An inventory of epigeal ants of the western Soutpansberg Mountain Range , South Africa

http://www.koedoe.co.za doi:10.4102/koedoe.v57i1.1244 The distribution, abundance and sensitivity of invertebrates to habitat change are largely unknown. Long-term monitoring of ecological gradients with standardised and comparable protocols can form the basis of a better understanding. Altitudinal gradients are particularly relevant within this context. Here we provide a check list and baseline data for ant species collected over a 5-year period across the Soutpansberg Mountain Range, South Africa. Standardised pitfall surveys across 11 sites yielded a total of 133 species in 38 genera and 6 subfamilies. Sample coverage of epigeal ants was 0.98 for the transect as a whole. Of these species, 21% were restricted to the southern slope of the mountain and 14% to the northern slope. Extrapolated richness estimates reached an asymptote for all, except for three sites. These were the only sites impacted by bush encroachment. Observed richness was the highest at a low-altitude mesic site that is exposed to considerable disturbance by megaherbivores and mechanical clearing of woody vegetation. Structural classification of vegetation was predictive of a broad-scale ant assemblage structure. On a smaller scale, however, structure was a function of elevation, space and temperature.


Introduction
Although invertebrates comprise the bulk of diversity in an ecosystem, they are often excluded from conservation initiatives.Cardoso et al. (2011) identified seven reasons for this.Three of the shortfalls had to do with: (1) the distribution of invertebrate species is unknown (Wallacean), (2) invertebrate abundance and their changes in space and time are unknown (Prestonian) and (3) invertebrates' way of life and sensitivity to habitat types are largely unknown (Hutchinsonian).These three shortfalls are particularly relevant to how ecologists practise their science, as the distribution, abundance in time and space, and sensitivities to habitat change are largely unknown for most species.These constraints can be resolved through inventories that follow standardised and comparable protocols (Cardoso et al. 2011).
Ants respond to the environment at small scales and considerable support exists for their use as indicators of disturbance and habitat degradation (see Andersen & Majer [2004] for review).Ants are easily collected and species or morpho-species level determination are aided through readily available taxonomic expertise (Ward 2010).In addition to being diverse, abundant and a dominant component of animal biomass in terrestrial ecosystems (Hölldobler & Wilson 1990), they also perform important functions in ecosystems, including pollination, myrmechory and nutrient cycling, among others (Wielgoss et al. 2014).
The Soutpansberg Mountain Range is a centre of endemism (Hahn 2002) and is the focal point for the newly proclaimed Vhembe Biosphere Reserve, one of the United Nations Educational, Scientific and Cultural Organisation (UNESCO) World Network of Biosphere Reserves (www.unesco.org).It has the highest plant generic and family level diversity among the 18 recognised Centres of Plant Endemism in southern Africa (Van Wyk & Smith 2001).Recent results suggest that this diversity and endemism is mirrored by spiders (Foord 2008) and reptiles (Kirchhof et al. 2010).As an inselberg that rises 1000 m above the surrounding plains, the Soutpansberg Mountain Range has provided a refuge for several taxa during periods of climate change and will once more play a biogeographic role in the region as regional climate changes take effect in response to elevated global greenhouse gases (Tshiala et al. 2011).
Transects can provide valuable diversity, distribution and abundance information as well as population trends.Altitudinal transects are particularly relevant within the context of contemporary climate changes.Here we summarise the results of a standardised inventory of epigeal ants over a 5-year period (2009)(2010)(2011)(2012)(2013), along a transect across the western Soutpansberg Mountain Range.We provide a checklist of epigeal ants with a sample coverage of 0.98 for the whole transect.A posteriori defines distinct ant assemblages across the transect and tests for differences between these groups.

Study area
The 16.5 km altitudinal transect extends over the highest point of the Soutpansberg Mountain Range, Lajuma (1748 m), in the Luvhondo Nature Reserve.It includes the northern and southern slopes of the mountain and consists of 11 sites, spaced at 200 m altitudinal intervals.It starts at 800 m a.s.l. in the Limpopo valley to the north of the mountain, ascends 900 m to Lajuma, and descends another 800 m to the plains south of the range (Figure 1).
Vegetation types along the plains and lower southern foothills are largely thicket (Figure 3d) and shrubland (Figure 3e), located on red loamy-clay soils dominated by woody species such as Dichrostachys cinerea, Acacia caffra and Olea europaea.
Structurally, the lowest site on the southern aspect (Figure 3e) is tall, closed shrubland (< 0.1% tree cover).It is dominated by a D. cinerea matrix with patches that are mechanically cleared of all woody plants.Site 10S (Figure 3d) is low thicket (10% -100% tree cover and > 10% shrub cover).The forests are found at mid-elevation, between 1200 m a.s.l. and 1400 m a.s.l. on the southern slope and include both short (Figure 3c) and tall (Figure 3b) forest sites with a crown-gap ratio of < 0.1.They are dominated by woody species such as Croton sylvaticus and Ekebergia capensis.The higher elevational zones of the southern aspect and summit consist of a sedgeland-herbland matrix with low shrubland patches that grow on shallow rocky soils on quartzite, dominated by Coleochloa species (Figures 2e-3a).

Ant sampling
Ants were sampled through standardised pitfall trapping (Munyai & Foord 2012) at each of the 11 sites.Sample units consisted of 10 pitfall traps (ø 62 mm) laid out in a sampling grid (2 x 5) with 10 m spacing between traps.
Pitfall traps were left open for 5 days, since it is considered representative and does not sample excessive ants (see Munyai & Foord [2012] and references therein).From September 2009 onwards (Appendix 1), sampling units were spatially replicated four times at each site.Replicates were separated by < 300 m to avoid pseudo-replication (Ness et al. 2004).The traps contained a 50% solution of propylene glycol because it neither attracts nor repels ants (Abensperg-Traun & Steven 1995).  (1994) was used to identify some specimens to genus level, while valid names were confirmed using AntCat (http:// www.antcat.org)and AntWeb (http://www.antweb.org).
Bothroponera, Mesoponera, Ophthalmopone and Pseudoponera were considered valid genera rather than being subsumed under Pachycondyla (Schmidt & Shattuck 2014).Subfamilies Aenictinae and Cerapachyinae were both considered invalid and therefore subsumed under Dorylinae, as treated by Brady et al. (2014).Reference collections are housed at the University of Venda and CSIRO Tropical Ecosystems Research Centre (Darwin, Australia).

Data analysis
Chao 2 and ICE, in EstimateS version 9.1.0(Colwell 2013), were used to estimate species richness.Sample coverage at each site and the transect as a whole was done with Both interpolation and extrapolation of samples was based on the Bernoulli product model as well as unconditional variance estimates for the rarefaction (interpolation) (Chao & Jost 2012) and extrapolation curves.This sample-based estimation provides both unconditional variance abundance data of this study and species richness predicted by a 1000 pitfall samples and were considered to be a fair reflection of the richness at a site.
Samples within each replicate were pooled over the period of the study for multivariate analysis.Ant assemblage structure was mapped using non-metric multidimensional scaling of Bray-Curtis similarity measures (Clarke & Gorley 2006).A posteriori groups of ant assemblages were defined based on hierarchical clustering with group average linking.Support for these groups was tested with the permutational procedure SIMPROF in PRIMER version 6.1.6(Clarke & Gorley 2006).

Results
In total, 85 759 ant specimens were collected during the 13 sampling periods (Appendix 2), comprising 133 species in 38 genera and 6 subfamilies.
Myrmicinae (with 72 species, 86% of the total abundance and 15 genera) was the most diverse and abundant subfamily, followed by Formicinae (27 species and 6 genera).Ponerinae (12 genera) had the second highest generic diversity.
Sample-based species rarefaction and extrapolation curves for the transect as a whole approximated an asymptote, indicating that most of the epigeal species were collected with a sample coverage of 0.98.Sample coverage for all sites was > 0.95, while the forest site 12S coverage was the lowest, at 0.955.Based on richness estimations (Table 1) and species richness extrapolations (Figures 4 and 5), ant richness peaked at mid-elevations on the northern aspect.Only at woodland site 14N was species richness significantly greater that of 17N, the summit (Figure 4).
The pattern on the southern aspect was more complex, with richness decreasing significantly from shrubland site 9S to forest site 12S (Figure 4) and then increasing at sedgeland sites 14S and 16S before decreasing again towards the summit (Figure 4).Based on predicted richness of sites (Figure 5), the ranking of sites changes only for 14N.Notably, the predicted curve for 16S also suggests a considerable increase in richness with increased sampling.
The sites with the highest and lowest variation in mean annual temperature variation (17N and 12S) also had the lowest species richness.
In total, 104 species were collected along the northern slope and 117 species on the southern slope.Twenty-eight species were restricted to the southern slope and nineteen species to the northern slope.and Technomyrmex pallipes Smith, occurred at all the sites (Appendix 2).
A total of 20 Ponerine species were sampled from both the southern and the northern aspects.Fifty per cent of all Ponerine species, namely Anochectus UOVC_01, Bothroponera ?strigulosaEmery, Bothroponera UOVC_02, Bothroponera UOVC_03, Hypoponera UOVC_01, Leptogenys intermedia Emery, Leptogenys schwabi Forel, Ophthalmopone UOVC_01, Plectroctena UOVC_02 and Pseudoponera UOVC_01, were restricted to the southern slope of the mountain.The other 10 species occurred on both slopes of the mountain.
The non-metric multidimensional scaling plot identified three major groupings: the high elevation sedgeland-herbland sites (14S -summit); woodlands (08N -14N); and shrubland, thicket and forest sites (09S -12S) (Figure 6).At Bray-Curtis similarity measures of 50, each of the 3 groups divides into 2 (Figure 6).The groupings reveal distinct assemblages associated with the summit (17N) at higher elevations, while 14N was distinct from all the lower elevation sites on the northern aspect.Low elevation (shrubland and thicket sites 09S -10S) and mid-elevation (forest sites -12S) of the southern aspect formed a distinct grouping.

Discussion
The level and higher taxon diversity of this study compares favourably with that of similar studies in the Cederberg and Drakensberg mountains, both of which employ a protocol similar to this study.The Cederberg transect located in the Cape Floristic Kingdom includes 17 sites over a distance of 160 km.That study recorded 86 species in 24 genera (Botes et al. 2006).The transect in the grasslands of the Drakensberg, with 9 elevational sites over a distance of 104 km along the Sani Pass, collected 98 species in 28 genera and 7 subfamilies (Bishop et al. 2014).Only one other study of ants along an altitude exists for the Savanna Biome in South Africa, where Schoeman & Foord (2012) found 104 species in 29 genera along a 1000 m elevational range in the Marakele National Park.With a higher generic plant diversity than the Cape Floristic Kingdom (Hahn 2006) and 90% of all spider families found in the Savanna Biome (Foord et al. 2011), the Soutpansberg is characterised by rich, higher taxon diversity.Myrmicinae and Formicinae were the most diverse subfamilies.The most speciose genera were Tetramorium, Camponotus, Pheidole and Monomorium.This conforms to results from other studies in South Africa (Botes et al. 2006).Monomorium is very diverse in Australia, while Camponotus and Pheidole are the most diverse in Brazil (Campos et al. 2011), although the highly diverse genus in Australia, Polyrhachis, is only represented by two species along this transect.Except for Lepisiota, genera found throughout the transect belong to cosmopolitan genera (Camponotus, Monomorium, Pheidole and Tetramorium).
Most of the species accumulation curves of predicted richness based on 1000 pitfall samples reached an asymptote.There were two exceptions: a woodland site 14N and sedgeland site 16S, both of which also had the lowest coverage.Although 14N is in the woodland vegetation type, it is near the ecotone between sedgeland and woodland.The large turnover of taxa at this site could be the result of mass effects from taxa found in sedgeland.The steep slope of the 16S accumulation curve might suggest a temporal change in the ant assemblages of this site.Based on photographic records at each of the pitfalls, the Coleochloa species that dominate this site are being invaded by short woody shrubs.Turnover at this site might be linked to this encroachment and could probably lead to increased abundance of taxa associated with sites that are structurally more complex, such as forest sites 12S and 12S2 and shrubland site 9S.This has particularly important implications for the management of the few remaining grassland habitats in the Soutpansberg.
The high diversity of Ponerinae species at the base of southern aspect might be due to more productive ecosystems associated with the basalt intrusions, and higher rainfall.Disturbance also seems to positively impact Ponerine richness, which peaks in shrubland site 9S.This site is inside a wildlife sanctuary that is stocked with buffalo and where the bush is mechanically cleared.The three major groupings of ant assemblages across the study area conform to the broad structural classification of the vegetation used in this study (Edwards 1983).

Conclusion
This study provides a thorough inventory of grounddwelling ants, with a sample coverage of 0.98.The sample coverage represents the proportion of the total number of individuals in the ant assemblage along the transect that belongs to the species represented in our sample (Chao & Jost 2012).Extrapolated richness for most sites reached an asymptote, except for three sites.All three of these sites can be considered to be sites that have experienced increased levels of bush encroachment.This is particularly concerning at 16S, which represents some of the last remaining grassland habitats in the Soutpansberg.
Ant taxa that act as indicators of bush encroachment should be identified in order to monitor the rate of these impacts.A single site on the lowest elevation of the southern aspect (09S) yielded 57% of the total ant species.This site experiences higher levels of disturbance through herbivory and bush clearing and provides support for the important role of disturbance as a driver of ant diversity in the Savanna Biome.The structural classification of vegetation in this study is a good predictor of broad-scale ant assemblage structure.Finer scale distinctions were linked to elevation, structural elements and temperature.

FIGURE 1 :
FIGURE 1: Location of the Soutpansberg transect in South Africa (inset) and the position of the 44 replicates with the 5 structural habitat types.
Soil temperature extremes were the highest on the summit of the mountain (mean annual range 10 °C -55 °C) and the lowest in the tall forests of the southern aspect (14 °C -32 °C).

FIGURE 5 :
FIGURE 5: Species accumulation curves and their 95% confidence intervals for 1000 samples per site.The solid circles represent reference samples; solid lines are rarefied samples and dashed lines are extrapolation based on Mau Tau estimation.(a) Northern aspect, (b) higher elevational sites and (c) low to mid-elevation southern aspect.SBT, accumulation curve for the whole transect.

FIGURE 6 :
FIGURE 6: Non-metric multidimensional scaling of ant assemblages for samples pooled within each of the 44 replicates across 11 sampling sites in the Soutpansberg Mountain Range.

TABLE 1 :
Observed number of species, abundance, richness estimates (Chao 2 and ICE, available in EstimateS 9.1.0package) and sample coverage for broad structural vegetation types and 11 sites with their geographical coordinates along the Soutpansberg transect.Extrapolated species richness for each site and 95% confidence intervals based on unconditional variance, from 1000 pitfall samples for each of the 11 sites across the Soutpansberg Mountain Range.

TABLE A2 :
Checklist of subfamilies and ant species collected during 13 surveys in different broad structural vegetation types across an elevational transect in the western Soutpansberg Mountain Range in South Africa.
Table A2 continues on the next page →

TABLE A2 (
Continues...): Checklist of subfamilies and ant species collected during 13 surveys in different broad structural vegetation types across an elevational transect in the western Soutpansberg Mountain Range in South Africa.Original Research UOVC, University of Venda Collection.*, Introduced species.

TABLE A2 (
Continues...): Checklist of subfamilies and ant species collected during 13 surveys in different broad structural vegetation types across an elevational transect in the western Soutpansberg Mountain Range in South Africa.