Skip to main content
  • Research article
  • Open access
  • Published:

Competitive assembly of South Pacific invasive ant communities

Abstract

Background

The relative importance of chance and determinism in structuring ecological communities has been debated for nearly a century. Evidence for determinism or assembly rules is often evaluated with null models that randomize the occurrence of species in particular locales. However, analyses of the presence or absence of species ignores the potential influence of species abundances, which have long been considered of major importance on community structure. Here, we test for community assembly rules in ant communities on small islands of the Tokelau archipelago using both presence-absence and abundance data. We conducted three sets of analyses on two spatial scales using three years of sampling data from 39 plots on 11 islands.

Results

First, traditional null model tests showed support for negative species co-occurrence patterns among plots within islands, but not among islands. A plausible explanation for this result is that analyses at larger spatial scales merge heterogeneous habitats that have considerable effects on species occurrences. Second, analyses of ant abundances showed that samples with high ant abundances had fewer species than expected by chance, both within and among islands. One ant species, the invasive yellow crazy ant Anoplolepis gracilipes, appeared to have a particularly strong effect on community structure correlated with its abundance. Third, abundances of most ant species were inversely correlated with the abundances of all other ants at both spatial scales. This result is consistent with competition theory, which predicts species distributions are affected by diffuse competition with suites of co-occurring species.

Conclusion

Our results support a pluralistic explanation for ant species abundances and assembly. Both stochastic and deterministic processes interact to determine ant community assembly, though abundance patterns clearly drive the deterministic patterns in this community. These deterministic patterns were observed at two spatial scales. Results indicate that abundance-based null models may be more sensitive in detecting non-random patterns in community assembly than species co-occurrences analyses.

Background

Opinions on the processes governing community assembly are polarized [1–3]. One school of thought maintains that ecological communities are assembled deterministically according to 'assembly rules', generated by biotic interactions [4–6]. An opposing view asserts that communities are assembled stochastically by chance dispersal of species with life history characteristics suited to local environmental conditions [7–9]. The relative importance of chance and determinism in structuring ecological communities has been fiercely debated for nearly a century and we are far from a resolution to the debate.

Diamond [5] hypothesized that interspecific interactions result in community assembly rules. He suggested that some pairs of species never coexist, either by themselves or as part of a larger combination. Evidence for assembly rules are often evaluated with null models that randomize the occurrence of species in particular locales. However, analyses of the presence or absence of species ignore the potential influence of species abundances, which have long been considered of major importance on community structure [10–13]. Abundance effects may be particularly important in determining the distribution of invasive species [14]. Less abundant invasive species likely have weaker ecological effects on community assembly than more abundant species. Nevertheless, the effects of species abundances have largely been ignored when testing for community assembly rules.

Patterns in community structure, and their underlying mechanisms, can differ among spatial scales [15], as can our understanding of the invasiveness of introduced species. For example, investigations over larger spatial scales often incorporate a variety of habitats, some of which may not be suitable for an invasive species and thus provide refuges for native species. Despite evidence that community assembly rules can be scale-dependent [16, 17], null model tests for community assembly rules are typically assessed at single scales [18].

Here, we test for community assembly rules in ant communities on islands in the South Pacific, by conducting abundance-based null model analyses at two spatial scales. Ant communities on Islands in the South Pacific are often composed of introduced species, which are an ideal group for the study of processes governing community assembly. Interspecific aggression [19], chemical warfare [20], dominance hierarchies and competitive displacement [21–23] have been documented in numerous ant species inhabiting different parts of the globe. In addition, many invasive ants are considered superior competitors [24]. Most islands in the South Pacific are volcanic in origin, so their flora and fauna are derived from over-water dispersal. Unlike most other island colonists, most ant species have colonized islands comparatively recently, with the aide of human travellers, and the process of island colonization and ant community assembly continues [25]. Because South Pacific ant communities are often comprised almost entirely of introduced species, they provide an ideal opportunity to study how ecological communities are assembled. Unfortunately, they might also provide a template for the future of many other ecological communities, as invasive species continue their global spread.

We censused ant communities in 39 plots on 11 islands and used null models to test for community assembly rules on two spatial scales (i.e. within and among islands). First, we randomised the presence of ant species among samples to test for negative co-occurrence patterns. Second, we randomised the presence of individual ants among samples to test whether ant species richness was inversely related to total ant abundances. Third, we randomised the presence of individual ants among samples to test whether the abundances of each ant species was inversely related to the abundance of all other ant species.

Methods

Study site, species and field sampling

Tokelau lies 483 km north of Samoa in the Pacific Ocean (approximately 9°45'S, 171°35'W) and is comprised of three low-lying coral atolls: Atafu, Nukunonu and Fakaofo. The atolls are 50–100 km apart and each is made up of 38–51 islands surrounding a shallow lagoon, one or two of which are permanently inhabited by people on each atoll (Fig. 1). The islands are generally small, with Tokelau's total land area being approximately 12 km2. Tokelau lies in the southeast trade wind belt and has a humid tropical climate that displays little seasonal variation (mean annual temperature 28°C, mean annual rainfall 3000 mm [26]). The islands of Tokelau are comprised of coral rubble of varying size with poorly developed soil overlying beach rock (Parham 1971). The islands are low-lying (~5 m above sea level) and narrow. The vegetation is low in diversity and typical of small Pacific atolls [26, 27].

Figure 1
figure 1

Tokelau and the islands used in our study. (A) A map of the South Pacific showing the relative location of Tokelau. (B) A map of Nukunonu Atoll showing the location of the 11 islands used in our study. (C) A photograph of Fakaofu Atoll of Tokelau. All three Tokelau Atolls are similar, with only one or two inhabited islands and many uninhabited islands such as in the background of this picture. The inset picture is of the yellow crazy ant (Anoplolepis gracilipes) consuming a dead hermit crab (Coenobita sp.).

Ant communities on islands in the Southern Pacific Ocean typically have few native species and are often comprised entirely of species that were brought their by human travellers [25]. Only 28 ant species have been recorded on Tokelau and of those only two are Pacific endemics, with no species endemic to Tokelau [28].

Censuses were conducted during three successive visits to the archipelago (November 2002, November–December 2004 and June–July 2005). A standard sampling design was used throughout [29–31]. Survey plots consisted of 15 × 15 m quadrats located at least 40 m apart in forested areas. The number of quadrats per island ranged from between two and nine over the period of 2002–2005. Four to five pitfall traps were haphazardly situated throughout the plot and one-third filled with Gault's solution, which is an insect killing agent and preservative [32]. Placement of traps was haphazard (rather than random) so as to avoid traps being placed beside or on individual ant nests, which could provide biased estimates of ant abundances. The pitfall traps were plastic containers 9 cm tall, 7.5 cm diameter at the top, tapered to 5 cm diameter at the bottom and placed flush with the ground surface. Traps were only left out for 24 hrs due to the speed at which they occasionally accumulated abundant ant species, such as Anoplolepis gracilipes. In total we placed a total of 184 pitfall traps, within 39 plots from 11 islands. Analyses at the "plot" scale refer to the pooled number and abundance of each species from all pitfall traps within each quadrat. Analyses at the "island" scale refer to the pooled number and abundance of each species from all quadrats within each island, sampled over the period of 2002–2005.

Species co-occurrences

To test for negative co-occurrence patterns, we used standard analyses in the freely available software package EcoSim [33]. In this analysis, there were 17 rows of species and either 11 or 39 columns for islands or plots, respectively. Negative co-occurrence patterns were quantified with three indices: the C-score index, the checkerboard score and the number of unique species combinations. The C-score is obtained by calculating the number of checkerboard units cu for each species pair. For example, at the island scale of our analyses: cu = (o i - s)(o j - s), where o i is the total number of islands occupied by species i, o j is the total number of islands occupied by species j, and s is the number of islands occupied by both species [34]. A single C-score, which describes community-level species co-occurrences, is obtained by averaging cu values of all species pairs. The checkerboard score is obtained by tallying the number of species pairs that never co-occur with one another. The number of unique species combinations is a count of all species pairs that co-occurred with one another.

To test for non-random co-occurrence patterns, the observed values of each index were compared to values generated in 5000 simulations of the observed species × plot presence-absence matrix using fixed row and column sums. The sequential swap technique was used to generate random permutations of the observed matrix. Miklós & Podani [35] show that this method biases against support for non-random co-occurrence patterns and provides a more conservative test of the co-occurrence assembly rule. If communities are structured deterministically according to assembly rules, observed communities should have higher C-scores, fewer unique species combinations, and more checkerboard species pairs than expected under the null model. Among-island and within-island analyses were conducted separately.

Species richness

Competition theory predicts that as the total number of individuals in an area increases, competition for resources leads to the exclusion of competitively subordinate species [10]. We tested this prediction by evaluating whether areas with higher numbers of ants had lower numbers of ant species. To test for an inverse relationship between species richness and overall ant abundances, we began by obtaining expected values of species richness for each sample locale (i.e. plot and island). Expected values of species richness were obtained using rarefaction, which estimates species richness on a per-individual, rather than a per-area, basis [2, 36, 37]. To generate rarefaction curves we used the programme RAREFACT 1.0 [38]. Different numbers of ants were randomly sampled from the total pool encountered during sampling using this computer simulation, which estimates species richness for a given number of individual ants sampled. Separate rarefaction curves were generated for each spatial scale. The total pool of ants sampled across the archipelago was used to calculate expected species richness values for each island and the total pool found on each island was used to calculate expected species richness values for each plot.

This application of rarefaction yields expected values of species richness if individuals were randomly distributed among samples. Differences between observed (O) and expected (E) species richness values (O - E) therefore indicate whether samples sites have higher or lower richness than expected by chance. If ants compete for resources, sites with greater total ant abundances will contain fewer species than expected, resulting in negative O - E values. Conversely, sites with lower overall ant abundances will contain more species than expected, resulting in positive O - E values. However, differences between observed and expected values increase passively with the magnitude of expected values. To remove this confounding effect we divided the difference by expected values [(O-E)E-1], which results in an unbiased estimates of deviations from expected species richness values.

General linear models were used to establish whether differences from expected species richness varied with total ant abundances. Standardised differences from expected species richness values were included in the model as a dependent variable. Total ant abundances were treated as a covariate. Separate analyses were conducted for each scale (i.e. plot or island). In the plot-scale analysis, island was included in the model as a random factor to account for the independence problem generated by multiple plots occurring on the same island. Data were log10 transformed to conform to normality and homoscedasticity assumptions at the island-scale. Data conformed to assumptions without transformation at the plot-scale. Analyses were conducted in SPSS [39].

Abundances

We generalized linear models to test whether the abundance of each ant species declines with the abundances of all other ant species. Observed abundances of each species were used as the dependent variable. The sum of all other ant species present in each sample (i.e. plots or islands) was treated as a covariate. The expected abundance of each species in each sample was treated as a second covariate (E ij ). To calculate the expected abundance (E) of each species (i) in each sample (j), we multiplied the total abundance (A) of that species ( ∑ j = 1 n A i MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaWaaabCaeaacqWGbbqqdaWgaaWcbaGaemyAaKgabeaaaeaacqWGQbGAcqGH9aqpcqaIXaqmaeaacqWGUbGBa0GaeyyeIuoaaaa@3556@ ) by the abundance of all ant species in that sample ( ∑ i = 1 n A j MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaWaaabCaeaacqWGbbqqdaWgaaWcbaGaemOAaOgabeaaaeaacqWGPbqAcqGH9aqpcqaIXaqmaeaacqWGUbGBa0GaeyyeIuoaaaa@3556@ ). This product was then divided by the total number of all ants found at that scale [( ∑ i = 1 n ∑ j = 1 n A i j MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaWaaabCaeaadaaeWbqaaiabdgeabnaaBaaaleaacqWGPbqAcqWGQbGAaeqaaaqaaiabdQgaQjabg2da9iabigdaXaqaaiabd6gaUbqdcqGHris5aaWcbaGaemyAaKMaeyypa0JaeGymaedabaGaemOBa4ganiabggHiLdaaaa@3DAB@ )] to obtain:

E i j = ( ∑ j = 1 n A i ) ( ∑ i = 1 n A j ) ∑ i = 1 n ∑ j = 1 n A i j MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaemyrau0aaSbaaSqaaiabdMgaPjabdQgaQbqabaGccqGH9aqpjuaGdaWcaaqaaiabcIcaOmaaqahabaGaemyqae0aaSbaaeaacqWGPbqAaeqaaaqaaiabdQgaQjabg2da9iabigdaXaqaaiabd6gaUbGaeyyeIuoacqGGPaqkcqGGOaakdaaeWbqaaiabdgeabnaaBaaabaGaemOAaOgabeaaaeaacqWGPbqAcqGH9aqpcqaIXaqmaeaacqWGUbGBaiabggHiLdGaeiykaKcabaWaaabCaeaadaaeWbqaaiabdgeabnaaBaaabaGaemyAaKMaemOAaOgabeaaaeaacqWGQbGAcqGH9aqpcqaIXaqmaeaacqWGUbGBaiabggHiLdaabaGaemyAaKMaeyypa0JaeGymaedabaGaemOBa4gacqGHris5aaaaaaa@59A4@

A significant effect of expected ant abundances would suggest stochastic processes shape community assembly. Support for community assembly rules would require negative relationships between the observed abundance of each species and the summed abundance of all other ant species. Ant species was included as a random factor in the island-scale analysis to account for the independence problem associated with including more than one value for each species. Both ant species and island were included as random factors in the plot-scale analysis to account for the independence problem associated with including more than one value for each species and each island. Separate analyses were conducted at each scale using the generalized linear model procedure with a Poisson distribution in SPSS [39].

Results

Seventeen species were found over the three years of sampling (Table 1). Total ant abundances varied over three orders of magnitude (5–5,000 ants) among plots. One ant species (Anoplolepis gracilipes) was particularly abundant. It accounted for an average of 50% of all ants sampled among islands and variation in total ant abundance was associated with A. gracilipes abundance. The total number of ants observed was highly correlated with the total number of A. gracilipes at both the quadrat (Pearson's r = 0.940;P < 0.001) and island scales (Pearson's r = 0.914;P < 0.001).

Table 1 Ant species with abundances found in plots and islands.

Species co-occurrences

Patterns of species co-occurrences were scale-dependent. No evidence for co-occurrence assembly rules were found at the analysis scale of among islands. The observed C-score (4.368) was not statistically different from null model expectations among islands (4.369, 0.006; mean, σ; P = 0.485). Similar results were obtained for the number of checkerboard species pairs (observed = 13, expected = 15.997, 9.985; P = 0.882) and the number of unique species combinations (observed = 11, expected = 10.999, 0.028; P = 0.998). In contrast, within-island analyses showed evidence for co-occurrence assembly rules. The observed c-score (41.007) was significantly greater than null model expectations (40.214, 0.151; P= 0.023). While the observed number of checkerboard species pairs (32) did not significantly differ from expected numbers (34.099, 11.918; P = 0.760), the observed number of unique species combinations (35) was lower than null model expectations (39.623, 0.376; P < 0.001).

Species richness

Observed ant species richness deviated markedly from expected richness values generated by rarefaction analyses at both scales. Differences between observed and expected species richness values were associated with total ant abundances. At the among-island scale, deviations from expected species richness values were negatively correlated with total ant abundances (F1,9 = 31.660, P < 0.001) (Figs. 2A &2B). At the within-island scale, deviations from expected values of species richness were again negatively correlated with total ant abundances (F1,27 = 13.330, P = 0.001) (Figs. 2C &2D). Therefore, samples (i.e. plots and islands) with large numbers of ants had fewer numbers of species than expected under the null model.

Figure 2
figure 2

Patterns in species richness of the Tokelau ant fauna. On the left, total species richness is plotted against the number of individuals sampled. Each point represents a single sampling locale. Solid lines are rarefaction curves reflecting relationships between cumulative species richness and the number of individuals encountered among samples and dashed lines are 95% confidence intervals. On the right, deviations from expected species richness vales [(O-E)E-1] are plotted against total ant densities. Among island analyses are shown on top (N = 11). Among plot analyses are shown on bottom (N = 39). The rarefaction curve for the island scale is shown for both island and plot graphs for aesthetic purposes, rather than calculating many individual, plot scale, rarefaction curves.

Abundances

Ant abundances were positively associated with null model expectations and negatively associated with the abundances of co-occurring species on both spatial scales. Analysis of ant abundances among islands showed that after accounting for species (Wald X2 = 11,925, df = 16, P < 0.001), abundances increased with expected values (Wald X2 = 52,802, df = 1, P < 0.001) (Fig. 3A) and decreased with the summed abundances of all other co-occurring species (Wald X2 = 342, df = 1, P < 0.001) (Fig. 3B). Similar results were found in within-island analysis. After accounting for both species (Wald X2 = 15,172, df = 16, P < 0.001) and island (Wald X2 = 9,516, df = 10, P < 0.001), abundances increased with expected abundance values (Wald X2 = 22,177, df = 1, P < 0.001) (Fig. 3C) and decreased with the summed abundances of all other co-occurring species (Wald X2 = 96, df = 1, P < 0.001) (Fig. 3D).

Figure 3
figure 3

Patterns of ant abundances in the Tokelau archipelago. The abundance of each species within each sample is plotted against null model expectations (left) and the total number of co-occurring ants (right). Among island analyses are shown on top (N = 11). Among plot analyses are shown on bottom (N = 39). On both spatial scales ant abundances covary with randomized distributions, and are also negatively associated with the abundance of all other species.

Discussion

The ant communities in the Tokelau Archipelago appear to be assembled deterministically. However, evidence for community assembly rules differed between abundance-based and community-based analyses. Species co-occurrences were distributed at random among islands, but evidence for negative co-occurrence patterns were observed within islands. The number of species present in samples at both spatial scales differed from randomized expectations; differences between observed and expected species richness values were negatively correlated with total ant abundances, indicating that areas with high ant abundances supported a reduced number of ant species. Population abundances of most ant species were correlated with both randomized expectations and the summed abundance of all other ant species. Therefore, consistent support for community assembly rules were found in abundance based analyses, but scale dependent results were found in analyses of species co-occurrences.

Co-occurrence analyses showed variable support for community assembly rules. Within islands, species tended to co-occur less frequently than expected by chance. Negative co-occurrence patterns are common in community ecology [40] and are commonly attributed to inter-specific competition. However, our results differ substantially from other analyses involving other invasive ants. Both red imported fire ants (Solenopsis invicta) and Argentine ants (Linepithema humile) have caused the disassembly of ant communities in other locales [17, 41]. Our study involved several invasive species including A. gracilipes, which is thought to have similar effects on ant communities as S. invicta and L. humile [24]. Conversely, between island analyses indicated that species co-occurrences were distributed at random. This result is consistent with several other studies which found that community assembly rules can be scale-dependent [16, 42]. Spatial scale thus appears to have a strong influence on the outcome of co-occurrence patterns. While there may be any number of hypotheses for this result, a plausible explanation is that analyses at larger spatial scales merge heterogeneous habitats that have considerable effects on species occurrences. Different species may have different habitat requirements. Determining the appropriate scale for co-occurrence analysis is probably community and habitat specific, and represents a difficult task.

Species richness analyses showed that the number of species present in samples declined with the total number of ants present. This result is consistent with theoretical predictions concerning how groups of competing species interact. Competition from groups of similar species may increase local extinction rates and decrease local colonization rates [5, 43]. Islands and plots with large numbers of ants housed fewer species than expected by chance, whereas samples with smaller numbers of ants housed more species. Although several previous studies have documented similar patterns in island communities (see Burns [18]; and references therein) null model analyses that attempt to link community level patterns of abundance to species are rare. Therefore, the generality of the abundance-species richness relationship observed here remains unclear.

Rarefaction can overestimate species richness when spatial distributions are clumped [44]. If species interactions (e.g., competition) cause the spatial segregation of species distributions, the number of species present in samples should be less than predicted by rarefaction. We found that deviations from expected richness were negatively related to total ant abundances, suggesting that densely populated areas are subject to a greater influence of deterministic processes.

Patterns of the abundance of most ant species were negatively associated with the abundances of all other species. This result is consistent with competition theory, which predicts species distributions are affected by diffuse competition with suites of co-occurring species [5]. Patterns of ant abundances were also associated with null model expectations, indicating a stochastic component to ant community assembly. Random dispersal events and/or stochastic population dynamics may therefore help to determine ant abundance patterns, supporting recent work highlighting the effects of 'neutral' processes on spatial patterns of biodiversity [9]. When both relationships are viewed jointly, overall results support a pluralistic explanation for ant population abundances. Under this view, both stochastic and deterministic processes interact to determine ant community assembly. A similar conclusion has been reached by others examining ants on islands (e.g. Cole [19]), though a novel feature of our results is how abundance clearly drives deterministic patterns in this community.

Why patterns indicative of deterministic processes were consistently observed in abundance-based analyses, yet inconsistently in analyses of species co-occurrences is unclear. Perhaps the qualitative nature of binary data was not sensitive enough to detect evidence for assembly rules at all scales. Suites of analyses across a variety of scales using both quantitative and qualitative data may provide more accurate tests for assembly rules.

One ant species, the yellow crazy ant Anoplolepis gracilipes, appeared to have a particularly strong effect on community structure. When present in low abundances it appeared to have little to no effect on ant communities. However, in high abundance, A. gracilipes was associated with reductions in the number of co-occurring species and their abundance. Similarly, on Christmas Island A. gracilipes has its strongest influence on the communities when in high densities [45]. High abundances may help invasive ants such as A. gracilipes obtain resources through numerical superiority, but are also associated with increased aggression [46]. Our experimental work on these islands has demonstrated that when in high abundance, A. gracilipes often dominates preferred resources and only those species with different foraging modes or food preference can co-occur with them [31, 47]. Spatial variation in the distribution of such dominant species also likely creates room for subordinate species, and the establishment of these subordinate species may have a large stochastic element [48]. Whatever the mechanisms for co-occurrence or exclusion, our results are consistent with competition theory, which predicts that interspecific competition and community structuring is abundance-dependent.

Conclusion

Overall results showed that ant communities in the Tokelau archipelago are assembled deterministically. Although some support for stochastic processes was observed, most of our results are consistent with the hypothesis that competition structures community assembly, both within and among islands. Because South Pacific ant communities are often comprised entirely of introduced species, they may provide a template for the future, given the continuing global spread of invasive species.

References

  1. Roughgarden J: The structure and assembly of communities. Perspectives in ecological theory. Edited by: Roughgarden J, May RM, Levin SA. 1989, Cambridge: Cambridge University Press, 203-226.

    Google Scholar 

  2. Gotelli NJ, Graves GR: Null models in ecology. 1996, Washington DC: Smithsonian Press

    Google Scholar 

  3. Crawley MJ: The structure of plant communities. Plant Ecology. Edited by: Crawley MJ. 1997, Oxford: Blackwell Scientific Publications, 475-531.

    Google Scholar 

  4. Clements FE: Plant succession: an analysis of the development of vegetation. Publication of the Carnegie Institute, Washington. 1916, 242: 1-512.

    Google Scholar 

  5. Diamond JM: Assembly of species communities. Ecology and evolution of communities. Edited by: Cody ML, Diamond JM. 1975, Cambridge: Harvard University Press, 342-444.

    Google Scholar 

  6. Silvertown J, Poulton P, Johnston E, Edwards G, Heard M, Biss PM: The Park Grass Experiment 1856–2006: Its contribution to ecology. Journal of Ecology. 2006, 94: 801-814. 10.1111/j.1365-2745.2006.01145.x.

    Article  CAS  Google Scholar 

  7. Gleason HA: The structure and development of the plant association. Bulletin of the Torrey Botanical Club. 1917, 44: 463-481. 10.2307/2479596.

    Article  Google Scholar 

  8. Connor EF, Simberloff D: The assembly of species communities: chance or competition?. Ecology. 1979, 60: 1132-1140. 10.2307/1936961.

    Article  Google Scholar 

  9. Hubbell SP: The unified neutral theory of biodiversity and biogeography. 2001, Princeton: Princeton University Press

    Google Scholar 

  10. Lack D: The natural regulation of animal numbers. 1954, London: Oxford University Press

    Google Scholar 

  11. Sutherland J: From Individual Behaviour to Population Ecology. 1996, Oxford: Oxford University Press

    Google Scholar 

  12. Harte J, Conlisk E, Ostling A, Green JL, Smith AB: A theory of spatial structure in ecological communities at multiple spatial scales. Ecological Monographs. 2005, 75: 179-197. 10.1890/04-1388.

    Article  Google Scholar 

  13. Ulrich W, Zalewski M: Abundance and co-occurrence patterns of core and satellite species of ground beetles on small lake islands. Oikos. 2006, 114: 338-348. 10.1111/j.2006.0030-1299.14773.x.

    Article  Google Scholar 

  14. Colautti RI, MacIsaac HJ: A neutral terminology to define 'invasive' species. Diversity and Distributions. 2004, 10: 135-141. 10.1111/j.1366-9516.2004.00061.x.

    Article  Google Scholar 

  15. Levin SA: The problem with pattern and scale in ecology. Ecology. 1992, 73: 1943-1967. 10.2307/1941447.

    Article  Google Scholar 

  16. Gotelli NJ, Ellison AM: Assembly rules for New England ant assemblages. Oikos. 2002, 99: 591-599. 10.1034/j.1600-0706.2002.11734.x.

    Article  Google Scholar 

  17. Sanders NJ, Gotelli NJ, Heller NE, Gordon DM: Community disassembly by an invasive species. Proceedings of the Natural Academy of Sciences, USA. 2003, 100: 2474-2477. 10.1073/pnas.0437913100.

    Article  CAS  Google Scholar 

  18. Burns KC: Patterns in the assembly of an island plant community. Journal of Biogeography. 2007, 34: 760-768. 10.1111/j.1365-2699.2006.01625.x.

    Article  Google Scholar 

  19. Cole BJ: Assembly of mangrove ant communities: patterns of geographical distribution. Journal of Animal Ecology. 1983, 52: 339-347. 10.2307/4557.

    Article  Google Scholar 

  20. Andersen AN, Blum MS, Jones TH: Venom alkaloids in Monomorium "rothsteini" Forel repel other ants: is this the secret to success by Monomorium in Australian ant communities. Oecologia. 1991, 88: 157-160. 10.1007/BF00320805.

    Article  Google Scholar 

  21. Holway DA: Effect of Argentine ant invasions on ground-dwelling arthropods in northern California riparian woodlands. Oecologia. 1998, 116: 252-258. 10.1007/s004420050586.

    Article  Google Scholar 

  22. Palmer TM, Stanton ML, Young TP, Goheen JR, Pringle RM, Karban R: Breakdown of an ant-plant mutualism follows the loss of large herbivores from an African Savanna. Science. 2008, 319: 192-195. 10.1126/science.1151579.

    Article  CAS  PubMed  Google Scholar 

  23. Savolainen P, Vespäläinen K: Niche differentiation of ant species within territories of the wood ant Formica polyctena. Oikos. 1989, 56: 3-16. 10.2307/3566082.

    Article  Google Scholar 

  24. Holway DA, Lach L, Suarez AV, Tsutsui ND, Case TJ: The causes and consequences of ant invasions. Annual Review of Ecology and Systematics. 2002, 33: 181-233. 10.1146/annurev.ecolsys.33.010802.150444.

    Article  Google Scholar 

  25. Wilson EO, Taylor RW: The ants of Polynesia (Hymenoptera: Formicidae). Pacific Insects Monograph. 1967, 14: 1-109.

    Google Scholar 

  26. Mueller-Dombois D, Fosberg FR: Vegetation of the tropical Pacific islands. 1998, New York: Springer-Verlag

    Book  Google Scholar 

  27. Parham BEV: The vegetation of the Tokelau Islands with special reference to the plants of Nukunonu Atoll. New Zealand Journal of Botany. 1971, 9: 576-609.

    Article  Google Scholar 

  28. Abbott KL, Sarty M, Lester PJ: The ants of Tokelau. New Zealand Journal of Zoology. 2006, 33: 157-164.

    Article  Google Scholar 

  29. Lester PJ, Tavite A: Long-legged ants (Anoplolepis gracilipes) have invaded Tokelau, changing the composition and dynamics of ant and invertebrate communities. Pacific Science. 2004, 58: 391-401. 10.1353/psc.2004.0031.

    Article  Google Scholar 

  30. Abbott KL, Greaves SNJ, Ritchie PA, Lester PJ: Behaviourally and genetically distinct populations of an invasive ant provide insight into invasion history and impacts on a tropical ant community. Biological Invasions. 2007, 9: 453-463. 10.1007/s10530-006-9052-2.

    Article  Google Scholar 

  31. Sarty M, Abbott KL, Lester PJ: Community level impacts of an ant invader and food mediated coexistence. Insectes Sociaux. 2007, 54: 166-173. 10.1007/s00040-007-0927-8.

    Article  Google Scholar 

  32. Walker AK, Crosby TK: The preparation and curation of insects. 1988, Auckland: DSIR Information Series 163 Entomology Division

    Google Scholar 

  33. Gotelli NJ, Entsminger GL: EcoSim: Null models software for ecology. Version 7. 2007, Acquired Intelligence Inc. & Kesey-Bear. Jericho, VT 05465, USA, [http://garyentsminger.com/ecosim/index.htm]

    Google Scholar 

  34. Stone L, Roberts A: Competitive exclusion, or species aggregation? An aid in deciding. Oecologia. 1992, 91: 419-424. 10.1007/BF00317632.

    Article  Google Scholar 

  35. Miklós I, Podani J: Randomization of presence-absence matrices: comments and new algorithms. Ecology. 2004, 85: 86-92. 10.1890/03-0101.

    Article  Google Scholar 

  36. Sanders HL: Marine benthic diversity: a comparative study. American Naturalist. 1968, 102: 243-282. 10.1086/282541.

    Article  Google Scholar 

  37. Gotelli NJ, Colwell RK: Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters. 2001, 4: 379-391. 10.1046/j.1461-0248.2001.00230.x.

    Article  Google Scholar 

  38. RAREFACT 1.0. [http://people.hofstra.edu/j_b_bennington/research/paleoecology/rarefact.html]

  39. SPSS: SPSS for Windows, version 16.0.1. 2007, Chicago: Lead Technologies;

    Google Scholar 

  40. Gotelli NJ, McCabe DJ: Species co-occurrence: A meta-analysis of J.M. Diamond's assembly rules model. Ecology. 2002, 83: 2091-2096.

    Article  Google Scholar 

  41. Gotelli NJ, Arnett AE: Biogeographic effects of red fire ant invasion. Ecology Letters. 2000, 3: 257-261. 10.1046/j.1461-0248.2000.00138.x.

    Article  Google Scholar 

  42. Sanders NJ, Gotelli NJ, Wittman SE, Ratchford JS, Ellison AM, Jules ES: Assembly rules for ant communities across spatial scales and habitats. Journal of Biogeography. 2007, 34: 1632-1641. 10.1111/j.1365-2699.2007.01714.x.

    Article  Google Scholar 

  43. Keddy PA: Competition. 1989, Chapman and Hall, London, UK

    Chapter  Google Scholar 

  44. Collins MD, Simberloff D: Rarefaction and non-random spatial dispersion patterns. Environmental and Ecological Statistics. 2008

    Google Scholar 

  45. O'Dowd DJ, Green PT, Lake PS: Invasional 'meltdown' on an oceanic island. Ecology Letters. 2003, 6: 812-817. 10.1046/j.1461-0248.2003.00512.x.

    Article  Google Scholar 

  46. Sagata K, Lester PJ: Behavioural plasticity associated with propagule size, resources, and the invasion success of the Argentine ant Linepithema humile. Journal of Applied Ecology. 2009, 46: 19-27. 10.1111/j.1365-2664.2008.01523.x.

    Article  Google Scholar 

  47. Sarty M, Abbott KL, Lester PJ: Habitat complexity facilitates coexistence in a tropical ant community. Oecologia. 2006, 149: 465-473. 10.1007/s00442-006-0453-9.

    Article  CAS  PubMed  Google Scholar 

  48. Andersen AN: Not enough niches: non-equilibrial processes promoting species coexistence in diverse ant communities. Austral Ecology. 2008, 33: 211-220. 10.1111/j.1442-9993.2007.01810.x.

    Article  Google Scholar 

Download references

Acknowledgements

We thank John Haywood, Nate Sanders, Dan Simberloff and three anonymous reviewers for commenting on this manuscript. This project was funded by the Royal Society of New Zealand Marsden Fund, The Pacific Conservation and Development Trust, the US National Geographic Society, and Victoria University of Wellington.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Philip J Lester.

Additional information

Competing interests

The authors declare that there are no competing interests.

Authors' contributions

PJL designed the study, took part in the sampling, analyzed the data and drafted the manuscript. KLA and MS did the majority of sampling, helped analyze the data and write the manuscript. KCB lead aspects of the data analysis, helped develop the theoretical background and assisted in drafting the manuscript. All authors read and approved the final manuscript.

Authors’ original submitted files for images

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and permissions

About this article

Cite this article

Lester, P.J., Abbott, K.L., Sarty, M. et al. Competitive assembly of South Pacific invasive ant communities. BMC Ecol 9, 3 (2009). https://0-doi-org.brum.beds.ac.uk/10.1186/1472-6785-9-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/1472-6785-9-3

Keywords