|Home | About | Journals | Submit | Contact Us | Français|
Arboreal ants are both highly diverse and ecologically dominant in the tropics. This ecologically important group is particularly useful in ongoing efforts to understand processes that regulate species diversity and coexistence. Our study addresses how polydomy can influence patterns of nest occupation in competing arboreal ants. We examined the spatial structure of nest occupation (nest distance, abundance and density) in three polydomous co-occurring twig-nesting ant species (Pseudomyrmex simplex, P. ejectus and P. PSW-53) by mapping twigs occupied by ants from each species within plots in our study site. We then used two colony structure estimators (intraspecific aggression and cuticular hydrocarbon variation) to determine the relative degree of polydomy for each species. All work was conducted in coffee agroforests in Chiapas, Mexico. Our results revealed that the two species with highest abundance and nest density were also highly polydomous, where both species had either single or multiple non-aggressive colonies occupying nests on a large spatial scale (greater than the hectare level). Our results also indicate that the species with the lowest abundance and density is less polydomous, occupying several overlapping and territorial colonies at the hectare level in which multiple colonies never co-occur on the same host plant. These results contribute evidence that successful coexistence and highly polydomous colony structure may allow ants, through reduced intraspecific aggression, to successfully occupy more nests more densely than ant species that have multiple territorial colonies. Furthermore our study highlights the importance of considering intraspecific interactions when examining community assembly of ants.
Arboreal ants are some of the most numerically dominant arthropods in the tropics largely due to their ability to access plant resources in the form of extrafloral nectar and hemipteran honeydew (Hölldobler and Wilson 1990, Davidson 1997; Blüthgen 2000, Davidson et al. 2003). As such they are an ideal group of insects to study patterns of species distribution and community assembly. Moreover, understanding the distribution patterns of ant communities is critical as ants are frequently involved in ecosystem services. For example, ants are common seed dispersers, pollinators, and often engage in plant protection mutualisms that benefit agricultural systems (Dejean et al 2007; Liere and Larsen 2010; Vandermeer et al. 2010; Matthews et al. 2011, De Vega et al. 2014). Early research on arboreal ants suggested that the structure of ant assemblages is largely determined by interspecific competitive interactions (Room 1971; Majer 1972; Majer et al. 1994), but this notion is now widely criticized (Ribas & Schoereder 2002; Blüthgen and Stork 2007, Sanders et al. 2007). Recent work has focused on how interactions with resources can shape the structure of ant assemblages (Davidson 1997; Blüthgen et al. 2004; Blüthgen and Stork 2007; Powell et al. 2011; Dejean et al. 2008, 2015; Jiménez-Soto and Philpott 2015). However, particularly in studies with cavity-dwelling ants where polydomy is common, colony scale and intraspecific aggression are infrequently examined.
Ants display high levels of diversity in spatial organization strategies and colony size, ranging on a continuum from establishment of a single nest (monodomy) to colonies that establish multiple nests (polydomy) in a given area (Debout et al. 2007). Polydomy as a nesting strategy has many benefits including increased foraging area, foraging efficiency, and brood rearing capacity when nesting sites are spatially finite (reviewed in Debout et al. 2007), as well as a reduction in costly energy expenditure on territorial fights (Holway 1998; Holway et al. 2002, Sanders et al. 2003). However, having spatially separated nests can impose costs, including queen-worker conflicts over both resource and sex allocation (Herbers et al. 2001) and could also hamper colony recognition among colony members that occupy distant nests (Vander Meer and Morel 1998).
Estimating colony size and structure in polydomous ants is notoriously difficult, particularly when nest sites are ephemeral and often the only method to determine colony size is by observing trails between nesting sites (e.g. Debout et al. 2007; Buczkowski 2012); however such trails are rare or difficult to follow for many twig nesting ants. Social insects use low-volatility chemical cues found on their cuticle (generally hydrocarbons) to distinguish colony members from non-colony members (Vander Meer and Morel 1998; Howard and Blomquist 2005). Nestmates share a common chemical signature through a combination of genetic similarity and the colony members creating an admixture through physical contact, social grooming, and trophallaxis (Crozier and Dix 1979; Foitzik et al. 2007). If an ant's chemical signature deviates from the admixed template (i.e. chemotype), the individual is then recognized as a non-colony member and frequently attacked (Tsutsui 2004). High cuticular hydrocarbon similarity is typically associated with low levels of aggression (eg. Suarez et al. 2002; Torres et al 2007). Furthermore, cuticular hydrocarbons are also species and caste specific (Howard and Blomquist 2005). Thus, aggression assays and chemical analysis of cuticular hydrocarbons are useful tools in assessing the colony structure of ants (Vander Meer and Morel 1998; Howard and Blomquist 2005).
In coffee agroecosystems in southern Mexico, a diverse community of arboreal ants nest in hollow twigs on coffee plants (Philpott and Armbrecht 2006; Livingston and Philpott 2010). The hollow coffee twigs are a naturally occurring, ephemeral and limited resource in the ecosystem (Philpott and Foster 2005; Jiménez-Soto and Philpott 2015). Furthermore, interspecific aggression, while present at varying degrees between species, does not impact the structure of ant assemblages in this system (Livingston and Jackson 2014). Here we investigate the colony spatial structure of three sympatric species of polygynous, polydomous Pseudomrymex ants (Pseudomyrmex simplex, Pseudomyrmex ejectus, and an un-described Pseudomyrmex sp. hereafter referred to as Pseudomyrmex PSW-53) that occupy twig-nests and compare the nesting strategies of these species.
This study uses a robust approach, combining spatial mapping of nests, aggression assays and cuticular hydrocarbon analysis of three sympatric polydomous ant species in the genus Pseudomyrmex to assess and compare social association between nests. This is the first study to use these approaches to examine the social structure of polydomous twig-nesting ants. Furthermore, the use of multiple approaches to estimate social association also makes it possible to assess the reliability of each method in evaluating colony spatial structure. Our findings highlight the importance of investigating colony spatial structure of polydomous ants to fully understand community dynamics.
We sampled ants on Finca Irlanda, a 300 ha, shaded coffee plantation in the Soconosco region of Chiapas, Mexico (15°11’N, 92°20’W). The plantation is located between 950-1150 m elevation and receives approximately 4,500 mm of rain each year. The farm contains approximately 100 species of shade trees, the most common being trees in the genus Inga (Vandermeer et al. 2008). Canopy cover at the time of ant sampling ranged from 35-70% in the 45 ha area and was around ~90% in the 7 ha area.
In the study site, there are at least 60 arboreal ant species including 40 that nest in dry, hollow coffee twigs (Philpott & Foster 2005; Livingston & Philpott 2010). Heavy rains, natural processes of decomposition, and farm workers moving through the coffee plantations frequently break these twigs from the coffee plants, making the nesting sites ephemeral. There are nine species of Pseudomyrmex that nest in coffee twigs, and the three most common at the study site are P. simplex (occupies ~35% of all occupied coffee twigs), P. ejectus (~20% of twigs), and P. PSW-53 (~6% of twigs) (Philpott and Foster 2005). Pseudomyrmex simplex range from Florida to Paraguay and are commonly collected from forest habitats (AntWeb 2015). Pseudomyrmex simplex nests are commonly found without queens, suggesting a high level of polydomy (Ward 1985). Pseudomyrmex ejectus are also commonly found in forests and coffee plantations, and range from Florida to Costa Rica (Ward 1985; AntWeb 2015). Pseudomyrmex PSW-53 have been collected from Baja California to Costa Rica, and commonly inhabit forests and woodlands (AntWeb 2015). The majority of nests collected for this study were queenless, accounting for 76% of P. simplex nests (n=143), 66.1% of P. ejectus nests (n=96), and 65.0% of the P. PSW-53 nests (n=92). These three ant species differ in their abilities to gain access to nesting sites. In nesting colonization trial experiments P. simplex won 37% of paired trials with other twig-nesting species, P. ejectus won 24% of trials, and P. PSW-53 won 70% of nest competition trials (Yitbarek & Philpott, unpublished data). We chose to work with these three species because of the differences they exhibit in their relative abundance in dry, hollow coffee twigs, and also because of their different apparent success in winning nest sites in competitive encounters.
During 2013 we surveyed twig-nesting ants in randomly selected 20×20 m plots within established permanent plots at the study site (Fig. 1). We sampled twig-nesting ants on every coffee plant (~100-250 coffee plants) within each 20×20 m plot and removed all dry twigs, counted hollow and occupied twigs and identified all occupant ants. The twig nests are most easily sampled destructively by breaking into the twigs to determine their occupants (Philpott and Foster 2005). We mapped the coordinates of each coffee plant within 20×20 m plots, and noted the plot location within each sampled hectare. We took all twigs occupied by P. simplex, P. ejectus, and P. PSW-53 back to the lab where we freeze-killed the ants and placed them at -18°C until extraction
For each twig that contained at least three ant workers (n=197), we collected three individuals for CHC extraction (270 ants for P. simplex, 207 ants for P. ejectus and 114 ants for P. PSW-53). Each ant was freeze-killed and submerged in 100-200 μL of hexane for 10 minutes. The ants were then removed, the hexane was evaporated, and the samples were stored at −18°C until they were transported to the lab. Once in the lab, samples were placed at −20°C until processing. Directly before processing, we resuspended each sample in 200°L of hexane. We filtered samples through a 1.5 cm hexane-rinsed silica column to remove non-hydrocarbon material and flushed the column with 300 μL of hexane to maximize sample recovery. We then used nitrogen gas to reduce our samples to a 60 μL volume. To analyze the samples, we injected 2 μL of the 60 μL into the GC/MS.
For each sample, we acquired electron impact mass spectra (70 eV) with an Agilent 5975 C mass selective detector interfaced to an Agilent 7890A gas chromatograph fitted with a DB-5 column (30 m × 0.32 mm i.d., Agilent Technologies). Extracts were analyzed in splitless mode, with a temperature program that started at 100°C for 1 min, then increased by 20°C min−1 until it reached 150°C, and then increased by 5°C min−1 until it reached 325°C where it was held for 5 min. Injector and transfer line temperatures were kept at 325°C and 280°C respectively. We used MSD ChemStation Software for Windows (Agilent Technoogies, Palo Alto, CA, USA) for data acquisition.
We collected ants from twig nests for aggression trials during June 2015. As before, we collected ants from coffee plants within 20×20m plots in different hectares within the permanent plots. Sample plots were were separated by a minimum of 100m (see Table S1 for exact distances between each sampled hectare). We surveyed areas of the farm that had been surveyed in 2013 to locate nests of P. simplex, P. ejectus, and P. PSW-53, but were not able to locate each of the three species in the same locations, and thus added two additional sample plots (8 and 9, Fig.1). P. simplex and P. ejectus were both collected from plots 2, 4, 6, 7, and 8. P. PSW-53 were collected from plots 6, 8, and 9 (Fig. 1). Occupied twigs were placed individually into ziplock bags and bags were marked with x, y coordinates within the 45-ha plot. We did not measure the size of each coffee plant sampled, or the height from which ants were collected. However, most coffee plants in this farm that have dry twigs are between 1.0-1.5 m in height. Once back in the lab, twigs (and all ant occupants) were placed into plastic tubs and kept until ant aggression was tested (a maximum of 4 days).
We tested ant aggression by placing two individual ants of the same species in a petri dish (3.5cm diameter, 1cm high). We tested aggression in four combinations increasing from small to large scale differences – a) ants from the same twig, b) ants from two twigs on the same coffee plant, c) ants from different coffee plants (but from the same 20×20m plot), and d) ants from different hectares (>100 m apart). Ant interactions were observed in petri dishes for 5min. We recorded the highest of the following levels of aggression to take place: touch (i.e. antennation or other non-aggressive contact, scored as 1), avoid (i.e. mandible flair where one ant moves towards the other while opening the mandibles or one or both ants recoil and retreat in the opposite direction, scored as 2), bite (i.e. one or both ants bite each other during a <30 second attack, scored as 3), aggression (i.e. >30 second attack that may involve removing leg or antenna, scored as 4), and kill (i.e. one ant kills another, scored as 5) (modified from Suarez et al. 1999 and Torres et al. 2007). We only observed behaviors 1-4. To minimize observer bias, blinded methods were use when all behavioral data were recorded and analyzed.
For the twig nesting survey, ant abundance in the study area was measured by total number of twigs occupied by the ants and the percentage of occupied twigs occupied by the species. Density of each species was measured as the inverse of the average distances (in meters) between each occupied twig and its closest conspecific neighbor.
All compounds in the chromatograms identified as cuticular hydrocarbons were included in the analysis. Quantitative comparisons of spectra were based on peak integration of total ion chromatograms. For each sample, we only included compounds with >1% total abundance. When comparing species as a whole, only compounds with an average of >1% total abundance were included after standard deviation was added. All subsequent analyses were performed after peak areas were standardized to relative proportions. Individual hydrocarbon peaks were identified through a combination of (a) comparing retention times and mass spectra with those of synthetic standards to determine carbon chain length and calculate Kovat Indices following Katritzky and Chen (2000), (b) matching our spectra with those previously published and (c) studying fragmentation patterns. To determine whether species grouped by chemotype, we used Ward's hierarchical cluster analysis, and assessed the uncertainty of clusters using multiscale bootstrap resampling (R Development Core Team, 2013, pvclust). For each species, we used the entire data matrix containing all the compounds present in their workers for each hectare. All cuticular hydrocarbon compounds were included in the analysis, as previous experiments show that excluding compounds may increase error (Krasnec and Breed 2013). Chemotypes were divided based on statistically significant clusters (Ward's hierarchical cluster analysis), and individuals were assigned to each chemotype if the majority (two out of three) of the twig-mates were assigned to a particular cluster.
For aggression assay analysis we examined whether variation in aggressive behavior correlated with increasing distance between nests. To this end, we compared the aggression score for worker-worker interactions within twig, within plant, within 20×20m plot, and between hectares. We used a two-way ANOVA with AI as the dependent variable, and ant comparison (within twig, within plant, within 20×20m plot, and between hectares) and ant species (P. simplex, P. ejectus, and P. PSW-53) as factors.
Overall, the three species of ants differed in density and abundance within the sampling area. We found that Pseudomyrmex simplex occupied the most available nesting sites (n=214 or 36%), followed by Pseudomyrmex ejectus (n=104 or 17%) and Pseudomyrmex PSW-53 (n=43 or 7%). P. simplex nests were also in the highest density (1.528), followed by P. ejectus (0.814) and P. PSW-53 (0.164) respectively (Table 1).
For the three species, P. simplex, P. ejectus, and P. PSW-53, we identified 18, 7, and 12 hydrocarbons (or combinations of co-eluting hydrocarbons) respectively, each representing at least 1% of the total area of all compounds (Table 2). Compounds consisted of straight chain alkanes, monomethyl alkanes and some multimethyl alkanes. Compounds had chain lengths ranging from 20 to 36 carbons, with P. simplex containing, on average, compounds with longer carbon chains than P. ejectus and P. PSW-53. Three compounds (C25, C27, and 3me C27) were found in all three species (Fig. 2).
All three Pseudomyrmex species exhibited significantly different cuticular hydrocarbon profiles that clustered independently of one another (Ward's hierarchical cluster analysis, P =0.04). P. simplex and P. ejectus did not reliably cluster into chemotypes beyond the species level, either within or between hectares. However, P. PSW-53 individuals exhibited substantial spatial variation both within and between hectares (Fig. 3). Individual ants of P. PSW-53 divided initially into two primary clusters (P <0.05) and further divided into five distinct secondary clusters (P <0.05, Fig. 4).. Furthermore, although some coffee plants did contain multiple P. PSW-53 nests, individual coffee plants did not support more than one P. PSW-53 chemotype. In other words, all ants from twigs on the same plant shared the same chemotype. Yet, some ants found in twigs on plants within 1m of one another had different chemotypes (Fig. 4). At the primary cluster level, all P. PSW-53 individuals separate accurately into their clusters. Furthermore, individuals separated into their secondary clusters with 91% accuracy, which is a similar level of accuracy as that obtained with monodomous, monogynous ant species using CHC profiles of ants known to belong to different colonies (see Krasnec and Breed 2013).
Overall, aggression levels differed by species, and the intensity of aggression increased with spatial scale for P. PSW-53, but not for P. simplex and P. ejectus. Intraspecific aggression (according to the AI) was significantly higher for P. PSW-53 overall compared with P. simplex and P. ejectus (F2, 125=12.32, P <0.001, pairwise comparisons P <0.01 for both P. simplex and P. ejectus). Aggression between P. simplex and P. ejectus individuals did not increase with spatial scale (P. ejectus, F3, 6 = 11.41, P =0.99; P. simplex, F3, 6=11.41, P =0.98). However, P. PSW-53 aggression increased with spatial scale, where ants from the same twigs and plants were significantly less aggressive than ants interacting from within the plots or between hectares (F3, 6=11.41,P < 0.01) (Fig. 5). Thus, P. PSW-53 engaged in more aggressive behavior than the other two species, especially at greater distances.
Our analysis revealed striking differences in the strategies of social organization of the three species of Pseudomyrmex ants that we examined, which appears to be regulated by a combination of spatial, chemical and behavioral parameters. Ants discriminate between nestmates through antennal contact and perception of cuticular hydrocarbons prior to aggression (Vander Meer and Morel 1998). Both P. simplex and P. ejectus exhibited cuticular hydrocarbon profiles that remained homogeneous through our entire study area and almost no aggression was observed between ants of these species regardless of distance, thus suggesting that either both species only have one colony within the study area or colonies exhibit non-aggressive behaviors to one another. Further work comparing aggression levels and cuticular hydrocarbon profiles of both P. simplex and P. ejectus nests from our sites to those at even greater distances would help elucidate whether these species are truly non-aggressive or if they can simply belong to very large colonies.
This is in contrast to the ant P. PSW-53, whose worker aggression increases with increasing spatial scale and cuticular hydrocarbon profiles can be broken reliably into five distinct chemotypes with the number of chemotypes increasing with distance. These results indicate that there are multiple territorial colonies of P. PSW-53 ants present within our study site that overlap and span across hectares. Interestingly, while P. PSW-53 workers with different chemotypes could be found on plants only 1m apart, P. PSW-53 ants on the same plant were always within the same chemotype. This is in accordance with our behavioral data, which showed that P. PSW-53 ants did not have a higher aggression index at the plant level than ants interacting at the twig level. Additionally, there were no physical attacks (i.e. bites) between workers from the same plant.
Our finding that both P. ejectus and P. simplex are either one large colony or a group of non-aggressive colonies was unexpected given the size of the study area. A previous study that sampled 77 nests of P. ejectus in Florida reported pronounced intercolonial aggression in lab experiments between P. ejectus colonies and estimated maximum colony size to be approximately 500 workers (Klein 1987). We sampled 71 twigs containing a total of 1217 P. ejectus workers, where workers exhibited low cuticular hydrocarbon variation between nests, and P. ejectus workers across our study area exhibited little intraspecific aggression. The contrast between our findings and those of Klein (1987) further highlight that P. ejectus is capable of intercolonial aggression in the laboratory and help rule out that our results can be explained merely by P. ejectus being a non-aggressive species. Additionally, our results in combination with Klein (1987) indicate that colony size may vary greatly between regions and cannot be generalized.
While P. simplex and P. ejectus ants from across several hectares have similar hydrocarbons and will not attack their conspecifics, it is possible that workers of these ant species do not regularly travel such distances or exchange food resources across the full extent of this range. A study examining colony boundaries in the polydomous ant Lasius niger found that average area in which ants remained non-aggressive was much larger than the average area in which ants exchanged food resources (Buczowski 2012). Other studies with the invasive argentine ant, Linepithema humile, show that these ants have reduced aggression and similar cuticular hydrocarbon profiles over many hundreds of miles in their invasive range (Tsutsui et al. 2003; Torres et al. 2007) yet workers only share food resources within a radius of up to 50 meters (Heller et al. 2008).
Although P. ejectus and P. simplex display interspecific aggression, both species are less successful in winning direct interspecific competitions over nest-sites than P. PSW-53 (Livingston and Philpott 2010). Despite the success of P. PSW-53 in competition for nest site colonization, nests of both P. simplex and P. ejectus are more abundant and more densely aggregated than those of P. PSW-53 in our study area. Nest sites are a limited resource for the Pseudomyrmex species, as hollow, unoccupied twigs are rare and occupation rate of experimentally placed artificial twigs is high (Livingston and Philpott 2010; Jiménez-Soto and Philpott 2015). Most likely, low cuticular hydrocarbon variation and reduced aggression between conspecifics facilitates the success of P. simplex and P. ejectus regardless of whether these nests are a part of the same colony or multiple non-aggressive colonies. Studies with Temnothorax ants report that species with reduced nestmate recognition have higher nest density than those with more efficient nestmate recognition (Foitzik et al 2007). Additionally, nest site limitation can trigger cooperative colony founding of unrelated queens (i.e. pleometrosis) (Feldhaar et al. 2005), which provides a competitive advantage in colony founding between species that compete for nesting sites (Vasconcelos 1993; Izzo et al. 2009). Cooperative colony founding is a common phenomenon in ants, where several queens might be present at the early stages of colony founding and early workers will admix cuticular hydrocarbons through physical contact, social grooming, and trophallaxis (e.g. Helms and Helms Cahan 2012; Overson et al. 2014). Additionally, the presence of multiple unrelated queens in mature colonies (i.e. oligogyny) also facilitates chemical similarity and reduced aggression between workers. Callow (i.e. freshly pupated) workers often lack colony cuticular hydrocarbons (Lenoir et al. 2001), and can result in unrelated ants ultimately possessing the same cuticular hydrocarbon admixture (Lenoir et al. 1999; Tsutsui 2004). This may also explain the high chemical similarity and reduced aggression in both P. simplex and P. ejectus.
Regardless of the mechanism that controls social structure, we observed two different strategies for social organization within the three species of Pseudomyrmex that may help explain the sharp differences in the nest abundance and density of these species within coffee agroecosystems. Previous research has examined in detail how both interspecific competition (e.g. Room 1971; Majer 1972; Taylor 1977; Jackson 1984; Majer et al. 1994) and environmental factors (e.g. Sanders et al. 2007; Powell et al. 2011; Jiménez-Soto and Philpott 2015) impact the structure of local arboreal ant assemblages. However, particularly in studies with cavity-dwelling ants where polydomy is common, colony scale and intraspecific aggression are infrequently considered (but see Kautz et al. 2012). Our study highlights the importance of considering intraspecific interactions when examining the community assembly of ants.
G. Lopez Bautista, G. Domíngez-Martínez, and F. Sánchez-López assisted with field collection and sample processing. We thank Finca Irlanda for allowing us to conduct research on the farm. We thank SEMARNAT (Secretaria de Medio Ambiente y Recursos Naturales) for permission to collect and export samples and J. Rojas and E. Chamé Vasquez for facilitating the process of acquiring permits. N. Tsutsui provided helpful comments on the manuscript. Funding was provided by a Packard Foundation Grant to SRR, National Science Foundation DEB-1262086 to SMP, and National Science Foundation GRFP DGE 1106400 to KAM.