Introduction

Organisms have developed an endless variety of strategies to exploit what the most productive seasons offer and to cope with what the harshest seasons dictate. Relatively mobile animals have opportunities to use resources and avoid environmental stressors by strategically moving across the seasonally changing landscapes [1], [2]. Nevertheless, there is one hard boundary condition for all: the length of the year [3]–[5]. Whether sedentary or migratory, to cope with the seasonal changing environments, adult vertebrates each year go through several consecutive life-cycle stages during which they adjust their morphology, physiology and behavior [6], [7]. To maximize reproductive opportunities, animals will generally try to complete all necessary seasonal activities within the period of a year [5]. Small species with shorter life-cycle stages, or animals with fewer such stages per year, are predicted to be relatively flexible in the timing of these stages [4], [7]. For some animals, particularly larger ones, it will take longer to complete a full cycle of life-history stages. This may also apply when mutually exclusive activities, such as breeding and migration, are necessary or when maintenance of plumage or pelage (e.g. flight feather molt) requires dedicated time [4].

Some bird species carry out molt at the same time as other life-cycle stages [7], [8]. Molt can also be interrupted and spread over a longer time [9], and the rates and extent of molt can be adjusted to individual circumstances [10], [11]. Birds that accelerate molt incur costs in terms of increased daily energy costs for molt, flight and thermoregulation, decreased flight abilities and therefore increased predation danger, and a lower quality of feathers grown, resulting in negative effects on e.g. pigmentation or ornaments, length, resistance to wear, and insulation [10], [12]–[21]. In long-distance migrant shorebirds, i.e. habitat specialists with tightly timed annual cycles [22], [23], wing molt rarely overlaps with breeding, and even more rarely with migration [12], [24]. North temperate wintering shorebirds have to complete molt before the onset of severe weather conditions in late autumn [12], [25] when they may be faced with time constraints on the completion of molt.

To examine possible trade-offs between breeding and migration, and wing molt, we investigated timing and duration of primary molt in red knots Calidris canutus islandica in comparative and experimental ways. These birds breed on the High Arctic tundra in Greenland and Canada, and molt and winter in the Dutch Wadden Sea, the UK, and the French Atlantic coast [26]. Our study focused on red knots wintering in the Dutch Wadden Sea. We compared their performance with those of birds not spending time on breeding and migration. These include second-year birds that remain in the Wadden Sea the whole summer [24], [27] and birds kept in captivity in outdoor aviaries. The latter two categories were not constrained in time or food, and served as a reference point against which actual molt of possibly time-pressed reproducing adults were compared.

Materials and Methods

Results

We predicted that wing molt in breeding adults would be time-constrained due to a trade-off between breeding and migration. The nonbreeding over-summering second-years and captive adults, not faced with this trade-off, would be predicted to use more time to molt than the reproductively active adults. Consistent with these expectations wing molt duration was shortest in free-living adults, ca. 71 days, and longer –and mutually comparable– in second-years and captive adults, varying from 81 to 93 days (Fig. 1C). For both free-living males and females, the best models were the models that estimated all molt parameters separately per age class, as indicated by the lowest AIC for these models (Table 1). Hence molt duration differed significantly between adults and second-years. Molt duration was ca. 22% longer in second-years than in free-living adults, and ca. 27% longer in captive adults. In the Supporting Information is shown how the differences in molt duration were established by presenting the individual primary molt durations for all groups (Fig. S2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Primary molt start and end date (A, B), and molt duration (C), in free-living and captive red knots.

Symbols: closed symbols, males; open symbols, females. Group abbreviations: SYM, second-year males; SYF, second-year females; FAdM, free-living adult males; FAdF, free-living adult females; CAdM, captive adult males; CAdF, captive adult females.

https://doi.org/10.1371/journal.pone.0053890.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of molt parameters between ages (within sex) or sex (within age) in free-living adult and second-year red knots.

https://doi.org/10.1371/journal.pone.0053890.t001

As shown by the model comparison (Table 1), not only duration, but also timing of wing molt differed between groups. Second-years were the first to start molt (ca. 21 June), more than four weeks before the captive adults, and almost seven weeks before free-living adults started in early August (Fig. 1A, see Table S6 for details). Adults completed molt at comparable dates, mid- to late October, independently of living conditions (Fig. 1B). Second-years completed molt a month earlier.

Molt parameters differed between sexes in second-years as indicated by the lowest AIC for the model that estimated all parameters separately for males and females (Table 1, w_i = 0.99). In free-living adults, two models were equally plausible, one that estimated start date and SD of start date separately for the sexes and one that estimated all parameters separately (w_i = 0.63 and w_i = 0.37, respectively). The model estimating start date and SD of start date separately was 1.7 times (0.63/0.37) more likely than the model estimating all parameters separately. In free-living adults and second-years, females started molt before males and molted longer than males (Fig. 1). Although captive females also tended to take longer to molt than captive males, this difference was not significant (ANOVA, F_1,30 = 4.014, P = 0.054). Start date did also not differ between the sexes in captive adults (ANOVA, F_1,30 = 0.055, P = 0.817). Combining sexes we checked if molt parameters differed between aviaries, which was not the case (ANOVA, duration, start and end date, all P>0.4).

Discussion

Primary molt duration of adult red knots in the Wadden Sea was similar to adult C. c. islandica in Scotland (77 days; [47]), but much shorter than in adult C. c. canutus knots in South Africa (95 days; [47]). Primary molt duration of second-years and captive adults resembled however more that of the C. c. canutus than of the Scottish free-living adult C. c. islandica. The sexual differences in molt parameters are in accord with body size differences and behavioral differences. Female knots are somewhat larger than males (Table S5; [29], [37]) and therefore have more feather mass to replace. A longer primary molt duration in females (Fig. 1) is thus not unexpected. As in many Arctic-breeding shorebirds, female and male knots both incubate the eggs, but only males care for the young. Adult females leave the breeding grounds earlier than males and arrive earlier in the wintering areas [28]–[30]. Stable isotope data indicate that knots start molt ca. 4 days after arrival in the Dutch Wadden Sea [27], and adult females are thus expected to start molt earlier than adult males (Fig. 1). That the average difference in start date (6 days) is smaller than the young-caring period (17–18 days; [31]) may be because of earlier departure of males that lose their clutch or brood, or where unable to find a mate [45].

Primary molt duration was much shorter (ca. 22%) in free-living adults than in the non-breeding second-years. This is also the case, with a similar percentage, in north temperate wintering grey plovers Pluvialis squatarola [48], which are slightly larger than red knots and have a comparable timing of molt and other life-cycle stages, including the skipping of breeding in their second year of life. Interestingly, when adult red knots were brought into captivity, molt duration increased to equal that of second-years (Fig. 1). The short molt duration in free-living adults, their ability to slow progression of molt when brought into captivity, and the rapid onset of molt after arrival in the Wadden Sea [27] suggest that molt is seriously time-constrained in free-living north-temperate wintering adult red knots. This time constraint may well relate to (1) a restriction on the onset of molt because adults do not arrive before late July in the Wadden Sea and do not molt primaries or body feathers during breeding or in transit, and (2) a restriction on the time during which molt needs to be completed because of increases in thermostatic costs and stormy weather in the course of the year. Indeed, second-years finished molt well before October, when thermoregulatory costs started to increase rapidly (Fig. 2). Free-living adults started molt when thermoregulatory costs were still low, but these showed a steep increase during the last part of molt. Precipitation, not included in the calculation of cost-levels, also increased from mid-October onwards, thus adding to the pressures later in the year. Adult molt was finished well before thermoregulatory costs had reached maximal winter levels. An important additional reason for adult knots to complete molt before thermostatic costs have increased too much is that their winter plumage provides better insulation than their summer plumage [38], [49].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. The average temporal distribution of primary molt in free-living adult and second-year red knots, and captive adult red knots (sexes combined, grey blocks).

The dashed vertical grey lines indicate when the knots had completed growth of primary 5. Also plotted are the estimated weekly maintenance costs for a red knot living in the Wadden Sea, calculated using the model of [63] (equation 5; using mudflat conductances) from daily wind, ambient temperature and solar radiation data obtained over the same period as the molt data (1998–2006; data from the Royal Netherlands Meteorological Institute (KNMI) weather station at Hoorn on the island Terschelling in the Dutch Wadden Sea, 52°23′N 5°21′E). Note that basal and thermoregulatory costs are included in maintenance costs, but molt costs (feather synthesis and increased thermoregulation) are not. Maintenance costs are indicated with circles; open circles indicate the period during which adult red knots are migrating or in the breeding areas [63]. In addition, daylength in the Wadden Sea is given (solid line, right Y-axis, data from KNMI, 2006). Arrows indicate the start of the rapid increase in number of peregrines in the Wadden Sea (P; [54]), and when the diet of free-living knots changes from shellfish to less profitable mudsnails (D; [56], [58]).

https://doi.org/10.1371/journal.pone.0053890.g002

There may yet be other factors. The gap in the wing during molt has negative effects on flight abilities and flight costs [50].The larger the gap, the larger this effect will be [50], and the more vulnerable will the birds be to predation [51]. The increasing precipitation, wind speed and considerably stronger wind gusts in autumn may add to the negative effects of molt on flight ability and costs. However, birds could compensate for these effects with a decrease in body mass and an increase in pectoral muscle mass [50], [52], just as captive knots decreased body mass while maintaining stable pectoral muscle mass when exposed to raptor models [53]. Free-living adult knots would have completed molt in the first five primaries before the number of peregrines Falco peregrinus in the Dutch Wadden Sea started to increase steeply from ca. 15 in September to ca. 45 in October ([54]; Fig. 2), and probably relied on their predation avoidance tactics during the remainder of primary molt [55]–[57]. Second-year birds were close to finishing molt at that time. Note also that the preferred and most profitable foods of knots become scarcer in September when knots tend to switch from bivalves to the less profitable mudsnail, Hydrobia ulvea [56], [58].

Flight feather quality is likely to be very important to a long-distance migrant that makes non-stop flights of many thousands of kilometers [23], [28]. The ubiquitous finding that fast growth, i.e. relative short molt duration, leads to lower quality feathers [13]–[21], [59]–[62] (but note that in the sedentary house sparrow, Passer domesticus, this effect was condition dependent [19]) that deteriorate more between molts [16]–[18], implies that the speedy wing molt carried out by northerly wintering knots represents a cost of reproduction. This trade-off may drive the timing of molt in adults in such a way that the endpoint is as late as possible, so that accumulated feather wear during northward and post-breeding southward migration is as small as possible to minimize the increase in flight costs due to feather wear. This may explain why captive adults timed molt such that the endpoint coincided with that of free-living adults and not of second-years.

That reproduction affects molt and vice versa has been shown before. Generally, reproductive activities tend to delay molt, e.g. resulting in reduced lean tissue mass during the pre-migratory period [59], a lower insulation [21], or a decrease in feather quality [62]. Interestingly, the carry-over effects on molt were only found in late (or experimentally delayed) breeders: i.e., some birds managed to escape the timing effect. Here we have an example where all individual northerly wintering migrating red knots are caught between a rock (the breeding season in the High Arctic and the time needed to travel there and back) and a hard place (steeply deteriorating environmental conditions in the course of autumn), simply by having a life-history that involves a migration towards and from High Arctic breeding locations. The general finding that fast growth leads to lower quality feathers implies that the speedy wing molt shown by birds that have reproduced in the High Arctic represents an unavoidable and routine cost of reproduction.

Interestingly, Hedenström [4], on the basis of a review of body size scaling relationships of different annual cycle components, predicted that only in birds reaching masses of several kilograms, a year would be too short to include breeding, molt, and migration. That large birds encounter problems with the timing of molt is also demonstrated by [45], who show that molt becomes incomplete in birds over 1 kg that maintain flight during molt. To this we can now add that even in birds an order of magnitude smaller, temporal imperatives impose the adoption of non-optimal life-cycle routines in the entire actively breeding population.

Supporting Information

Acknowledgments

Catching knots involved the efforts of many people, often working in volunteer capacity. We are especially grateful to Bernard Spaans, Anne Dekinga, and Maarten Brugge. Maarten Brugge also took care of the captive knots including the measurement of molt. We thank the crew of the research vessel MS Navicula for their hospitality and helping hands during our field trips. During the analyses, Danny Rogers provided regular feedback, and Peter Pyle and two anonymous reviewers provided constructive feedback.