Introduction
Biological invasions are important drivers of global change and pose a major challenge to biodiversity and human welfare (Vitousek 1990, exotic species differ from native species is central to the study of invasion biology (Lemoine 2016, understanding why some exotic species excel at dominating recipient communities requires a biogeographical comparison between the native and introduced ranges of exotic species (Hierro 2005).
The biotic resistance hypothesis has been widely cited to explaining invasive species success, as regions with high species diversity are thought to be more resistant to exotic species than depauperate regions such as island communities (Elton 1958). The reason for this is that exotic species are less likely to have access to available niche space in recipient communities (Shea 2002). Evidence for biotic resistance by native ant communities, however, has been mixed. In the introduced range of Linipithema humile, climatic and habitat factors have been attributed to the expansion range of L. humile populations in California (Roura-Pascual 2009, in the introduced range of Australia studies have found that native ant resistance by Irodomyrmex species constrained L. humile populations in Australia
The impact of exotic ants on native biodiversity has been widely documented in the literature (Holway 2002, Lach 2009). Consequences of exotic ant invasions in the introduced range include the local displacement and reduction of native ant diversity (Wittman 2014). Interspecific competition is a major driving factor structuring ecological communities and regarded as an important determinant for invasive success (Hölldobler 1990, Sakai 2001) . One mechanism through which exotic ants can affect recipient ant communities is through resource competition(Kenis 2008). In native ant communities, resource trade-offs are thought to lead to species coexistence in which some ant species excel at resource discovery while others thrive at dominating resources. (Fellers 1987, ant species in displacing local ant species has been attributed to their superior discovery and dominance abilities (Holway 1998, the invasive little fire ant Wasmannia auropunctata, showed no significant differences in resource discovery and dominance abilities between its native and introduced range (Yitbarek et al. in press).
Understanding how exotic ants are organized within local ant assemblages in their native range can allows us to make predictions about their invasion success in the introduced range (Calcaterra 2016). Relatively few studies have examined biotic interactions within the native range of exotic ant species which have mainly focused on L. humile and S. invicta
populations.(Suarez 1999, lack studies comparing the invasion dynamics of other globally important exotic species in their native and introduced ranges. It’s especially important that we examine how exotic ants are organized within local species assemblages.
W. auropunctata  is considered to be a globally important exotic ant species, posing a serious threat to biodiversity and human health as well as being a major pest species in agricultural ecosystems(Breton 2003, decades invaded tropical and subtropical regions of the world including Caribbean and Pacific islands, the United States (Florida), and Western parts of Africa (Gabon and Cameroon) spread to parts outside of the tropics such as in the Mediterranean region of Israel, suggesting that a tropical species can successfully adapt to colder climates (Vonshak 2009). The evolutionary history of W. auropunctata suggests that two genetically distinct clades within W. auropunctata populations coincided in central Brazil, with clade A distributed to Central America and the northern parts of South America while clade B was distributed to the southern parts of South America suggests that Caribbean populations are non-native and likely underwent multiple introductions presumably from the northern regions of South America (Foucaud 2010). Within the native range of W. auropunctata, demographic and reproductive differences distinguish between dominant clonal populations occupying human-disturbed habitats and non-dominant sexual populations that occur in natural forests (Fournier 2005, Foucaud 2009, populations have all been found to be clonal and originate from native dominant populations (Foucaud 2010). While the success of W. auropunctata has been partially attributed to its clonal reproductive system, we lack a comparative ecological analysis of how W. auropunctata ranks relative to native competitors in invaded regions and areas of origins.
The aim of this study was to compare the ecological dynamics of W. auropunctata populations in shaded agricultural coffee ecosystems between their native range of Mexico and introduced range of Puerto Rico. Shaded agricultural coffee ecosystems represent a useful replicated model system for comparing invasion patterns because they have similar vegetation patterns around the globe and are situated within important biodiversity hotspots (Moguel 1999, (Yitbarek et al. in press) found no differences in resource discovery rates in W. auropunctata populations between Mexico and Puerto Rico. However, that study did not address the community context of species interactions.
We studied the competitive interactions between W. auropunctata and native competitors in recipient communities. We compared the foraging behavior and competitive interactions around food baits during short-term (80 minutes) and long-term (14 days) experiments. Within the context of biotic resistance, the expectation is that competitively dominant native species in Mexico will inhibit W. auropunctata populations from expansion. In contrast, we expect that in the introduced range of Puerto Rico W. auropunctata will competitively displace local ant species. Comparing the relative rank of exotic species in local assemblages will allows us to make predictions about invasion patterns.

Materials and Methods

\label{materials-and-methods}

Study Sites

\label{study-sites}
This study was conducted on an organic shaded agricultural coffee ecosystems within the native range of W. auropunctata populations in Mexico and in the introduced region of Puerto Rico. Both regions experience annual wet and dry seasons, as is common in many tropical regions. Data were collected during the wet season (summer) between the months of June and July in 2012, 2013, and 2014. In the Mexico site, a 45-ha plot was surveyed to map the spatial distribution of W. auropunctata colonies in a medium shaded organic coffee farm in the state of Chiapas in southern Mexico (15.1735835, -92.3382748), a 30-h a plot was surveyed in a low-shade conventional coffee farm (15.172465, -92.3301377), and finally a 6-ha plot was surveyed in a rustic part of the coffee farm with relatively high shade levels. We observed a total of 4 colonies in the 45-ha plot and 1 colony in the 30-ha plot ranging from 0.25-1 ha in size (Figure 1). In Puerto Rico we surveyed 10 small coffee farms (mostly < 5 ha) in the mountainous regions of Puerto Rico within the municipalities of Orocovis, Lares, Adjuntas, and Utuado (18.175850 -66.4155700). The largest coffee farm (5-ha) where W. auropunctata populations occurred was located in Orocovis and this farm had the highest shade level (Figure 2).

Field surveys

\label{field-surveys}
Surveys were conducted to map the spatial distributions and abundance of dominant ant species. We walked dense trail systems by placing tuna baits on coffee plants approximately every 4 meters. Baits were checked every 30 minutes for the presence of ant species. A total of 1181 baits were placed within the 45-ha plot, 992 baits within the low-shade coffee farm, and another 229 baits within the rustic 6-ha plot (Figure 1). Detailed descriptions of the 45-ha coffee farm in Mexico have been reported elsewhere (Vandermeer 2010).
In Puerto Rico, we placed a total of 664 baits across all 10 coffee farms (32 baits per ha, approximately 20 ha in total farm size). Site locations in both Mexico and Puerto Rico were selected so that native ant species, both arboreal and ground foraging ants, overlapped in their ranges with W. auropunctata populations. Our baiting method did not set out to sample the entire community, but instead focused on dominant species that engage in competitive interactions (Armbrecht 2003). Differences in farm size reflected the fact that coffee farms in the Soconoscu regions of Mexico are relatively large in size (300-ha) as compared to the central region of Puerto Rico, where the average farm size is considerably small (2-5 ha).

Short-term competition experiment

\label{short-term-competition-experiment}
We conducted short-term competition experiments between W. auropunctata and native ant species. In Mexico, competition experiments were performed on the ground between W. auropunctata and native ants Solenopsis geminata and Pheidole protensa. Arboreal competition experiments in turn were conducted with Solenopsis picea and Pheidole synanthropica species. In Puerto Rico, we conducted short-term experiments between W. auropunctata and the native ground-foraging ant Solenopsis invicta.
For our terrestrial competition experiments, we recorded the timing of resource discovery, recruitment, and the total number of individual workers at baits by placing ten baits (i.e. tuna) in a 1-m line right at the border between ant colonies and subsequently checked the baits at 30-sec intervals for up to 80 minutes was separated by 10 cm within sites where species co-occurred. The intent of this experimental setup was to sample right at the border between monospecific patches were competition took place. For our arboreal competition experiments in the field, we placed five baits on bamboo branches between coffee trees inhabiting ant nests. Each of the five sites was separated by 10 cm located between arboreal nesting sites. We recorded the discovery time, recruitment time, and the total number of individual workers at baits.

\label{section}

Long-term competition experiments

\label{long-term-competition-experiments}
For the long-term competition experiments in Mexico, we used platforms to connect plastic containers inhabiting W. auropunctata and S. picea nests in a common foraging arena for 14-days. Holes were drilled in the plastic containers to allow ants from leaving the nest to the platform were the resources were placed. To prevent ants from escaping we placed the containers in a water bath. Prior to our laboratory experiment, we conducted field trials to identify potential dominant arboreal ants in the field. We selected S. picea nests due to their dominance in coffee plants and relative ease of nest collection. Caution was taking to collect nests that had sufficient amount of queens, workers, and brood in order to keep the nests viable in the lab. Each day the ants were supplied with sugar water and tuna in the lab to enable both species to compete for resources. We used 6 replicates in the experiment with different pairs of unique nest that had not previously been exposed to each other. Controls consisted of nests pairs that were not connected (n =2 pairs) with a platform. In Puerto Rico, we followed the same protocol for assessing the long-term competitive dynamics between W. auropunctata and the arboreal ant L. iniquum . We used 8 connected pairs of W. auropunctata and L. iniquum nests in the experiment and 2 pairs that served as control replicates.

Analysis

\label{analysis}
For all the competition experiments, we analyzed the differences among species with respect to total number of workers at baits. Prior to conducting statistical analyses, we plotted the data to test for normality using the Shapiro-Wilk test. Because of the non-normality, we used the non-parametric Kruskal-Wallis rank sum test with the library stats(R Core Team 2013). Nemenyi’s post-hoc tests were applied for pairwise multiple comparisons using the PMCMR package in R, v. 2.15. ANOVA analysis was performed to detect differences in the number of live ant workers found between treatments in the long-term competition experiments.

\label{section-1}

Results

\label{results}
Within the native range, we found a low overall abundance of W. auropunctata surveyed in Mexico relative to its introduced range of Puerto Rico, where W. auropunctata was found to occur in much higher abundances. In the 45-hectare plot in Mexico (medium-shade organic coffee), we documented a total of 84 morpho-species, 3.4 % of which consisted of W. auropunctata found at baits. In the 30-hectare plot (low-shade conventional farm), we documented a total of 58 morphospecies of which only 0.4% occupied baits by W. auropunctata. In the 6-hectare plot (rustic coffee, within high shaded coffee), we detected W. auropunctata at 7.9 and 3.9% respectively at arboreal baits, whereas ground-baits occupancy by W. auropunctata was only 0.5%. While W. auropunctata distribution on the farm was limited in the native range, we detected a large cluster of W. auropunctata colonies that presumably make up part of a larger super-colony (Figure 1). In Puerto Rico, we detected a total of 16 morpho-species throughout our surveys on 10 smaller-sized coffee farms (ranging from 1-6 ha). On average, we found 41.7 % of arboreal trees occupied by W. auropunctata. In most cases, W. auropunctata was widely distributed across all farms reaching high densities in patches it dominated thereby excluding other ant species from occupying nearby trees.
Overall, we found that W. auropunctata was patchily distributed between Mexico and Puerto Rico. The main difference, however, is that patches dominated by W. auropunctata included a greater diversity of native ant species in Mexico. For example, native species diversity in the largest patch (1 ha) ranged anywhere from 30-50 native ant species, with only 40 % of baits occupied by W. auropunctata (Figure 1). In Puerto Rico, farms had on average less than 20 native ant species present, with the largest patch only containing 2 dominant arboreal ant species (Figure 2).

Short-term competition: Mexico and Puerto Rico

\label{short-term-competition-mexico-and-puerto-rico}
Competition between W. auropunctata and native ant species revealed important differences during contests. Interactions with S. geminata showed that W. auropunctata was able to rapidly recruit workers to baits during the first 20 minutes of the experiment (Figure 3 a), after which S. geminata took over and dominated the majority of baits by the end of the experiment  (\(x^{2}=\)55.976, p < 0.0001​). Competitive interactions between W. auropunctata and P. protensa revealed oscillatory dynamics (Figure 3 b). W. auropunctata rapidly increased at baits within the first 25-minutes after which P. protensa dominated for some time until getting surpassed by W. auropunctata (\(x^{2}\)​​​=15.641, p < 0.0001​). Arboreal competitive interactions between W. auropunctata and S. picea were equally intense. S. picea initially dominated the baits but was overtaken by W. auropunctata after 30-minutes (Figure 3 c). Despite S. picea maintaining a constant recruitment rate throughout the experiment, W. auropunctata steadily increased recruitment of workers enabling it to dominate most baits (\(x^{2}\)= 9.8396, p=0.0017​). However, the arboreal ant P. synanthropica was competitively superior against W. auropunctata (Figure 3 d) during the experiment resulting in the complete dominance of baits (\(x^{2}\)​​​= 69.03, p < 0.0002​). In Puerto Rico, we compared competitive dynamics with W. auropunctata and the dominant ant competitors S. invicta and L. iniquum. W. auropunctata initially increased its recruitment rate in response to S. invicta but was quickly overtaken by S. invicta as more of its workers dominated baits. Although we didn’t detect a significant interaction in the competition experiment (\(x^{2}=\)3.485, p=0.062) S. invicta held on to more baits and its rapid recruitment of workers to baits enabled them to push out W. auropunctata workers (Figure 4).

Long-term competition: Mexico and Puerto Rico

\label{long-term-competition-mexico-and-puerto-rico}
W. auropunctata workers were superior against S. picea during the long-term experiment in Mexico (Figure 5). We observed greater survival of W. auropunctata workers in the 14-day experiment (F=5.43, p=0.013). During the first couple of days, S. picea recruited higher number of workers to the foraging arena and displaced W. auropunctata workers at baits in the majority of replicates. As aggression intensified, S. picea preempted W. auropunctata workers from accessing resources. In response, small groups of W. auropunctata workers attacked S. picea individuals from the arena. In most of the replicates, we observed that W. auropunctata had a clear numerical advantage over S. picea. In five of our replicates, W. auropunctata invaded the nest boxes of S. picea resulting in a high worker density of W. auropunctata including many brood and queens. In one replicate (#3), we observed a nest invasion by W. auropunctata workers into a S. picea nest, with queens and brood moved into the invaded nest. Further inspection into nests with a tiny camera revealed that W. auropunctata workers frequently dispersed their brood and queens within nests. In Puerto Rico, the number of live W. auropunctata workers was not significantly higher than L. iniquum (F=1.87, p=0.17). In the first two days, L. iniquum was much faster at recruiting workers to baits compared to W. auropunctata. In about one quarter of the replicates, few L. iniquum workers were found alive in the nest boxes as they were killed by W. auropunctata workers. We did observe occasional fighting between species, but L. iniquum was quick enough to avoid W. auropunctata workers in the foraging arena (Figure 6).

 

\label{section-2}