at SciVerse ScienceDirect

Animal Behaviour 84 (2012) 835e842

Contents lists available

Animal Behaviour

journal homepage: www.elsevier .com/locate/anbehav

Shorebird incubation behaviour and its influence on the risk of nest predation

Paul A. Smith a,*, Ingrid Tulp b, Hans Schekkerman c, H. Grant Gilchrist d, Mark R. Forbes a

aDepartment of Biology, Carleton University, Ottawa, ON, Canadab Institute for Marine Resources and Ecosystem Studies, IJmuiden, The Netherlandsc Sovon, Dutch Centre for Field Ornithology, Nijmegen, The Netherlandsd Science and Technology Branch, Environment Canada, Ottawa, ON, Canada

a r t i c l e i n f o

Article history:Received 9 March 2012Initial acceptance 25 April 2012Final acceptance 20 June 2012Available online 9 August 2012MS. number: A12-00188R

Keywords:biparental incubationincubationnest predationnest survivalpredation riskshorebirduniparental incubation

* Correspondence and present address: P. A. SmEcological Research Ltd, 772e7th Conc. South, Pakenh

E-mail address: paul_smith@smitheco.ca (P. A. Sm

0003-3472/$38.00 � 2012 The Association for the Stuhttp://dx.doi.org/10.1016/j.anbehav.2012.07.004

Both nest survival and incubation behaviour are highly variable among shorebirds (Charadrii), and wetested whether more conspicuous incubation behaviour increased the risk of nest predation. During2000e2006, we monitored nest fate at 901 shorebird nests at three study sites across the circumpolarArctic. Using miniature video recorders and nest temperature sensors, we obtained 782 days ofbehavioural data for 161 nests of 11 species. We related nest fate to the rate and duration of adults’ nestabsences or restless movements on the nest, as well as the total proportion of each day that adult birdsengaged in these activities. Nest predation was positively related to the proportion of time that eachspecies left the nest unattended. After controlling for species effects, the likelihood of a successfulnesting attempt was lower for individuals that spent more time off the nest, but among failed nests, thenumber of days that a nest survived prior to depredation was not significantly predicted by measures ofincubation behaviour. To control for weather or seasonal effects, we paired observations from nests thatwere ultimately depredated with observations from successful nests of the same species on the sameday. In this paired sample (dominated by two species: red phalaropes, Phalaropus fulicarius, and littlestints, Calidris minuta), both incubation recesses and restless movements were more numerous amongfailed versus successful nests. Our results suggest that more conspicuous incubation behaviour is indeedrelated to a higher risk of nest predation, and that this relationship may underlie patterns of nest survivalwithin and among shorebird species.� 2012 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Parentsmust balance the costs of providing carewith the benefitsto their offspring of being cared for (Clutton-Brock 1991; Stearns1992). A fundamental parental care behaviour among birds is theincubation of eggs, and because a large proportion of eggs do notsurvive until hatch in most species (e.g. Ricklefs 1969), parent birdscould realize benefits by increasing their investment in incubation.In most previous studies of incubation behaviour, the costs toparents are measured in terms of time and energy, and balancedagainst the need to maintain the eggs at a suitable temperature forembryonic development (Mallory & Weatherhead 1993; Williams1996; Tulp & Schekkerman 2006). However, because predation isthe primary cause of nest failure in almost all avian species studied todate (e.g. Martin 1993), incubation behaviour also may be modifiedto reduce the risk of nest predation.

Increased activity of parent birds around the nest can increase therisk of predation if predators locate nests by sight (Skutch 1949). Forspecies where one mate feeds the other while on the nest(i.e. ‘incubation feeding’), more frequent feeding trips have been

ith, Smith and Associatesam, ON, K0A 2X0, Canada.ith).

dy of Animal Behaviour. Published

linked to reduced nest survival, and feeding trips are suspendedwhen parents are faced with an immediate risk of nest predation(Martin & Ghalambor 1999; Ghalambor & Martin 2002; Martin et al.2000). For species without incubation feeding, more frequent incu-bation recesses may increase the risk of predation (Cresswell et al.2003; Smith et al. 2007a), but this effect has not yet been demon-strated directly.

Shorebirds do not perform incubation feeding, but do varydramatically in their incubation behaviour. Within days orthroughout the season, shorebird incubation behaviour has beenshown to vary in response to environmental conditions and energeticdemands (Norton 1972; Cartar &Montgomerie 1987; Cresswell et al.2004; Tulp & Schekkerman 2006; Smith et al. 2012). However,incubation behaviour is constrained at a higher level by matingsystem; ecologically similar species breeding in sympatry showstrategies ranging from completely uniparental incubation, byfemales or by males, to incubation shared evenly or unevenlybetween the sexes (Pitelka et al. 1974; Székely & Reynolds 1995; StClair et al. 2010). Uniparental incubators leave the nest morefrequently to feed than do members of a biparental pair (e.g. Norton1972; Reneerkens et al. 2011), and someprevious studies suggest thatuniparental birds may suffer higher rates of nest predation in most

by Elsevier Ltd. All rights reserved.

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842836

years (Smith et al. 2007a; Smith & Wilson 2010). There is consider-able variation in incubation behaviour evenwithin groups of specieswith uni- or biparental incubation (Tulp & Schekkerman 2006;Reneerkens et al. 2011), and how this variation may contribute tointerspecific variation in nest survival remains unknown.

We simultaneously monitored the nest survival and incubationbehaviour of shorebirds to determine whether more conspicuousbehaviour increases the risk of nest predation both within andamong species. Data were collected on the rate and duration ofincubation recesses, as well as the rate and duration of restlessmovements on the nest for 11 shorebird species varying widely inincubation behaviour, with both uniparental and biparental incu-bation systems. We related patterns in behaviour among species tointerspecific patterns in nest survival. Within species, we askedwhether successful and failed nests differed in the conspicuousnessof adult behaviour on the nest. We then paired observations fromnests that eventually failed with observations from successful nestsof the same species on the same day to evaluate behaviouraldifferences while controlling for seasonal or weather-related vari-ation in incubation behaviour.

METHODS

Study Area

Fieldworkwas conducted between 2000 and 2006, at three sitesacross the circumpolar Arctic. At Coats Island, Nunavut (62�510N,

200km

Canada

(a) East Bay

(b) Coats Island

Figure 1. Study sites at (a) East Bay, Nunavut, Canada, (b) Co

82�290W; Fig. 1), work was carried out from the beginning of Juneuntil the end of July, 2004e2006. At East Bay, Nunavut (63�590N,81�400W; Fig. 1), we worked from late May until late July in 2002,2005 and 2006. At both of these sites, workers searched for nestsover an area of 12 km2 inwet lowlands, upland heath tundra, raisedbeach ridges and coastal habitat types typical for these latitudes.The third site was located at Medusa Bay, on the Taimyr Peninsula,Russia (73�200N, 80�300E; Fig. 1). Work here was carried out fromearly June to early August, 2000 and 2001, in a 4 km2 area of hillytundra interspersed with wet sedge meadows and scattered, stonyridges. The physiography of the field sites are described in moredetail elsewhere (Canada: Smith et al. 2007a; Siberia: Tulp 2007).

The sample includes behavioural data from 11 species: semi-palmated plover, Charadrius semipalmatus, black-bellied plover,Pluvialis squatarola, American golden-plover, Pluvialis dominica,red phalarope, Phalaropus fulicarius, ruddy turnstone, Arenariainterpres, dunlin, Calidris alpina, curlew sandpiper, Calidris ferru-ginea, semipalmated sandpiper, Calidris pusilla, white-rumpedsandpiper, Calidris fuscicollis, little stint, Calidris minuta, andpectoral sandpiper, Calidris melanotos. The relative abundance ofnests found at each of the sites and the composition of the sampleof behavioural data appear in Table 1. At the Canadian sites, wecollected behavioural data for all of the species that breed insignificant numbers, together representing more than 95% of allbreeding individuals. At Medusa Bay, behavioural data werecollected only for species with uniparental incubation (see below),and thus some abundant species are not represented in the

200km

Russia

(c) Medusa Bay

ats Island, Nunavut, Canada and (c) Medusa Bay, Russia.

Table 1Sample sizes of nests found (i.e. those included in estimates of nest survival at the species level), nests observed to quantify incubation behaviour and days of observationscollected

East Bay (2002, 2005, 2006) Coats Island (2004, 2005, 2006) Medusa Bay (2000, 2001) Incubationsystem

Movementdata?

Nestsfound

Nestsobserved

Nest-daysof records

Nestsfound

Nestsobserved

Nest-daysof records

Nestsfound

Nestsobserved

Nest-daysof records

Semipalmated plover 37 2 3 Biparental YesBlack-bellied plover 45 4 7 11 Biparental YesAmerican golden-plover 1 24 6 12 Biparental YesRed phalarope 40 14 96 36 23 147 13 2 13 Uniparental YesRuddy turnstone 89 8 14 10 Biparental YesDunlin 6 71 6 12 54 Biparental YesCurlew sandpiper 29 15 91 Uniparental NoSemipalmated sandpiper 162 14 26 Biparental YesWhite-rumped sandpiper 36 9 33 18 13 67 Uniparental YesLittle stint 204 38 197 Uniparental NoPectoral sandpiper 3 12 7 64 Uniparental No

For each species, we also list whether incubation is uniparental or biparental, and whether the behavioural data include information on restless movements on the nest.

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842 837

sample (ringed plover, Charadrius hiaticula, Pacific golden-plover,Pluvialis fulva, and dunlin).

All species share many basic features of reproductive ecology,with a typical clutch of four eggs laid in a simple scrape on theground and incubated for approximately 3 weeks (19e26 days), butthey differ in incubation system. Incubation is carried out solely bythemale in the polyandrous red phalarope (Tracy et al. 2002), and bythe female in the curlew sandpiper, white-rumped sandpiper andpectoral sandpiper (Tulp 2007; Smith et al. 2012). Little stint clutchesare incubated uniparentally, but by either the male or the female, asa result of the double-clutch breeding system of this species (Hildén1978; Tulp et al. 2002). The remaining six species are monogamousbiparental incubators. The organization of incubation behaviour iscomplex; all species considered here alter their incubationthroughout the day and in response to variable weather conditions(Tulp & Schekkerman 2006), and these responses may differbetween uniparental and biparental species, or between nests inearly and late incubation (Smith et al. 2012). Overall, uniparentalspecies take more and longer recess, and leave the nest uncoveredfor a greater proportion of each day than biparental species, althoughvariation among species is pronounced (Smith et al. 2012).

Shorebird Nest Finding, Ageing and Monitoring

Methods for finding and monitoring nests are described in detailelsewhere (Smith et al. 2007a; Tulp 2007). Nests found withcomplete clutches were aged with the egg flotation method(Liebezeit et al. 2007), which provides accuracy� 4 days or less inmost instances. Nests were considered successful if one ormore eggshatched, and failed if they were depredated or abandoned. Nestswere considered abandoned if the eggs were cold on two consecu-tive visits spanning at least 4 days. We considered nests depredatedif their contents disappeared before they could possibly havehatched, and if no small eggshell fragments indicating hatchingwerepresent in the empty cup (Mabee 1997). In the cases where nestsdisappeared but the fate could not be assigned with certainty, werecorded the fate as unknown and omitted nests from analyses.

Species-specific estimates of daily nest survival were calculatedusing Mayfield’s (1961) method, with standard errors calculated asper Johnson (1979). We calculated an estimate of the dailymortality rate for each species using all nests found at all sites overthe years of this study (see Table 1).

Monitoring Incubation Behaviour

The equipment used tomonitor incubation behaviour differed forspecies with uniparental and biparental incubation. For uniparental

species, we placed thermistor probes amongst the eggs to capturethe temperature changeswhen birds departed from the nest (Norton1972; Erckmann 1981). At Coats Island and East Bay, we constructedthe probes using a 10 kU Curve-G thermistor on a 15 m, 24 AWGcable, with a 10 kU (�1%) reference resistor loop, on a 2.5 mm stereojack, and attached these probes to a Hobo H8 data logger (OnsetInstrument Corporation, Pocasset, MA, U.S.A.). Loggers were placed15 m from the nest in a camouflaged, waterproof housing, and thecable between them was buried or concealed. At Medusa Bay, weobtained similar probes commercially and attached these to GeminiTiny Tag data loggers (Gemini Data Loggers Inc., West Sussex, U.K.),which were hidden beneath a moss tussock within 1 m from thenest. The placement procedure lasted less than 10 min (for allrecording devices, including those below), and observations fromportable blinds (24 h total) on two nests at East Bay before and afterdeployment of the logger systems revealed that the probes had nodetectable effect on incubator behaviour.

Temperature readings were taken every 30 s at East Bay andCoats Island, and every minute at Medusa Bay. The thermistorsaccurately captured departures of the incubator (i.e. drops intemperature) within this time frame, and visual observations sug-gested that nest absences shorter than 1 minwere rare. At East Bay,we based interpretation of the temperature records on the 24 h ofvisual observations mentioned above. We defined any temperaturechanges of 2 �C or less as continuous incubation. Any dropexceeding 9 �C was considered a recess. The beginning and end ofincubation recesses (�30 s) were identified by examining graphs oftemperature over time. Any sudden change of >2 �C and �9 �C inthe recorded temperature was defined as movements such as eggrolling, cup maintenance, or restless incubation. At Medusa Bay,large drops in temperature were classified as recesses but no effortwas made to identify movements on the nest or restless incubation(see Tulp & Schekkerman 2006 for details).

Temperature readings were generally unambiguous because theprobes were placed such that they were near to the warm broodpatches when the bird was incubating. Occasionally, however, thetip of the probe was displaced and temperature readings becamedifficult to interpret. Readings also became erratic when eggs beganto pip and chicks began to thermoregulate. Erratic temperaturedata were discarded prior to analysis.

The incubation of biparental species cannot be monitoredadequately with the above method because change-overs betweenpair members could happen too rapidly to be detected by a drop intemperature. Instead, we used portable video-recording systemsconsisting of a small (10� 2 cm), weatherproof, low-lux, CMOScamera, connected to a 40 GB digital video recorder. Using a 21 Ah(w10 kg) lead acid battery, these comparatively economical systems

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842838

could record full-motion video of incubating birds for continuousperiods ofmore than 36 h. Cameraswere placed approximately 10 maway from nests, and were elevated 30e40 cm above the groundwith wire tripods. The wire legs of the tripods were pressed into theground to stabilize the camera. The battery and recorder were con-cealed in a camouflage-painted waterproof housing and placedanother 10 m further from the nest. To analyse videos, we watchedthem at 4� speedwith DivX� computer software, and pausedwherenecessary to record behaviours� 1 s. These video systems were onlydeployed at East Bay and Coats Island.

Although the methods of data collection differed among bipa-rental and uniparental species, we derived similar measures ofincubation behaviour for both. For all species, we calculated (1) thefrequency of nest absences, (2) the duration of nest absences and(3) the proportion of time that nests were left unattended. Forbiparental species, these absences could end with the return ofeither parent; in some cases the sexes or individuals could bedistinguished while in others they could not. At Coats Island andEast Bay, we also quantified the (4) frequency and (5) duration ofmovements, and (6) the proportion of each day occupied by thesemovements. For uniparental species, these were defined on thebasis of the temperature records (see above). For biparental species,‘movements’ were detected from the videos and constituted nestmaintenance, egg rolling, feeding from the incubating position,preening and repositioning.

Fog and strong winds occasionally reduced the quality of thevideo and impaired our ability to record the full suite of incubationmovements. In these instances, the value for number of movementsrepresents a minimum. If the quality of the recording was so poorthat recesses and change-overs could not be reliably identified, wediscarded the video from all analyses. Incubation behaviour can beerratic during laying (Norton 1972), so we monitored behaviouronly once clutches were complete.

Data Analysis

Incubation behaviour is highly variable, and the effects of interesthere could be masked by the complex interactions with time of day,weather, nest age and species. Also, predation rates are known tovary substantially among years, but year, site and species areconfounded in our data set to varying degrees. To reduce the risk ofspurious conclusions, we relied on multiple lines of evidence toquantify the relationship between incubation behaviour and the riskof nest predation (Table 2). To account for diel patterns in incubationbehaviour, all analyses are based on records spanning 24 h.

Interspecific differences in nest survival are strong, and we firstasked whether these differences in the risk of predation wererelated to interspecific differences in incubation behaviour. Weentered the species-specific daily mortality rate as the dependentvariable in a general linear model (GLM) and tested the influence ofthe metrics of behaviour as covariates. Some are correlated (e.g.

Table 2Summary of variables considered and tests conducted to evaluate the influence of incub

Response variable Candidate explanatory variables Method

Species-specific dailymortality rate

Incubation system, six metrics ofincubation behaviour, taxonomic family

General linear

Nest fate Year, species, six metrics of incubationbehaviour (random effect)

Generalized limixed model

Time to failure (days) Year, species, six metrics of incubationbehaviour (random effect)

Generalized limixed model

Six metrics of incubationbehaviour from paireddata for successfuland failed nests

Species, nest fate General linearpaired t tests

time per recess � recess frequency ¼ the total proportion of timethat the nest is unattended), but because we were interested inidentifying the strongest behavioural correlates of nest survival, weused a forward stepwise procedure and type I sums of squares. Wehad data on rates and durations of movements on the nest for 8 of11 species; we conducted one analysis on all species using onlyrecess data, and a second analysis on these eight species using thethree measures of movements on the nest. Both nest survival andincubation behaviour may vary by mating system (Smith & Wilson2010; Smith et al. 2012), or as a result of shared phylogeny. Wetherefore tested whether a fixed effect of uniparental versus bipa-rental incubation, or taxonomic family, changed the interpretationof the model results.

We then asked whether the incubation behaviour of birds atnests that were ultimately successful differed from that of birds atnests that eventually failed. We used generalized linear mixedmodels (GLMM, logit link function) with nest fate as the dependentvariable, and the six metrics of incubation behaviour as predictors.We included species and year as fixed effects to account for annualand species-specific differences in nest survival (year and site wereconfounded, so that a year*site interaction was largely redundant).Because we had a variable number of repeated observations fromindividual nests, we included nest as a random effect. As above, weconducted one analysis including all 11 species and the threemeasures of incubation recesses, and a second analysis for the eightspecies for which the threemeasures of incubationmovementswereavailable. Some nests fail faster than others, and we tested whetherthe nests that failed the fastest had themost conspicuous incubationbehaviour. Analyses were limited to failed nests, and time to failure(in days) was treated as the dependent variable (Gaussian linkfunction). Nest was included as a random effect, andwe included themetrics of behaviour as covariates. Because interspecific differencesin nest survival could create interspecific differences in time tofailure, we added species to the model as a fixed effect. Similarly,interannual variability was controlled for by including a year effect.

Finally, we conducted analyses on paired records, whereobservations for depredated nests werematched with observationsfrom similar, but successful, nests. For each day of observation ona failed nest, we selected an observation from the same site, yearand day, for a successful nest of the same species. Nest age wasmatched as closely as possible (mean � SE difference ¼ 1.8 � 0.4days), and records from successful nests were used more than onceif unsuccessful nests outnumbered successful nests on a given dayof observation. While greatly reducing sample size, this approacheliminates the confounding influence of day-to-day variation inincubation behaviour. We then assessed whether incubationbehaviour differed between successful and failed nests in thesepaired records using a GLM including species and nest fate ascovariates. Within species, metrics of behaviour were comparedamong successful and failed nests with paired t tests. All analyseswere conducted with program R (R Development Core Team 2012)

ation behaviour on survival of shorebird nests

Question addressed

model Is variation in nest survival among species explainedby differences in incubation behaviour?

near Within species, is more conspicuous incubation behaviourassociated with nest failure?

near Within species, is more conspicuous incubation behaviourassociated with more rapid nest failure?

model, After controlling for seasonal, weather and species effects,does incubation behaviour differ between failed and successful nests?

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842 839

and means are displayed � SE. To better satisfy the assumptions ofnormality, we log-transformed the counts of recesses and move-ments and arcsine transformed the proportions of time that birdsspent engaged in these activities prior to most analyses. Allresearch in Canada was approved through permits from theCanadianWildlife Service, the Territory of Nunavut and the KivalliqInuit Association.

RESULTS

Nest Survival and Behaviour among Species

We found and monitored a total of 901 nests of 11 species. Five ofthese species have uniparental incubation,while incubation is sharedin the remaining six (Table 1). This sample was used to generatespecies-specific estimates of nestmortality.We obtained behaviouraldata for 161 nests and a total of 782 nest-days of observation(Table 1), and this sample forms the basis of our behavioural metrics.

Of 852 nests for which fate was known, 522 (61%) were depre-dated and only 23 (3%) were abandoned. Estimates of the dailymortality rate varied among species, froma lowof 0.040 � 0.007 forthe black-bellied plover to a high of 0.155 � 0.040 for the curlewsandpiper. Rates of nestmortality varywithin and among years, andinterannual variation in predation is explored in detail elsewhere(Tulp & Schekkerman 2001; Schekkerman et al. 2004; Smith 2009;Smith & Wilson 2010).

Among the 11 species, daily mortality rate was positively relatedto the duration of incubation recesses (F1,9 ¼ 12.2, P< 0.01) and alsoto the proportion of time that the nest was left unattended (arcsinetransformed: F1,9 ¼ 44.3, P< 0.0001), but not to the number ofrecesses (Fig. 2aec). Whether a species showed uniparental versusbiparental incubationwas significantly related to daily mortality rate(F1,9 ¼ 12.1, P< 0.01), but not whether a species was a sandpiper(Scolopacidae) or plover (Charadriidae). Only the proportion of timespent off the nest remained significant when effects were combined,suggesting that the strong effect of this variable (R2 ¼ 0.83) super-seded effects of time per recess or mating system per se. Thus, someof the large variation in nest predation among species was related toincubation behaviour, with those species spendingmore time off thenest suffering greater rates of predation.

In contrast, we found no evidence that the rate or duration ofmovements, or the total proportion of time spent performingmovements at the nest influenced nest predation (Fig. 2d, e). Amongthe eight species for which we had movement data, no measures ofmovements on the nest were significant predictors of daily mortalityrate in the GLM (all Ps > 0.4).

Incubation Behaviour at Successful versus Failed Nests

We had incubation recess or movement data for 161 nests, andfate was known for 149 of these. Among these, 93 (62%) failed. Thisvalue is nearly identical to the proportion of nest failures observedin the full sample of nests (61% nest failure among 852 nests),suggesting that the deployment of data-logging devices at nestsites did not increase the risk of predation.

In logistic regression (GLMM) analyses, the fate of a nestingattempt (successful versus unsuccessful) varied marginally amongyears (with more nest failures in 2002, and fewer in 2004). Althoughsignificant differences in the rate of nest predation among specieswere evident in the larger sample of nests, interspecific variationwasnot strong for the sample of 161 nests for which we had behaviouraldata. In contrast, more frequent incubation recesses (z¼ 2.09,P¼ 0.037), and a greater amount of time spent off the nest (z¼ 2.60,P¼ 0.009)were significantlyassociatedwithnest failure. Yeardidnotremain significantwhenadded to eithermodel. For the reduced set of

eight species for which we had movement data, nest fate was notrelated to the rate of movements on the nest.

Time to Failure

For the sample of 93 failed nests for which we had behaviouraldata, we measured the time from the onset of incubation until thenest was depredated. In 80 cases, the exact time of predation wasknown, but for the remaining 13 we assumed that predationhappened midway between the penultimate and final visit. Thistime to failure varied widely, from 2 days to 26 days, with a mean of12.1 � 0.6 days.

The variation in this value among the restricted sample of nestswas not related to species, nor was it related to year, despite bothyear and species effects being detected in the daily risk of nestmortality in the larger sample of nests (see above). None of the sixmetrics of incubation behaviour (transformed as appropriate) wasrelated to time to failure (all ts < 1.3, Ps > 0.05).

Analyses of Paired Observations

Incubation behaviour can vary dramatically among species anddays, and the use of paired records from successful and failed nests ofa given species on a given day is an effective means of accounting forthis variation. However, sample size is limited dramatically by thisapproach, and the sample of paired records included 85 nests anda total of 211 pairs of observation days, with some species poorlyrepresented (Table 3). After controlling for species effects, we foundthat nest fate was a significant predictor of the rate of both nestrecesses and movements, with successful nests associated withlower values for both measures (recess rate: log transformed:bfate� SE¼ �0.29� 0.15, t1 ¼ �2.0, P ¼ 0.047; movement rate: logtransformed: bfate � SE ¼ �0.096� 0.045, t1 ¼ �2.1, P¼ 0.033). Thesample is dominated by records for red phalaropes and little stints(Table 3). For both of these species,measures of incubation behaviourwere larger for individuals whose nests failed, suggesting that moreconspicuous behaviour may have increased the risk of predation. Forred phalaropes, paired t tests suggested that incubators whose nestsfailed had significantly higher rates of nest recesses (t88 ¼ �2.78,P¼ 0.007) and spent more time off the nest (t88 ¼ �2.81, P¼ 0.006).For little stints, only the rate of nest recesses differed significantlybetween failed and successful nests (t67 ¼ �2.86, P¼ 0.006). Resultsfor other species, particularly those with very small samples, weremore variable.

DISCUSSION

Nesting in the Arctic is energetically challenging (e.g. Piersma &Morrison 1994; Piersma et al. 2003; Cresswell et al. 2004), andshorebirds must leave the nest to feed frequently in order to balancetheir energy budget. These breaks in incubation have traditionallybeen viewed as the means by which the costs of incubation arerecouped. However, our results suggest these recesses also entailcosts, with more frequent or longer incubation recesses increasingthe risk of nest predation.

Variation in nest survival among species of shorebirds is large,and it has been shown previously that for some species, this vari-ation is related towhether the species has uniparental or biparentalincubation (Smith et al. 2007a; Smith & Wilson 2010). Here, ourresults suggest that this is not caused by shared versus single-parent incubation per se, but is rather a product of increased riskof predation for species that take longer incubation recesses and/orleave the nest uncovered for a greater proportion of time.

We found similar effects within species, where the fate of a nestacross all species could be predicted by the rate of incubation

0.2(a)

(b)

(c) (f)

(e)

(d)

0.15

0.1

0.05

00.2 0.4 0.6 0.8 1 1.2 1.4 1.6

0.2

0.15

0.1

0.05

00 5 10 15 20

0.2

0.15

0.1

0.05

00 0.05 0.1 0.15

%Time off the nest %Time moving

Mean duration of movements (min)Mean recess duration (min)

No. recesses /h

Dai

ly m

orta

lity

rat

e

No. movements /h

0.2 0 0.01 0.02 0.03

0.2

0 0.40.2 0.6 0.8 10

0.05

0.1

0.15

0.2

0.2

0.15

0.1

0.05

00 0.5 1 1.5 2 2.5

0.15

0.1

0.05

0

Figure 2. (aef). Relation between six measures of incubation behaviour (untransformed) and daily mortality rates of shorebird nests. Each data point represents a species; (def)behavioural data include information about movements on the nest for only 8/11 species.

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842840

recesses and the total proportion of time spent off the nest. Evenamong paired records, where the confounding influence of varia-tion in environmental conditions was controlled for, we found thatrecesses andmovements were significantly more numerous amongfailed versus successful nests. Thus, through a variety of analyses,we demonstrated that the rate and duration of nest recesses, andthe total proportion of time that a nest is left unattended, playimportant roles in determining nest fate. Although these metrics ofincubation are correlated, the mechanism by which they couldinfluence the risk of predation might vary.

Table 3Sample sizes andmeans (�SE) for the six metrics of incubation behaviour from the samplesame year, species and day, from successful nests

Species Fate N nests N paired records(days)

Recesses/h � SE Moves

Little stint Succ. 8 68 1.27�0.03Failed 18 1.41�0.06

Red phalarope Succ. 10 89 0.86�0.03 0.86�0Failed 18 1.04�0.05 1.00�0

We present data only from red phalaropes and little stints, the two species that domina

Frequent trips to and from the nest could disclose its location topredators that use visual cues. Skutch (1949) first hypothesized thisrelationship in the context of trips to feed altricial nestlings, and theeffect has since been demonstrated in a variety of species (e.g.Martin & Ghalambor 1999; Martin et al. 2000; but see Roper &Goldstein 1997). In our analyses of paired records, for example,red phalaropes incubating nests that would ultimately succeedtook 17% fewer recesses than those with nests that ultimatelyfailed. Restless movements while incubating (e.g. egg rolling, nestmaintenance, preening), may have the same negative effect, and

where observations from unsuccessful nests were pairedwith observations from the

/h � SE % Off nest � SE % Moving � SE Time/recess(min) � SE

Time/movement(min) � SE

0.19�0.01 9.90�0.940.20�0.01 10.81�1.44

.13 0.16�0.01 0.01�0.00 11.70�0.86 0.66�0.03

.14 0.19�0.01 0.01�0.00 14.05�1.58 0.97�0.20

ted this paired data set.

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842 841

were associated with increased risk of predation in some of ouranalyses. These conspicuous behaviours may act as cues for pred-ators, directly disclosing the location of the nest.

Time spent off the nest may also lead to nest failure if thepredators observe birds while they are off the nest and followthem back to the nest location. Shorebird nest sites are typicallymore concealed than the exposed habitats in which the adultsforage (e.g. Smith et al. 2007a), and more time spent in exposedhabitats, such as pond edges, may increase the likelihood of beingseen by a predator. One of the techniques that we as human‘predators’ employ while searching for nests is to watch foragingbirds from a distance and follow them back to the nest site; itseems intuitive that predators could profit from a similarapproach. The proportion of time that a nest was left unattended,and then returned to, varied from 3% to 19%, and accounted for83% of the variation in nest survival among species. Becauseincubation is shared for biparental species, they undertake muchof their foraging while their mate is incubating. As experiencednest searchers, we recognize that it is less profitable to watcha foraging bird if it is a species with biparental incubation (becauseit may be the off-duty pair member). Whether predators that usevisual cues discern between shorebird species and focus on thosenests that are easier to find is unknown, but may also contribute tothe differences observed in nest survival.

Whether incubation is biparental or uniparental, leaving the nestunattended could also increase the risk of predation by leaving theeggs uncovered or by compromising a bird’s ability to defend its nest.By sitting on the nest, birds with cryptic plumage and conspicuouseggs can camouflage their nest contents (Martin 1992 and referencestherein). Studies with artificial nests mimicking those of shorebirdshave shown inconsistent results, with both higher (Erckmann 1981;Smith 2009) and lower (Ashkenazie & Safriel 1979; Safriel 1980) ratesof nest survival for unattended versus real, active nests. However,shorebird eggs have muted background colours and heavy mottling;common adaptations among ground-nesting birds to provide cam-ouflage and reduce nest predation (e.g. Montevecchi 1976; Solís & deLope 1995; Lloyd et al. 2000). It therefore seems unlikely that eggswould be markedly more conspicuous than an incubating parent.Studies of nest, egg and incubator crypsis in this system are ongoing,and we cannot at present discount the possibility that eggs areconspicuous and that leaving them uncovered increases the risk ofnest predation.

By remaining on the nest, birds also ensure that they are presentto defend it when predators approach. Nest defence is common ina wide variety of bird species (Montgomerie & Weatherhead 1988),and shorebirds perform defence behaviours ranging from aggres-sive mobbing to distraction displays (Gochfeld 1984). Aggressivenest defence is most prevalent among biparental shorebirds(Larsen 1991; Larsen et al. 1996; Smith & Wilson 2010), and of thespecies considered here, black-bellied plovers and ruddy turn-stones are most likely to pursue avian predators aggressively(P. Smith, personal observation), while other species more charac-teristically use distraction displays or no defence. The effectivenessof nest defence is dependent upon early detection of predators(e.g. Martin 1992; Götmark et al. 1995), and limiting time awayfrom the nest ensures that a parent is able to respond appropriatelywhen predators are nearby.

The behavioural responses of parents to a perceived risk ofpredation may be more subtle than aggressive defence (review inLima & Dill 1990). For example, the incubation feeding hypothe-sized by Skutch (1949) to be a risky behaviour is suspended whenpredators are nearby (Ghalambor & Martin 2002) and increaseswhen predators are experimentally removed (Fontaine & Martin2006). Similarly, for red phalaropes at the East Bay study site,individuals nesting within the protective umbrella of aggressive

Sabine’s gulls, Xema sabini, behaved less cryptically, taking moreand longer recesses, than those in areas with a higher risk ofpredation (Smith et al. 2007b). Because the results presented hereare based upon observed behaviour and observed risk of predation,responses by shorebirds to modify their behaviour could influencethe interpretation of our results.

For example, some studies suggest that incubation recesses areshorter as nests age, and shorter than the improvement inweather conditions alone predict (Smith et al. 2012; but see Tulp &Schekkerman 2006). The value of a nesting attempt to parentsincreases as nests approach hatch, and shorter incubation recessesfor older nests could reflect the desire for parents to remain nearerto their nests. At the same time, the intensity of nest defenceincreases with nest age, and perhaps as a result of these behav-ioural changes, higher nest survival is realized (Smith & Wilson2010). An adaptive reduction in the duration of nest absencesthroughout the season would strengthen the relationshipbetween recess duration and nest survival as measured here.Further study of patterns in incubation behaviour among yearswith varying predation pressure, or with experimental manipu-lation of energetic requirements or risk of predation arewarranted.

The nature of predators’ hunting behaviour, visual versusolfactory, could also influence the manner in which incubationbehaviour influences the risk of predation. The primary nestpredators at the three study sites are Arctic foxes, Alopex lagopus,and parasitic- and long-tailed jaegers, Stercorarius parasiticus andStercorarius longicaudus, with a variety of other predators includingpomarine jaegers, Stercorarius pomarinus, gulls, common ravens,Corvus corax, and ermine, Mustela erminea, encountered lessfrequently. Avian predators hunt by sight, and jaegers are routinelyobserved flying low over the tundra, searching for eggs, lemmingsor birds (Maher 1974; Wiley & Lee 1999). Foxes are known to useolfaction primarily (and often scent-mark nests after depredation),but as opportunistic foragers, may use any visual or olfactory cuesavailable to them to locate food. Moreover, increased activity nearthe nest in the form of movements or recesses could reinforce scentmarks. More detailed study of the nature of predation events couldclarify the role played by incubation behaviour in increasing therisk of nest predation.

Similarly, a better understanding of when and over how large anarea predators hunt for shorebird nests could help to explain thevariable time to failure that we observed. We found a link betweenbehaviour and risk of predation, but not between behaviour andtime to failure. This could simply be a product of a small sample forthe latter test. However, if predators hunt over large areas and covera given area irregularly or infrequently, or if they focus on shorebirdnests only at specific times during the nesting season, time tofailure might not vary as a direct function of risk of predation.

We have demonstrated that more frequent or longer breaks inincubation, leaving the nest unattended for a greater proportion oftime, and/or showing more restless movement on the nest canincrease the risk of nest predation for ground-nesting shorebirds.This result suggests that shorebirds could increase their nestsurvival by reducing the conspicuousness of their incubationbehaviour. However, shorebirds’ ability to do so might be con-strained by several facets of their ecology and life history. Specieswith uniparental incubation (typically polygamous) must leave thenest to feed, and this energetic constraint imposes a limit on thepossible levels of nest attendance. Energetic limitation is less severeamong biparental species (each member of a monogamous pair canfeed while its mate is incubating), but incubation behaviour mustbe coordinated between pair members. The potential for plasticityin incubation behaviour, and the ability of parent birds to reducetheir risk of predation by modifying it, therefore varies among

P. A. Smith et al. / Animal Behaviour 84 (2012) 835e842842

species and represents an additional selective pressure to beconsidered in studies of the evolution of shorebirds’ varied lifehistory strategies.

Acknowledgments

We are indebted to the many field assistants responsible forfinding and monitoring nests. Field work in Canada was funded byEnvironment Canada, and logistical support provided by the PolarContinental Shelf Project. Financial support for the expeditions toMedusa Bay was provided by the former Dutch Ministry for Agri-culture, Nature Management and Fisheries (currently Ministry ofEconomic affairs, Agriculture and Innovation). I.T. received grantsfrom the Association for the Study of Animal Behaviour, The DutchOrganisation for the Netherlands Organisation for Scientific Research(NWO) and the European Science Foundation. We thank two anon-ymous referees for helpful comments on a previous draft of thismanuscript.

References

Ashkenazie, S. & Safriel, U. N. 1979. Time-energy budget of the semipalmatedsandpiper Calidris pusilla at Barrow, Alaska. Ecology, 60, 783e799.

Cartar, R. V. & Montgomerie, R. D. 1987. Day-to-day variation in nest attentivenessof white-rumped sandpipers. Condor, 89, 252e260.

Clutton-Brock, T. H. 1991. The Evolution of Parental Care. Princeton, New Jersey:Princeton University Press.

Cresswell, W., Holt, S., Reid, J. M., Whitfield, D. P. & Mellanby, R. J. 2003. Doenergetic demands constrain incubation scheduling in a biparental species?Behavioral Ecology, 14, 97e102.

Cresswell, W., Holt, S., Reid, J. M., Whitfield, D. P., Mellanby, R. J., Norton, D. &Waldron, S. 2004. The energetic costs of egg heating constrain incubationattendance but do not determine energy expenditure in the pectoral sandpiper.Behavioral Ecology, 15, 498e507.

Erckmann, W. J., Jr. 1981. The evolution of sex-role reversal and monogamy inshorebirds. Ph.D. thesis, University of Washington.

Fontaine, J. J. & Martin, T. E. 2006. Parent birds assess nest predation risk andadjust their reproductive strategies. Ecology Letters, 9, 428e434.

Ghalambor, C. K. & Martin, T. E. 2002. Comparative manipulation of predation riskin incubating birds reveals variability in the plasticity of responses. BehavioralEcology, 13, 101e108.

Gochfeld, M. 1984. Antipredator behavior: aggressive and distraction displays ofshorebirds. In: Behaviour of Marine Animals. Vol. 5. Shorebirds: Breeding Behaviorand Populations (Ed. by J. Burger & B. Olla), pp. 289e377. New York: Plenum.

Götmark, F., Blomqvist, D., Johansson, O. C. & Bergkvist, J. 1995. Nest siteselection: a trade-off between concealment and view of the surroundings?Journal of Avian Biology, 26, 305e312.

Hildén, O. 1978. Occurrence and breeding biology of the little stint Calidris minutain Norway. Anser, Supplement, 3, 96e100.

Johnson, D. H. 1979. Estimating nesting success: the Mayfield method and analternative. Auk, 96, 651e661.

Larsen, T. 1991. Anti-predator behaviour and mating systems in waders: aggressivenest defence selects for monogamy. Animal Behaviour, 41, 1057e1062.

Larsen, T., Sordahl, T. A. & Byrkjedal, I. 1996. Factors related to aggressive nestprotection behaviour: a comparative study of Holarctic waders. BiologicalJournal of the Linnean Society, 58, 409e439.

Liebezeit, J. R., Smith, P. A., Lanctot, R. B., Schekkerman, H., Tulp, I., Kendall, S. J.,Tracy, D., Rodrigues, R. J., Meltofte, H., Robinson, J. A. R., et al. 2007. Assessingthe development of shorebird eggs using the flotation method: species-specificand generalized regression models. Condor, 109, 32e47.

Lima, S. L. & Dill, L. M. 1990. Behavioural decisions made under the risk ofpredation: a review and prospectus. Canadian Journal of Zoology, 68, 619e640.

Lloyd, P., Plaganyi, E., Lepage, D., Little, R. M. & Crowe, T. M. 2000. Nest-siteselection, egg pigmentation and clutch predation in the ground-nestingNamaqua sandgrouse Pterocles namaqua. Ibis, 142, 123e131.

Mabee, T. J. 1997. Using eggshell evidence to determine nest fate of shorebirds.Wilson Bulletin, 109, 307e313.

Maher, W. J. 1974. Ecology of pomarine, parasitic, and long-tailed jaegers innorthern Alaska. Pacific Coast Avifauna, 37, 1e148.

Mallory, M. L. & Weatherhead, P. J. 1993. Incubation rhythms and mass loss ofcommon goldeneyes. Condor, 95, 849e859.

Martin, T. E. 1992. Interaction of nest predation and food limitation in reproductivestrategies. Current Ornithology, 9, 163e197.

Martin, T. E. 1993. Nest predation among vegetation layers and habitat types:revising the dogmas. American Naturalist, 141, 897e913.

Martin, T. E. & Ghalambor, C. K. 1999. Males feeding females during incubation. I.Required by microclimate or constrained by nest predation? AmericanNaturalist, 153, 131e139.

Martin, T. E., Scott, J. & Menge, C. 2000. Nest predation increases with parentalactivity: separating nest site and parental activity effect. Proceedings of the RoyalSociety B, 267, 2287e2293.

Mayfield, H. F. 1961. Nesting success calculated from exposure. Wilson Bulletin, 73,255e261.

Montevecchi, W. A. 1976. Field experiments on the adaptive significance of avianeggshell pigmentation. Behaviour, 58, 26e39.

Montgomerie, R. D. & Weatherhead, P. J. 1988. Risks and rewards of nest defenceby parent birds. Quarterly Review of Biology, 63, 167e187.

Norton, D. W. 1972. Incubation schedules of four species of calidridine sandpipersat Barrow, Alaska. Condor, 74, 164e176.

Piersma, T. & Morrison, R. I. G. 1994. Energy expenditure and water turnover ofincubating ruddy turnstones: high costs under High Arctic conditions. Auk, 111,366e376.

Piersma, T., Lindström, Å., Drent, R. H., Tulp, I., Jukema, J., Morrison, R. I. G.,Reneerkens, J., Schekkerman, H. & Visser, G. H. 2003. High daily energyexpenditure of incubating shorebirds on High Arctic tundra: a circumpolarstudy. Functional Ecology, 17, 356e362.

Pitelka, F. A., Holmes, R. T. & MacLean, S. T. 1974. Ecology and evolution of socialorganization in Arctic sandpipers. American Zoologist, 14, 185e204.

R Development Core Team 2012. R: a Language and Environment for StatisticalComputing. Vienna, Austria: R Foundation for Statistical Computing. http://www.R-project.org/.

Reneerkens, J., Grond, K., Schekkerman, H., Tulp, I. & Piersma, T. 2011. Douniparental sanderlings Calidris alba increase egg heat input to compensate forlow nest attentiveness? PLoS ONE, 6, e16834.

Ricklefs, R. E. 1969. An analysis of nesting mortality in birds. Smithsonian Contri-butions to Zoology, 9, 1e48.

Roper, R. J. & Goldstein, R. R. 1997. A test of the Skutch hypothesis: does activity atnests increase nest predation risk? Journal of Avian Biology, 28, 111e116.

Safriel, U. N. 1980. The semipalmated sandpiper: reproductive strategies andtactics. Ibis, 122, 425.

St Clair, J. J. H., Küpper, C., Herrmann, P., Woods, R. W. & Székely, T. 2010. Unusualincubation sex-roles in the rufous-chested dotterel Charadrius modestus. Ibis,152, 402e404.

Schekkerman, H., Tulp, I., Calf, K. M. & DeLeeuw, J. J. 2004. Studies on BreedingShorebirds at Medusa Bay, Taimyr, in Summer 2002. Report 922. Wageningen:Alterra.

Skutch, A. F. 1949. Do tropical birds rear as many young as they can nourish? Ibis,91, 430e455.

Smith, P. A. 2009. Variation in shorebird nest survival: proximate pressures andultimate constraints. Ph.D. thesis, Carleton University, Ottawa.

Smith, P. A. & Wilson, S. D. 2010. Intraseasonal patterns in shorebird nest survivalare related to nest age and defence behaviour. Oecologia, 163, 613e624.

Smith, P. A., Gilchrist, H. G. & Smith, J. N. M. 2007a. Effects of nest habitat, food,and parental behaviour on shorebird nest success. Condor, 109, 15e31.

Smith, P. A., Gilchrist, H. G., Smith, J. N. M. & Nol, E. 2007b. Annual variation in thebenefits of a nesting association between red phalaropes (Phalaropus fulicarius)and Sabine’s gulls (Xema sabini). Auk, 124, 276e290.

Smith, P. A., Dauncey, S. A., Gilchrist, H. G. & Forbes, M. R. 2012. The influence ofweather on shorebird incubation. In: Studies in Avian Biology. Vol. 43: VideoSurveillance of Nesting Birds (Ed. by C. A. Ribic, F.R. Thompson III & P. J. Pietz),pp. 89e104. Berkeley: University of California Press.

Solís, J. C. & de Lope, F. 1995. Nest and egg crypsis in the ground nesting stonecurlew Burhinus oedicnemus. Journal of Avian Biology, 26, 135e138.

Stearns, S. 1992. The Evolution of Life Histories. Oxford: Oxford University Press.Székely, T. & Reynolds, J. D. 1995. Evolutionary transitions in parental care in

shorebirds. Proceedings of the Royal Society B, 262, 57e64.Tracy, D. M., Schamel, D. & Dale, J. 2002. Red phalarope (Phalaropus fulicarius). In:

The Birds of North America Online. No. 698 (Ed. by A. Poole). Ithaca, New York:Cornell Lab of Ornithology, http://bna.birds.cornell.edu/bna/species/698. http://dx.doi.org/10.2173/bna.698.

Tulp, I. 2007. The Arctic pulse: timing of breeding in long-distance migrantshorebirds. Ph.D. thesis, Rijksuniversiteit Groningen, The Netherlands.

Tulp, I. & Schekkerman, H. 2001. Studies on Breeding Shorebirds at Medusa Bay,Taimyr, in Summer 2001. Report 451. Wageningen: Alterra.

Tulp, I. & Schekkerman, H. 2006. Time allocation between feeding and incubationin uniparental Arctic-breeding shorebirds: energy reserves provide leeway ina tight schedule. Journal of Avian Biology, 37, 207e218.

Tulp, I., Schekkerman, H., Chylarecki, P., Tomkovich, P., Soloviev, M., Bruinzeel, L.,van Dijk, K., Hildén, O., Hötker, H., Kania, W., et al. 2002. Body mass patterns oflittle stints Calidris minuta during incubation and chick-rearing at different lati-tudes. Ibis, 144, 122e134.

Wiley, R. H. & Lee, D. S. 1999. Parasitic jaeger (Stercorarius parasiticus). In: The Birdsof North America Online. No. 445 (Ed. by A. Poole). Ithaca, New York: Cornell Labof Ornithology, http://bna.birds.cornell.edu/bna/species/445. http://dx.doi.org/10.2173/bna.445.

Williams, J. B. 1996. Energetics of avian incubation. In: Avian Energetics andNutritional Ecology (Ed. by C. Carey), pp. 375e415. London: Chapman & Hall.