Geographic variation in animal phenotypes is common, yet we know surprisingly little about how communication varies across populations. Instead, most communication research focuses on one or a few populations and assumes recognition behavior is consistent across a species’ range. Here, we show that individual identity signals and individual recognition vary across the geographic range of Polistes fuscatus wasps. P. fuscatus in MI and NY have variable facial patterns that signal individual identity and are used by receivers for individual recognition. However, P. fuscatus from Rothrock, PA lack individual identity signals, as they have less variable color patterns than P. fuscatus from MI. Further, P. fuscatus from PA are not capable of individual recognition. PA P. fuscatus do not learn and remember individual conspecifics during social interactions or during training. The MI and PA populations are genetically differentiated, but the differentiation is most likely geographic and not driven by differences in recognition. Overall, both signals and receiver responses vary across populations of P. fuscatus. Our results suggest that communication systems may rapidly evolve to produce variation in signals and receiver responses across a species’ geographic range.

Polistes fuscatus foundresses (nest-founding queens) were collected from two locations. Ann Arbor, Michigan (42.2997° N, 83.6627° W) and Roth Rock, PA (40.637052° N, -78.074672° W). After collection, wasps were housed in the laboratory in individual containers with ad lib water, sugar, and caterpillars. To guage individual recognition ability, we staged contests between pairs of wasps with and without a prior history of social interactions following methods in previous studies (Sheehan & Tibbetts, 2008; Tibbetts et al. 2019a; Tibbetts et al. 2019b). We quantified wasps’ ability to learn and remember conspecific faces using a negative reinforcement training methods established in our lab (Tibbetts et al. 2019a; Tibbetts et al. 2019b; Tibbetts et al. 2018).  Wasps must learn to approach the face image associated with safety to avoid being shocked. To assess variation in facial patterns of wasps, we quantified the proportion of each facial area that is brown, black, or yellow. We used ‘DNA barcoding’ with the mitochondrial cytochrome c oxidase subunit I (COI) gene to confirm species identities. We evaluated population genomics using a double digest RAD-sequencing library (Peterson et al., 2012), in which EcoR1 and MSE were used as the digestion enzymes (single-end 150 bp reads which had undergone size selection for 350-450 bp). Bioinfomatics was conducted with STACKs (Catchen et al. 2013). For genomic data, tests of geographic subdivision were done with STRUCTURE (v. 2.3.4;Pritchard 2000; Pritchard and Falush 2009) and via a PCA. Tests for loci under selection were done via OutFlank (Whitlock and Lotterhos 2015) and via a custom script in R, and quantification of genetic drift was also done in STRUCTURE (Pritchard 2000).