Discussion

Phorid fly presence significantly affected competitive interactions between W. auropunctata and L. in- iquum. Phorid flies significantly affected the recruitment of L. iniquum workers to tuna baits, which signifi- cantly increased the overall abundance levels of W. auropunctata workers. In the absence of phorid flies, L. iniquum workers had a higher recruitment rate to baits and gained a competitive advantage over W. auropunctata.  In the presence of phorids, W.  auropunctata  workers were able to overtake  baits through  a combination of direct interactions (i.e. competition) and indirect attacks by phorids on L.  iniquum.  This double-whammy led to the rapid decline of L. iniquum workers over time. The effect of phorids on ant community structure depends on the spatial scales of interaction. Our results indicate that phorid searching efficiency depends on the spatial scale of both the phorid fly and distribution of L. iniquum nests. It took much longer for phorids to locate L. iniquum nests within the patch of W. auropunctata as compared to areas where L. iniquum was more prevalent. As distance increased from densely populated areas of L. iniquum, phorid attack rates diminished because of the time it took to locate nests. Although we don’t claim to have information on the population density of phorids at local spatial scales, it appears that phorid flies increase their local population densities nearby areas dominated L. iniquum. Phorid flies are known to use chemical cues to locate their hosts (Brown and Feener 1991, Morrison 1999, LeBrun and Feener 2002, Hsieh and Perfecto 2012). During competitive encounters, ants release alarm pheromones resulting in the rapid response by phorid flies. One study found that phorid attack rates increased when Linipethema was in the presence of dominant ant competitors (Orr et al. 2003). Thus, a behavioral response by Linipethema dur- ing competitive encounters elicits a positive feedback by increasing phorid attack rates which increases the number of foragers of its opponents (Hsieh and Perfecto 2012, Philpott 2005). At larger spatial scales,
we found that phorid   attack rates diminished at greater distances from L. iniquum clusters.   Interestingly, L. iniquum nests found within W.   auropunctata clusters were temporarily relieved from phorid attacks.    There was a five-fold reduction in phorid attack rates in isolated L. iniquum trees within W. auropunctata clusters which suggests   that L. iniquum species disperse to   patches dominated W. auropunctata in   order    to receive protection against phorid parasitoids.      The effects of phorids in mediating ant   competition      may vary  temporally.  Our results demonstrate that L. iniquum abundances were reduced in   the presence      of phorids during the experiment.  However,  ants can adapt the   timing of their foraging activities as a way   of avoiding phorid attacks (Philpott et al. 2004). For  example,  the ground forager ant Atta cephalotes  primarily forages during the night time in order to avoid  phorids during the day time that tend to build  up  their populations outside of A. cephalotes nest colony entrances (Orr 1992). Several Linipethema species construct mobile nests that are often   disguished from phorids making it more difficult to find (Markin    1970).  In our  system,  L. iniquum species have  many  different shallow nest entrances throughout coffee  trees making it even more difficult for phorids to figure out where to attack. Therefore, species differences in nesting ecology and habitat complexity may play an important role in phorid   attacking abilities (Orr et         al. 2003) (Philpott 2005). Another important factor in assessing TMII on competition is resource size (Mehdiabadi et al. 2004, Philpott et al. 2004). In our study, both L. iniquum and W. auropunctata were  provided with large resources. However,  this may not have  given L. iniquum enough time to defend itself  from phorids as they   recruit large number of workers to uptake resources.  In our experiment, L. iniquum   was rapid at resource   discovery,  as compared to W. auropunctata,   but spent longer time recruiting work-   ers to carry baits. Although not enough is known about species response to variation in resources, small resources distributed over larger spatial scales may enable L. iniquum to avoid persistent phorid attacks. (LeBrun 2004) showed that phorids take a much longer time locating ants consuming smaller resources. As fewer ants are recruited to smaller resources, less pheromones are being released   that phorids can exploit      to detect ant foraging activities (Folgarait and   Gilbert 1999) (Feener  and Brown 1992).      These findings have important implications for the invasion dynamics of W. auropunctata in a coffee agricultural ecosystem. In the absence of phorids, L. iniquum reaches higher abundance  levels resulting in increased  access to resources. This pattern changes however  when phorids come into the scene and L. iniquum abundance diminish rapidly enabling W. auropunctata to increase its abundance levels.  At  the population  level, L. iniquum builds up locally dense clusters but undergoes increased attacks by phorids, which provides  an opportunity for W. auropunctata to invade. To counter this two-fold attack, L. iniquum disperses into patches dominated by W. auropunctata, providing temporary protection from phorid attacks. These factors taking   together show clear effects of phorids on ant competition.   Studies of ant competition have  shown   the existence of competitive trade-offs among species can structure ecological communities and promote coexistence (Fellers 1987, Savolainen et al. 1988, Perfecto and Vandermeer 2013). The trade-off implies that some species have strong resource discovery abilities as opposed to other species that have strong resource domination abilities. TMII mediated competitive interactions by phorids can break the discovery-dominance trade-off and enable invasive species to spread (Hsieh and Perfecto 2012). In the absence of phorids, L.  iniquum is a faster resource discoverer and gains a competitive advantage over  W. auropunctata species,  which are known to be poor resource   discoverers (Bertelsmeier et al.   2015).   However,  phorids can tip  the balance and limit the rapid acquisition of resources by L. inquum thereby providing W. auropunctata  species with ample time to dominate resources and spread. A  similar pattern was found in a pine-oak  woodland ecosystem involving Pheidole host species that in the absence of phorids increased their discovery abilities, thereby breaking the discovery-dominance trade-off. The presence of phorids reduced the com- petitive ability of the host species to the level of the ant assemblage reinforcing the discovery-dominance trade-off (Lebrun and Feener  2007).  Phorid incuced TMII  attacks on L. iniquum enable to W. auropunctata  to spread in Puerto   Rico.        A key question that remains is what   potential factors limit the expansion        of W.   auropunctata in its native range.   To  our knowledge no phorid parasitoids   have  been recorded in    the native range of W.  auropunctata.  However,  previous research in the native range of Mexico  suggests   that dominant ant species can limit the expansion of W. auropunctata in a Mexican agricultural ecosystem, providing further support for the biotic resistance hypothesis  (Yitbarek et al.  in press).   In the case of   another invasive ant species S. invicta, phorid induced TMII and   interspecific competition influenced the
ecological success of the invasive ant S. invicta (Feener et al. 2008). However, this pattern was found to vary geographically depending on the presence of phorid parasitoids in the region. In the case of Puerto Rico, we find that a combination of release from competitors in the native range and TMII contributes to the expansion of W.  auropunctata.   Our study on the role of phorid   parasitoids mediating competitive ant interactions has important consequences for biodiversity maintenance. Competitive dynamics between
W. auropunctata and L.  iniquum for resources suggest that two  different trade-offs operate.  The  ability of L. iniquum to arrive at resources quickly and the ability for W. auropunctata to hold on to acquired resources shows that a discovery-dominance maintains species coexistence in the absence of phorids. The discovery-dominance vanishes in the presence of phorids and in place a new trade-off emerges between the ability to compete for resource versus the ability to defend against parasitoids (Lebrun and Feener 2007). This so-called competition-defense trade-off reduced L. iniquum workers at resources while W. auropunc- tata was able to increase its workers to baits. The interplay between these trade-offs can amplify within complex ecological networks whereby the addition of a parasitoid reduces the dominance of a species to the level of the entire assemblage resulting in coexistence. From a practical standpoint, W. auropunctata has been found to be an effective biological control against the coffee berry borer pest in Mexico (Gonthier et al. 2013). Indirect effects by phorids enables W. auropunctata to expand in shaded areas with relatively high densities of the coffee berry borer. One potential mechanism for limiting the expansion of W. au- ropunctata on coffee farms is through the management of shade trees. Pruning of trees provides enough sunlight to attract ant competitors that can potentially limit the expansion of W. auropunctata while also providing ecosystem services. These findings open up a new set of inquiry on the effects of TMII on ant community structure across temporal and spatial scales. While phorids reduced the abundance level of L. iniquum resulting in a competitive advantage for W. auropunctata, we know very little about the long-term population level consequences of TMII on ant communities. In particular, it’s important to explore whether parasitoid phorids attacks on L. iniquum have  implications for colony growth (Mehdiabadi et al. 2004). Although phorids reduce the number of workers of host species this may not necessarily affect the colony as a whole and therefore future investigations should evaluate the long-term dynamics between ants and phorids (Philpott 2005). Spatial considerations between ants and phorids need to take into account the foraging range of phorids. While phorids appear to build up their populations nearby L. iniquum clusters it’s not clear how far the range of phorids extends. Within W. auropunctata clusters several colonies of L. iniquum have been established in order to limit phorid attacks. This feedback between phorid range and  ant competition leads to the formation of spatial clusters. In this regard, spatially explicit models can serve us useful guides to disentangle complex direct and indirect interactions between ants and phorids that give rise to spatial pattern formation and have important consequences for biodiversity (Li et al. 2016).