct groups emerged here, the first in mountain and mountain-ecotone sites, and the other in the four remaining plains vegetation types and ecotones. In the mountain-ecotone group, local (alpha) diversity was comparatively high, and the species composition of these assemblages was distinct from those of the plains group. In addition, grasshopper species showed strong affinity to habitat types, with a strong dominance pattern mirroring the observed compositional groups: the plains vegetation types were dominated by the same species (Acrotylus diana), and the mountainecotone group another species (Amesotropis sp. 1).

Alpha diversity
The southern Kalahari is known as a region of high diversity and endemism (Davids et al., 2010;Dippenaar-Scoeman et al., 2018;Moritz et al., 2008;van Noort et al., 2021), but is relatively under-sampled in terms of grasshopper diversity. Previously, Barker (1983Barker ( , 1984 found 35 species of grasshoppers in the Kgalagadi Transfrontier Park, a much larger reserve to the north of TKR. This figure is comparable to the number of species that was found in TKR: the higher species richness in this study is likely a result of more intensive sampling across a greater number of sites than the sampling conducted in KTP. Also, the KTP survey was conducted after a prolonged drought, likely leading to the sparsity of populations sampled (Barker, 1983). Here, the first sampling period was conducted after a period of exceptional summer rains, promoting an increase in both the numbers of individuals and species present at TKR.
Grasshopper diversity in TKRand evidently in the southern Kalahari in generalis also comparable with that of other more productive habitats and biomes elsewhere in South Africa. For example, communities at sites in the grassland biome comprise 32 to 48 species (Bazelet & Samways, 2011Crous et al., 2014), although as many as 58 species were found at sites in grassland remnants within a timber plantation matrix (Theron et al., 2022a(Theron et al., , 2022b. In the semiarid Karoo biome, species richness varies between 23 and 28 at sites (Gebeyehu & Samways, 2002, and across eight reserves in the Cape Floristic Regiona biodiversity hotspot -, 86 species were recorded, yet species richness of the individual reserves ranged between 20 and 44 (Matenaar et al., 2015). Although this is not a comprehensive list of all grasshopper diversity studies in this country, these examples do allow for comparisons to be drawn regarding the diversity sampled in TKR. Furthermore, despite that sampling methods and duration of these, and other, studies differ, most reported that species accumulation curves had reached some form of asymptote and thus qualitative comparisons are possible. Thus, it is reasonable to conclude that grasshopper diversity in TKR is comparatively high and is in fact almost double that of other arid systems such as the Karoo (Gebeyehu & Samways, 2002.
Within TKR, alpha diversity varied, being significantly higher in both the Korannaberg-Langeberg Mountain Bushveld and Koranna-Olifants Ecotone than in plains vegetation types. These differences emerged in terms of both the species richness, and Shannon's Entropy values, which account for variations in species' relative abundances.
Grasshoppers are sensitive to changing vegetation conditions (Crous et al., 2014;Pronk et al., 2017;Theron et al., 2022a), which likely explains the difference in alpha diversity between the mountainecotone and plains groups: the habitats of the latter lacked the rocky

Community composition
The difference between mountain and plains habitats also influenced species composition, leading to a high degree of species turnover within TKR. Compositional turnover has been reported in a wide range of orthopteran studies in the country. In a large-scale study in the Cape Floristic Region where the furthest sites were separated by roughly 500 km, Matenaar et al. (2015) found that grasshopper species turnover between reserve pairs (Renkonen Index) was between 91.04% and 91.42%. In comparison, this study conducted over a much smaller spatial scale, with roughly 20 km separating the most eastern and western sites, revealed between 96% and 97% turnover in species composition from Korannaberg-Langeberg Mountain Bushveld to the plains vegetation types. On a similar spatial scale in the grassland biome, Bazelet and Samways (2011) found high beta diversity between sites and attributed this to grasshopper sensitivity to environmental differences first, and geographic distance second. This is likely the case here, considering that both vegetational and environmental conditionsboth sets of data being highly correlateddiffered so vastly between mountain and plains vegetation types.
Indeed, dissimilarity between the Koranna-Olifants Ecotone and the Mountain vegetation type was also relatively high (74%), suggesting a stepwise turnover in species composition from plains to mountain vegetation as a result of mixed ecological conditions and vegetation types at the base of the mountains.
All assemblages sampled, irrespective of alpha or beta diversity levels, showed steeply-concave rank-abundance curves, indicating a ubiquitous pattern of dominance by a small number of species in each.
Such dominance may be indicative of either priority effects or competitive dominance (Nadeau et al., 2021;Weidlich et al., 2020). Priority effects refer to the order in which species arrive at a site, and how these arrivals may positively or negatively impact the establishment of species arriving later by excluding or facilitating access to niche space (Helsen et al., 2016,). Competitive dominance, on the other hand, implies that superior competitors will limit niche opportunities for others, and so will always establish themselves as dominant regardless of dispersal patterns (Kunte, 2008). The structure of grasshopper communities in arid grasslands in New Mexico show evidence for both processes (Rominger et al., 2009). In this homogeneous environment, similarity of communities decreased with increasing distance between sites, indicating dispersal limitations. However, grasshopper community composition also varied across an elevational gradient, reflecting that species differed in their ability to take advantage of different niches.

Species-vegetation associations
In the present study, Acrotylus diana dominated assemblages in all vegetation types within the plains. Given the relative homogeneity of condi-

Conservation value of mountain assemblages
Species turnover from plains to mountain habitats reflects the underlying topology of TKR, with the plains vegetation types occurring across the permanent sand dunes in the west, while the mountain-ecotone grouping encompasses the Korannaberg mountain range that bisects the reserve from north to south. Similar to the results for grasshoppers, Davids et al. (2010) identified three distinct dung beetle assemblages in TKR: northern plains and dunes, southern plains, and dunes and hills. Although grasshoppers assemblages in TKR do not exhibit the same turnover along a north-south axis, both sets of data emphasise the contribution of mountains and their ecotones to landscape wide heterogeneity. Elsewhere, the influence of topology on grasshopper diversity has been associated with the size of hills. In the Succulent Karoo biodiversity hotspot, small hills support a greater diversity of grasshopper families than do medium and large ones (Gebeyehu & Samways 2006). Gebeyehu and Samways (2006) note that smaller hills have less vegetation cover, and increased cragginess, which likely provide a wider array of microhabitats for grasshoppers to utilise.
Although size of individual peaks and hills comprising the Korannaberg sample was not considered, rockiness was found to increase at the base of the hills, and does not extend into the plains vegetation types. Grass, as well as total vegetation, cover also decreases from this point. This transitional zone comprised a mixed grasshopper assemblage, which became increasingly unique further up the mountain. This, coupled with the strong species associations with habitat types suggests that grasshopper species found in the mountain and ecotone sites represent somewhat specialised adaptions to these environments. Consequently, while diverse grasshopper assemblages (and those of other taxa such as dung beetles; Davids et al., 2010) are maintained by natural heterogeneity of the entire conserved landscape, the highest levels of taxonomic diversity are supported by mountain habitats.

CONCLUSIONS
Grasshoppers were found to be considerably diverse within the Tswalu Kalahari Reserve, mirroring the high diversity of other animal taxa in the region. Landscape-level variation in grasshopper diversity provide some insights into how such high levels of biodiversity are maintained within this arid, unproductive environment. Variation in environmental and vegetational conditions across the landscape had non-significant impacts on diversity, except along a gradient from plains to mountain sites. Mountains supported unique and highly diverse grasshopper assemblages not found elsewhere, implying that landscape topology is a necessary feature for the maintenance of biodiversity of the region. While the effects of species composition on community assembly requires more detailed analysis of functional trait diversity, it is clear that the (relatively small) mountains and hills in the Kalahari system are islands of diversity that must be considered when executing conservation measures beyond the boundaries of protected areas.

AUTHOR CONTRIBUTIONS
Aileen C. van der Mescht and Daryl Codron conceptualised the study.
Aileen C. van der Mescht, conducted the fieldwork and specimen identifications, as well as the statistical analyses with input from Daryl Codron. The initial manuscript was written by Aileen C. van der Mescht, with Daryl Codron contributing to its revision.