In many animal species, males compete for access to fertile females. The resulting sexual selection leads to sex differences in morphology and behaviour, but may also have consequences for physiology. Pectoral sandpipers are an arctic breeding polygynous shorebird in which males perform elaborate displays around the clock and move over long distances to sample potential breeding sites. We examined the oxygen carrying capacity of pectoral sandpipers, measured as the volume percentage of red blood cells in blood (haematocrit, Hct). We found a remarkable sex difference in Hct levels, with males having much higher values (58.9 ± 3.8 SD) than females (49.8 ± 5.3 SD). While Hct values of male pectoral sandpipers are notable for being among the highest recorded in birds, the sex difference we report is unprecedented and more than double that of any previously described. We also show that Hct values declined after arrival to the breeding grounds in females, but not in males, suggesting that males maintain an aerobic capacity during the mating period equivalent to that during trans-hemispheric migration. We conclude that sexual selection for extreme physical performance in male pectoral sandpipers has led to exceptional sex differences in oxygen carrying capacity.

Study site and general procedures

We studied pectoral sandpipers at a ~2 km² site at the northern tip of the Alaskan Arctic Coastal Plain near Utqiagvik, Alaska (71°18′ N, 156°44′ W) in the years 2004-2009, 2012, 2014, and 2018. We caught pectoral sandpipers during the breeding season using hand-held mist nets (males and females) or nest traps (females only). We assigned each bird a metal leg band and a unique combination of colour leg bands, weighed them (to the nearest 0.1 g), measured their tarsus (to the nearest 0.1 mm), and sampled 200–300 μl of blood using brachial venepuncture. We collected blood in 70 μl heparinized microhematocrit capillary tubes and centrifuged the samples at 5,000 rpm for 10 min. on the day of collection, thus separating plasma from cellular blood. Haematocrit levels were measured for each full capillary as the percentage of packed red blood cells over the total blood sample. For statistical analyses, we used the mean value from all capillary tubes obtained from an individual during a given capture. Red blood cells were kept and stored in Queen’s lysis buffer for subsequent molecular sexing. In total, we obtained 778 blood samples from males and 262 from females. For 38 males we obtained multiple samples within a breeding season (i.e., two (N = 34), three (N = 3), or four (N = 1)). Additionally, for eight males we obtained samples across multiple breeding seasons (i.e., two (N = 6) or three (N = 3)). We attempted to find all nests on the study site by 1) observing foraging females until they went to their nest to incubate or 2) flushing females off their nest by systematically searching or rope dragging the area.

Comparative data

We obtained estimates of haematocrit values from other bird species from Minias (2020). Minias (2020) compiled a dataset of 611 Hct estimates from 279 species based on data available in the published literature. This dataset includes only non-experimental studies, and only studies on wild birds or birds kept in outdoor aviaries (i.e. studies on birds kept indoors were not included). From this dataset, we extracted all studies that reported a separate estimate for males and females from the same age class (juvenile or adult). We only included studies when the Hct estimate for each sex was based on at least ten individuals, resulting in a dataset consisting of 63 estimates of male and female Hct values from 35 different species.

We additionally obtained sex-specific Hct estimates from five species at our study site which we had collected as part of parallel research projects. These samples were obtained and processed in the same way as those of pectoral sandpipers. Thus, we included data from American golden plover Pluvialis dominica (n = 10 males, 12 females), dunlin Calidris alpina (n = 10 males, 10 females), long-billed dowitcher Limnodromus scolopaceus (n = 26 males, 21 females), red phalarope Phalaropus fulicarius (n = 257 males, 312 females), and semipalmated sandpiper Calidris pusilla (n = 71 males, 67 females).

The combined set of data extracted from Minias (2020) and collected by ourselves (excluding pectoral sandpipers) thus consisted of 68 estimates of male and female mean Hct values from 40 different species.

Statistical analyses

All statistical analyses were performed with R (version 4.2.2, www.r-project.org). We performed (generalised) linear mixed models using the lme4 package (Bates et al. 2014). We used the R package multcomp (Hothorn et al. 2008) to obtain P-values corrected for multiple testing.

                To test for a difference in Hct levels between male and female pectoral sandpipers, we performed a t-test. To test for a difference in average Hct levels between males and females across different bird species (68 male and female estimates from 40 species), we performed a paired t-test.

To examine factors associated with variation in Hct values in pectoral sandpipers, we performed linear mixed-effect models with Hct value as the response variable. Because factors affecting Hct levels may differ between the sexes, and to avoid having to fit 3-way interactions, we ran a model for males and females separately. We included date, body mass (g), and tarsus length (mm) as fixed effects. For females, we also included breeding status, i.e., whether they nested locally at out study site during a given season. We also tested for interaction effects of breeding status with date, body mass, and tarsus length, respectively, but only retained these interaction effects in the model if they explained significant variation (p<0.01). Breeding status was not included in the model for males, because males mate opportunistically at different breeding sites and can all be considered potential breeders. Previous work showed that males who sired offspring locally did not differ in Hct values from males that did not (Kempenaers & Valcu 2017). Numeric variables were mean-centred, so that the model intercept was estimated for the mean value of the explanatory variables. Body mass and tarsus length correlated positively, but weakly (Pearson’s correlation coefficients; males: r=0.24, females: r=0.13) and both were thus included as explanatory variables in the models. Year was included as a random effect in both models. Because we had more than one sample for some males, we also included individual identity as a random intercept in the model for males.