Lower vulnerability to predation should increase the capacity of prey populations to maintain positive population growth rate in regions characterized by high predation pressure. Some arctic nesting shorebirds nest almost exclusively in areas where predation pressure is regularly released. The few species that can breed within the entire distribution range of the Arctic fox (Vulpes lagopus), the main nest predator in the arctic tundra, are supposedly less sensitive to predation. However, empirical data supporting this hypothesis are scarce and mechanisms driving interspecific variation in vulnerability to nest predation are poorly documented. We monitored nest success of two shorebird species with contrasting breeding distribution and nesting habitat. We found that i) when co-existing at the same breeding site, the widely distributed Ringed Plover nesting along stony shores showed a higher nest survival rate than the Golden Plover nesting in mesic tundra, and ii) such differences in nest survival were at least partly driven by the nesting habitat type per se, with lower predation risk in stony shores than in adjacent mesic tundra. We suggest that the use of safer nesting habitat by some shorebird species can contribute to maintaining viable breeding populations over a broader distribution range.

Shorebirds nest monitoring

This study was conducted over three years (2014-2016) on the southwest plain of Bylot Island, Nunavut, Canada (73° 08’ N, 80° 00’ W), located in Sirmilik National Park. The landscape is dominated by mesic tundra on the uplands and both mesic tundra and wetlands in the lowlands. The mesic tundra is covered with relatively lush vegetation for this latitude, mainly composed of low shrubs (Salix, Vaccinium) and forbs (e.g. Cassiope, Dryas) with grasses and mosses. The American Golden Plover (Pluvialis dominica) and the Ringed Plover (Charadrius hiaticula) are both common breeding shorebird species within the study area. The American Golden Plover prefers to breed in the low vegetation of mesic tundra, while the Ringed Plover breeds on stony and sandy shores and gravel bars with scarce vegetation along coasts and rivers. Each summer, nesting plovers were monitored along a 50-km-long coastline. We found nests either opportunistically or through systematic search (line transect surveys) from mid-June to mid-July. Nests were marked with a 10 cm wooden stick (medical tongue depressor) and a natural object (rock or feather) placed 5 m and 7 m from the nest respectively. We typically visited nests every 3-5 days or every 2-3 days when nearing the estimated hatching date. The incubation stage was estimated for each nest using the flotation method. Incubation lasts 26 days for the American Golden Plover and 24 days for the Ringed Plover. A nest was considered successful if at least one egg hatched or if one of the following criteria were met: 1) < 5 mm of residual egg shell was found in the nest material close to the estimated hatching date 2), the nest was hatching (starred or pipped) on the last visit and was empty on the next visit, and 3) the nest was empty on the last visit and the banded adult was later seen with chicks. Shorebirds chicks generally leave the nest within 24 hours of hatching. Predation is the main cause of shorebird nest failure in our study area and other sources of mortality are marginal (nest abandonment or nest flooding was confirmed in only 3 out of 236 plover nests with known fate).

Artificial nest experiments

Artificial nest experiments were conducted during the plover incubation period in 2015 and 2016. Each year, a total of 60 paired artificial nests were deployed in suitable habitat used by the two selected plover species: 30 artificial nests were placed on stony shores (nesting habitat of the Ringed plover) and 30 nests in mesic tundra (nesting habitat of the American Golden Plover). The two paired nests were separated by 150-200 m and the distance between artificial nests positioned in each habitat was within the range of distances recorded for real nests on Bylot Island (average distance: 877 m and 547 m, range 341-1431 m, 52-1845 m, for artificial and real nests respectively). Artificial nests were deployed on 12 July 2015 and 10 July 2016 and were visited after 1, 2, 4 and 6 days. Each artificial nest consisted of four quail eggs (Coturnix japonica) placed in a shallow depression in the ground; quail eggs are similar to plover eggs in colour and size. The artificial nests were marked as the real nests (see above), but a nail wrapped in fluorescent tape was hidden beneath the eggs so depredated nests could be located easily. To reduce human scent, all eggs and pieces of material used were handled with latex gloves, and researchers used the sole of their rubber boots to make the nest depression. The Arctic fox is the main predator of both real and artificial shorebird nests in our study area. However, artificial nests are more vulnerable to avian predators than real nests, which may result from the absence of parental nest defence and because uncovered artificial nests can be more easily detected by avian predators than nests covered by the incubating adult. To examine if habitat-driven patterns in predation risk are affected by the exclusion of avian predators, we conducted another experiment using artificial nests covered with a patch of lichen commonly found in the study area (genus Bryoria or Gowardia). The lichen covers were maintained on top of the eggs using a wooden stick inserted in the ground in the middle of the nest. We deployed 20 paired covered artificial nests on 28 June 2016 (20 nests in stony shores and 20 in mesic tundra), and visited them after 2, 6 and 15 days of exposure. Using motion-triggered cameras, we confirmed that covered artificial nests were only depredated by the Arctic fox in our study area (Bêty and Léandri-Breton, unpublished data, N = 82 depredated nests monitored with cameras in 2015 and 2016). Covered nests also allowed us to control for potential differences in quail egg crypsis associated with habitat types.

Statistical analyses

For real shorebirds nests, we compared the daily nest survival of the two plover species using the logistic-exposure method. Year was included as a random factor to control for inter-annual variations. The logistic-exposure method is a generalized linear model with a binomial response distribution (1 when the interval nest fate is a success and 0 when depredation occurred) using a logit link function to account for variations in the length of observation intervals. This method is advantageous as it allows the inclusion of random effects (e.g. “year”) and requires no assumptions about when nest losses occur. We assumed a constant daily nest survival during the nesting period. For graphical representation, daily nest survival was estimated through a logistic exposure model per species and per year. The nest success probability over the entire incubation period was obtained by raising the daily nest survival to the power of the mean incubation period of each species (26 days for the American Golden Plover and 24 for the Ringed Plover). We used Cox proportional hazards regression (R package “survival” 2.40-167) to assess the effect of habitat types on the survival of artificial nests. Cox model tests for a relationship between Kaplan-Meier survival estimates, a nonparametric statistic commonly used to estimate survival over time, and explanatory variables. Cox model is very appropriate for artificial nests because it allows for right-censoring data when nests survive past the end of the experiment. Year was treated as a random variable. The proportion hazards assumption was verified by calculating the correlation between scaled Schoenfeld residuals and survival time. For visualization purposes, predation risk was estimated by fitting a Kaplan-Meier survival probability curve for each species and "year" without random effect. All analyses were carried out in R version 3.2.270.