• assemblage structure;
  • behavioural dominance;
  • competition;
  • numerical dominance;
  • savanna


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Competition is considered a key factor structuring many communities, and has been described as the ‘hallmark’ of ant ecology. Dominant species are thought to play a key role structuring local ant assemblages through competitive exclusion.
  • 2
    However, while there have been many studies demonstrating competitive exclusion and consequently reduced richness at baits, it is not clear whether such regulation of ‘momentary’ diversity at clumped food resources can scale up to the regulation of richness at the site or assemblage level.
  • 3
    In this study, ant assemblages were sampled in three different savanna habitats in South Africa using both baiting and pitfall trapping.
  • 4
    As has been found in previous studies, there was a unimodal relationship between dominant ants and species richness at baits, with high abundances of dominant ants regulating species richness through competitive exclusion. Analysis of pitfall samples revealed strong convergence in pattern, and results from null model co-occurrence analyses supported the findings.
  • 5
    The importance of competition in structuring local ant assemblages was, however, only apparent at one of the three savanna habitats suggesting that a full range of extreme environments is needed to produce the full unimodal relationship at the assemblage level.
  • 6
    Although the relative importance of competition varied with habitat type, the study demonstrated that in some habitats, dominant ants can control species richness at the assemblage level.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The determinants of local species richness remain the subject of ongoing debate. Although it is now widely appreciated that regional scale processes determine the potential membership of local assemblages (Ricklefs 1987; Gaston 2000; Godfray & Lawton 2001), how membership of an assemblage is finally determined remains contentious. Membership may be a consequence of highly stochastic processes such as random colonization (see Gaston 2000), or it may be the outcome of more deterministic processes such as the operation of a set of assembly rules (Weiher & Keddy 1999). Depending on the taxon and system concerned, interspecific competition may be an important local process structuring the assemblage, may be negligible relative to other processes, or may simply not occur.

Interspecific competition is thought to be a key mechanism structuring local ant assemblages, and as such has been referred to as the ‘hallmark of ant ecology’ (Hölldobler & Wilson 1990). Support for the importance of interspecific competition includes spatial ant mosaics (Room 1971; Majer, Delabie & Smith 1994), behavioural dominance hierarchies (Savolainen & Vepsäläinen 1988), territoriality (Fox, Fox & Archer 1985; Andersen & Patel 1994), agonistic behaviour (Andersen, Blum & Jones 1991), and results from simulation modelling (Parr et al. 2005) (but also see Ribas & Schoereder 2002; Gibb & Hochuli 2004).

The regulation of small-scale diversity by dominant ants (regulation of ‘momentary’ diversity or the numbers of species in attendance at a bait at a given time) has been found in several baiting studies, including those in tropical savannas of Australia and Africa (Andersen 1992; Andersen & Patel 1994; Parr et al. 2005), in tropical French Polynesia (Morrison 1996), in the boreal taiga biome (Savolainen & Vepsäläinen 1988), and in tropical agro-ecosystems (Perfecto & Vandermeer 1996). At baits, it has been shown that the full relationship between species richness and dominance is unimodal: species richness is low at very low abundances of dominant ants, and as the abundance of dominant ants increases, species richness also increases until a point is reached after which species richness declines as dominance increases (Andersen 1992; Parr et al. 2005). While the ascending portion of the curve is thought to correspond to increasing habitat favourability and reduced stress for ants, as well as constraint imposed by abundance frequency distributions, the descending part of the dominance–richness relationship is due to an increase in the abundance of dominant ants, to such an extent that they reduce species richness via competitive exclusion (Andersen 1992; Parr et al. 2005).

However, the use of baits to emphasize the importance of competition in structuring ant assemblages is problematic because it represents only one scale of the dominance–species richness relationship, and exclusion from baits does not necessarily imply that competitive exclusion by dominants is taking place at the population level, and therefore that competition should be considered a key mechanism structuring assemblages (see Hairston 1981; Andersen & Patel 1994; Sanders & Gordon 2000; Ribas & Schoereder 2002; Gibb & Hochuli 2003).

Although competitive exclusion and the regulation of richness at baits appear to be the rule (e.g. Fellers 1987; Morrison 1996; Bestelmeyer 2000), to-date no studies have demonstrated that this also translates to control of resources and the regulation of richness by dominant ants at the assemblage level. Indeed, the available evidence suggests that this does not occur (Andersen & Patel 1994; Ribas & Schoereder 2002; Gibb & Hochuli 2003). Dominant species may control bait resources, but because at a broader scale, resources may either not be limited or species in the community use different resources, it may not be possible to predict with confidence whether competition is also the main structuring force for the local assemblage. Other factors including predators (Gotelli 1996), parasitoids (LeBrun & Feener 2002), environmental fluctuations (Cerdá, Retana & Cros 1997) and patchy distributions may also prevent dominant species from regulating assemblage structure.

This study investigates whether competitive exclusion at baits scales up to competition at the assemblage level by comparing dominance–species-richness patterns found with baits and pitfall traps (i.e. do temporal relationships at baits also hold true at the assemblage level?). Baits represent localized richness or small-scale diversity as a result of interactions at a resource (‘momentary diversity’, Andersen 1992), while pitfall traps represent general foraging activity and richness at the assemblage level. Thus, the study explores the extent to which competition and dominant ants play a role in regulating assemblage level richness.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

study sites

The study was carried out in Kruger National Park, South Africa in three savanna types: Mopane woodland (Mopane, 23°33′S, 31°26′E), Acacia savanna (Satara, 24°26′S, 31°46′E) and Terminalia woodland (Pretoriuskop, 25°12′S, 31°23′E).

The Mopane woodland savanna is dominated by Colophospermum mopane J. Kirk ex J. Léonard, with very few other woody species (Low & Rebelo 1996). The altitude of the Mopane area ranges from 300 to 340 m, and the mean annual rainfall is 450 to 500 mm (Gertenbach 1983). The savanna at Satara is a mixed Acacia nigrescens Oliv. and Sclerocarya birrea A. Rich. Hochst. savanna in the Sweet Lowveld Bushveld (Low & Rebelo 1996). The mean annual rainfall in this area is approximately 550 mm, and the altitude ranges from 240 to 320 m (Gertenbach 1983). Satara and Mopane share basalt-derived clay soils (Gertenbach 1983). The vegetation of the Pretoriuskop area is Sour Lowveld Bushveld (Low & Rebelo 1996). This is an open-tree savanna dominated primarily by Terminalia sericea Burch. ex DC, but also Combretum collinum Fresen. and Dichrostachys cinerea (Linnaeus) R. Wight & Arnott. Soils are sandy and granitic-derived. Mean annual rainfall is 700 mm, and altitude ranges from 560–640 m (Van Wilgen et al. 2000). Table 1 presents information on vegetation cover, tree density, soils and temperature for each savanna habitat.

Table 1.  Vegetation, ground layer, soil characteristics, and mean maximum and minimum temperature (mean ± SE) in the three savanna habitats (Mopane, Satara and Pretoriuskop) in Kruger National Park, South Africa (temperature data provided by WeatherSA, Erasmusrand, Pretoria, South Africa, and soil data from Mills & Fey 2004)
Percentage of cover (mean ± SE)
 Grass38·8 (4·1)42·2 (5·3)55·4 (3·8)
 Litter and dead grass23·3 (2·7)11·4 (2·0)16·1 (6·3)
 Bare ground35·0 (7·2)43·8 (8·1)15·4 (5·5)
 Forbs2·4 (0·5)2·0 (0·5)12·0 (0·9)
 Tree density (per hectare)93·011·8163·8
Percentage of soil characteristics
 Clay18·3 (2·3)11·7 (0·6)7·6 (0·9)
 Silt22·4 (3·8)24·0 (3·4)6·3 (2·1)
 Sand59·8 (2·4)64·5 (4·5)86·5 (3·4)
Mean maximum temperature: December (°C)31·932·632·5
Mean minimum temperature: December (°C)20·020·919·5
Mean maximum temperature: July (°C)25·026·325·3
Mean minimum temperature: July (°C)7·910·010·6

As part of a broader study (see Parr et al. 2004, 2005), ant sampling was undertaken on a series of experimental burn plots that form part of a long-term burning experiment initiated in 1954; unburned plots have had no fires since then (see Parr et al. 2004, 2005). All plots are approximately 7 ha in size (Trollope et al. 1998).

ant sampling: baiting

The relative dominance of species was quantified on four plots in each of the three habitats, using observation of ants at fish baits (see Fellers 1987; Savolainen & Vepsäläinen 1988; Andersen 1992; Parr et al. 2005). Plots were an average of 800 m apart. Baiting was conducted in the summer months when ants are typically most active (late November 2001–early February 2002). In each habitat, baiting was conducted in the early morning, midday and late afternoon, and was repeated at least three times for each time period. Baiting across a range of temperature regimes, in addition to different plot and habitat types, enabled a wide range of stress levels to be sampled (see Parr et al. 2005).

For each bait session, fifteen bait stations were set out at 5-m intervals along a 70-m transect. A teaspoon of cat food (≈ 3 g, pilchards) was placed on a small white piece of paper (60 mm × 50 mm) to aid observations at the bait, and all species present at each bait after 60 min were recorded. To avoid problems of data non-independence, the location of transects was randomized on each of the plots such that consecutive baiting sessions were separated by a distance of more than 50 m, and the locations of any two baiting sessions were never closer than 50 m. Abundances of all ants were scored according to a six-point scale: 1 = 1 ant, 2 = 2–5 ants, 3 = 6–10 ants, 4 = 11–20 ants, 5 = 21–50 ants and 6 ≥ 50 ants (following Andersen 1997; Parr et al. 2005).

A species’ abundance per baiting session was defined as the total of its abundance scores summed across the 15 baits. The total abundance of dominant ants during a baiting session was the sum of each dominant species’ total abundance for each bait station. The total maximum abundance of any ant species per baiting session was 90 (abundance score of 6 × 15 baits).

Ground surface temperature was recorded adjacent to each bait after 60 min (in order to be representative of the environmental conditions for foraging) using a digital thermometer with a 20 SWG Type K (chromel-alumel) thermocouple. Equilibration time was 1 min ± 30 s.

ant sampling: pitfall trapping

Pitfall trapping in both Mopane and Satara was carried out on fourteen plots, but was restricted to six plots in Pretoriuskop due to logistic constraints. The plots included those used for the baiting trials, with 10 additional plots at both Mopane and Satara, and two additional plots at Pretoriuskop. On each plot, 20 pitfall traps were laid out in a grid (5 × 4) with 10-m spacing. All grids were situated at least 50 m from the plot edge to reduce the possibility that ants from adjacent areas were collected in the traps. Pitfall traps had a diameter of 62 mm, and contained 50 ml of a 50% solution of propylene glycol. All traps were left open for 5 days in January 2002. Voucher specimens of ants are held at the Iziko Museum of Cape Town, South Africa. Although the sampling efficiency of pitfall traps can be affected by trap size (Abensperg-Traun & Steven 1995), length of time open, and species’ susceptibility to capture (Marsh 1984), a pilot experiment indicated that the chosen pitfall sampling method provided a good representation of the ant fauna present at the plot level (Parr & Chown 2001).

dominance analyses

In this study, a combination of numerical and behavioural dominance is used to define dominant ants. Thus, numerically dominant ant species are considered to be those that occurred at a large proportion of baits, numerically dominated and monopolized many of the baits where they occurred, and had high mean abundance scores (Andersen 1992; Morrison 1996; Cerdáet al. 1997; Davidson 1998; Parr et al. 2005).

Behavioural dominance was assessed using individual interactions between taxa (see Fellers 1987; Cerdáet al. 1997; Cerdá, Retana & Manzaneda 1998; Bestelmeyer 2000; Retana & Cerdá 2000), and involved measuring the number of times that individuals of a given taxon were behaviourally dominant or subordinate in interactions with other taxa. An individual was considered dominant to another if it behaved aggressively and caused the other to retreat and abandon the bait (following Bestelmeyer 2000; Retana & Cerdá 2000). Aggression included biting, charging and use of chemical compounds from the end of the gaster. Dominance scores were calculated as the percentage of encounters in which a taxon was dominant (i.e. aggressive) in all its interspecific encounters (see Fellers 1987; Bestelmeyer 2000; Retana & Cerdá 2000). Each bait station was observed for any agonistic interactions during a 30-s period at 5, 15, 30 and 60 min, and recorded only if individuals were on the bait paper.

Dominant species were identified for each habitat on the basis of the following descriptors (following Andersen 1992, 1997; Morrison 1996; Cerdáet al. 1997; Parr et al. 2005): number of baits monopolized with > 20 individuals of a species where they occurred, high mean abundance score, and dominance scores > 50% (Retana & Cerdá 2000). Dominant species monopolized many of the baits where they occurred, had high mean abundance scores (> 3) and dominance scores > 50%. The abundance of dominant species in pitfall traps (determined using baits) was summed for each site.

statistical analyses

Species-richness values at baits and at the assemblage level (pitfalls) were individually regressed against the abundance of dominant ant species at both baits and pitfall traps respectively in both linear and nonlinear (logarithmic and quadratic) models. If more than one of the models revealed a significant relationship, they were compared using Akaike's Information Criterion (AIC). This technique is used to compare the goodness-of-fit of statistical models. In addition, to provide further understanding of the role of dominance as a causal factor influencing diversity at baits, the proportion (indicated as percentage) of species at baits relative to the total species pool was regressed against the abundance of dominant ant species at baits. If competition by dominant ants reduces richness at baits, then the proportion of ant species at baits will decrease as the abundance of dominant ants increases. The total species pool was determined by combining species from baiting and pitfall trapping, but excluding species that would not be attracted to baits (e.g. cryptic species such as Pyramica, and specialist predators such as Aenictus).

Null model analysis using pitfall data at the habitat scale (Gotelli & Graves 1996; Gotelli 2000) was used to test for nonrandom patterns of species co-occurrence. For each habitat (Mopane, Satara and Pretoriuskop), a presence–absence matrix was produced where each row was a different species, and each column was a different plot within that area (pitfall samples combined for each plot). Stone and Roberts’ (1990) C-score was used to quantify co-occurrence. In competitively structured communities, this index is significantly larger than expected by chance (i.e. there is less co-occurrence that would be expected by chance). Conversely, with a lower C-score value, there is more co-occurrence than would be expected by chance (i.e. the community is random or aggregated). For each presence–absence matrix, 5000 random matrices were produced using a fixed–fixed algorithm. This null model retains the row and column sums of the original matrix, and has been shown to have good statistical properties with low probability of type 1 errors (Gotelli 2000). C-score values were calculated for each random matrix, and then the tail probability for the observed matrix was determined by comparing it with a histogram of the simulated values. All null model analyses were conducted with ecosim Version 6·0 (Gotelli & Entsminger 2001).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

dominant species

A total of 69 ant species was recorded at the baits with 38, 36 and 49 species occurring in the Mopane, Satara and Pretoriuskop areas of Kruger National Park, respectively. Pheidole (including Pheidole sp. nr. megacephala) and Crematogaster spp., Myrmicaria natalensis (Smith), Monomorium emeryi Mayr, and Anoplolepis custodiens (Smith) were identified as numerically dominant species because they were responsible for most of the monopolization at baits and had the highest mean abundance scores per habitat (see Parr et al. 2005, Table 2).

Table 2.  Occurrence of numerically dominant ants and subordinate ants at baits in Kruger National Park
 Percentage of baits recordedPercentage of baits monopolizedMean abundance score*
  • *

    Mean abundance scores range from a possible minimum of 1 (always a single ant recorded whenever the species occurred) to a possible 6 (always > 50 ants whenever the species occurred). M = Mopane, S = Satara, P = Pretoriuskop.

 Pheidole spp.27·553·017·835·760·840·03·804·063·21
 Crematogaster spp.6·91·130·054·350·023·04·094·933·62
 Myrmicaria natalensis11·214·519·364·533·333·33·364·223·30
 Anoplolepis custodiens1·75·188·926·13·934·06
 Monomorium emeryi8·855·63·88
 Camponotus spp.11·85·114·03·301·61·451·531·55
 Polyrhachis spp.5·911·018·70001·461·521·39

Species of Pheidole and Crematogaster, and M. emeryi were behaviourally dominant [dominance scores of 95·2% (n = 60), 89·5% (n = 19), and 100% (n = 5) respectively]. Pheidole and Crematogaster spp. were observed aggressively displacing other ants from the baits in all areas, but were never displaced themselves. There were few observations of A. custodiens and M. natalensis expelling other species because they had patchy distributions. Nevertheless, A. custodiens won all encounters (except with Crematogaster), and M. natalensis was dominant in all encounters except with Pheidole and Crematogaster. For these reasons and their high numerical dominance, these species were still considered dominants. Non-dominant ants included species of Camponotus [particularly. C. cinctellus (Gerstaecker)] and Polyrhachis (see Parr et al. 2005) (Table 2). A total of 164 species, and 30 261 individuals were collected with pitfall trapping with 69, 111 and 89 species recorded in the Mopane, Satara and Pretoriuskop, respectively (Parr et al. 2004).

dominance–richness relationships

The full relationship between abundance of dominant ants and species richness at baits across all three sites is best described as unimodal (quadratic) (F2,108 = 20·41, P < 0·0001, R2 = 0·330) compared to linear or logarithmic relationships (F1,108 = 0·72, P = 0·398, R2 = 0·007 and F1,108 = 2·23, P = 0·138, R2 = 0·020 respectively) (Fig. 1a). At the assemblage level, data from pitfall traps indicated that the full relationship between the abundance of dominant ants and species richness across all three sites was also best described as a unimodal (quadratic) relationship (F2,33 = 18·24, P < 0·0001, R2 = 0·541, AIC = 138·66) compared to linear or logarithmic relationships (F1,32 = 4·126, P = 0·051, R2 = 0·114, AIC = 158·44 and F1,32 = 17·38, P < 0·0001, R2 = 0·352, AIC = 148·21 respectively) (Fig. 1b).


Figure 1. Relationship between the abundance of dominant ants and species richness (including dominant ant species) across three savanna habitats in Kruger National Park, using (a) baiting data (quadratic regression: y = −0·003x2 + 0·27x + 2·25, R2 = 0·33, P < 0·0001), and (b) pitfall trap data (quadratic regression: y = −0·0001x2 + 0·11x + 13·74, R2 = 0·541, P < 0·0001). For the baiting data, each data point represents total species richness and total abundance of dominant ants for 15 baits per baiting session. For the pitfall data, each data point represents total species richness and total abundance of dominant ants per plot for a grid of 20 pitfall traps, and may represent several dominant species.

Download figure to PowerPoint

Thus, the dominance–species richness relationship can be considered unimodal at both local ‘momentary’ bait and assemblage levels; highest richness occurs at intermediate dominance levels, and as the abundance of dominant ants increases, species richness declines. With bait data, there was a greater spread of dominance values within each site than with pitfall data. At the assemblage level (pitfall data), Mopane data points cluster to the left-hand side of the graph indicating overall low dominance levels and low species richness. Species richness is highest at Pretoriuskop where dominance levels are intermediate; while at Satara, data from pitfall trapping exhibit an overall negative relationship indicating that at high dominance level, species richness is low (Fig. 1b). The full unimodal relationship is thus only evident when data from the three habitats are combined, and Satara is the only site with declining richness matching increasing dominance (Fig. 1b).

For all three habitats combined, the full relationship between abundance of dominant ants and percentage of species at baits relative to the species pool across all three sites is best described as unimodal (quadratic) (F2,108 = 15·36, P < 0·0001, R2 = 0·221) (Fig. 2a). This unimodal pattern is driven by Satara (quadratic regression: F2,45 = 21·57, P < 0·0001, R2 = 0·489), with Mopane and Pretoriuskop showing no significant trends (Fig. 2b). The ascending part of the curve at Satara is due primarily to high temperature stress (Fig. 2b). Baiting sessions with mean ground surface temperature > 36 °C represent high physiological stress for most ant species in this area and foraging activity is consequently reduced; above 40 °C, the majority of ant species cease foraging entirely (see Cerdáet al. 1997; Bestelmeyer 2000). When excluding high-temperature baiting sessions, the relationship at Satara is negative linear (F1,33 = 22·25, P < 0·0001, R2 = 0·403, AIC = 38·82), compared to quadratic relationship (F2,32 = 11·72, P < 0·0001, R2 = 0·423, AIC = 39·63) (Fig. 2b).


Figure 2. Relationship between the abundance of dominant ants at baits and the percentage of species at baits relative to the available species pool (a) across three savanna habitats in Kruger National Park (quadratic regression: y = −0·004x2 + 0·287x + 5·68, R2 = 0·221, P < 0·0001), and (b) for Satara Acacia savanna only, showing baiting sessions with mean ground surface temperature above and below 36 °C (all baiting data combined: quadratic regression, y = − 0·003x2+ 0·288x + 2·20, R2 = 0·489, P < 0·0001; below 36 °C only: linear regression, y = −0·062x + 9·75, R2 = 0·403, P < 0·0001). For the baiting data, each data point represents the percentage of species at baits relative to the available species pool and total abundance of dominant ants for 15 baits per baiting session.

Download figure to PowerPoint

Co-occurrence analyses at the habitat scale indicated that ant assemblages in Satara had significantly less co-occurrence than expected by chance (large observed C-scores); while in contrast, ant assemblages in the Mopane and Pretoriuskop savanna habitats were random (Table 3).

Table 3.  Patterns of species co-occurrence for ant assemblages in each savanna habitat using pitfall data. For an assemblage to be competitively structured, the C-score should be significantly larger than expected by chance (Gotelli 2000). Statistically significant P values for observed C-scores greater than simulated are in bold
HabitatObserved C-scoreMean of simulated C-scoreVariance of simulated C-scoreP value


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The relationship between dominance and species richness both at baits (‘momentary diversity’) and pitfall traps (assemblage level) revealed strong convergence in pattern with data from baiting and from pitfall traps both producing the full unimodal relationship, although clearly the richness and dominance values are rather different (Fig. 1). Importantly, this is the first study to show that the full unimodal dominance–richness relationship found at baits (Fig 1a, see also Andersen 1992; Parr et al. 2005) can also be found with pitfall trap data representing the broader assemblage level pattern (Fig. 1b). This finding suggests that at the assemblage level, different parts of the dominance–richness relationship face the same constraints as the relationship at baits, with the ascending part due to both environmental stress and particular abundance frequency distributions, while the descending part is due to interspecific competition (see Parr et al. 2005). This is most noteworthy because it demonstrates that ant behaviour at baits and at the assemblage level can produce a similar outcome, irrespective of scale. The importance of competition as the causal mechanism reducing species richness is further suggested at Satara where, under a favourable temperature regime, the proportion of ant species at baits relative to the species pool decreases as the abundance of dominant ant species increases (Fig. 2b).

Alternative explanations and mechanisms to that of competitive exclusion resulting in reduced assemblage richness when abundances of dominant ants are very high are hard to reconcile. Top-down control by predation is not a satisfactory alternative explanation: not only are there few known predators of ants in this savanna system, but of those that are known, dominant ants with conspicuous nests are the favoured prey, rather than subordinate species [e.g. A. custodiens constitutes an important part of the pangolin's (Manis temminckii Smuts) diet, Swart, Richardson & Ferguson 1999]. The finding that the dominance–richness pattern at baits scales up to the assemblage level, indicates that in this savanna system patterns of diversity and behavioural dominance at baits are likely to reflect broader patterns of foraging activity and changes at the population level. This therefore underscores the importance of competition in structuring local ant assemblages.

When looking at the full unimodal dominance–richness relationship, species richness is influenced both by environmental stress (the ascending portion of the curve) and competition (descending portion) (Andersen 1992). Importantly, with bait data, the wider range of dominance values at each site compared with assemblage level patterns is driven primarily by temperature influencing momentary diversity. Ants are considered thermophilic (Hölldobler & Wilson 1990; Kaspari 2001), with temperature an important influence affecting foraging behaviour (Cerdáet al. 1997). When ground surface temperatures are considered physiologically high for ants (i.e. above 36 °C), typically at midday, the majority of ant species reduce foraging activity (C. L. Parr, personal observation). Given baiting was conducted at different times throughout the day (including midday) to cover a range of temperatures, it is not therefore surprising that a negative linear relationship between abundance of dominant ants and percentage of species at baits relative to the species was not found at Satara, and instead a unimodal relationship is apparent (Fig 2). High temperature acts as a significant stress factor reducing the abundance of all ants at baits (including dominant species), and as a result, species richness declines (Fig. 2b). Under more favourable temperature regimes, dominant species can control species richness through competitive exclusion at baits (descending part of relationship), as is evidenced by the data from Satara (Fig. 2b). At Mopane and Pretoriuskop, weather conditions during baiting meant air temperatures, and consequently ground surface temperatures were slightly lower than Satara. This resulted in few baiting periods with stressful conditions for ants, and consequently the full unimodal relationship was not apparent (Figs 1 and 2).

At the assemblage level however, the dominance–richness pattern was slightly different with the different savanna sites making up different parts of the unimodal dominance–richness pattern: low and ascending at Mopane (indicative of a stressful habitat, see Andersen 1992; Parr et al. 2005), ascending at Pretoriuskop, and descending at Satara (Fig. 1b). The importance of competition in structuring assemblage richness within each site therefore varied considerably. At the assemblage level, control of species richness by dominant ants was restricted to the Satara Acacia savanna where very high abundances of dominant ants resulted in low assemblage richness (Figs 1b and 2b). In this regard, co-occurrence analyses indicating nonrandom co-occurrence patterns in Satara provide further support for competition as a key mechanism structuring ant assemblages in this savanna habitat (see Gotelli & Ellison 2002) (Table 3). It should be noted that previous work at the same study sites (Parr et al. 2004) demonstrated that the nonrandom co-occurrence patterns found at the plot level in Satara cannot be attributed to differences in fire history between the plots (the plots represent several different burning treatments as part of a long-term fire experiment with several burning treatments).

In contrast to Satara, the low dominance levels at Mopane are likely to be related to local environmental stresses, such as reduced food resources and limited nest availability due to heavy clay soils (maximum 23% clay content, mean > 18% clay; Mills & Fey 2004) which reduce ant abundance (see Andersen 2000); the reduced importance of competition was further evidenced through random co-occurrence patterns (Table 3). In less favourable sites, competition therefore does not appear to be an important factor in structuring ants at the assemblage level, and patterns at baits do not scale up.

Because the importance of competition in structuring assemblages varies with habitat type, it should not, therefore, automatically be assumed that the dominance–richness relationship found here with assemblage level data occurs in other locations. In order to detect the full unimodal relationship between dominant ants and species richness, it is likely that a wide range of environments need to be surveyed (e.g. from low- to high-productivity sites, from stressful to more favourable sites), and partial sampling may only produce part of the full relationship (e.g. the ascending portion of the curve). Indeed, limited sampling (spatially or temporally) may explain why in some previous studies, the regulation of richness by dominant ants has been found at baits but not at the assemblage level. For example, Andersen & Patel (1994) showed that northern meat ants, Iridomyrmex sanguineus Forel, reduced ant abundance and species richness at baits but similar patterns could not be detected at the assemblage level, possibly because of the short-term nature of the study (< 2 months). Likewise, where exclusion experiments remove only one dominant ant species, when there may be several dominant species in a system, the effects of competition at the assemblage level may be masked (e.g. Gibb & Hochuli 2004).

Further work is needed to elucidate the factors that determine under what conditions dominant ants are able to play a key role in determining species richness and assemblage structure, and those conditions where other factors play a more important role regardless of the abundance of dominant ants. For example, although theory and field experiments suggest that dominant species increase where resources are abundant and resource quality is high (Davidson 1997; Palmer 2003; MacNally & Timewell 2005), do the resultant high numbers of dominant ants always exert a competitive influence? Finally, given that both natural and anthropogenic disturbance can promote dominant ant species (Gibb & Hochuli 2003; Hoffmann & Andersen 2003), what might the implications of this be for processes such as ecological restoration?


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Thanks to Alan Andersen, Heloise Gibb and Lori Lach for their comments on earlier drafts of this manuscript. Fieldwork assistance by members of the Department of Zoology, University of Pretoria was greatly appreciated. Identification assistance was provided by Hamish Robertson. I am also very grateful to Steven Chown for the support offered while undertaking fieldwork. Two anonymous reviewers provided constructive comments. Funding was provided through the Office of the Director of Research, University of Pretoria, and the Trapell Fund, University of Oxford. South African National Parks provided substantial logistical support.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Abensperg-Traun, M. & Steven, D. (1995) The effects of pitfall trap diameter on ant species richness (Hymenoptera: Formicidae) and species composition of the catch in semi-arid eucalypt woodland. Australian Journal of Ecology, 20, 282287.
  • Andersen, A.N. (1992) Regulation of ‘momentary’ diversity by dominant species in exceptionally rich ant communities of the Australian seasonal tropics. American Naturalist, 140, 401420.
  • Andersen, A.N. (1997) Functional groups and patterns of organisation in North American ant communities: a comparison with Australia. Journal of Biogeography, 24, 433460.
  • Andersen, A.N. (2000) Global ecology of rainforest ants: functional groups in relation to environmental stress and disturbance. Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D.Agosti, J.D.Majer, L.E.Alonso & T.R.Schultz), pp. 2534. Smithsonian Institution Press, Washington, DC.
  • Andersen, A.N. & Patel, A.D. (1994) Meat ants as dominant members of Australian ant communities: an experimental test of their influence on the foraging success and forager abundance of other species. Oecologia, 98, 1524.
  • Andersen, A.N., Blum, M.S. & Jones, T.M. (1991) Venom alkaloids in Monomorium rothsteini’ Forel repel other ants: is this secret to success by Monomorium in Australian ant communities? Oecologia, 88, 157160.
  • Bestelmeyer, B.T. (2000) The trade-off between thermal tolerance and behavioural dominance in subtropical South American ant community. Journal of Animal Ecology, 69, 9981009.
  • Cerdá, X., Retana, J. & Cros, S. (1997) Thermal disruption of transitive hierarchies in Mediterranean ant communities. Journal of Animal Ecology, 66, 363374.
  • Cerdá, X., Retana, J. & Manzaneda, A. (1998) The role of competition by dominants and temperature in the foraging of subordinate species in Mediterranean ant communities. Oecologia, 117, 404412.
  • Davidson, D.W. (1997) The role of resource imbalances in the evolutionary ecology of tropical arboreal ants. Biological Journal of the Linnean Society, 61,153181.
  • Davidson, D.W. (1998) Resource discovery versus resource domination in ants: a functional mechanism for breaking the trade-off. Ecological Entomology, 23, 484490.
  • Fellers, J.H. (1987) Interference and exploitation in guild of woodland ants. Ecology, 68, 14661478.
  • Fox, B.J., Fox, M.D. & Archer, E. (1985) Experimental confirmation of competition between two dominant species of Iridomyrmex (Hymenoptera: Formicidae). Australian Journal of Ecology, 10, 105110.
  • Gaston, K.J. (2000) Global patterns in biodiversity. Nature, 405, 220227.
  • Gertenbach, W.P.D. (1983) Landscapes of the Kruger National Park. Koedoe, 26, 921.
  • Gibb, H. & Hochuli, D.F. (2003) Colonisation by dominant ant facilitated by anthropogenic disturbance: effects on ant assemblage composition, biomass and resource use. Oikos, 103, 469478.
  • Gibb, H. & Hochuli, D.F. (2004) Removal experiment reveals limited effects of behaviorally dominant species on ant assemblages. Ecology, 85, 648657.
  • Godfray, H.C.J. & Lawton, J.H. (2001) Scale and species numbers. Trends in Ecology & Evolution, 16, 400404.
  • Gotelli, N.J. (1996) Ant community structure: effects of predatory ant lions. Ecology, 77, 630638.
  • Gotelli, N.J. (2000) Null model analysis of species co-occurrence patterns. Ecology, 81, 26062621.
  • Gotelli, N.J. & Ellison, A.M. (2002) Assembly rules for New England ant assemblages. Oikos, 99, 591599.
  • Gotelli, N.J. & Entsminger, G.L. (2001) Swap and fill algorithms in null model analysis: rethinking the knight's tour. Oecologia, 129, 281291.
  • Gotelli, N.J. & Graves, G.R. (1996) Nulls Models in Ecology. Smithsonian Institute Press, Washington, DC.
  • Hairston, N.G. (1981) An experimental test of guild: salamander competition. Ecology, 62, 6572.
  • Hoffmann, B.D. & Andersen, A.N. (2003) Responses of ants to disturbance in Australia, with particular reference to functional groups. Austral Ecology, 28, 444464.
  • Hölldobler, B. & Wilson, E.O. (1990) The Ants. Springer, Berlin, Heidelberg, New York.
  • Kaspari, M. (2001) Taxonomic level, trophic biology and the regulation of local abundance. Global Ecology & Biogeography, 10, 229244.
  • LeBrun, E.G. & Feener, D.H. (2002) Linked indirect effects in ant-phorid interactions: impacts on ant assemblage structure. Oecologia, 133, 599607.
  • Low, B. & Rebelo, A.G. (1996) Vegetation of South Africa, Lesotho and Swaziland: A Companion to the Vegetation Map of South Africa, Lesotho and Swaziland. Department of Environmental Affairs and Tourism, Pretoria, South Africa.
  • MacNally, R. & Timewell, C.A.R. (2005) Resource availability controls bird-assemblage composition through interspecific aggression. The Auk, 122, 10971111.
  • Majer, J.D., Delabie, J.H.C. & Smith, M.R.B. (1994) Arboreal ant community patterns in Brazilian cocoa farms. Biotropica, 23, 173181.
  • Marsh, A.C. (1984) The efficacy of pitfall traps for determining the structure of desert ant community. Journal of the Entomological Society of Southern Africa, 47, 115120.
  • Mills, A.J. & Fey, M.V. (2004) Frequent fires intensify soil crusting: physicochemical feedback in the pedoderm of long-term burn experiments in South Africa. Geoderma, 121, 4564.
  • Morrison, L.W. (1996) Community organisation in recently assembled fauna: the case of Polynesian ants. Oecologia, 107, 243256.
  • Palmer, T.M. (2003) Spatial habitat heterogeneity influences competition and coexistence in an African acacia ant guild. Ecology, 84, 28432855.
  • Parr, C.L. & Chown, S.L. (2001) Inventory and bioindicator sampling: testing pitfall and winkler methods with ants in a South African savanna. Journal of Insect Conservation, 5, 2736.
  • Parr, C.L., Robertson, H.G., Biggs, H.C. & Chown, S.L. (2004) Response of African savanna ants to long-term fire regimes. Journal of Applied Ecology, 41, 630642.
  • Parr, C.L., Sinclair, B.J., Andersen, A.N., Gaston, K.J. & Chown, S.L. (2005) Constraint and competition in assemblages: cross-continental and modeling approach for ants. American Naturalist, 165, 481494.
  • Perfecto, I. & Vandermeer, J. (1996) Microclimate changes and the indirect loss of ant diversity in tropical agroecosystem. Oecologia, 108, 577582.
  • Retana, J. & Cerdá, X. (2000) Patterns of diversity and composition of Mediterranean ground ant communities tracking spatial and temporal variability in the thermal environment. Oecologia, 123, 436444.
  • Ribas, C.R. & Schoereder, J.H. (2002) Are all ant mosaics caused by competition? Oecologia, 131, 606611.
  • Ricklefs, R.E. (1987) Community diversity: relative roles of local and regional processes. Science, 235, 167171.
  • Room, P.M. (1971) The relative distributions of ant species in Ghana's cocoa farms. Journal of Animal Ecology, 40, 735751.
  • Sanders, N.J. & Gordon, D.M. (2000) The effects of interspecific interactions on resource use and behaviour in desert ant. Oecologia, 125, 436443.
  • Savolainen, R. & Vepsäläinen, K. (1988) Competition hierarchy among boreal ants: impact on resource partitioning and community structure. Oikos, 51, 135155.
  • Stone, L. & Roberts, A. (1990) The checkerboard score and species distributions. Oecologia, 85, 7479.
  • Swart, J.M., Richardson, P.R.K. & Ferguson, J.W.H. (1999) Ecological factors affecting the feeding behaviour of pangolins (Manis temminckii). Journal of Zoology, 247, 281292.
  • Trollope, W.S.W., Potgieter, A.L.F., Biggs, H.C. & Trollope, L.A. (1998) Report on the experimental burning plots (EBP) trial in the major vegetation types of the Kruger National Park. South African National Parks Report No. 9/98, Skukuza, South Africa.
  • Van Wilgen, B.W., Biggs, H.C., O’Regan, S.P. & Marè, N. (2000) Fire history of the savanna ecosystems in the Kruger National Park, South Africa, between 1941 and 1996. South African Journal of Science, 96, 167178.
  • Weiher, E. & Keddy, P. (1999) Ecological Assembly Rules: Perspectives, Advances, Retreats. Cambridge University Press, Cambridge, UK.