UC Berkeley

In my dissertation, I have explored behavioral, chemical and genetic aspects of a unique nesting symbiosis called parabiosis. In parabiosis, two unrelated ant species share a nest and foraging trails in a potentially mutualistic association. I have focused on the Neotropical parabiosis between Camponotus femoratus (Subfamily: Formicinae) and Crematogaster levior (Subfamily: Myrmicinae), which occur in ant- gardens throughout Amazonia. These two ants share a common nest but keep their brood in separate chambers. Behavioral tradeoffs suggest that the relationship is a mutualism: both species build the carton nest and forage, but Cr. levior is superior in finding food sources, and Ca. femoratus carries the epiphyte seeds required to give the nest structural support. Like any mutualism, the relationship is vulnerable to exploiters and cheaters, so reliable recognition systems would help to maintain the relationship.

In Chapter 1, I examine the nestmate recognition behaviors of Cr. levior and Ca. femoratus living in parabiotic ant garden nests. By using pairwise behavioral assays in neutral arenas, I assayed the proportion of ants exhibiting aggressive behaviors when paired with nestmate and non-nestmate ants. We expect for ants to aggress non-nestmates by biting, stinging, spraying formic acid or otherwise attacking to exclude these intruders. I also sampled the cuticular hydrocarbon chemistry of ants in these nests to determine whether recognition behavior was related to these compounds. Cuticular hydrocarbons (CHC) are often used as recognition cues amongst social insects. I found that there were three different CHC phenotypes in my study population in French Guiana. There were two sympatric chemotypes of Cr. levior, with very little overlapping chemistry. Within each nest, there was only a single chemotype of Cr. levior, and neither chemotype shared chemical cues with Ca. femoratus. Despite sharing a nest, Ca. femoratus exhibited a single chemotype throughout the population, and did not chemically match its Cr. levior nestmates. Both species maintain intraspecific recognition abilities, and Cr. levior shows some evidence of being able to distinguish amongst its Ca. femoratus nestmates. However, despite their strong chemical divergence, there was noevidence Ca. femoratus could distinguish between the Cr. levior chemotypes. My findings suggest that selection to maintain reliability in conspecific recognition can potentially constrain the evolution of interspecific cooperation.

In Chapter 2, I delve further into the details of the cuticular chemistry of these parabiotic ants. Incidentally, there are three species living within the ant garden nests -- Cr. levior, Ca. femoratus and a tiny Solenopsis thief ant, called Solenopsis picea. Do these multispecies nests still form a common colony `gestalt' odor? Are there differences in the chemical integration techniques of social mutualists and social parasites? I sampled individual ant CHCs to look for patterns of similarity within and between ant species, and within and between colonies. Both parabiotic species show some evidence of forming a single-species common colony odor, which is consistent with the gestalt hypothesis that nestmates share chemical cues. However, this cue sharing does not spread to allospecific nestmates. The two parabiotic species, Cr. levior and Ca. femoratus, share very few cues in common. In contrast, the social parasite S. picea shares cuticular chemistry with both of its host species. The specificity of this chemical cue similarity is limited, and S. picea is not chemically different whether it is nesting with Cr. levior Type A and Cr. levior Type B. These findings are the first to examine the CHC patterns in nests with three ant species, and highlight important differences in the chemical integration of social mutualists and social parasites.

In Chapter 3, I examine the genetic basis of the chemical phenotypes of Cr. levior and Ca. femoratus. Using the individual profiles of ants from Chapter 2, I compare the chemical phenotypes to genotypic information from both nuclear microsatellite loci and mitochondrial co-1. For both species, there is a correlation between chemical phenotypes and genotypes. In Ca. femoratus, there are positive correlations between genetic distances and both chemical and geographic distances of colony pairs. The genetic basis for chemotype includes a correlation between some alleles and the proportion of straight-chain alkanes, which are shorter than other hydrocarbons in the typical Ca. femoratus profiles. Likewise, there are correlations between chemical phenotypes and genotypes of Cr. levior ants, with a strong genetic distinction between the two Cr. levior chemotypes. There is no geographic partitioning of either chemical or genetic differences, which supports the observation of sympatry of the Cr. levior Type A and Type B. We find correlations between several alleles and the proportion of different chemical compounds. Specifically, there appear to be opposing genetic trends for the alkane and methyl branched compounds that dominate Cr. levior Type A profiles, and the unsaturated alkenes and alkadienes, which typify Cr. levior Type B profiles. Together this evidence supports the hypothesis that Cr. levior Type A and Type B are genetically distinct and potentially different cryptic species.

In Chapter 4, I attempt to resolve the relationships between geography, chemistry, genetics, and recognition behaviors of the parabiotic ants. In other systems, cuticular chemistry plays an important role in determining the outcome of recognition assays, with increased chemical dissimilarity usually resulting inincreased aggression. Similarly, more genetically distant non-nestmates are expected to be subject to more aggression. I find that for Ca. femoratus, conspecific aggression is related to genetic differences, but not to the measured chemical differences. This suggests potential kin-informative cues exist, that we have not yet measured. These genetic cues may also be used by Cr. levior to recognize their Ca. femoratus nestmates, and highlight the technical limitations of our current chemical machinery. In contrast, Cr. levior conspecific recognition is more strongly related to chemical differences, but mainly at the level of chemotypes. These results emphasize the importance of considering all levels of chemical and genetic differentiation when assessing nestmate recognition patterns. Both species of parabiotic ant uses chemical and associated genetic information to assess nest membership. The fully functioning recognition systems in parabiotic nests likely maintain the relationship by excluding exploiters of this unique cooperative relationship.

In sum, my dissertation research characterizes the recognition behaviors, chemical cues, and population genetics of an uncommon but abundant ant-ant mutualism. My findings support the hypothesis that recognition behaviors, although proximately mediated by chemical similarity, are ultimately controlled by genetic factors. I find that both chemical integration techniques and nestmate recognition behaviors differ for mutualistic and parasitic nesting symbioses. Together my dissertation research highlights the unique properties of the parabiotic nesting association, and supports its status as the only ant-ant mutualism.