The fire ant Solenopsis saevissima and habitat disturbance alter ant communities

The fire ant Solenopsis saevissima and habitat disturbance alter ant communities

Biological Conservation 187 (2015) 145–153 Contents lists available at ScienceDirect Biological Conservation journal homepage: www.elsevier.com/loca...

1MB Sizes 2 Downloads 53 Views

Biological Conservation 187 (2015) 145–153

Contents lists available at ScienceDirect

Biological Conservation journal homepage: www.elsevier.com/locate/biocon

The fire ant Solenopsis saevissima and habitat disturbance alter ant communities Alain Dejean a,b,c,⇑, Régis Céréghino a,b, Maurice Leponce d, Vivien Rossi e, Olivier Roux f, Arthur Compin a,b, Jacques H.C. Delabie g,h, Bruno Corbara i,j a

Université de Toulouse, UPS, INP, Ecolab, 118 route de Narbonne, 31062 Toulouse, France CNRS, Laboratoire Écologie Fonctionnelle et Environnement (UMR-CNRS 5245), 31062 Toulouse, France CNRS, Écologie des Forêts de Guyane (UMR-CNRS 8172), Campus agronomique, BP 316, 97387 Kourou cedex, France d Aquatic and Terrestrial Ecology, Royal Belgian Institute of Natural Sciences, Rue Vautier 29, B-1000 Brussels, Belgium e CIRAD, Biens et services des écosystèmes forestiers tropicaux (CIRAD-UR 105), BP 2572, Yaoundé, Cameroun f IRD, MIVEGEC (IRD 224-CNRS 5290-NUM) Équipe ESV, 911 avenue Agropolis, BP 64501, 34394 Montpellier Cedex 5, France g Laboratório de Mirmecologia, CEPEC-CEPLAC, 45600-970 Itabuna, BA, Brazil h UESC-DCAA, 45662-900 Ilhéus, BA, Brazil i CNRS, Laboratoire Microorganismes, Génome et Environnement (UMR-CNRS 6023), Université Blaise Pascal, Complexe Scientifique des Cézeaux, 63177 Aubière cedex, France j Université Clermont Auvergne, Université Blaise Pascal (LMGE), 63000 Clermont-Ferrand, France b c

a r t i c l e

i n f o

Article history: Received 26 September 2014 Received in revised form 4 April 2015 Accepted 18 April 2015 Available online 16 May 2015 Keywords: Ant community Species coexistence Fire ants Supercoloniality Invasive species

a b s t r a c t The fire ant Solenopsis saevissima is a major pest frequent in human-disturbed areas of its native range where it forms ‘supercolonies’. We determined that its natural habitat in French Guiana is likely the sporadically flooded riparian forest and aimed to evaluate this ant’s impact on the abundance and diversity of other ants by comparing different habitats at two sites. We noted a significant decrease in ant species richness between the rainforest and human-disturbed habitats (but not between the former and the naturally disturbed riparian forest), and between extreme habitats and all others. The number of ant nests per surface unit (i.e., quadrats of equal surface area), a proxy of ant abundance, globally followed this pattern. S. saevissima was absent from pristine rainforest (as expected) and from extreme habitats, showing the limits of its adaptability, whereas some other native ants can develop in these habitats. Ant species richness was significantly lower in the presence of S. saevissima in the riparian forest, forest edges and meadows, illustrating that this ant species has a negative impact on the ant communities in addition to the impact of natural- and man-made disturbances. Only some ant species can develop in its presence, and certain of these can even thrive. Because it has been recorded in Africa, Guadeloupe and the Galápagos Islands, we concluded that, due to the increasing volume of global trade and forest destruction, S. saevissima could become a pantropical invasive species. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Tropical rainforest loss and fragmentation continue to progress at an alarming rate and only some species can cope with humanaltered habitats while others cannot, leading to their extinction. So, the structure of the ecosystem is simplified and species richness is generally reduced (Fahrig, 2003; Broadbent et al., 2008; Hansen et al., 2008; Stuble et al., 2009; Laurance et al., 2011; Michel and Sherry, 2012). In addition, by reducing native species ⇑ Corresponding author at: Écologie des Forêts de Guyane, Campus Agronomique, BP 316, 97379 Kourou cedex, France. Tel.: +33 594 594 32 93 00; fax: +33 594 594 32 43 02. E-mail address: [email protected] (A. Dejean). http://dx.doi.org/10.1016/j.biocon.2015.04.012 0006-3207/Ó 2015 Elsevier Ltd. All rights reserved.

diversity and abundance, anthropogenic disturbance plays a major role in the establishment and persistence of invasive species, with both combining to lower global biodiversity (Seabloom et al., 2003; MacDougall and Turkington, 2005). In their introduced range, many invasive species benefit from the absence of natural enemies and from superior competitive abilities vis-à-vis native species, so that they drive changes in the community structure through resource pre-emption (Holway et al., 2002; Bruno et al., 2005; MacDougall and Turkington, 2005). For animals, the impact of introduced, invasive species is amplified because their better ability to deplete resources is enhanced by interspecific territoriality and competition, which affects species assemblages (Bruno et al., 2005).

146

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

Because ants generally constitute the largest fraction of the animal biomass, play different roles in the food web (they can be herbivores, generalists, scavengers or predators) and respond to stress on a finer scale than many other animals, they occupy a central place in the functioning of tropical ecosystems (King et al., 1998; Andersen et al., 2002; Graham et al., 2004, 2008; Delabie et al., 2009). So, there are numerous advantages to studying ants rather than other arthropods and vertebrates when examining habitat disturbance in terrestrial ecosystems. Some ant species were carried to different parts of the world through human transport, successfully establishing themselves and becoming invasive in these areas where they were introduced. Many of the most damaging invasive ants develop supercolonies extending over large distances in their introduced range, are numerically dominant and exhibit a high level of interspecific aggressiveness. By lowering the abundance of native ant species through exploitation and interference competition facilitated by the disparity between their respective population densities, they can disrupt arthropod community structure with subsequent repercussions on the entire ecosystem (Holway et al., 2002; O’Dowd et al., 2003; Sanders et al., 2003; Helanterä et al., 2009). Yet, this is apparently not the case for the monogyne form of the fire ant Solenopsis invicta whose development is facilitated by habitat degradation even in its native range (i.e., the monogyne form has one queen per mound and mounds separate from each other as opposed to the polygyne form with mounds belonging to the same colony that are interconnected by galleries) (Stuble et al., 2011). Indeed, experimental studies suggest that the success of this monogyne form in areas where it was introduced could be due to the negative impact of human disturbance on native ants rather than to this ant’s superior competitive ability (King and Tschinkel, 2006, 2008, 2013a; Tschinkel and King, 2013). These results have stirred controversy as other studies have shown that the presence of this ant is associated with declines in native ant species richness (Stuble et al., 2009, 2013; LeBrun et al., 2012 and papers cited therein; see also King and Tschinkel, 2013b; Hill et al., 2013). Consequently, the challenge is to distinguish the specific role human disturbance and fire ants each play in changes in ant species richness and abundance in different habitats. On the other hand, it is now recognized that the ecological and/ or evolutionary changes underlying biological invasions might occur within the native range of introduced species, reiterating the importance of studies in their native range (Facon et al., 2006; Lee and Gelembiuk, 2008; Valéry et al., 2008; for ants see: Calcaterra et al., 2008; Orivel et al., 2009; Fournier et al., 2012). Accordingly, this study focused on a population of the fire ant Solenopsis saevissima (Smith, 1855) in French Guiana, which is part of its native range extending from Suriname and Amazonia in the north to Argentina in the south (Shattuck, 2014). This species, frequently found in human-disturbed areas (Trager, 1991; Delabie et al., 2009; Zeringóta et al., 2014) and introduced into Africa, Guadeloupe and the Galápagos Islands (Peck et al., 1998; Causton et al., 2006; Wetterer, 2014), belongs to a clade of recently diverged taxa, including the well-known invasive fire ants S. invicta, Solenopsis richteri and Solenopsis geminata (Ross et al., 2010; Martins et al., 2014). Although it is considered a major pest in its native range (where it forms supercolonies extending over up to 54 km), the ecology of S. saevissima has been little studied (Trager, 1991; Taber, 2000; Martin et al., 2011; Lenoir et al., 2015). We were interested in examining the impact of S. saevissima visà-vis native ant species by comparing its presence in different habitats with increasing levels of disturbance (i.e., the undisturbed rainforest, the naturally disturbed riparian forest, forest edges, meadows, lateritic soils and bare white sands). Our aim was three-fold: (1) to determine the natural habitat of S. saevissima in our study area; (2) to verify to what extent habitat degradation

negatively affects ant species richness and abundance by comparing natural rainforests with areas presenting increasing degrees of disturbance; and (3) to determine if ant species richness and abundance are lower in the presence of S. saevissima. 2. Materials and methods After completing preliminary surveys in the same area, this study was conducted in French Guiana between January 2010 and 2014 mostly around the HYDRECO field station situated next to the Petit Saut dam (05°030 4300 N 53°030 0000 W) with clayey ferralitic soils (hereafter ‘Petit Saut’), and around Paracou, closer to the coast (05°180 N 52°530 W), with nutrient-poor, white-sand substrates (hereafter ‘Paracou’). 2.1. The search for the natural habitat of S. saevissima S. saevissima is a rapid colonizer of naturally disturbed (e.g., riverbanks, animal trails) and human-disturbed areas of tropical to warm temperate South American lowlands (Trager, 1991; Taber, 2000) which enters forests and nests in forest clearings (Majer et al., 1997). We therefore looked for its presence in natural habitats represented by the rainforest and the riparian forest. Our staff has worked in the Guianese rainforest over the past 22 years, corresponding to 300–2000 h of field surveys per year (including along streams; see Orivel et al., 2009). These studies were mostly conducted around Petit Saut and the Nouragues field station (05°010 0000 N, 52°420 0000 W), the Chutes Voltaire (05°030 100 N, 54°050 1800 W), and on Kaw Mountain (04°330 5000 N, 52°120 2200 W). We also searched for S. saevissima in the riparian forest by exploring a ca.30-m-wide swath between the riverbanks and the forest during the rainy season when the S. saevissima mounds are easily visible. In this way, we examined 3 km along the Sinnamary River (downstream from Petit Saut), 2 km along the Kourou River (upstream from Degrad Saramaca; 05°010 0000 N, 52°420 0000 W) and 6 km along the Maroni River (upstream from Maripasoula; 03°370 6000 N, 54°010 6000 W). 2.2. Natural and human disturbance versus the presence of S. saevissima To study the impact of various types of disturbance on ant communities, we compared the undisturbed rainforest with several habitats disturbed to different degrees: (1) the riparian forest (naturally disturbed by recurrent flooding, the Petit Saut area only; in Paracou, mangroves line the riverbanks), (2) ‘‘artificial’’ forest edges (along dirt roads where bulldozers scraped away the topsoil down to the laterite and lined by ditches to prevent the heavy rains from eroding the roadbed; these roads are poorly maintained), (3) meadows (mowed once a year) and (4) a highly-disturbed habitat. In Petit Saut, the latter corresponded to a zone cleared by bulldozer during the construction of the hydro-electric dam, so that the laterite was laid bare. In Paracou, it corresponded to white sands laid bare due to the construction of both Route Nationale 1 and the installation of high-tension power lines, public works dating back to 1987. In both cases, only very sparse vegetation, principally Cecropia pioneer trees, has been able to develop. So, these human-disturbed areas have remained ca. the same since we began our survey or over the ensuing 21 years (1988–2009). We assessed the impact of human disturbance by delimiting 20 plots (6 m  3 m; 18 m2) in each of the compared habitats. In the rainforests and along the forest edges, these 6-m-long quadrats were positioned lengthwise parallel to the forest edge every 10 m along two and four transects, respectively. The 10-m buffer distance prevented interference between ant colonies from one

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

quadrat to the next. These quadrats were placed ca. 1 m from the base of the first trees, outside the forest. The very nature of the riparian forest, meadows and highly disturbed habitats obliged us to choose sites according to the topography of the area; two neighboring quadrats were separated by 10 m or more. For habitats where S. saevissima is present (i.e., riparian forest, forest edges, meadows), we also determined the position of the quadrats so that there was a mound in the center (hereafter, ‘S. saevissima quadrats’ as opposed to ‘control quadrats’ for those without such mounds). Quadrat size (6.1  3.2 m) took into consideration the surface area occupied by the S. saevissima mounds and their surroundings estimated at ca. 70 cm in radius (i.e., the mean of 20 measurements), resulting in a surface area of 19.52  1.54 m2 = 17.98 m2 (18 m2) available for occupation by other ant species. Along the forest edges, the rectangular shape of the quadrats permitted us to encompass the S. saevissima galleries that interconnect the mounds and run along this zone where the workers mostly forage (Martin et al., 2011). Each of these quadrats was ‘paired’ with another without S. saevissima situated less than 400 m away (generally less than 80 m away in the riparian forest, meadows). The orientation and structure of the vegetation was comparable in the two kinds of quadrats, which were as similar as possible. Sampling was standardized by spending one man-hour per quadrat carefully searching for ant nests in all suitable microhabitats: the leaf litter including all hollow, rotten twigs; dead wood; humus and the bare ground. We counted the number of nests (or calies in the case of the polycalic species, Wasmannia auropunctata) per quadrat for each ant species recorded (worker visitors were not taken into consideration in this study dealing with ant nesting habits). Voucher specimens of the ants were deposited in the Laboratório de Mirmecologia collection (acronym CPDC), Cocoa Research Centre (Ilhéus, Bahia, Brazil). 2.3. Statistics Diversity statistics were calculated using EstimateS 9.1.0 software with 500 randomizations of the sampling order without replacement (Colwell, 2013) with the extrapolation of the rarefaction curves. Rarefied and extrapolated species richness values were calculated with Eqs. (17) and (18) in Colwell et al. (2012), respectively. Differences in species richness between habitats were tested by inspecting if the 95% confidence intervals overlapped. No overlapping is a conservative condition indicating a significant difference in means at P < 0.05 (Colwell, 2013). To compare species richness between habitats where differences in sampling completeness and ant density between surveys had little influence, species richness values were standardized for a fixed number of occurrences (fixed here arbitrarily at 100) (Gotelli and Colwell, 2001). Using a generalized linear model with a Poisson distribution (GLM Poisson), we modeled the link between (1) the numbers of ant species in the quadrats and the habitat type (i.e., rainforest, riparian forest, forest edges, meadows, lateritic soils or white sands), and (2) the presence of an S. saevissima colony and the site location. The results by habitat type were compared using Tukey’s honest significance test. We also compared the number of nests for ant species co-occurring with S. saevissima between control and S. saevissima quadrats. For each kind of quadrat and each ant species, the number of nests was modeled through a GLM Poisson as a function of the presence of an S. saevissima colony, habitat type and site location (Paracou or Petit Saut). The effects of these three factors were assessed with a Wald test (R v. 2.14.2 software; R Development Core Team, 2011). To bring out the relationships between ant species assemblages, degrees of habitat disturbance, and S. saevissima occurrence, we used the Self-Organizing Map algorithm (SOM, neural network)

147

as detailed in Céréghino et al. (2005). The SOM provides a non-linear projection of the data space onto a two-dimensional map (Kohonen, 2001; Park et al., 2003; Céréghino and Park, 2009; Dejean et al., 2015a). First, we classified the quadrats from each site (160 quadrats in Petit Saut; 120 quadrats in Paracou) according to ant species presence-absence data. The input layer of the SOM was composed of n ant species (n = 120 for Petit-Saut and n = 101 for Paracou) connected to the quadrats representing the habitat types (i.e., rainforest, riparian forest, forest edge, meadow and extreme habitat, plus cases where S. saevissima was present). The output layer, after testing for quantization and topographical errors (see Céréghino and Park, 2009), was composed of 66 neurons (Petit Saut) or 54 neurons (Paracou) displayed as hexagonal cells (maps composed of 11 or 9 rows and 6 columns). The hexagons act as virtual quadrats and approximate the probability density function of the input data. During the learning process, an ant assemblage is therefore computed for each virtual quadrat. The iterative learning procedure for species occurrence data was described in Céréghino et al. (2005). At the end of training, an ant assemblage is known for each output neuron of the SOM, and each quadrat is set in the corresponding hexagon of the SOM map. Quadrats that are neighbors on the grid are expected to represent neighboring clusters of quadrats; consequently, quadrats that are very far from each other (according to the ant assemblages) are expected to be distant in the feature space. In other words, the quadrats that are in the same hexagon are very similar in terms of ant assemblage, and the quadrats that are distant in the grid have greater differences in their ant species assemblages. A kmeans algorithm was applied to cluster the hexagons of the trained map, and the clusters were justified according to the lowest Davis Bouldin Index (i.e., low variance within clusters and high variance between clusters; Céréghino et al., 2003). Using the qualitative presence/absence dataset for each ant species, the model calculated continuous quantitative values between 0 and 1. In other words, the connection intensity between input and output layers calculated during the learning process can be considered the probability of occurrence for each species in the area concerned. The occurrence probability for each species in a given area in the form of the connection intensity is presented on the SOM map using a color scale (blue = low probability of occurrence; brown = high probability), and therefore allowed us to analyze the effect of the presence of each ant species on the patterning input dataset (quadrats), and to predict the occurrence of each species in subsets of quadrats or clusters. Second, to bring out the relationships between habitat types and the occurrence of S. saevissima with the different ant species, we used a mask function to assign a null weight to S. saevissima during the above-mentioned training, whereas all the other ant species were assigned a weight of ‘1’. So, S. saevissima served as an explanatory variable, whereas the ordination process was based on all ant species except S. saevissima. Setting the mask value to zero for a given component (here S. saevissima) removes the effect of that component on the map (Sirola et al., 2004). The values for S. saevissima were thus represented on the SOM trained through the other ant species.

3. Results 3.1. The search for the natural habitat of S. saevissima We never recorded S. saevissima nests in the Guianese rainforest; however, during this study, we recorded 22 colonies in the riparian forest, each represented by several interconnected mounds. Of them, two (9%) were present in hilly areas where the riverbanks had become eroded, triggering a landslide where little by little pioneer vegetation developed; six (27%) were situated

148

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

next to rapids, which are areas often subject to flooding; and 14 (64%) in areas where the vegetation permitted sunrays to reach the ground. 3.2. Effect of disturbance on ant communities We recorded a total of 1239 ant nests (738 in Petit Saut; 501 in Paracou) of 146 ant species (120 in Petit Saut; 101 in Paracou; S. saevissima included). S. saevissima was present only in the riparian forest, along the forest edges and in the meadows (Appendix 1). At both sites, the ant species richness decreased with an increasing level of disturbance. The richness (mean ± SD) decreased from 49.4 ± 4.0 (rainforest) to 46.2 ± 4.8 (riparian forest), 41.0 ± 4.2 (forest edges), 20.6 ± 2.3 (meadows) and 9.7 ± 1.7 (laterite) in Petit Saut, and in Paracou from 56.0 ± 3.9 (rainforest) to 46.9 ± 3.3 (forest edges), 21.0 ± 1.2 (meadows) and 7.5 ± 1.1 (white sands) (Fig. 1). Yet, the differences in species richness were not significant between the rainforest, the riparian forest and the forest edges (Fig. 1). The number of species per quadrat followed similar patterns, but this time the differences between the rainforest and all human-disturbed habitats were significant; they were not with the riparian forest, illustrating differences between naturally and human-disturbed habitats (Fig. 2A). The differences between the riparian forest, the forest edges and the meadows were not significant and the differences between the extreme habitats and all other habitats were significant (Fig. 2A). The number of ant nests per quadrat also decreased from the rainforest to the different disturbed habitats, but the differences were not significant between the rainforest, the riparian forest, the forest edges and the meadows (Fig. 2B). Also, the differences between the three latter habitats and the lateritic soils were not significant, while the number of ant nests per quadrat in the white sands was significantly lower than in all other habitats, showing that this environment is particularly drastic (Fig. 2B). 3.3. Effect of the presence of S. saevissima on ant communities That the presence of S. saevissima was negatively correlated with ant species richness is illustrated in Fig. 1. Its presence corresponded to significantly lower species richness in the riparian forest (Petit Saut: 46.2 ± 4.8 vs. 30.2 ± 3.2), and the effect was even more dramatic in the two human-disturbed habitats where it can

develop: forest edges (Petit Saut: 41.0 ± 4.2 vs. 10.2 ± 1.0; Paracou: 46.9 ± 3.3 vs. 10.1 ± 0.7) and meadows (Petit Saut: 20.6 ± 2.3 vs. 12.2 ± 2.2; Paracou: 21.0 ± 1.2 vs. 11.3 ± 1.8) (Fig. 1). We did not note significant differences for either the number of species or the number of nests in the riparian forest (Table 1). Yet, the differences were significant for both the forest edges and the meadows, illustrating again the differences between naturally and human-disturbed habitats. 3.4. Ant species co-occurring with S. saevissima Among the 21 ant species frequently co-occurring with S. saevissima, seven were significantly more frequent in the S. saevissima quadrats than in the control quadrats (i.e., Brachymyrmex patagonicus, Camponotus blandus, Crematogaster tenuicula, Pheidole fallax, Rogeria foreli, S. geminata, and W. auropunctata), while the contrary was true for six others (i.e., Atta sexdens, Crematogaster sp.6, Dorymyrmex pyramicus, Odontomachus haematodus, Pheidole sp.4, and Pseudomyrmex termitarius). We did not note an effect for the remaining eight species (Table 2). A habitat effect exists as some species were noted in the riparian forest rather than along the forest edges or between the forest edges and the meadows or vice versa, while no effect was noted for the other ant species (Table 2). Also, a ‘site effect’ concerned 11 species (Table 2) with, for the most extreme case, D. pyramicus and S. geminata present only in Paracou (likely due to a difference in the soil substrate). 3.5. Relationships between ant species, degree of habitat disturbance and S. saevissima occurrence The SOM illustrates that, in both Petit Saut and Paracou, the ant species distribution and degrees of habitat disturbance showed congruent patterns (Figs. 3 and S1). In both cases, four clusters of quadrats were delineated on the SOM according to the ant species that characterize them, with all S. saevissima quadrats grouped in cluster D (although we assigned a weight of ‘‘0’’ to S. saevissima itself) (Figs. 3a and S1a). In Petit Saut, cluster A (Fig. 3a) groups together different cases of disturbance, from the riparian forests (upper left part of the cluster, typically Ectatomma tuberculatum; see Fig. 3b) to the meadows and bare laterite (mid- and lower left parts of the cluster

Fig. 1. Comparison of rarefied species richness in plots in the absence (F: forest, R: riparian forest; E: forest edge, M: meadow, L: laterite, S: white sands) or presence of S. saevissima mounds (Rs; Es; Ms) at two sites (A: Petit Saut; B: Paracou). Dots indicate estimated species richness values and dotted lines a 95% confidence interval. Note that the confidence intervals for naturally disturbed riparian forest (Rs vs. R) and human-disturbed habitats with and without S. saevissima (Es vs. E, Ms. vs. M) do not overlap, indicating a significant difference in species richness.

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

149

Fig. 2. Comparisons of the number of ant species (A) and of ant nests (B) per quadrat. Data from Paracou and Petit Saut were pooled because the difference in site location was not significant according to the GLM Poisson parameter test. Statistical comparisons: different letters indicate significant differences at P < 0.05 (post hoc Tukey’s HSD test on GLM Poisson). Box and whiskers plots (median, first and third quartiles; whiskers show 1.5 times the difference between the third and first quartiles, when necessary, reduced to data range).

Table 1 Comparison of the number of ant species and nests per quadrat in the absence versus presence of Solenopsis saevissima (Student’s t-test). Petit Saut

No. ant species per quadrat

t value

df

P

Riparian forest Forest edges Meadows

4.65 ± 0.48 vs. 3.9 ± 0.49 4.9 ± 0.38 vs. 3.5 ± 0.3 4.95 ± 0.38 vs. 3.6 ± 0.38

1.1 2.7 2.4

38 38 38

0.27; NS 0.008 0.02

Paracou Forest edges Meadows

4.8 ± 0.37 vs. 3.0 ± 0.2 4.05 ± 0.3 vs. 2.7 ± 0.16

4.2 3.8

38 38

0.0001 0.0004

Petit Saut Riparian forest Forest edges Meadows

No. ant nests per quadrat 4.65 ± 0.48 vs. 4.4 ± 0.55 5.05 ± 0.53 vs. 3.6 ± 0.33 5.15 ± 0.43 vs. 3.8 ± 0.57

0.34 2.7 2.4

38 38 38

0.73; NS 0.008 0.02

Paracou Forest edges Meadows

4.55 ± 0.44 vs. 2.9 ± 0.25 4.15 ± 0.32 vs. 2.7 ± 0.28

4.2 3.8

38 38

0.0001 0.0004

characterized by P. termitarius) and some cases of forest edges and rainforest (right part of the cluster). The formation of cluster B is mostly due to species from the forest edge, such as Acanthostichus brevicornis (Fig. 3b), with some cases of species from the riparian forest and the rainforest. Finally, the formation of cluster C is mostly due to forest species with numerous species from the genera Odontomachus and Pachycondyla (Ponerinae) and Pheidole (Myrmicinae) (see Appendix 1); it includes only one quadrat from the forest edge. Species able to co-occur with S. saevissima were rather ubiquitous and, except for W. auropunctata, able to establish and maintain colonies in extreme habitats (Fig. 3c). Relatively similar results were noted in Paracou where D. pyramicus was well represented (see Fig. S1).

4. Discussion We show that S. saevissima colonies do not live in the rainforest or in extreme habitats, illustrating the limited possibilities this

species has to nest. Our results support the hypothesis that the presence of S. saevissima is negatively correlated with ant species richness and nest abundance. Yet, this was not always true for the naturally disturbed habitat represented by the riparian forest. Also, some native ant species can co-occur with its mounds. 4.1. The search for the natural habitat of S. saevissima In French Guiana, S. saevissima is a species naturally occurring in riparian forests, which are subject to flooding. S. saevissima’s occurrence is favored by the presence of rapids or of areas where a landslide occurred due to riverbank erosion; so, a relatively disturbed habitat. In these areas, pioneer vegetation progressively develops; ferns, Cecropia obtusa (Cecropiaceae) and Goupia glabra (Goupiaceae) are very abundant (AD; pers. obs.). One can note that human-disturbed areas such as forest edges along dirt roads that penetrate into the rainforest, meadows mowed once a year (this study) and roadsides (Martin et al., 2011) present some similarities with the naturally disturbed habitats represented by the riparian forest, and thus favor S. saevissima’s dispersion. This is in keeping with previous data indicating that this species is adapted to disturbed habitats, including naturally disturbed riverbanks (Taber, 2000; Adis et al., 2001). Also, S. saevissima seems unable to nest in the Guianese rainforest whatever the soil structure (Delabie et al., 2009; this study), but has occasionally been recorded in clearings in the Bahian part of the Atlantic forest of Brazil (Majer et al., 1997; but see Fox et al., 2012 as it could be a cryptic species due to the geographical position of that study). Therefore, S. saevissima, which is present in some areas of the riparian forest, forest edges and meadows (Appendix 1), characterizes naturally and human-disturbed habitats of the Amazon Basin (Trager, 1991; Delabie et al., 2009; this study). Furthermore, we have shown the limits of its habitats as, in addition to its absence from the pristine rainforest, it is unable to live on lateritic soils or white sands. On the contrary, Ectatomma brunneum, Ps. termitarius, B. patagonicus, C. blandus, Cr. tenuicula, D. pyramicus, and P. fallax

150

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

Table 2 Comparison of the number of nests belonging to ant species able to co-occur with Solenopsis saevissima along forest edges or in meadows; (+) and (): species presence facilitated or not by the presence of S. saevissima. Effects of the riparian forest compared to the forest edges, the latter compared to meadows and site (different soil substrates in Petit Saut and Paracou). Statistical comparisons (the reference was the Paracou forest edge): different effects were assessed with a Wald test. Species

Effect of Solenopsis saevissima

Effect of riparian forest (+) vs. forest edge ()

Effect of meadow (+) vs. forest edge ()

Effect of Petit Saut (+) vs. Paracou ()

Atta sexdens Brachymyrmex patagonicus Camponotus blandus Camponotus melanoticus Cephalotes sp. Crematogaster tenuicula Crematogaster sp. 6 Dorymyrmex pyramicus Gigantiops destructor Hypoponera opaciceps Hypoponera sp. 2 Nylanderia sp.a Odontomachus haematodus Pseudoponera stigma Pheidole fallax Pheidole sp. 4 Pseudomyrmex termitarius Rogeria foreli Rogeria sp. 1 Solenopsis geminata Wasmannia auropunctata

() P < 0.01 (+) P < 0.0001

() P < 0.01 () P < 0.001

No effect (+) P < 0.0001

No effect (+) P < 0.0001

(+) P < 0.0001 No effect

No effect No effect

(+) P < 0.0001 () P < 0.0001

No effect () P < 0.0001

No effect (+) P < 0.0001

No effect No effect

No effect No effect

No effect () P < 0.0001

() P < 0.0001 () P < 0.05 No effect No effect No effect No effect () P < 0.0001

() P < 0.0001 No effect No effect No effect No effect No effect (+) P < 0.0001

No (+) No No No No (+)

(+) P < 0.0001 Only in paracou No effect No effect No effect () P < 0.0001 () P < 0.05

No effect (+) P < 0.0001 () P < 0.0001 () P < 0.0001

No effect () P < 0.0001 () P < 0.0001 No effect

No effect (+) P < 0.0001 () P < 0.0001 (+) P < 0.0001

No effect (+) P = 0.0001 (+) P = 0.0001 () P < 0.05

(+) No (+) (+)

No No No No

No effect No effect No effect () P < 0.0001

No effect No effect Only in paracou (+) P = 0.0001

P < 0.0001 effect P < 0.01 P < 0.001

effect effect effect effect

(the five latter species otherwise frequently co-occurring with S. saevissima) are well represented in these extreme habitats (Figs. 3 and S1; Appendix 1). Because our study deals with ant nesting and territoriality, we worked at the scale of nests and recorded a total of 146 species, with rarefaction curves showing signs of saturation. This value is comparatively low because this method eliminates foraging workers, including those from arboreal species, which frequently forage on the ground and in the leaf litter (284 species sampled by sifting litter in the French Guianese forest of the Nouragues; Groc et al., 2013). 4.2. Impact of natural and human disturbance on ant communities Through both rarefaction curves and comparisons of the number of species per quadrat, we have shown the differences between natural and man-made disturbances. The latter have a negative impact on ant species richness, but ant abundance was affected only in extreme habitats (Figs. 1 and 2). Although we have confirmed that ant species richness decreases with a higher level of disturbance (Figs. 1, 2A; King et al., 1998; Graham et al., 2004, 2008), specific ant assemblages, each represented by emblematic species, characterize the degree of disturbance better than ant species richness does. Indeed, the SOM, which clearly separates the presence of S. saevissima in cluster D, illustrates that, generally, each habitat has its own set of species (Figs. 3 and S1; Appendix 1; see also Delabie et al., 2009). Yet, some overlapping can occur in the same cluster as most ant species have adapted and can develop in several habitats (for instance: riparian forest, forest edge and meadows in cluster A; forest edge, riparian forest and forest in cluster B; Fig. 3a). 4.3. Direct impact of S. saevissima on ant communities In the naturally disturbed habitat of its native range, represented by riparian forests, the presence of S. saevissima was

effect P < 0.0001 effect effect effect effect P < 0.05

negatively correlated with ant species richness in only one study area out of two. This is likely due to the buffer effect of B. patagonicus, A. sexdens, C. blandus, Cr. tenuicula and P. fallax already known to co-occur with S. saevissima (Roux et al., 2013; Dejean et al., 2015b), certain of them even able to thrive in its presence (Table 2). On the contrary, in human-disturbed habitats such as forest edges and meadows, the presence of S. saevissima is negatively correlated with ant species richness and abundance (Figs. 1 and 2; Table 1; see also Trager, 1991; Delabie et al., 2009). Another frequent Guianese ant species, W. auropunctata, is adapted to living in naturally disturbed areas represented by flood-prone forest zones situated along small streams and has therefore evolved under habitat constraints, a prerequisite explaining why it became expansive in its natural range and then invasive in areas of its introduction (Orivel et al., 2009). Consequently, S. saevissima has the potential to become an invasive species if transported elsewhere to areas where it is able to reproduce, as has already happened (Peck et al., 1998; Causton et al., 2006; Wetterer, 2014). With its mounds interconnected by galleries over wide areas (Martin et al., 2011), S. saevissima is rather comparable to the polygyne form of S. invicta. This colony structure makes it very difficult to conduct field experiments involving the destruction of colonies, or, on the contrary, the introduction of supernumerary colonies over several years as did King and Tschinkel (2006, 2008) with the monogyne form of S. invicta. Note that we worked on ‘‘stabilized situations’’ because the forest edges, meadows and extreme habitats we studied resulted from the construction of the Petit Saut dam and Route Nationale 1 which date back to 1987. S. saevissima invaded all these new human-disturbed habitats from the start, but never entered the rainforest itself, not even very large clearings. So, this situation where both S. saevissima and other ant species are native to French Guiana differs from that of the polygyne form of S. invicta in its introduced range; in Texas, native ant abundance and species richness firstly declined in its

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

151

Fig. 3. (a) Distribution of the 160 quadrats on the Self-Organizing Map (SOM; 66 hexagons) according to the number of ant nests per quadrat for the 121 ant species recorded in Petit Saut. The codes correspond to the quadrats (F: forest; R: riparian forest; FE: forest edge; M: meadow; L: lateritic soil) or cases when Solenopsis saevissima is present (So). The different clusters (A–D, outlined in bold) were derived from the k-means algorithm. (b and c) Gradient analysis of the number of nests per quadrat for selected typical ant species on the trained SOM presented in ‘‘a’’. The gradient is represented by a color scale (blue = low probability of occurrence, brown = high probability of occurrence). Each small map (i.e., distribution map), corresponding to one ant species, can be superimposed on map ‘‘a’’, thus showing the average nest abundance for each ant species within each cluster of quadrats. (b) Distribution map representing typical cases related to the riparian forest (upper left part of cluster A: Ectatomma tuberculatum), the meadows and lateritic soils (mid-left part of cluster A: Pseudomyrmex termitarius), mostly forest edges (upper part of cluster B: Camponotus atriceps and Acanthostichus brevicornis), the forest (cluster C), and Solenopsis saevissima presence (concentrated in cluster D). (c) Distribution maps corresponding to seven ant species frequently cooccurring with S. saevissima; all are relatively widely distributed, but only marginally present in the forest (clusters B and C). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

152

A. Dejean et al. / Biological Conservation 187 (2015) 145–153

presence (Porter and Savignano, 1990), but largely recovered a decade later although S. invicta remained the most abundant ant species (Morrison, 2002). 4.4. Ant species able to co-occur with S. saevissima In addition to being rather ubiquitous and, for certain of them, also able to develop in extreme habitats (Figs. 3c, d, S1c and d; Appendix 1), the species frequently co-occurring with S. saevissima (Table 1) are accepted by the latter due to different sets of biological and ecological traits (Dejean et al., 2015b). They are accepted, for example, due to their submissive behavior (mostly P. fallax) and/or appeasing cuticular compounds (Roux et al., 2013). Also, coexistence with S. saevissima may be facilitated by different rhythms of activity that limit interactions; this is the case for C. blandus and D. pyramicus, which forage during the hottest hours of the day, whereas, S. saevissima is mostly nocturnal (Orivel and Dejean, 2002). 4.5. Impact of habitat favorability and S. saevissima dominance on ant species richness Although interspecific competition is frequently considered a key factor in structuring many communities, its significance at the local scale over short periods of time contrasts with that occurring at a regional scale over evolutionary time (Ricklefs, 2004). Indeed, null models show that many patterns attributed to local interactions can occur independently and rather result from regional-scale processes (Gotelli, 2000, 2001). In ants, the importance of interspecific competition is best illustrated by the existence of ‘dominant species’ and considered the ‘hallmark of ant ecology’. The relationships between dominant ants and ant species richness were thought to follow the ‘dominanceimpoverishment rule’ where a behaviorally-dominant species (or two) with large, aggressive colonies lowers ant species richness in a local community (Hölldobler and Wilson, 1990). The impact of dominant ants on ant species richness was first tested through baiting experiments showing the unimodal shape of this relationship (Andersen, 1992), something confirmed by later studies (Parr et al., 2005). More importantly, a similar pattern was shown over a wider scale (regional) when the baiting was complemented by pitfall sampling, and null model co-occurrence analyses supported these findings (Parr, 2008). This unimodal shape corresponds to the combination of better habitat quality and the impact of dominant ants. In extreme habitats, both species richness and dominant ant abundance are low. As habitat favorability improves, ant species richness and abundance (including the abundance of dominant ants) increase. The ascending portion of the curve corresponds to an increase in both species richness and dominant ant abundance provided that conditions continue to improve. As the abundance of dominant ants increases, species richness continues to increase until it peaks after which species richness decreases whilst dominance increases, forming the descending part of the curve (Parr et al., 2005; Parr, 2008). A polynomial curve applied to our data as a rule of thumb also shows a unimodal shape if we consider the extreme habitat (lateritic soil) and the species richness in the S. saevissima quadrats of the disturbed habitats (Fig. S2). So, because it is adapted to perturbed habitats, the dominant ant species, S. saevissima, plays a direct role in quantitatively regulating assemblage-level richness in these habitats, while, qualitatively, some species are able to establish and maintain colonies in its presence. We have seen that S. saevissima, which presents the supercolony syndrome in its native range (Martin et al., 2011; Lenoir et al., 2015) and is adapted to living in naturally disturbed areas (here the riparian forest, something it shares with W. auropunctata,

which is known as a major pest among invasive ants; Holway et al., 2002; Orivel et al., 2009), can displace most ant species likely able to develop in human-disturbed areas (this study). Only a few ant species are adapted to living in its presence. This fire ant therefore possesses the principal prerequisites to becoming an invasive species. In addition, if introduced into other areas, which is already the case (Peck et al., 1998; Causton et al., 2006; Wetterer, 2014), it will be released from its natural enemies, increasing its potential to displace native ants (see cases in Holway et al., 2002). Indeed, like most other fire ants, S. saevissima rather occurs in disturbed habitats, something facilitating its dispersion through maritime harbors or airports, while its transport could be facilitated by the increasing extent and volume of global trade as was the case for S. invicta (Ascunce et al., 2011). It would therefore be of great interest to keep up with its progress wherever its presence has been or will be reported in the Tropics. Acknowledgments We are grateful to Andrea Yockey-Dejean for proofreading the manuscript, Frédéric Azémar, Sardadebie Loza and Lucie Toussaint for technical help, and the Laboratoire Environnement de Petit Saut for furnishing logistical assistance. Financial support for this study was partially provided by the Programme Convergence 2007–2013, Région Guyane from the European Community (Project Bi-Appli, 115/SGAR-DE/2011/052274). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biocon.2015.04. 012. References Adis, J., Marques, M.I., Wantzen, K.M., 2001. First observations on the survival strategies of terricolous arthropods in the northern Pantanal wetland of Brazil. Andrias 15, 127–128. Andersen, A.N., 1992. Regulation of ‘momentary’ diversity by dominant species in exceptionally rich ant communities of the Australian seasonal tropics. Am. Naturalist 140, 401–420. Andersen, A.N., Hoffmann, B.D., Müller, W.J., Griffiths, A.D., 2002. Using ants as bioindicators in land management: simplifying assessment of ant community responses. J. Appl. Ecol. 39, 8–17. Ascunce, M.S., Yang, C.C., Oakey, J., Calcaterra, L., Wu, W.-J., Shih, C.-J., Goudet, J., Ross, K.G., Shoemaker, D.D., 2011. Global invasion history of the fire ant Solenopsis invicta. Science 331, 1066–1068. Broadbent, E., Asner, G.P., Keller, M., Knapp, D., Oliveira, P., Silva, J., 2008. Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biol. Conserv. 141, 1745–1757. Bruno, J.F., Fridley, J.D., Bromberg, K.D., Bertness, M.D., 2005. Insights into biotic interactions from studies of species invasions. In: Sax, D.F., Stachowicz, J.J., Gaines, S.D. (Eds.), Species Invasions: Insights into Ecology, Evolution and Biogeography. Sinauer Associates, Sunderland, pp. 9–40. Calcaterra, L.A., Livore, J.P., Delgado, A., Briano, J.A., 2008. Ecological dominance of the red imported fire ant, Solenopsis invicta, in its native range. Oecologia 156, 411–421. Causton, C.E., Peck, S.B., Sinclair, B.J., Roque-Albelo, L., Hodgson, C.J., Landry, B., 2006. Alien insects: threats and implications for conservation of Galàpagos Islands. Ann. Entomol. Soc. Am. 99, 121–143. Céréghino, R., Santoul, F., Compin, A., Mastrorillo, S., 2005. Using self-organizing maps to investigate spatial patterns of non-native species. Biol. Conserv. 125, 459–465. Céréghino, R., Park, Y.S., 2009. Review of the self-organizing map (SOM) approach in water resources: commentary. Environ. Model. Software 24, 945–947. Céréghino, R., Park, Y.S., Compin, A., Lek, S., 2003. Predicting the species richness of aquatic insects in streams using a limited number of environmental variables. J. North Am. Benthol. Soc. 22, 442–456. Colwell, R.K., 2013. EstimateS: statistical estimation of species richness and shared species from samples. Version 9. User’s Guide and Application. . Colwell, R.K., Chao, A., Gotelli, N.J., Lin, S.Y., Mao, C.X., Chazdon, R.L., Longino, J.T., 2012. Models and estimators linking individual-based and sample-based rarefaction, extrapolation and comparison of assemblages. J. Plant Ecol. 5, 3–21.

A. Dejean et al. / Biological Conservation 187 (2015) 145–153 Dejean, A., Azémar, F., Céréghino, R., Leponce, M., Compin, A., Corbara, B., Orivel, J., Compin, A., 2015a. The dynamics of ant mosaics in tropical rainforests characterized using the self-organizing map algorithm. Insect Sci. http:// dx.doi.org/10.1111/1744-7917.12208. Dejean, A., Corbara, B., Céréghino, R., Leponce, M., Roux, O., Rossi, V., Delabie, J.H.C., Compin, A., 2015b. Traits allowing some ant species to nest syntopically with the fire ant Solenopsis saevissima in its native range. Insect Sci. 22, 289–294. Delabie, J.H.C., Céréghino, R., Groc, S., Dejean, A., Gibernau, M., Corbara, B., Dejean, A., 2009. Ants as biological indicators of Wayana Amerindians land use in French Guiana. C. R. Biol. 332, 673–684. Facon, B., Genton, B.J., Shykoff, J., Jarne, P., Estoup, A., David, P., 2006. A general ecoevolutionary framework for understanding bioinvasions. Trends Ecol. Evol. 21, 130–135. Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Evol. Syst. 34, 487–515. Fournier, D., Tindo, M., Kenne, M., Mbenoun Masse, P.S., Van Bossche, V., De Coninck, E., Aron, S., 2012. Genetic structure, nestmate recognition and behaviour of two cryptic species of the invasive big-headed ant Pheidole megacephala. PLoS One 7, e31480. Fox, E.G.P., Pianaro, A., Delabie, J.H.C., Vairo, B.C., Machado, E.A., Bueno, O.C., 2012. Intraspecific and intracolonial variation in the profile of venom alkaloids and cuticular hydrocarbons of the fire ant Solenopsis saevissima Smith (Hymenoptera: Formicidae). Psyche. http://dx.doi.org/10.1155/2012/398061. Gotelli, N.J., 2000. Null model analysis of species co-occurrence patterns. Ecology 81, 2606–2621. Gotelli, N.J., 2001. Research frontiers in null model analysis. Global Ecol. Biogeogr. 10, 337–343. Gotelli, N.J., Colwell, R.K., 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecol. Lett. 4, 379–391. Graham, J.H., Hughie, H.H., Jones, S., Wrinn, K., Krzysik, A.J., Duda, J.J., Freeman, D.C., Emlen, J.M., Zak, J.C., Kovacic, D.A., Chamberlin-Graham, C., Balbach, H., 2004. Habitat disturbance and the diversity and abundance of ants (Formicidae) in the Southeastern Fall-Line Sandhills. J. Insect Sci. 4, 30, . Graham, J.H., Krzysik, A.J., Kovacic, D.A., Duda, J.J., Freeman, D.C., Emlen, J.M., Zak, J.C., Long, W.R., Wallace, M.P., Chamberlin-Graham, C., Nutter, J.P., Balbach, H.E., 2008. Species richness, equitability, and abundance of ants in disturbed landscapes. Ecol. Indicators 141, 1717–1725. Groc, S., Delabie, J.H.C., Fernández, F., Leponce, M., Orivel, J., Silvestre, R., Vasconcelos, H.L., Dejean, A., 2013. Leaf-litter ant communities in a pristine Guianese rainforest: stable functional structure versus high species turnover. Myrmecol. News 19, 43–51. Hansen, M.C., Stehman, S.V., Potapov, P.V., Loveland, T.R., Townshend, J.R.G., DeFries, R.S., Pittman, K.W., Arunarwati, B., Stolle, F., Steininger, M.K., Carroll, M., DiMiceli, C., 2008. Humid tropical forest clearing from 2000 to 2005 quantified by using multitemporal and multiresolution remotely sensed data. Proc. Nat. Acad. Sci. USA 105, 9439–9444. Helanterä, H., Strassmann, J.E., Carrillo, J., Queller, D.C., 2009. Unicolonial ants: where do they come from, what are they and where are they going? Trends Ecol. Evol. 24, 341–349. Hill, J.K., Rosengaus, R.B., Gilbert, F.S., 2013. Invasive ants – are fire ants drivers of biodiversity loss? Ecol. Entomol. 38, 539. Hölldobler, B., Wilson, E.O., 1990. The ants. Springer-Verlag, Berlin. Holway, D.A., Lach, L., Suarez, A.V., Tsutsui, N.D., Case, T.J., 2002. The causes and consequences of ant invasions. Ann. Rev. Ecol. Syst. 33, 181–233. King, J.R., Andersen, A.N., Cutter, A.D., 1998. Ants as bioindicators of habitat disturbance: validation of the functional group model for Australia’s humid tropics. Biodiv. Conserv. 7, 1627–1638. King, J.R., Tschinkel, W.R., 2006. Experimental evidence that the introduced fire ant, Solenopsis invicta, does not competitively suppress co-occurring ants in a disturbed habitat. J. Animal Ecol. 75, 1370–1378. King, J.R., Tschinkel, W.R., 2008. Experimental evidence that human impacts drive fire ant invasions and ecological change. Proc. Nat. Acad. Sci. USA 105, 20339– 20343. King, J.R., Tschinkel, W.R., 2013a. Experimental evidence for weak effects of fire ants in a naturally invaded pine-savanna ecosystem in north Florida. Ecol. Entomol. 38, 68–75. King, J.R., Tschinkel, W.R., 2013b. Fire ants are not drivers of biodiversity change: a response to Stuble et al. (2013). Ecol. Entomol. 38, 543–545. Kohonen, T., 2001. Self-Organizing Maps, third ed. Springer, Berlin. Laurance, W.F., Camargo, J.L.C., Luizão, R.C.C., Laurance, S.G., Pimm, S.L., Bruna, E.M., Stouffer, P.C., Williamson, G.B., Benítez-Malvido, J., Vasconcelos, H.L., Van Houtan, K.S., Zartman, C.E., Boyle, S.A., Didham, R.K., Andrade, A., Lovejoy, T.E., 2011. The fate of Amazonian forest fragments: a 32-year investigation. Biol. Conserv. 144, 56–67. LeBrun, E.G., Plowes, R.M., Gilbert, L.E., 2012. Imported fire ants near the edge of their range: disturbance and moisture determine prevalence and impact of an invasive social insect. J. Anim. Ecol. 81, 884–895. Lee, C.L., Gelembiuk, G.W., 2008. Evolutionary origins of invasive populations. Evol. Appl. 1, 427–448.

153

Lenoir, A., Devers, S., Touchard, A., Dejean, A., 2015. The Guianese population of the fire ant Solenopsis saevissima is unicolonial. Insect Sci. (in press). MacDougall, A.S., Turkington, R., 2005. Are invasive species the drivers or passengers of change in degraded ecosystems? Ecology 86, 42–55. Majer, J.D., Delabie, J.H.C., McKenzie, N.L., 1997. Ant litter fauna of forest, forest edges and adjacent grassland in the Atlantic rain forest region of Bahia, Brazil. Insect. Soc. 44, 55–66. Martin, J.-M., Roux, O., Groc, S., Dejean, A., 2011. A type of unicoloniality within the native range of the fire ant Solenopsis saevissima. C. R. Biol. 334, 307–310. Martins, C., Fernando de Souza, R., Bueno, O.C., 2014. Molecular characterization of fire ants, Solenopsis spp., from Brazil based on analysis of mtDNA gene cytochrome oxidase I. J. Insect Sci. 14, 50, . Michel, N.L., Sherry, T.W., 2012. Human-altered mesoherbivore densities and cascading effects on plant and animal communities in fragmented tropical forests. In: Sudarshana, P., (Ed.). Tropical forests. . Morrison, L.W., 2002. Long-term impacts of an arthropod-community invasion by the imported fire ant, Solenopsis invicta. Ecology 83, 2337–2345. O’Dowd, D.J., Green, P.T., Lake, P.S., 2003. Invasional ‘meltdown’ on an oceanic island. Ecol. Lett. 6, 812–817. Orivel, J., Dejean, A., 2002. Ant activity rhythms in a pioneer vegetal formation of French Guiana (Hymenoptera: Formicidae). Sociobiology 39, 65–76. Orivel, J., Grangier, J., Foucaud, J., Le Breton, J., Andres, F.X., Jourdan, H., Delabie, J.H.C., Fournier, D., Cerdan, P., Facon, B., Estoup, A., Dejean, A., 2009. Ecologically heterogeneous populations of the invasive ant Wasmannia auropunctata within its native and introduced ranges. Ecol. Entomol. 34, 504–512. Park, Y.S., Céréghino, R., Compin, A., Lek, S., 2003. Applications of artificial neural networks for patterning and predicting aquatic insect species richness in running waters. Ecol. Model. 160, 265–280. Parr, C.L., 2008. Dominant ants can control assemblage species richness in a South African savanna. J. Anim. Ecol. 77, 1191–1198. 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. Am. Naturalist 165, 481–494. Peck, S.B., Heraty, J., Landry, B., Sinclair, B.J., 1998. Introduced insect fauna of an oceanic archipelago: the Galápagos Islands, Ecuador. Am. Entomol. 44, 218– 237. Porter, S.D., Savignano, D.A., 1990. Invasion of polygyne fire ants decimates native ants and disrupts arthropod community. Ecology 71, 2095–2106. R Development Core Team, 2011. R: A language and environment for statistical computing. Vienna. . Ricklefs, R.E., 2004. A comprehensive framework for global patterns in biodiversity. Ecol. Lett. 7, 1–15. Ross, K.G., Gotzeck, D., Ascunce, M.S., Shoemaker, D.D., 2010. Species delimitations: a case study in a problematic ant taxon. Syst. Biol. 59, 162–184. Roux, O., Rossi, V., Céréghino, R., Compin, A., Martin, J.-M., Dejean, A., 2013. How to coexist with fire ants: the roles of behaviour and cuticular compounds. Behav. Proc. 98, 51–57. Sanders, N.J., Gotelli, N.J., Heller, N.E., Gordon, D.M., 2003. Community disassembly by an invasive species. Proc. Nat. Acad. Sci. USA 100, 2474–2477. Seabloom, E.W., Harpole, W.S., Reichman, O.J., Tilman, D., 2003. Invasion, competitive dominance, and resource use by exotic and native California grassland species. Proc. Nat. Acad. Sci. USA 100, 13384–13389. Shattuck, S.O., 2014. . Sirola, M., Lampi, G., Parviainen, J., 2004. Using self-organizing map in a computerized decision support system. In: Pal, N., Kasabov, N., Mudi, R., Pal, S., Parui, S. (Eds.), Neural Information Processing. Springer-Verlag, Berlin, pp. 136–141. Stuble, K.L., Chick, L.D., Rodriguez-Cabal, M.A., Lessard, J.-P., Sanders, N.J., 2013. Fire ants are drivers of biodiversity loss: a reply to King and Tschinkel (2013). Ecol. Entomol. 38, 540–542. Stuble, K.L., Kirkman, L.K., Carroll, C.R., 2009. Patterns of abundance of fire ants and native ants in a native ecosystem. Ecol. Entomol. 34, 520–526. Stuble, K.L., Kirkman, L.K., Carroll, C.R., Sanders, N.J., 2011. Relative effects of disturbance on red imported fire ants and native ant species in a longleaf pine ecosystem. Conserv. Biol. 25, 618–622. Taber, S.W., 2000. Fire ants. Texas A&M University Press, College Station, PA. Trager, J.C., 1991. A revision of the fire ants, Solenopsis geminata group (Hymenoptera: Formicidae: Myrmicinae). J. N. Y. Entomol. Soc. 99, 141–198. Tschinkel, W.R., King, J.R., 2013. The role of habitat in the persistence of fire ant populations. PLoS One 8, e78580. Valéry, L., Fritz, H., Lefeuvre, J.C., Simberloff, D., 2008. In search of a real definition of the biological invasion phenomenon itself. Biol. Invasions 10, 1345–1351. Wetterer, J.K., 2014. A South American fire ant, Solenopsis nr. saevissima, in Guadeloupe, French West Indies. Biol. Invasion 16, 755–758. Zeringóta, V., Monteiro De Castro, M., Castro Della Lucia, T.M., Prezoto, F., 2014. Nesting of the fire ant Solenopsis saevissima (Hymenoptera: Formicidae) in an urban environment. Florida Entomol. 97, 668–673.