Many closely related populations are distinguished by variation in sexual signals and this variation is hypothesized to play an important role in reproductive isolation and speciation. Within populations, there is considerable evidence that sexual signals provide information about the incidence and severity of parasite infections, but it remains unclear if variation in parasite communities across space could play a role in initiating or maintaining sexual trait divergence. To test for variation in parasite-associated selection, we compared three barn swallow subspecies with divergent sexual signals. We found that parasite community structure and host tolerance to ecologically similar parasites varied between subspecies. Across subspecies we also found that different parasites were costly in terms of male survival and reproductive success. For each subspecies, the preferred sexual signal(s) were associated with the most costly local parasite(s), indicating that divergent signals are providing relevant information to females about local parasite communities. Across subspecies, the same traits were often associated with different parasites, indicating that parasite-sexual signal links are quite flexible and may evolve relatively quickly. This study provides evidence for 1) variation in parasite communities and 2) different parasite-sexual signal links among three closely related subspecies with divergent sexual signal traits, suggesting that parasites may play an important role in initiating and/or maintaining the divergence of sexual signals among these closely related, yet geographically isolated populations.

Field work for H.r. erythrogaster was conducted May- September 2012 in Colorado, USA where we collected data on 173 males from 23 breeding colonies (typically barns or bridges) in Boulder, Jefferson, and Weld counties. Field work for H.r. rustica took place April-August 2013 in Jihocesky, Czech Republic where we followed 84 males at 6 breeding colonies near the towns of Třeboň and Lužnice. Field work for H.r. transitiva took place January-June 2015 in northern Israel, where we collected data on 42 males in the villages of Ami’ad, Hoquq, and Kahal. We had smaller sample sizes in Israel because barn swallows breed at lower densities there.  Breeding sites within each subspecies were located geographically near one another and there is likely gene flow between them.

At each breeding site, we attempted to capture all adults using mist nests. Each adult was given a numbered band and a unique combination of color bands. We measured morphology (wing length, mass, tail streamer length), and collected ventral plumage samples and blood samples (Safran et al. 2008; Jenkins et al. 2013; Hubbard et al. 2015). Using behavioral observations, we identified social pairs and determined their nest. All nests were monitored every three days until the end of incubation, at which point, nests were monitored daily to determine hatch date. When nestlings were 12 days old, we banded them and measured morphology (wing length and mass), and collected blood and plumage samples (Hubbard et al. 2015). For each subspecies, we measured parasites of adults and offspring, cost of parasites to male survival and reproductive success, and associations between the expression of ornamental traits and parasite infections. We used the same methods for all three subspecies to enable direct comparisons and the same people (AKH, JKH) led the field work and performed the measurements in all three countries.

Measurement of parasites

We quantified all ectoparasites infecting adults using a loupe (3.5x magnification) at the start of the breeding season (all ectoparasites were quantified by the same person, AKH, in all three subspecies) (Møller 1991). Nest parasites were quantified 12 days after nestlings hatched (supplement 1.1 for details). Haemosporidian blood parasites were detected via nested PCR targeting cytochrome b using the Hellgren et al. (2004) protocol, followed by Sanger sequencing of positive samples (supplement 1.2 for details). Given that this method cannot tell us about infection intensity, data for Haemosporidians for each individual was coded as present or absent in our analyses. We used non-lethal field sampling methods in wild birds and were therefore unable to measure all possible parasites (i.e. gut parasites, bacteria, viruses). Our parasite community is thus limited to the parasites that we could quantify via visual and molecular methods using samples collected from living animals (see supplement 1.3 for parasite terminology).

For two subspecies, in Colorado and the Czech Republic, we were able to collect data on a subset of parasites across multiple years and thus test if parasite abundance and variance was relatively consistent within a population across years, as well as ensure that our data from one year was representative of typical parasite infections for that population. We compared parasite abundance across years within subspecies with repeated measures using general linear mixed models (Poisson distribution) with year as a fixed effect. We also compared the variance within a population between years using a Kruskal-Wallis test. We measured ectoparasites in Colorado in 2012 (n=173), 2013 (n=196), and 2014 (n=103), and in the Czech Republic in 2013 (n=84) and 2014 (n=81).

Measures of parasite costs

For each subspecies, we tracked the following measures of reproductive success: onset of breeding, where early onset is associated with increased offspring survival, quality, longevity, and seasonal reproductive success in temperate breeding birds, including barn swallows (Dubiec and Cichoñ 2001; Gurney et al. 2012; Saino et al. 2012; Öberg et al. 2014), the number of nestlings successfully fledged from a male’s first brood, and whether that male attempted a second brood during that season (Møller 1990a; Saino and Møller 1995). We also measured male survival to the next breeding season for the Czech and Colorado populations, but were unable to collect these data for the Israeli population. Adult barn swallows have high breeding site fidelity from year to year, so presence the following year is considered a good measure of survival (Safran 2004; Schaub and Von Hirschheydt 2009).

Measures of sexual signals

Expression of all three ornamental traits (streamer length, throat color, and breast color) was measured in each subspecies. Unbroken tail streamer length was measured in millimeters from the base to the tip of the feather (Safran and McGraw 2004; Wilkins et al. 2016). Feather samples from the throat and breast region were collected and stored in envelopes. These feathers were later mounted on cards and analyzed using a spectrophotometer (Wilkins et al. 2016). We used brightness and hue as measures of color expression (supplement 1.4 for details). We confirmed within our dataset that the expression of sexual traits varied between subspecies (supplement 2.1), and that the signal(s) used by females within each subspecies was associated with reproductive success (supplement 2.2). These results supported the presence of divergent sexual selection and corroborated previous experimental and comparative studies of barn swallows (Scordato and Safran 2014; Safran et al. 2016a; Romano et al. 2017).

This is a combine datasheet that contains data from both the sexual selection portion of the study and the repeated parasites measures across years (for a subset of parasites). For the Colorado 2013, 2014 individuals and the Czech Republic 2014 individuals, we only include parasites information. Missing information for other individuals was due to our inability to collect this information in the field (i.e. nest was found after eggs were laid, so clutch initiation date was impossible to determine, or sample was lost or missing). For some subspecies certain parasites were not present (i.e. flies in Colorado or Haemosporidians in Israel), so these values are left blank for these subspecies. We were also unable to collect survival data for Israel.