Tropical arboreal twig-nesting ants formed a
linear dominance hierarchy in our study system. We found that M.
mexicana was ranked the highest, using Elo-rating method, while
P. ejectus species ranked the lowest in the hierarchy. We further
estimated the level of uncertainty associated with the hierarchy and
found an intermediate level of steepness in dominance suggesting a
relatively stable hierarchy. The randomized Elo-rating repeatability was
0.93 and remained very stable independently of the ratio of interactions
to individuals. To verify that this pattern was not due to a lack of
sampling effort, we correlated the two halves of the interaction dataset
and found that our sampling effort was 25.8, falling above the minimal
recommended range of 10-20 species. Overall, we find that the
probability of a higher ranked individual winning a contest against a
lower ranked is relatively high, which corroborates our qualitative
finding of the hierarchy steepness. Motif analysis of the network
further revealed a significant excess of transitive interactions. Thus,
the degree of orderliness in the hierarchy is maintained at higher
levels of interactions, or in other words, dominance is relatively
linear, and transitive, indicating that there is a strong, structured
hierarchy in nest-site competition for arboreal twig-nesting ants in
coffee
agroecosystems.
A question stemming from these results is what effect this might have on structuring the distribution and relative abundance patterns of ants under natural conditions. In Mediterranean ecosystems, dominant and subordinate ants are partitioned on the basis of their life-history traits (Arnan et al. 2012). Dominant ant species had more abundant colonies and displayed increased defense for resources in contrast to subordinate ant species. Meanwhile, subordinate ants exemplified greater tolerance to higher temperatures (Cros
et al. 1997; Cerdá et al.
1998b ). The outcome of interspecific interactions within the dominance hierarchy was contingent on environmental conditions (Arnan et al. 2012). In a temperate forest ecosystem of North Carolina, dominance was context dependent (Stuble et al. 2017). Rankings on the basis of bait monopolization revealed that dominance correlated positively with relative abundances since the most abundant species were ranked higher in the dominance hierarchy. In contrast, rankings based on aggressive encounters did not correlate with abundance. In some habitats, dominance patterns are largely determined by the time of day that foraging occurs (Cerdá
et al. 1998b ; Bestelmeyer
2000). In the North Carolina system, the most abundant ant species Aphaenogaster rudis was most active during the morning hours, whereas the cold-tolerant Prenolepis imparis species was dominant during the night hours (Stuble et al. 2017). The ranking of species also depends on the size of resources. In an assemblage of woodland ants, smaller-sized ants were more efficient at acquiring and transporting fixed resources. Bigger sized solitary ants excelled at retrieving smaller food that were mobile during competitive interactions. However, the introduction of phorid parasitoids in this system reduced the transitive hierarchy facilitating the coexistence of subdominant ants (LeBrun 2005; Lebrun and Feener 2007). In our study on competition for nesting sites in the lab, we were able to used fixed resources and to a certain degree control variation in colony size.
Regardless of the ecological factors driving dominance hierarchies under natural conditions, it’s important to note that ranking methods vary considerably among studies (Stuble et al. 2013). Traditionally, field studies have quantified dominance relationships on the basis of proportion of contests won. Other studies have use more sophisticated methods to account for competitive reversals (Vries 1998) or have updated rankings based on relative wins and losses during contests (Colley 2002). In this study, we used the Elo-rating by randomizing the sequence of interactions and calculating the mean of individual Elo-ratings (Sánchez‐Tójar
et al.
2017). By calculating the repeatability of the individual Elo-rankings, we estimated the uncertainty in the rankings by obtaining confidence intervals for each species. This uncertainty measure has allowed us to detect an intermediate dominance hierarchy. By further considering ant communities as networks of interacting species, we have found that twig-nesting ants are overwhelming self-organized into a transitive dominance hierarchy.
More broadly, a high proportion of stable transitive relations have been observed in other animal systems for both dyadic and triadic level interactions (McDonald and Shizuka 2013; Shizuka and McDonald 2015). For example, dominance contest over free-ranging African elephants in Kenya showed that between-group dominance structure is highly transitive (Wittemyer and Getz 2007). Despite the wide geographical distribution of resources and infrequent contests among elephants, the potential cost for conflict were sufficiently high resulting in strong winner and loser effects. Among wild chimpanzees, both male and females formed dominance hierarchies. However, for high ranking female chimpanzees, dominances was associated with reproductive fitness due to greater access to food resources (Wittig and Boesch 2003). Miller et al. 2017 partnered with citizen scientists to examine social hierarchies among bird species at the continental scale. They used extensive catalogs of interspecific interactions at bird feeders across North America and observed that, across the continent, birds at feeders formed a linear dominance hierarchy encompassing an ecologically diverse range of species. Social dominance within a hierarchy was found to be strongly associated with body size with higher ranked species having preferential access to food resources at feeders.
While our study shows that nest site competition is important for structuring twig-nesting ant communities, dominance hierarchies are often context-dependent and ranking of the same species varies across geographical regions or disturbance regimes (Andersen 1997; Feener et al. 2008, Sensenig et al. 2017). Previous research involving ant competition for variable resources in temperate ecosystems showed that intransitive competition at local spatial scales mediates ant coexistence (Sanders and Gordon 2003). Microclimatic factors also disrupt dominance hierarchies (Cerdá
et al.
1998b ). For instance, environmental
variation in agricultural coffee systems is likely to influence
dominance hierarchies (Philpott and Foster 2005a ; Perfecto and
Vandermeer 2011). Likewise, occurrence of fire can disrupt dominance
hierarchies in specialist ants in Acacia trees resulting in
increased abundance of subordinate ants (Sensenig et al. 2017)
Further, top down processes such as predation and parasitism are likely to mediate twig-nesting ant competition in natural communities (Philpott
et al. 2004; Feener et al. 2008b ; Hsieh and Perfecto
2012). In addition, competition and disturbance from ground- and carton-nesting ants may influence the colonization and community composition of arboreal twig-nesting ants (Philpott
et al. 2004a ; Ennis and Philpott
2017). Therefore, more comparative research is needed to examine how geographical differences or disturbance affects the hierarchy and ultimately the distribution and relative abundance of different arboreal, twig-nesting ant species.
A myriad of other factors may drive the distribution and relative abundance of arboreal ants (Yamaguchi
1992a ; Palmer et al.
2000a ). Variation in life-history trade-offs can influence dominance patterns. For example, competition-colonization trade-offs have been identified between competitive colonies expanding unto nearby trees and foundress queens establishing new nest sites (Stanton
et al.
2002a ). Twig-ant communities are strongly influenced by canopy structure and habitat complexity (Philpott et al. 2018). Tree size correlates positively with ant abundance (Yusah and Foster 2006), species richness (Klimes et al. 2015), and composition (Dejean et al. 2008). Canopy connectivity, in turn, impacts local species coexistence as lower connectivity decreases species richness (Powell et al. 2011). Canopy connections serve as pathways by which arboreal ants access tree resources. Limited access to nesting cavities hampers growth and reproduction of arboreal ants (Philpott
and Foster
2005). Differences in nest entrances can affect both the abundance and richness of arboreal ant species that are competing for cavities resources (Jiménez‐Soto
and Philpott
2015). A study examining the effects of twig diversity on litter-nesting ant species found that twigs derived from different trees species harbored greater ant diversity as compared to twigs obtained from a single tree species (Armbrecht et al. 2004). In some ant species belonging to the genus Cephalotes, ant size and nest entrance size impacted survival and colony fitness (Powell 2009). Although our artificial twigs featured standardized cavity entrances, variations in ant sizes and cavity entrance size in natural conditions will be important in determining arboreal ant dominance. Thus, future studies should consider variation in nest entrance size to guide our understanding in twig dominance hierarchies.
Based on our findings, we find that tropical twig-nesting ants competing for nesting resources are arranged in a dominance hierarchy. We find that the arboreal ants Myrmelachista
mexicana was the highest ranked species in the hierarchy, while
Pseudomyrmex ejectus was ranked as the lowest in the hierarchy.
To more reliably infer a hierarchy in our dataset, we accounted for
uncertainty in the hierarchy by estimating the probability that a higher
ranked defeats a subordinate species. We found that an intermediate
steepness level characterized the hierarchy after corroborating an
adequate sampling effort. Moving beyond simple pair-wise interactions,
we used motif analysis to infer higher order interactions. Transitive
interactions were significantly over-represented in the network which
further illustrates that twig-nesting ants are organized in a linear
hierarchy. While we find that twig-nesting ants form a dominance
hierarchy in this tropical agricultural system, it’s likely to expect
variation in domination patterns across ecosystems and habitats.
Subsequent studies should link dominance patterns with relative
abundance patterns in the field in order to assess if particular species
traits are important in structuring local communities. While competitive
outcomes in our experiment are static (winner and loser), dominance
hierarchies exhibit considerable variation and field studies should
therefore include spatial and temporal variation. Dominance hierarchy
studies are typically designed to assess antagonistic interactions, but
less focus has been placed on collecting data with neutral interactions.
Difference in food preference and temporal foraging patterns suggest
that species don’t necessarily interact in an antagonistic fashion.
Therefore, more studies noting neutral interactions will shed greater
light on the prevalence of dominance hierarchies under natural
conditions.