To compare whether the proportion of occupied nests differed with nest strata or the diversity of nest entrance sizes available, we used generalized linear mixed models with ‘glmer’ in the ‘lme4’ package in R . We compared two models. In the first, we included nest strata , nest size treatment , and the interaction between the two as fixed factors, the vegetation complexity index as a covariate, and site as a random factor. In the second, we removed the VCI. To select the best model, we used the Akaike’s Information Criterion computed with the ‘mass’ package . For both models, we used the binomial error distribution with the logit link. Instead of using the proportion data directly, we used the ‘cbind’ function with number of nests occupied and number of nests that were not occupied as input variables. To examine whether species richness differed with nest strata or the diversity of nest sizes available, we used two methods. First, we compared the mean species richness of ants occupying nests on a plant with GLMMs with ‘glmer’ in the ‘lme4’ package in R . We compared two models. In the first, we included nest strata , nest size treatment , and the interaction between the two as fixed factors, the vegetation complexity index as a covariate, and site as a random factor. In the second, we removed the VCI. To select the best model, we used the Akaike’s Information Criterion computed with the ‘mass’ package . For both models, we used a Poisson error distribution with the log link. Second, we created sample-based species accumulation curves, scaled to the number of individuals, to compare richness in coffee plants vs. trees and diverse vs. uniform nest size treatment plants with EstimateS . We used the number of ant colonies encountered instead of the number of individuals, growing blackberries in containers as ants are social organisms and better captured by number of colonies .
We examined curves for both observed species richness and plotted 95% confidence intervals to statistically compare species richness between treatments. To compare whether community composition of ants differed with strata and with nest size treatment, we used two methods. We used non-metric multidimensional scaling and analysis of similarities in PAST to visually and statistically compare species composition of the ants occupying nests in coffee vs. shade trees and in uniform vs. diverse nest treatments. The ANOSIM compares the mean distance within groups to the mean distance between groups, and can statistically determine separation in species composition between the plots in different treatment groups. For the NMDS and ANOSIM we used the Bray-Curtis similarity index as the similarity measure. Finally, we examined whether common ant species more frequently colonized nests of a certain entrance size or vegetation strata. To compare if nests with certain entrance sizes were more frequently occupied by ants we used an ANOVA followed by a Tukey’s test to compare the mean proportion of nests of each entrance size that were occupied. We only used data from the diverse treatment plants to calculate differences in nest colonization. To compare if certain ant species more frequently occupied certain nest sizes or strata we performed Chi-squared analysis which is recommended for categorical data and tests the likelihood that an observed distribution is due to chance .Ecological studies strive to understand local and regional factors that influence community assembly and species coexistence . Some factors important in for assembly of arboreal twig-nesting ants include presence of a canopy dominant species and resource access through canopy connectivity . Previous studies have found that diversity of nesting resources influences the colonization process of leaf-litter twig-nesting ants and of tropical arboreal ants and that the abundance of nesting resources may impact colonization of arboreal twig-nesting ants .
The study of assembly in ant communities in a spatial context reveals that species sorting, by which different species specialize in a particular habitat, and mass effects, in which species disperse from less to more suitable habitats, are likely important for common and rare species, respectively, in agroecosystems — habitats embedded in landscape mosaics were local communities interact through dispersal . Our study is novel in that we examined colonization in a managed ecosystem looking at two factors and their importance in colonization. In this study, we suggest that nesting resource utilization, specifically different frequencies of occupation of specific nest entrance sizes and specific nesting strata are important drivers of community assembly. Overall, l we found that nesting strata and the diversity of nest entrance sizes did not significantly influence the proportion of occupied artificial nests. Thus, ants use newly available cavities for colonization and nesting resources are somewhat limiting for the community of twig-nesting ants in the habitat studied. In comparison to our study, Powell et al. found that total nest occupancy was higher with higher nest cavity diversity in the Brazilian savanna . It is possible that such distinct results derive from differences in overall nest availability, differences in vegetation and differences in the abundance of particular genera . However, it is important to consider that the near-saturation found in the present study could be a result of adding only one cavity of each size per plant, which could mean that there were not enough nests to be colonized, once the “preferred” sizes were used on every plant —hence not available for other species to occupy. Differences in nest saturation and the proportion of nest occupation between both studies could be due to differences in the number of cavities per size used in the experiment. Thus, it is difficult to say that differences in nest limitation are due to the agroecological context, since previous studies in coffee plantations have found that the community of twig nesting ants are limited by nesting resources , as are ants in natural ecosystems . In addition, differences in occupation dynamics of artificial nests during the colonization phase could potentially change with length of the study. A clear contrast is that the present study lasted three months, a third of the previous study, this difference in time could potentially influence competition for “preferred” cavities during colonization, as these are available for a longer period of time during the colony life cycle. Very little information is available about the reproductive phenology of arboreal twig-nesting ants. The evidence collected from our nests indicates that all common species were producing larvae and pupae, and that most species nests did contain alate males. Two of the common species collected from nests in the present study do experience queen flights during this time period , but information is lacking for the other species. Thus, timing of nest placement may have affected the colonization processes, but it is important to note that many twig-nesting species expand by colony budding, and not only nuptial flights. Changes in the occupation dynamics —i.e. proportion of occupied nests, changes in diversity and species interactions—through time, could be the focus of future studies. Even though diversity of nest entrance sizes did not influence the percentage of occupation overall, frequency of occupation of nests by ants did differ for particular sizes. Higher occupancy was found in middle sizes , these results are similar to Powell et al. in which middle sizes were the most frequently occupied.
The specificity in the use of particular sizes is important in two ways: first, the evolution of ecological specialization underlies the evolution of morphological specialization in ant soldiers, Powell showed that for different species of Cephalotes anincrease in ecological specialization corresponded to a higher head specialization ; in that same study C. persimilis uses cavities that match the size of one soldier’s head and it has also evolved a highly specialized complete headdisc, while less ecologically specialized Cephalotes species, square pot like C. pusilus have evolved a domed-head. Second, such size specialization maximizes individual nest survival and is likely to have a positive effect on overall colony reproduction as shown previously for C. persimilis, which more frequently nests in cavities that fit its head size . On the other hand Cephalotes ants using cavities larger than their soldier’s head, allows them to protect the nest using cooperative blocking . The present study supports the former hypothesis , in that the Cephalotes species present in our study , a domed-headed soldier morphotype, was more frequently found in the largest size , an entrance size much larger than the ant’s head maximum-recorded width . Other Cepahlotes species prefer natural nest sizes between four and up to ten times their head size . If C. basalis shows a similar preference, and if we assume a maximum head size of ~5 mm , than its preferred size might be between the 16 mm2 and 32 mm2 nests offered in this study.Mean species richness was not different in artificial nests on coffee plants and trees, however the diversity of nest entrance sizes increased mean species richness on individual trees and coffee plants. In contrast to a previous study in which diversity of nest cavities did not significantly affect the number of ant species per tree , we did find that providing a diverse array of twig entrance sizes promoted local ant species richness in both coffee shrubs and shade trees. This supports the idea that making a diversity of resources available in both strata supports a more diverse mix of arboreal twig nesting ants. That we found more species richness per tree when providing a higher diversity of nest sizes could indicate that competition for resources might happen more intensively at the local scale, rather than at larger spatial scales. Namely, in a study of leaf-litter twig-nesting ants in shade coffee plantations in Colombia, 80% more species were found when providing a diverse mix of twigs rather than a monospecific collection of twigs showing that diversity of twig-nesting ants is influenced by other aspects of diversity of nesting resources. We found that certain ant species more frequently occupied particular sizes and this may be in part, an explanation for why we found higher species richness on individual plants with a diversity of nest entrance sizes. Armbrecht et al. showed the importance of a diverse mix of twigs for species richness, however the driver in their study was not preference of different ant species for a different species of twigs, but rather an emergent property of a diverse mix of twigs. In our study, we provide evidence that species sorting along a size gradient likely explains the differences observed in mean species richness in uniform vs. diverse treatments. The frequency of occupation differed between sizes for certain ant species, largely following differences in ant head sizes . As small ants can occupy a nest with a wide array of entrance sizes, larger ants can only occupy nests with entrances sizes larger than the workers. Thus providing a wider diversity of nest sizes may allow for greater niche differentiation in the ant community. This outcome might increase the overall richness of the ant community or on individual plants. In our study, larger ants seem to be more size limited than smaller ants, likely because larger ants simply cannot fit into the nests with smaller entrance sizes, and thus are directly constrained by the availability of twigs that fit their body dimensions . In vastly different systems, similar properties operate. For example, in aquatic systems, water temperatures can limit temporal and spatial distribution of certain species as morphological constraints can significantly limit species’ access to suitable habitats . Alternatively, models of exploitative competition have suggested that when two species compete for one limiting resource the result of such competition is determined by the species more capable to attain the lowest equilibrium resource concentration possible, R* . In other words, R* becomes a factor that is the lowest extent to which a certain species can survive in a certain area. Community composition varied between plants with uniform vs. diverse nest entrance sizes, as well as in coffee plants and shade trees. Our results are consistent with previous studies that have investigated ant stratification in the rainforest, where there is a strong partitioning of ant species in the leaf litter, lower vegetation and canopy . Likewise, tropical ant activity is often higher in the canopy than in the litter environment, and species composition differs between the canopy and litter assemblages .