Ants constitute one of the most important animal groups in terms of biomass and local abundance (Brandão et al., 2012Brandão, C.R.F., R.R. Silva, and J.H.C. Delabie. 2012. Neotropical ants (Hymenoptera) functional groups: nutritional and applied implications Pp. 213-236. In: Insect bioecology and nutrition for integrated pest management (A. R. Panizzi, J. R. P. Parra, Eds.). CRC, Boca Raton.; Widhiono et al., 2016Widhiono, I.R., D. Pamdhani, D.E. Riwidiharso, S. Santoso and L. Prayoga. 2017. Ant (Hymenoptera: Formicidae) diversity as bioindicator of agroecosystem health in northern slope of Mount Slamet, Central Java, Indonesia. Biodiversitas 18: 1475.1480.). The ants mediate key ecological processes, like nutrient cycling, mixing of organic matter, and soil enrichment and aeration (Folgarait, 1998Folgarait, P. J. 1998. Ant biodiversity and its relationship to ecosystem functioning: a review. Biodiversity and Conservation 7: 1221-144. ; del Toro et al., 2015Del Toro, I., R. R Silva and A. M. Ellison. 2015. Predicted impacts of climatic change on ant functional diversity and distributions in Eastern North American forests. Diversity and Distribution 21: 781-791. ). They also can play important roles as predators, defoliators, scavengers, decomposers, seed dispersers and pollen collectors (Brandão et al., 2012Brandão, C.R.F., R.R. Silva, and J.H.C. Delabie. 2012. Neotropical ants (Hymenoptera) functional groups: nutritional and applied implications Pp. 213-236. In: Insect bioecology and nutrition for integrated pest management (A. R. Panizzi, J. R. P. Parra, Eds.). CRC, Boca Raton.; Kwong et al., 2014Kwong, T.S., C.M. Lee and J.H. Sung. 2014. Diversity decrease of ant (Formicidae, Hymenoptera) after a forest disturbance: different responses among functional guilds. Zoological Studies 53: 1-11.). On the other hand, ants are among the most problematic invasive species (Bertelsmeier et al., 2015aBertelsmeier, C. A., O. Avril, H. Blight, H. Jourdan and F. Courchamp. 2015a. Discovery-dominance trade-off among invasive ant species. Ecology and Evolution 5: 2673-2683. , bBertelsmeier, C. A. , O. Avril, A. Bilgth, L. Confais, H. Diez, J. Jourdan, N. Orivel and F. Courchamp. 2015b. Different behavioral strategies among seven highly invasive ant species. Biological Invasions 17: 2491-2503., 2016Bertelsmeier, C., O. Bligth and F. Courchamp. 2016. Invasions of ants (Hymenoptera: Formicidae) in light of global change. Myrmecological News 22: 25-42. ). They are capable to monopolize space, along with other resources, and thus displace native species, specially in insular areas (Miravete et al., 2014Miravete, V., N. Roura-Pascual, V. Dunn and R.R. Gómez. 2014. How many and which ant species are being accidentally moved around the world? Biological Letter 10: 20140518. Dx.doi.org/10.1098/rsbl.2014.0518.; Roura-Pascual et al., 2016Roura-Pascual, N., N.J. Sanders and C. Hui. 2016. The distribution and diversity of insular ants. Do exotic species play by different rules? Global Ecology and Biogeography 25: 642-654.). In addition, these insects are considered suitable bioindicators of habitat quality and environmental changes (Rivas et al., 2014Rivas, S.P., H. Carrillo-Ruiz, A. Bonilla, D.M. Figueroa-Castro and A.R. Andrés-Hernández. 2014. Effect of disturbance on the ant community in a semiarid region of central Mexico. Applied Ecological Environmental Research 12: 703-716.; Bharti et al., 2016Bharti, H., M. Bharti and M. Pfeiffer. 2016. Ants as bioindicators of ecosystem health in Shivalik Mountains of Himalayas: assesment of species diversity and invasive species. Asian Myrmecology 8: 1-15.).

Ants have been poorly studied in coastal dunes systems. These systems are affected from different sources of natural and anthropogenic disturbances, which include hurricanes, invasive species, global sea-level rise, urbanization, and inadequate management (Rojas et al., 2014Rojas, P., C. Fragoso and W.P. Mackay. 2014. Ant Communities along a Gradient of Plant Succession in Mexican Tropical Coastal Dunes. Sociobiology 61: 119-132.; Chen et al., 2015Chen, X., B. Adams, C. Bergeron, A. Sabo and L. Hooper-Bui. 2015. Ant community structure and response to disturbances on coastal dunes of Gulf of Mexico. Journal of Insect Conservation 19: 1-13. ). This scenario is also the case for Cuban sandy coastal dunes systems, specially those present in areas of intense human activity, like the beaches located northeast from Habana City, known as “Eastern Beaches”; and Varadero Beach, along the Northern coast of the Peninsula of Hicacos.

The typical Cuban sandy coastal vegetation is classified as “sandy coast vegetation complex” and is composed mainly by herbaceous species like Sesuvium portulacastrum, Panicum amarum, Ipomea pes-caprae and Canavalia rosea (Ricardo et al., 2009Ricardo, N., P.B. Herrera, F. Cejas, J. A. Bastard and T. Regalado. 2009. Tipos y características de las formaciones vegetales de Cuba. Acta Botánica Cubana 203: 1-42. ; Álvarez and Ricardo, 2011Álvarez, A. and N. Ricardo. 2011. Flora y vegetación de Playas del Este, Ciudad de la Habana, Cuba II. La vegetación de las dunas. Acta Botánica Cubana 210: 35-44.). Another typical vegetable component of sandy coasts is the sea-grape, Coccoloba uvifera. This species grows, as a shrub formation, behind the herbaceous components of the vegetation complex (Ricardo et al., 2009Ricardo, N., P.B. Herrera, F. Cejas, J. A. Bastard and T. Regalado. 2009. Tipos y características de las formaciones vegetales de Cuba. Acta Botánica Cubana 203: 1-42. ), representing thus a distinctive habitat, continuous or patchy, between the sandy beaches and other plant formations or urbanizations of different kinds.

As field observations suggest, the sea-grape shrubs represent a suitable habitat for ants, providing shelter and resources, such as nectar, fruits and Hemiptera known as “scale insects”. These insects are attended by ants in demanding of the so called “honey dew”, a sugar and amino acid-rich bodily secretion (Newton et al., 2011Newton, J.S., J. Glasier and H.E. Footti. 2011. Ants and subterranean Sternorrhyncha in a native grassland in east-central Alberta, Canada. Canadian Entomologists 143: 518-523. ). In addition, the sea-grape leaf-litter supplies organic matter and retains humidity. Sea-grapes shrubs are usually utilized by persons in pursuit of shadow. The resulting leftovers contribute to the proliferation of ants, mainly invasive ones, and other undesirable organisms, like flies ant rats.

The known Cuban ant fauna is composed by 168 species, belonging to 46 genera and nine subfamilies. The proportion of endemic species is 44.0% and of cosmopolitan species is 18.0% (Fontenla and Alfonso-Simonetti, 2018Fontenla, J.L. and J. Alfonso-Simonetti. 2018. Classification of Cuban ants (Hymenoptera: Formicidae) into functional groups. Poeyana 506: 21-30.). Bertelsmeier et al. (2016)Bertelsmeier, C., O. Bligth and F. Courchamp. 2016. Invasions of ants (Hymenoptera: Formicidae) in light of global change. Myrmecological News 22: 25-42. recognize 19 invasive ant species, and 10 of them are present in Cuba. Some of these species, like Wasmannia auropunctata and Pheidole megacephala are very abundant in Cuban agroecosystems and man-modified environments (Fontenla and Matienzo, 2011Fontenla, J.L. and L. Matienzo. 2011. Hormigas invasoras y vagabundas de Cuba. Fitosanidad 15: 253-259.).

Ants are an ideal group for testing patterns of assemblage organization, because their assemblages tend to be organized by interspecific interactions (Wittman and Gotelli, 2011Wittman, S.E. and N.J. Gotelli. 2011. Predicting community structure of ground-foraging ant assemblages with Markov models of behavioral dominance. Oecologia 166: 207-2019.; Camarota et al., 2016Camarota, F., S. A. Powell, S. Mello, G. R. Priest, J. Marquis and H. L. Vascocelos. 2016. Co-occurrence patterns in a diverse arboreal ant community are explained more by competition than habitat requirements Ecology and Evolution 6: 8907-8918.). Understanding the environmental drivers of community organization in natural ecosystems is of great importance, specially in those severely threatened (Silva et al., 2017Silva, L.C., R.M. Souza, R.R. Solar, C. de Siqueira and F. Neves. 2017. Ant diversity in Brazilian tropical dry forests across multiple vegetation domains. Environmental Research Letter 12: 035002.). In addition, visualizing the structure of ant communities may be useful for understanding the effects of urbanization and for evaluating the qualities of man-modified environments (Santos et al., 2019Santos, M.N., J.H.C. Delabie and J.M. Queiroz. 2019. Biodiversity conservation in urban parks: a study of ground-dwelling ants (Hymenoptera: Formicidae) in Rio de Janeiro City. Urban Ecosystems doi.org/10.1007/s11252-019-00872-8.).

The goal of this paper is to assess species composition and co-occurrence patterns of ant assemblages in sea-grape shrub formations along two sandy beaches in the northern coast of Western Cuba.

Ant sampling. The study was developed by sampling at 12 sites in Eastern Beaches and at four sites in Varadero Beach during the years 2017-2018. The sites were only a reference to locate the quadrats, the level at which the analyses were conducted. Sampling sites in Eastern Beaches were located between the coordinates 23^° 10^' 48.53^''N, -82^° 12^' 15.35^''W, and 23^° 10^' 26.88^''N, -82^° 05^' 57.01^''W. The sites in Varadero Beach were located between the coordinates 23^° 09^'17.68^''N, -81^° 14^' 53.08^''W, and 23^° 12^' 11.19^''N, -81^° 09^' 10.09^''W. The distance from the first site to the last site in Eastern Beaches was about 10 km, and the corresponding distance in Varadero Beach was about 11.5 km. The distance from the last site in Eastern Beaches to the first site in Varadero Beach was about 90 km (Fig. 1). The average annual temperature and rainfall in Eastern Beaches is 25.0 ^°C and 1272 mm, respectively (https://es.climate-data.org/location/874868/), which is similar to Varadero Beach, with 25.2 ^°C and 1264 mm respectively (https://es.climate-data.org/location/30137/).

There were established 20 randomly located 1X1m quadrats in each site, for a total of 240 quadrats at Eastern Beaches and 80 quadrats at Varadero Beach. Each quadrat was separated from each other by at least five meters. In each quadrat, sea-grape litter and sandy soil were examined, as well as trunk and branches of sea-grapes up to 2.0 m high above the ground. Only species frequencies were taken into account as a surrogate for ant abundance. Incidence data, instead of number of individuals, avoids biases due to sampling near nests or trails (Croc et al., 2014Croc, S., J. H. C. Delabie, F. Fernández, M. Leponce, J. Oribel, R. Silvestre, H. S. Vasconcelos and A. Dejean. 2014. Leaf-litter ant communities (Hymenoptera: Formicidae) in a pristine Guianese rainforest: stable functional structure versus high species turnover. Myrmecological News 19: 43-51. ). The ant species were assigned to functional groups (Fontenla and Alfonso-Simonetti, 2018Fontenla, J.L. and J. Alfonso-Simonetti. 2018. Classification of Cuban ants (Hymenoptera: Formicidae) into functional groups. Poeyana 506: 21-30.). For identification, specimens were compared at the ant collection of the Institute of Ecology and Systematic, Cuba.

Data analysis. The completeness of sampling effort across quadrats was assessed using two non-parametric species richness estimators: 1. Model (h) (ICE). It assumes that detection probabilities are hetererogeneous among species. 2. Model (th). It assumes that detection probabilities vary not only among species, but also among samples (quadrats). This analysis was conducted by the SPADE program (Chao and Shen, 2009Chao, A. and T.J. Shen. 2009. Program Spade (Species prediction and diversity estimation). Available in http://chao.stat.nthu.edu.tw/software . Last access: September 17 of 2019.).

The confidence limits (95.0%) of variables were computed by the adjusted percentile method with 9 999 bootstrapping replicates. Correlations were running by Spearman’s rank-order correlation coefficient (rs). Statistical significance was assessed by 9999 random replicates of Monte Carlo permutation test. The program used was PAST 3.15 (Hammer, 2017Hammer, Ø., D.A.T. Harper y P. D. Ryan. 2001. PAST: Paleontological Statistics software package for education and data analysis. Paleontologia Electronica 4(1): 9 pp.).

A rarefaction test was applied to compare ant species richness between the two beaches. There were ran 1000 iterations with independent sampling to rarefy the larger assemblage down to the frequency level of the smaller, and checked if the observed species richness of the species poorer assemblage fell within the 95% confidence limit. This would mean that both assemblages do not differ significantly in species richness. This analysis was made by the program EcoSim version 7.0 (Gotelli and Entsminger, 2012Gotelli, N.J. and G.L. Entsminger. 2012. EcoSim 7.72. Acquired Intelligence, Inc. Available in http://www.uvm.edu/~ngotelli/EcoSim/EcoSim.html . Last access: September 17 of 2019.).

Total beta dissimilarity was given by: $\text{βto}=\text{b}+\text{c}/\text{a}+\text{b}+\text{c}$ , where b is the number of species exclusive of one beach; c is the number of species exclusive of the second beach, and a is the number of shared species between beaches. Beta dissimilarity was decomposed into species replacement plus differences in species richness $\left(\text{βto}=\text{βre}+\text{βri}\right)$ . $\text{βre}=2\text{min}\left(\text{b},\text{c}\right)/\text{a}+\text{b}+\text{c}$ . $\text{βri}=\left[\text{b}-\text{c}\right]/\left(\text{a}+\text{b}+\text{c}\right)$ . The relative importance of replacement was calculated as: $\text{Tre}=\text{βre}/\text{βto}$ , and the relative importance of difference in species richness by: $\text{Tri}=\text{βri}/\text{βto}$ (Tonial et al., 2012Tonial, M.L., S.H. Silva, I.J. Tonial, M.C. Costa, N.J. Silva and J.A. Diniz-Filho. 2012. Geographical patterns and partition of turnover and richness components of beta-diversity in faunas from Tocantins river valley. Brazilian Journal of Biology 72: 497-504.; Podani y Schmera, 2016Podani, J. and D. Schmera.2016. Once again on the components of pairwise beta diversity. Ecological Informatics 32: 63-68.).

Based on the total frequency of species, it was calculated as a diversity index the “effective number of species” ${\text{e}}^{\text{H}}$ , where H is the Shannon-Wiener index of entropy. The assemblages’ frequency evenness was calculated as ${\text{e}}^{\text{H}}/\text{S}$ , where S is the species richness of a beach (Jost, 2010Jost, L. 2010. The relation between evenness and diversity. Diversity 2: 207-232. ; Aisling et al., 2018Aisling J. D., J. M. Baetens and B. de Baets. 2018. Ecological Diversity: Measuring the unmeasurable. Mathematics 6: 119. Doi:10.3390/math6070119. ). The frequency contribution of the most frequent species was specified by the Berger-Parker index. These indices were compared by running a permutation test of 10 000 random matrices PAST 3.15 (Hammer et al., 2001Hammer, Ø., D.A.T. Harper y P. D. Ryan. 2001. PAST: Paleontological Statistics software package for education and data analysis. Paleontologia Electronica 4(1): 9 pp.).

Co-occurrence patterns across quadrats were analyzed by the C-score (checkerboard species pairs) and the number of unique species combinations. These indices were compared with the average simulated values from 5000 randomly assembled matrices by a swapping algorithm and the fixed-fixed model. The analyses were conducted with EcoSim 7.0 software. In addition, there were identified the species pairs that showed significant positive or negative spatial associations by applying the togetherness index with 1000 simulated distribution of the fixed-fixed null model. The program used was PAIRS (Ulrich, 2008Ulrich, W. 2008. Pairs - a FORTRAN program for studying pair-wise species associations in ecological matrices. Version 1.0. Available in www.uni.torun.pl/~ulrichw . Last access: September 17 of 2019).

A nested configuration is a pattern of spatial order where common species tend to occur in any sites, while rare species tend to occur only in the richest sites. The degree of nestedness was calculated with the NODF index, based in overlap and decreasing fill of the species in the matrix. It was applied the fixed-fixed null model with 1000 iterations using the NODF Program (Ulrich, 2010Ulrich, W. 2010. NODF- a FORTRAN program for nestedness analysis Version 1.0 [Unpublished]. Nicolaus Copernicus University in Toruń. 9 pp.). Nestedness was calculated for the whole incidence matrix (NODFm), sites (NODFc) and species (NODFr). Site (quadrats) nestedness occurs when species present at species-poor sites are a subset of the species present at species-rich sites. Species nestedness occurs when the occurrences of species occupying few sites are a subset of the occurrences of species occupying more sites (Novak et al., 2011Novak, M.J., W. Moore and R. Leidy. 2011. Nestedness patterns and the dual nature of community reassembly in California streams: a multivariate permutation-based approach. Global Change Biology 17: 3714-3723. ).

The softwares mentioned above offer a standardized (z-transformed) effect size (SES), which expresses the direction and degree of deviation from the null model. SES values ≥ [2.0] point to statistical significance. If SES is positive indicates less co-occurrences, more unique combinations and more positive spatial associations than expected by chance alone, as well as absent of nestedness (segregation). If SES is negative indicates a higher co-occurrence, less unique species combinations and more negative spatial associations than expected by chance alone as well as nestedness (aggregation).

There were observed 21 species in Eastern Beaches, and 13 species in Varadero Beach, for a total of 23 species between the two beaches (Supplementary material 1, 2). The 21 species observed in Eastern Beaches fell within 92.5%-93.3% of the estimated species richness for that locality, and the 13 species observed in Varadero Beach fell within 88.4%-89.0% of the expected species richness for that beach. According to these values, sampling effort was adequated in both beaches (Table 1). Species richness rarefaction pointed out that the confidence interval of Eastern Beaches showed values from 13 species (lower limit) to 19 species (upper limit). The 13 species observed at Varadero Beach fell within the lower limit of this interval. Therefore, the two beaches did not differ in species richness.

Myrmicinae was the species richest and most frequent subfamily at the two localities, followed by Formicinae. Most of the species with the highest frequency (above upper confidence limit) were common between the two localities, like Wasmannia auropunctata, Pheidole megacephala and Paratrechina longicornis. Total species frequencies were significantly correlated between the two assemblages (r_s= 0.72, pep= 0.0001). The cosmopolitan species represented 43.0% of the species richness and 69.4% of total assemblage frequency in Eastern Beaches, whereas comprised 54.0% of the species richness and 61.0% of total assemblage frequency in Varadero Beach. The dominant omnivorous of ground and vegetation, ground and vegetation opportunists, and arboreal functional groups, accounted to the highest frequency and species richness in the two beaches (Table 2).

Both beaches shared 11 species. Species composition dissimilarity yielded a beta value of 52.2%. The difference in species richness component contributed the most to the dissimilarity value, with 34.8%. The replacement component contributed with 17.4%. The relative importance of difference in species richness was 66.7%, and the relative importance of replacement was 33.3%.

There were observed between 1-4 species across quadrats in Eastern Beaches and between 1-3 species in Varadero Beach (Supplementary material 1, 2), with similar mean species per quadrats. Both of the areas also showed similar values of species diversity (effective number of species). The main difference between species compositions were evenness and the frequency of the most frequent species. According to the confidence limits, the evenness in Varadero Beach assemblage was significantly higher than in Eastern Beaches. On the other hand, the contribution of the most frequent species in Eastern Beaches (W. auropunctata) was significantly higher than the contribution of the most frequent species in Varadero Beach (P. megacephala) (Table 3).

Species co-occurred less and there were less species combinations than expected by chance in both beaches, respectively. The values of the standard Effect Size were high for the two indexes, especially in Varadero Beach, with over nine standard deviations, either for the C-Score or for the species combinations (Table 4).

The pattern of species-pairs association among quadrats showed that four species pairs had significant negative association and five species pairs significant positive associations at Eastern Beaches. At Varadero Beach, six species pairs had a pattern of significant negative association, and four species pairs significant positive associations. The species pair Pheidole megacephala-Wasmannia auropunctata displayed the most intense negative spatial association in both beaches. These two species, along with Paratrechina longicornis, were mostly involved in negative spatial interactions. The species more frequently involved in positive associations in both places was Brachymyrmex obscurior. The most intense positive associations corresponded to the species pair Paratrechina longicornis-Dorymyrmex pyramicus at Eastern Beaches, and to Wasmannia auropunctata-Camponotus planatus at Varadero Beach (Table 5).

According to confidence limits and SES values, the whole incidence matrix and quadrat species composition showed a significant nested pattern in both of the beaches. On the contrary, species distribution per quadrats pointed to a segregated tendency (positive sign) in Varadero Beach, which was statistically significant in Eastern Beaches (Table 6).

The 23 ant species detected in the two beaches represent 14.0% of the Cuban myrmecofauna. In a sea-grape formation located in a more complex vegetation matrix, between a subcoastal forest and a rocky coast vegetation complex, Fontenla (1993)Fontenla, J.L. 1993. Composición y estructura de comunidades de hormigas en un sistema de formaciones vegetales costeras. Poeyana 441: 1-19. found 27 ant species in only 20 quadrats, but Increasing in ant species richness is expected with a corresponding increasing in vegetation structure complexity (Rojas et al., 2014Rojas, P., C. Fragoso and W.P. Mackay. 2014. Ant Communities along a Gradient of Plant Succession in Mexican Tropical Coastal Dunes. Sociobiology 61: 119-132.; García-Martínez et al., 2015García-Martínez, M.A., D.L. Martínez-Tlapa, R. Pérez-Toledo, L.N. Quiroz-Robledo, G. Castaño-Meneses, J. Laborde and J.E. Valenzuela-González. 2015. Taxonomic, species and functional group diversity of ants in a tropical anthropogenic landscape. Tropical Conservation Science 8: 1017-1032. ). Several of the recorded species are shared with other sandy systems elsewhere. Cupul-Magaña (2006)Cupul-Magaña, F. G. 2006. Registro de hormigas (Hymenoptera: Formicidae) en un remanente de dunas costeras de Puerto Vallarta, Jalisco, México. Ciencia y Mar X: 25-31. reports Dorymyrmex pyramycus, Solenopsis geminata and Brachymyrmex obscurior from Mexican Pacific coastal dunes, while Rojas et al. (2014)Rojas, P., C. Fragoso and W.P. Mackay. 2014. Ant Communities along a Gradient of Plant Succession in Mexican Tropical Coastal Dunes. Sociobiology 61: 119-132. mention among the most frequent ants in Mexican Atlantic coastal dunes to S. geminata and Camponotus planatus.

With the exception of Brachymyrmex obscurior, the rest of the most frequent species in the two areas, the big-headed ant (Pheidole megacephala), the little red fire ant (Wasmannia auropunctata) and the long-horn crazy ant (P. longicornis), are cosmopolitan-invasive species (Wetterer, 2012Wetterer, J.K. 2012. Worldwide spread of the African big-headed ant, Pheidole megacephala (Hymenoptera: Formicidae). Myrmecological News 17: 51-62. , 2015Wetterer, J.K. 2015. Geographic origin and spread of cosmopolitanants (Hymenoptera: Formicidae). Halteres 6: 66-78. ; Bertelsmeier et al., 2016Bertelsmeier, C., O. Bligth and F. Courchamp. 2016. Invasions of ants (Hymenoptera: Formicidae) in light of global change. Myrmecological News 22: 25-42. ). Other cosmopolitan ants are present in the beaches, such as Monomorium pharaonis, Nylanderia bourbonica, Solenopsis geminata, and Tapinoma melanocephalum. These ants have become also major ecological, agricultural or household pest species (Wetterer, 2015Wetterer, J.K. 2015. Geographic origin and spread of cosmopolitanants (Hymenoptera: Formicidae). Halteres 6: 66-78. ), but they are not very frequent in the studied environment.

The predominant functional groups are the dominant omnivorous of ground and vegetation, and the ground and vegetation opportunists. Opportunist species was the most diverse functional group across the dune systems in the north coast of the Gulf of Mexico. This is related to the ability of opportunist ants to withstand natural disturbance on coastal dunes, such as sand burial and strong winds (Chen et al., 2015Chen, X., B. Adams, C. Bergeron, A. Sabo and L. Hooper-Bui. 2015. Ant community structure and response to disturbances on coastal dunes of Gulf of Mexico. Journal of Insect Conservation 19: 1-13. ). In the sea-grape shrub habitat studied here, this last factor can be less intense, because of the sheltering of the sea-grape shrub itself, although this habitat is also affected by human intrusion, which frequently results in sand burial and disturbance of ant colonies. Predominance of dominant and opportunist species can be facilitated by a lack of any major change in vegetation (Underwood and Fisher, 2006Underwood, E.C. and B.L. Fisher. 2006. The role of ants in conservation monitoring: If, when and how. Biological Conservation 132: 166-182. ), and by the relative homogeneity of the habitat (Croc et al., 2014Croc, S., J. H. C. Delabie, F. Fernández, M. Leponce, J. Oribel, R. Silvestre, H. S. Vasconcelos and A. Dejean. 2014. Leaf-litter ant communities (Hymenoptera: Formicidae) in a pristine Guianese rainforest: stable functional structure versus high species turnover. Myrmecological News 19: 43-51. ). Both evironmental conditions tipify the sea-grape habitat.

Species replacement contributes more to beta dissimilarity across different vegetation types (Silva et al., 2017Silva, L.C., R.M. Souza, R.R. Solar, C. de Siqueira and F. Neves. 2017. Ant diversity in Brazilian tropical dry forests across multiple vegetation domains. Environmental Research Letter 12: 035002.). The component that contributes more to beta dissimilarity in these ant assemblages is difference in species richness, whose relative importance is twice as much as that of the species replacement component. This pattern of dissimilarity should be expected across plots in homogeneous habitats.

The assemblages showed very similar frequency proportion of shared species and diversity. Structural differences in species composition between the beaches are centered in evenness and dominance values. Varadero Beach exhibits the highest evenness and lowest dominance index. Its most frequent species, Pheidole megacephala has a lesser proportional frequency that his similar in Eastern Beaches, Wasmannia auropunctata. In general terms, the presence of W. auropunctata is associated with a low evenness of ant communities, as a result of the great population density that this species is able to reach (Wauters et al, 2016Wauters, N., F. Dekoninck, H. Hendrickx, W. Herrera and D. Fournier. 2016. Habitat association and coexistence of endemic and introduced ant species in the Galápagos Islands. Ecological Entomology 41: 40-50. ), which matches the situations observed here.

High values of the checkerboard index, combined with fewer species combinations than expected by chance, point to a restricted coexistence pattern of species across quadrats. Like in this study, stable environments have been associated with significant negative species interactions (Ulrich et al., 2012Ulrich, W. , M. Zalewski and A. Uvarov. 2012. Spatial distribution and species co-occurrence in soil invertebrate and plant communities on northern taiga islands. Annals Zoological Fennici 49: 161-173. ) or checkerboard segregation (Fayle et al., 2015Fayle, T. M., P. Eggleton, A. Manica, K. M. Yusah and W. A. Foster. 2015. Experimentally testing and assessing the predictive power of species assembly rules for tropical canopy ants. Ecology Letters 18: 254-262. ). This kind of spatial pattern suggests that assemblages are structured mainly by strong species interactions or habitat constraints (Ulrich and Gotelli, 2007Ulrich, W. and N.J. Gotelli. 2007. Disentangling community patterns of nestedness and species co-occurrence. Oikos 116: 2053-2061. ). Ant assemblage structure at small or local spatial scale is regulated mainly by interspecific interactions (Olivier et al., 2014Olivier, B., J. Orgeas, F. Torre and E. Provost. 2014. Competitive dominance in the organization of Mediterranean ant communities. Ecological Entomology 39: 595- 602.; Silva et al., 2017Silva, L.C., R.M. Souza, R.R. Solar, C. de Siqueira and F. Neves. 2017. Ant diversity in Brazilian tropical dry forests across multiple vegetation domains. Environmental Research Letter 12: 035002.). Chen et al. (2016)Chen, X. , B. Adams, A. Sabo, T. Crupi and L. Hooper-Bui. 2016. Ant assemblages and co-occurrence patterns in Cypress-Tupelo Swamp. Wetlands DOI 10.1007/s13157-016-0795-y. assert that dominant ant species tend not to occur together in the same patch due to competition for resources. This spatial pattern is coherent with these results.

Negative spatial interactions must be more evident between dominant/invasive species, which usually show non-random mutually exclusive distributions (Sanders et al., 2007Sanders, N.J., N. J. Gotelli , S.E. Whittman, J.S. Tarchford, A.M. Ellison and E.S. Jules. 2007. Assembly rules of ground-foraging ant assemblages are contingent on disturbance, habitat and spatial scale. Journal of Biogeography 34: 1632-1641.; Chen et al., 2016Chen, X. , B. Adams, A. Sabo, T. Crupi and L. Hooper-Bui. 2016. Ant assemblages and co-occurrence patterns in Cypress-Tupelo Swamp. Wetlands DOI 10.1007/s13157-016-0795-y.). This kind of distribution are observed in this study mainly between dominant-invasive species, like Pheidole megacephala and Wasmannia auropunctata. This species-pair exhibits the highest negative SES values in both beaches. These ants are aggressive species (Holway et al., 2002Holway, D.A., L. Lach, A.V. Suarez, N.D. Tsutsui and J. Ted. 2002. The causes and consequences of ant invasions. Annual Review of Ecology and Systematics 33: 181-233.; Armbrecht and Ulloa-Chacon, 2003Armbrecht, I. and P. Ulloa-Chacon. 2003. The Little Fire ant Wasmannia auropunctata (Roger) (Hymenoptera: Formicidae) as a Diversity Indicator of Ants in Tropical Dry Forest Fragments of Colombia. Environmental Entomology 32: 542-547. ) that have been observed displaying mutually exclusive distributions (Xerdá et al., 2011Xerdá, X., E.S. Angulo and S. Caut. 2011. Ant community structure on a small Pacific island: only one native species living with the invaders Dugesiana 17: 113-124.; Bertelsmeier et al., 2015bBertelsmeier, C. A. , O. Avril, A. Bilgth, L. Confais, H. Diez, J. Jourdan, N. Orivel and F. Courchamp. 2015b. Different behavioral strategies among seven highly invasive ant species. Biological Invasions 17: 2491-2503.) or mosaic patterns of dominant ant species (Franken and Gasnier, 2010Franken, E.P. and T.R.J. Gasnier. 2010. Applicability of the ant mosaic theory on floor litter in a forest of Central Amazon, Brasil. Entomotropica 25: 37-42. ). These two species, along with Paratrechina longicornis, are involved in most of the negative interactions, and thus they do not tend to overlap their spatial distribution across quadrats, either among themselves or with other species.

The observed mutual spatial exclusion pattern by several species-pairs might be a consequence of competition for resources in the homogeneous environment represented by the sea-grape shrub formation. On the other hand, subordinate species can coexist with dominant species (Wittman and Gotelli, 2011Wittman, S.E. and N.J. Gotelli. 2011. Predicting community structure of ground-foraging ant assemblages with Markov models of behavioral dominance. Oecologia 166: 207-2019.; Castracani et al., 2014Castracani, C., Spoutti, F. A. and Grasso, D. A. 2014. A new exception to the dominance-discovery trade-off rule in ant communities. Journal of Zoology 97: 171-175. ). This assertion is also consistent with these results, since most of the positive interactions are displayed among opportunist species, such as Brachymyrmex obscurior, Cardiocondyla emeryi, Dorymyrmex pyramicus, and Paratrechina longicornis, or between a dominant species (Wasmannia auropunctata) and a habitat specialist (Camponotus planatus).

In nested assemblages, common species tend to occur in all sites and rare species in the richest sites (Ricotta and Pavoine, 2015Ricotta, C. and S. Pavoine. 2015. A multiple-site dissimilarity measure for species presence/absence data and its relationship with nestedness and turnover. Ecological Indicators 54: 203-206. ). This tendency is observed in both ant assemblages, where the majority of the quadrats (with only one species) are occupied by the most frequent and abundant species, like Wasmannia auropunctata or Pheidole megacephala. Nestedness is to be expected in sites within a homogeneous area (Ulrich et al., 2009Ulrich, W. , M. Almeida-Neto y N.J. Gotelli. 2009. A consumer’s guide to nestedness analysis. Oikos 118: 3-17. ), and across temporarily stable sites (Ulrich et al., 2012Ulrich, W. , M. Zalewski and A. Uvarov. 2012. Spatial distribution and species co-occurrence in soil invertebrate and plant communities on northern taiga islands. Annals Zoological Fennici 49: 161-173. ), which is also coherent with the characteristics of the studied sea-grape formations.

Nestedness is a pattern of aggregated species co-occurrence; therefore, negative interactions decreases the matrix nestedness (Ulrich et al., 2009Ulrich, W. , M. Almeida-Neto y N.J. Gotelli. 2009. A consumer’s guide to nestedness analysis. Oikos 118: 3-17. ). Nestedness (aggregation) and negative co-occurrences (segregation) represent alternatives patterns, but the same matrix can show both assemblage attributes (Ulrich and Gotelli, 2007Ulrich, W. and N.J. Gotelli. 2007. Disentangling community patterns of nestedness and species co-occurrence. Oikos 116: 2053-2061. ; Ulrich et al., 2009Ulrich, W. , M. Almeida-Neto y N.J. Gotelli. 2009. A consumer’s guide to nestedness analysis. Oikos 118: 3-17. ). This configuration is present in the species distribution component across quadrats, being a tendency in Varadero Beach and a significant pattern in Eastern Beaches.

Several of the most frequent species, such as Wasmannia auropunctata, Pheidole megacephala and P. longicornis, are involved in most of the negative interactions, and thus they do not tend to overlap their spatial distribution across quadrats, either among themselves or with other species. These interactions yield a segregated pattern of spatial distribution of species across quadrats, no matter the significant degree of nestedness of the whole incidence matrix and quadrat species composition.

In summary, the ant assemblages are characterized by high frequency of widespread dominant and opportunist species. Species richness and diversity are not significant different. Co-occurrence patterns are driving mainly by negative spatial interactions, especially between dominant-invasive species. It can be concluded that both ant assemblages exhibit similar ecological structure with significant nested pattern for the whole matrix of species incidence and species composition across quadrats.