high water roost selection by waders: maximizing feeding opportunities or avoiding predation?

Ibis

(2006),

148

, 88–97

© 2006 The Authors Journal compilation © 2006 British Ornithologists’ Union

Blackwell Publishing Ltd

High water roost selection by waders: maximizing feeding opportunities or avoiding predation?

SUSANA ROSA,* ANA L. ENCARNAÇÃO, JOSÉ P. GRANADEIRO & JORGE M. PALMEIRIM

Centro de Biologia Ambiental, Departamento de Biologia Animal, Faculdade de Ciências da Universidade de Lisboa,

Campo Grande, 1749-016 Lisbon, Portugal

During the highest spring tides the intertidal sediment flats of estuaries are fully inundatedat high water, and waders have no choice but to move to supratidal roosts, e.g. on open farm-land, saltpans or beaches. However, in many estuaries during the lowest neap or interme-diate tides there are sectors of upper intertidal sediment flats that remain exposed even atthe peak of high water, and so waders have the choice of roosting either there or in supratidalsites. In the Tagus Estuary, Portugal, as elsewhere, waders use both types of roosts duringhigh water. Our main objective was to understand what makes waders opt for one of thesetwo types of available roosts. We monitored wader use of saltpans and intertidal roosts fromspring to neap tides, and measured foraging and alarm behaviour, prey availability and dis-turbance by predators. Most of the wader species studied chose intertidal (mudflat) roostswhenever these were available, and only roosted in saltpans during the peak of spring tides.We hypothesized that this preference was explained either by an attempt (i) to continuefeeding into the high water period, or (ii) to minimize predation risk. Extending feeding timeinto the high water period did not seem to be very relevant for roost choice because bothprey availability and foraging activity were low in both types of roosts. However, predatordisturbance was several times higher in the saltpans than in the intertidal roosts, suggestingthat this factor may be the determinant in the choice of roost type.

Through most of the winter, estuaries harbour largeconcentrations of waders, which forage on the inter-tidal sediment flats during low tide. However, as thetide advances, birds are forced to move to high waterroosts, where they must remain until the recedingtide exposes their feeding grounds. During the highwater, birds congregate in these areas in very denseflocks and spend most of their time there sleeping,preening and, in some cases, feeding.

During spring tides the intertidal sediment flats ofthe estuaries are fully inundated at high waters andbirds have no other choice than to move to supratidalroosts. However, in many estuaries at neap or inter-mediate tides there are small sectors of upper inter-tidal sediment flats that remain exposed and thusavailable for waders even at the peak of high water.Whenever these intertidal sectors are available athigh water, waders have to choose between roostingin them or at other sites, such as saltpans.

Although these roosts are critical for waders, theyare frequently under great pressure because they aregenerally located in areas that are highly modified byhuman activity such as open farmland and saltpans.Such sites often have to be protected and managed inorder to maintain their attractiveness to birds; however,knowledge of the factors that underlie the choice ofroosts by waders is very limited (Rogers 2003).

Some wader species may spend a substantialpercentage of high water time feeding, in order tocomplement the low-tide food intake (Masero

et al

.2000, Luís

et al

. 2002). This suggests that in somecircumstances waders, unable to fulfil their energeticrequirements during low tide, may base their selec-tion of high water roosting sites on their feedingpotential. The presence or absence of predators is alsoan important factor in the choice of roost (Luís

et al

.2001). In fact, predator avoidance may also have astrong effect on wader behaviour, even when the riskof predation is low (Rogers 2003).

It is clear that both factors, extended feedingtime and predator avoidance, have the potential to

*Corresponding author.Email: [email protected]

© 2006 The AuthorsJournal compilation © 2006 British Ornithologists’ Union

High water roost selection by waders

89

influence the choice of roost. In this study we testedthe following predictions in order to determinewhich of the two factors – feeding potential or pred-ator avoidance – is more important in determiningthe choice of high water roost. If roost selection isdriven by the need to extend the feeding period, wewould predict that birds will continue to feed duringthe high water period. If predator avoidance is thedeterminant, we would predict that in the preferredsites there would be fewer predator visits andalarm flights, and that the birds will spend less timein vigilance activities.

We compared roosting sites in saltpans with thoselocated in the adjacent upper intertidal sedimentflats. In addition, we measured the use made bywaders of high water roosts from neap to spring tides,

in order to examine how tidal condition influencesroost selection by waders.

METHODS

Study area

This study was carried out in the Tagus Estuary, Portugal(38

°

45

′

N, 8

°

55

′

W) (Fig. 1), one of the most importantestuarine wetlands in Europe. Tidal amplitude rangesbetween 1 and 3.8 m, from neap to spring tides. In thePortuguese ordnance datum level, the annual mini-mum and maximum high water heights are 2.6 and4.2 m, respectively (referred to the hydrographic zero).

When the tides exceed 3.3 m, the high waterreaches the saltmarshes and the dykes that separate

Figure 1. Location of the study areas in the Tagus Estuary. The dotted line represents the limit of lowest neap high waters. The areabetween this line and the saltmarsh is not inundated in every tidal cycle.

90

S. Rosa

et al.


the estuary from farmland or built-up areas, leavingno space for waders on the sediment flats. Duringneap high waters (from 2.6 to 3.2 m), a band of sed-iment flats (of

c.

600 m in the lowest tides) remainsexposed even at high water (see Fig. 5). Some ofthese areas are used by waders during high water.Our study contrasted the most important supratidalwader roosts in the estuary (part of a large saltpancomplex) with three nearby areas delimited in theupper intertidal mudflats, where birds gather in neaphigh waters. We will refer to these areas as ‘saltpans’and ‘mudflat roosts’, respectively.

Number of waders at roost sites

The relationship between the height of the highwater and the number of waders using the two typesof roosts was examined by carrying out bird countswithin

±

30 min of peak high water simultaneouslyin the saltpans (in a 10-ha sector, where most of thewaders in the saltpan complex stay) and in threesectors in the mudflats roosts (with a total surfacearea of 40.1 ha; Fig. 1). Counts were carried outduring six or seven consecutive days, to cover thewhole tidal gradient (from neap to spring tides). Thisprocedure was repeated three times in February–March 2002. The three selected sectors (Fig. 1) rep-resent only a fraction of the total mudflat roost area.Consequently, the number of birds counted in thesesectors is lower than the total numbers in the salt-pans. A local regression technique was used toillustrate the relationship between tidal heightand wader abundances. We used the techniqueknown as lowess (locally weighted regression scat-terplot smoothing), which is a non-parametricregression method, where

y

-values are calculatedfrom a neighbourhood of

x

-values, whose weightsvary according to the distance to the central

x

-value(Venables & Ripley 2002).

Foraging and prey availability at roost sites

To determine the importance of feeding activity atthe roosts, we recorded the percentages of feedingbirds for all the species. We also estimated prey avail-ability in May 2002, using a total of 40 cores ofsediment, each having an area of 87 cm

2

, taken to adepth of 20 cm. We stratified our sampling by defin-ing five lines parallel to the coast. The first three lines(located at 200, 400 and 600 m from the coast)coincided with the mudflat roosts; the following

sampling lines (at 800 and 1000 m) were taken inthe adjacent lower area, lying below high water levelin neap tides. The superficial 5 cm of the sedimentcores were sieved using a 0.5-mm mesh size and theremaining 15 cm with a 1-mm mesh. The inverte-brates retained in the sieve were preserved in 70%ethanol, and later identified and counted.

In order to quantify the biomass available for birdswe measured the total length of the gastropod

Hydrobia ulvae

, the antero-posterior length of thebivalve

Scrobicularia plana

and the jaw length of theragworm

Hediste diversicolor

. These parameters werethen used to estimate ash-free-dry-weight (AFDW)from regression equations derived from local speci-mens (Moreira 1996, Santos

et al

. in press). Siphons oflarge

S. plana

are also consumed by waders (Moreira1995a), so we also calculated their AFDW, usingthe equations from Zwarts

et al

. (1994) and Moreira(1996). The AFDW of the remaining species, thecrustacean

Cyathura carinata

, the polychaete

Streblospio shrubsolii

and individuals of the ClassOligochaeta, were determined by loss-on-ignition at500

°

C for 2 h.In the saltpans most birds roosted on the com-

pacted dry and coarse substrate dykes, unsuitable fortypical prey, so we did not sample prey abundancethere. Only the invertebrates in the top 5-cm layer ofsediment in our cores were considered to be avail-able to birds (Bryant 1979, Kalejta 1993, Zwarts &Wanink 1993) and only

H. ulvae

smaller than 4 mmand the

S. plana

between 2 and 20 mm were con-sidered (Goss-Custard 1977, Worral 1984, Moreira1994, 1995b, Pérez-Hurtado

et al

. 1997, Dierschke

et al

. 1999a, 1999b, Scheiffarth 2001, Santos

et al

.in press). We also included

H. diversicolor

and

S. plana

from the full cores; the former because it ishighly mobile and makes extensive vertical move-ments, and the latter because the siphon thatextends to the surface is often consumed by birds(Moreira 1995a).

Risk of predation

We compared predator pressure on saltpans andmudflat roosts during high water using four para-meters: (i) number of raptors flying over the roosts,(ii) number of alarm flights caused by raptors, (iii)total number of alarm flights and (iv) proportion oftime spent in vigilance.

We measured the proportion of time spent invigilance by the commonest species, Dunlin

Calidrisalpina

and Grey Plover

Pluvialis squatarola

, using 4-min



91

focal observations, during which we recorded thenumber of scans and their durations. We consideredthat a bird was scanning when its head was raised andits beak was not touching the plumage (Roberts1995). These observations were carried out duringhigh water, in an area of

c.

10 ha in the saltpans inspring and neap tides and in 12 ha of the nearestmudflat roosting area in neap tides, during the win-ter and spring of 2002 and the spring of 2003.

RESULTS

Roost site characteristics and their use by waders

The great majority of the birds of most species pre-ferred mudflat roosts during tides up to 3.2 m high,i.e. about 3 days before and after the neap tides(Fig. 2). When tides are higher than 3.3 m, the water

Figure 2. Variation in wader abundance in saltpans and mudflat roosts in relation to tide heights. Below 3.2–3.3 m the abundance ofmost species is higher on mudflat roosts; above 3.2–3.3 m these roosts are submerged and wader abundance increases in saltpans.The shaded area represents the tidal heights when both types of roost are available. Closed dots: number of birds roosting in saltpans;open dots: number of birds roosting in mudflats; lines: overall tendency for saltpans (lowess lines).

92

S. Rosa

et al.


reaches the margins of the saltmarsh, birds are forcedto move to the saltpans and these mudflats becomesubmerged (see Fig. 5). Only Ringed Plover

Charadriushiaticula

and Kentish Plover

Charadrius alexandri-nus

did not show this trend (Fig. 2).Prey available for waders in the mudflat roosts,

which remained exposed during some high waters,was very low (Table 1). In fact, with the exceptionof

H. diversicolor

, AFDW of prey in these areas wasseveral times lower than that on the adjacent lowermudflats, inundated in all daily tidal cycles. Apart

from a few areas that had small numbers of insects,the soil of the dykes separating the tanks of thesaltpans had no prey. All the invertebrate speciesreported (Table 1) are potential prey for the studyspecies.

Foraging at roost sites

The overall proportion of waders feeding in theroosting areas was very low in both types of roost,but higher at the mudflats (13%) than at the saltpans

Figure 2. Continued.

Table 1. Biomass of invertebrate prey harvestable by waders (mean ± se AFDW (mg/m2)) in mudflat roosts and adjacent lower mudflats(non-consumed sizes were not included). Note that most prey types are far less abundant in the upper mudflats, which are not inundatedon all tidal cycles, than in the immediately adjacent lower mudflats. ***P < 0.001; ns, non-significant.

Species Mudflat roosts Lower mudflats Mann–Whitney U-test

Siphons of Scrobicularia plana 1.9 ± 1.9 113.2 ± 1.6 888.6***Hydrobia ulvae 3.4 ± 1.3 21.4 ± 4.7 1026.5***Hediste diversicolor 579.2 ± 186.9 759 ± 281.4 1652.5 nsStreblospio shrubsolii 1.3 ± 0.5 31.7 ± 4.3 107***Oligochaeta 1.4 ± 0.6 8.74 ± 2.5 709***Cyathura carinata 0 22.7 ± 9.2 1440 ns



93

(1%). However, there was a substantial variationamong species; although most species fed very rarely,a high proportion of Common Redshank

Tringa totanus

continued to feed at high water, especially whenusing the mudflat roosts (Fig. 3). When feeding inthe dykes that separate the tanks of the saltpans,Common Redshanks were probably eating insectsor other terrestrial invertebrates. Ruddy Turnstone

Arenaria interpres

did not feed in the mudflat roostsat high water. Kentish Plover and Ringed Plover didnot occur at all in mudflat roosts, despite a reasonableabundance of their preferred prey (

H. diversicolor

).The water in the tanks of the saltpans was too

deep for most species to reach the bottom, althoughthe species with longer legs (Black-tailed Godwit

Limosa limosa

and Pied Avocet

Recurvirostra avosetta

)

occasionally fed in the tanks. Pied Avocet wouldmost often feed while swimming.

Risk of predation

All the results indicate that predator pressure wassubstantially higher in the saltpans than in themudflat roosts (Table 2, Fig. 4). During the 54 h ofobservation, nine species of raptors flew over salt-pans and only five were observed over the mudflatroosts. During high water there were more individualraptors flying over the saltpans than over the mudflats(Mann–Whitney

U

-test:

U

= 196,

P

= 0.01, Table 2).On average, there were three times more raptorsflying above saltpans than mudflats.

The higher number of raptors in the saltpansresulted in more frequent flock alarm flights and

Figure 3. Proportion of waders feeding (mean ± se) in highwater in saltpans and mudflat roosts. Most birds did not feed athigh water, but mudflats were more important for foraging thansaltpans. Turnstone = Ruddy Turnstone, K. Plover = KentishPlover, R. Plover = Ringed Plover, Bar-t. Godwit. = Bar-tailedGodwit, Black-t. Godwit = Black-tailed Godwit, G. Plover = GreyPlover, Avocet = Pied Avocet, Redshank = Common Redshank.

Table 2. Number of raptors/h that flew over saltpans and mudflatroosts during high waters (35 h of observation in saltpans and19 h in mudflat roosts)

Species SaltpansMudflat roosts

Eurasian Marsh Harrier Circus aeruginosus 1.11 0.47Booted Eagle Hieraaetus pennatus 0.54 0Peregrine Falcon Falco peregrinus 0.17 0.05Short-toed Eagle Circaetus gallicus 0.09 0.05Common Buzzard Buteo buteo 0.09 0.05Hen Harrier Circus cyaneus 0.06 0Montagu’s Harrier Circus pygargus 0.03 0.05Common Kestrel Falco tinnunculus 0.03 0Osprey Pandion haliaetus 0.03 0Total 2.14 0.68

Figure 4. Disturbance indexes calculated for saltpans and mudflat roosts. (a) Number of raptors/h above roosts; (b) number of alarmflights/h (for all the birds present in the roosts); and (c) proportion of time spent in vigilance. Note that all disturbance measures werehigher in saltpans. Data are expressed as medians ± interquartile range.

94

S. Rosa

et al.


higher levels of vigilance. Both the total number offlock alarm flights and the number of alarm flightscaused by raptors were several times higher in salt-pans than in mudflat roosts (Mann–Whitney

U

-tests:

U

= 69.5,

P

< 0.001, and

U

= 194,

P

= 0.007, respec-tively). Dunlins spent 3.5 times more time in vigilancein the saltpans than in the mudflat roosts (Mann–Whitney

U

-test:

U

= 1093.5,

P

= 0.012), and thedifference was even greater for Grey Plovers, whichspent eight times more time in vigilance in thesaltpans than in the mudflat roosts (Mann–Whitney

U-test: U = 1471.5, P < 0.001).

DISCUSSION

Waders prefer to roost in upper mudflats than in saltpans

When both types of roost are available, most wadersclearly prefer to roost in mudflats than in saltpans.Kentish Plover and Ringed Plover were the only spe-cies that did not show this tendency, and simply didnot use any of the monitored mudflat roosts. Theyalways roosted in the saltpans, a choice that may berelated to their anti-predator strategy. In fact, theytended to be less reactive to the presence of predators,and often relied on immobility and their high degreeof crypsis to escape the predators’ attention. Sub-strate preferences may also play a role in roost-siteselection. In fact, both species are known to special-ize on fairly hard sandy substrates (Moreira 1995a,Granadeiro et al. 2004), and so they may not feelcomfortable roosting in the soft and muddy tidal flatroosts. Saltpans are mostly used when mudflat roostsbecome unavailable, which occurs at tidal heightsgreater than 3.3 m. Having established the clearpreference for roosts in mudflats, it becomes impor-tant to understand the reasons for the preference, asthis may shed light on the factors that make a poten-tial roosting site attractive to waders, informationthat is critical for the correct management of highwater roosts. This issue is of particular importancegiven that most species tend to show a high within-and between-year roost-site fidelity (e.g. Rehfischet al. 1996).

Do waders roost on upper mudflats to feed at high water?

We predicted that, if the need to feed during highwater drives the choice of roost, we should observea high proportion of the birds foraging while in their

preferred high water roosts, i.e. on the mudflats.However, with the exception of Common Redshank(and, to a lesser extent, Bar-tailed Godwit Limosalapponica), the number of birds feeding in mudflatroosts was very low. This suggests that bird choicebetween roosting sites was not driven by substantialforaging advantages in the mudflats.

These results are consistent with observations ofDunlins in the Yukon–Kuskokwim Delta, Alaska,where most of the birds slept, rather than feeding, inhigh water roosts (Handel & Gill 2004). In contrastto our observations, Kalejta (1992) and Masero andPérez-Hurtado (2001) found large numbers of win-tering waders feeding in their high water roosts, andsuggest that this may be explained by a need to com-plement the low tide foraging bout. One possibleexplanation for the behavioural difference is that thebirds that they studied had less profitable low tidefeeding grounds, or were subjected to energeticallymore demanding climatic conditions. An alternativeexplanation is that they continued foraging duringthis period because food was widely available in theroosts. In our study area food was quite scarce in bothtypes of roosts so that this option was unavailable.

Prey densities observed in the upper mudflat roosts,which remained exposed during the high water onconsecutive tidal cycles, were much lower than inthe adjacent mudflats, flooded on all tidal cycles.This difference in the immersion regimes seems tojustify the contrast in prey abundance. For example,one of the main prey items of many waders in ourstudy area, S. plana, was rare in the intertidal mud-flats used as roosts, probably because it cannot with-stand long periods of desiccation (Hughes 1970). Inmany estuaries, mudflat areas that have the potentialto serve as high water roosts for waders, because theyremain exposed during high water on consecutivetidal cycles (Fig. 5), are also likely to have prey den-sities depressed by the same factor.

The joint interpretation of the information nowavailable suggests that food availability was verylow and therefore that maintaining feeding activitywould be energetically uneconomic; we must thereforeseek alternative explanations for roosting preferences.

An alternative explanation for the preference ofmudflat over saltpan roosts might be the shorterdistance to feeding areas. However, the distancebetween the mudflat roosts and the saltpans wasvery small (c. 300 m), compared with the regularmovements of the birds to and from the feedingareas, and we often observed waders flying severalkilometres to roost at alternative saltpans, located in


High water roost selection by waders 95

the south of the estuary. Thus, the distance betweenmudflats and saltpan roosts appears to be irrelevantin relation to movements usually undertaken bywaders in estuaries (Piersma et al. 1993, Rehfischet al. 1996).

Do waders roost in upper mudflats to minimize the risk of predation?

All the predictions made under the hypothesis thatpredator avoidance is the determinant in the choiceof high water roost were corroborated by our results.At the sites preferred by waders for roosting, the fre-quency both of raptor visits and of alarm flights waslower, and birds spent less time in vigilance than atother roosting sites. Both vigilance and disturbanceevents can be energetically costly, because they forcebirds to perform more alarm flights and so suffer areduction in resting time and incur further energeticcosts. Thus, waders prefer to select roosting siteswhere they are less disturbed, and where they canminimize vigilance behaviour. In fact, it has beendemonstrated that disturbance, whether natural or

anthropogenic, may result in roost abandonment(Colwell et al. 2003).

Waders may be more vigilant in the saltpans thanon mudflats not only because of the frequent pres-ence of predators but also because of lower visibility.With lower visibility birds have less time to react toa predator attack (Whitfield 1985, Cresswell 1994,Lima 1995). The importance of visibility is supportedby the fact that on the mudflats studied waderstended to roost near the water line, where visibility isbetter than near the edge of the saltmarsh, althoughthis behaviour could also be explained by otherfactors.

All the disturbance indices measured in the studywere lower for mudflat roosts than for saltpans.These data seem to support the hypothesis thatwaders prefer to roost on mudflats because the pre-dation risk there is lower.

Relevance for management

Our results highlight the great importance of upperintertidal sediment flats as high water roosting sites

Figure 5. (a) Representation of mudflat roosts and intertidal area available in relation to tidal height. (b) Variation of total exposure time (hours)and maximum number of consecutive hours exposed with tide height. Mudflat roosts remain exposed on several consecutive days.

96 S. Rosa et al.


for waders on the Tagus Estuary. Although in generalless obvious to observers than other more easilyaccessible supratidal roosts (such as saltpans orfarmland), these areas sustained the great majority ofthe roosting birds of the Tagus Estuary during abouthalf of the tidal cycles. Birds roosting on mudflatswere under less predation pressure and probablyspent less energy during the roosting period. It islikely that this situation is similar to that found atmany other estuaries where mudflat roosts are avail-able. Regrettably, upper intertidal flats tend to beunder great pressure for land reclamation, as theyare usually close to the shore and are easier to drainbecause they are shallower. The disappearance ofthese preferred roosting areas reduces the number ofdays in which waders are able to stay in high-qualityroosts, and may result in lower winter survival.

The finding that disturbance by predators maydetermine the choice of roost underlines the impor-tance of minimizing all sources of disturbance atroosts. In our study area, human disturbance was nothigh at the saltpans because the saltpans had beenabandoned and access to them was quite difficult.The mudflat roosts are particularly undisturbedbecause of the maintenance of a broad edge coveredby saltmarsh. However, on other estuaries distur-bance by humans, pets, or vehicles is likely to havean effect similar to that of potential predators, andultimately force birds to change their selection ofroosting site (Pfister et al. 1992, Burton et al. 1996).

Several authors have pointed out the importanceof saltpans for the maintenance of wader popula-tions in many estuaries (Rufino et al. 1984, Maseroet al. 2000). Our finding of a preference by wadersfor roosting in upper intertidal flats in no way mini-mizes the importance of roosts in saltpans or othersupratidal sites. Our results demonstrate that in theTagus Estuary, tidal flat roosts become unavailablearound spring tides and waders must roost in salt-pans. This is likely to occur in many other estuaries,but the availability of this type of supratidal habitatis rapidly declining in many of them. Consequently,it is critical to protect saltpans and other types ofsupratidal roosts, and to minimize their anthropo-genic disturbance.

We thank R.C. Martins, M.P. Dias, M. Lecoq, C.D.Santos and several volunteers for their help in the field.M.P. Dias, C.D. Santos and P. Segurado provided commentson an early version of the manuscript. P. Segurado andM.P. Dias provided the drawing and the map, respectively.This study was supported by Fundação para a Ciência eTecnologia through grants SFRH/BD/6221/2001 to S.R.

and SFRH/BPD/11544/2002 to J.P.G., and research projectSat-Tagis (Contract PDCTM/C/MAR/15256/1999), bothco-financed by the EU ERDF.

REFERENCESBryant, D.M. 1979. Effects of prey density and site character

on estuary usage by overwintering waders (Charadrii).Est. Coast. Mar. Sci. 9: 369–384.

Burton, N.H.K., Evans, P.R. & Robinson, M.A. 1996. Effectson shorebird numbers of disturbance, the loss of a roost siteand its replacement by an artificial island at Hartlepool,Cleveland. Biol. Conserv. 77: 193–201.

Colwell, M.A., Danufsky, T., Fox, N.W., Roth, J.E. & Concklin, J.R.2003. Variation in shorebird use of diurnal, high tide roosts:how consistently are roosts used? Waterbirds 26: 484–493.

Cresswell, W. 1994. Age-dependent choice of redshank (Tringatotanus) feeding location: profitability or risk? J. Anim. Ecol.63: 589–600.

Dierschke, V., Kube, J., Probst, S. & Brenning, U. 1999a.Feeding ecology of Dunlins Calidris alpina staging in thesouthern Baltic Sea, 1. Habitat use and food selection.J. Sea Res. 42: 49–64.

Dierschke, V., Kube, J. & Rippe, H. 1999b. Feeding ecology ofDunlins Calidris alpina staging in the southern Baltic Sea, 2.Spatial and temporal variations in the harvestable fraction oftheir favourite prey Hediste diversicolor. J. Sea Res. 42: 65–82.

Goss-Custard, J.D. 1977. Predator responses and prey mortal-ity in redshank, Tringa totanus (L.) and a preferred prey,Corophium volutator (Pallas). J. Anim. Ecol. 46: 21–35.

Granadeiro, J.P., Andrade, J. & Palmeirim, J.M. 2004. Model-ling the distribution of shorebirds in estuarine areas usinggeneralized additive models. J. Sea Res. 52: 227–240.

Handel, C.M. & Gill, R.E. 2004. Roosting behavior of premigra-tory Dunlins (Calidris alpina). Auk 109: 57–72.

Hughes, R.N. 1970. Population dynamics of the bivalve Scrobic-ularia plana (Da Costa) on an intertidal mud-flat in northWales. J. Anim. Ecol. 39: 333–356.

Kalejta, B. 1992. Time budgets and predatory impact of wadersat the Berg River Estuary, South Africa. Ardea 80: 327–342.

Kalejta, B. 1993. Intense predation cannot always be detectedexperimentally: a case study of shorebird predation on nereidpolychaetes in South Africa. Neth. J. Sea Res. 314: 385–393.

Lima, S.L. 1995. Collective detection of predatory attack bysocial foragers: fraught with ambiguity? Anim. Behav. 50:1097–1108.

Luís, A., Goss-Custard, J.D. & Moreira, M.H. 2001. A methodfor assessing the quality of roosts used by waders during hightide. Wader Study Group Bull. 96: 71–73.

Luís, A., Goss-Custard, J.D. & Moreira, M.H. 2002. The feed-ing strategy of the Dunlin (Calidris alpina L.) in artificial andnon-artificial habitats at Ria de Aveiro, Portugal. Hydrobiologia475/476: 335–343.

Masero, J.A. & Pérez-Hurtado, A. 2001. Importance of thesupratidal habitats for maintaining overwintering shorebirdpopulations: how Redshanks use tidal mudflats and adjacentsaltworks in Southern Europe. Condor 103: 21–30.

Masero, J.A., Pérez-Hurtado, A., Castro, M. & Arroyo, G.M.2000. Complementary use of intertidal mudflats and adjacentsalinas by foraging waders. Ardea 88: 177–191.


High water roost selection by waders 97

Moreira, F. 1994. Diet, prey-size and intake rates of Black-tailedGodwits Limosa limosa feeding on mudflats. Ibis 136: 349–355.

Moreira, F. 1995a. A utilização das zonas entre-marés doestuário do Tejo por aves aquáticas e suas implicações paraos fluxos de energia na teia trófica estuarina. PhD thesis,University of Lisbon.

Moreira, F. 1995b. The winter feeding ecology of Avocets Recur-virostra avosetta on intertidal areas. I. Feeding strategies.Ibis 137: 92–98.

Moreira, F. 1996. Diet and feeding behaviour of Grey PloversPluvialis squatarola and Redshanks Tringa totanus in asouthern European estuary. Ardeola 43: 145–156.

Pérez-Hurtado, A., Goss-Custard, J.D. & Garcia, F. 1997. Thediet of wintering waders in Cádiz Bay, southwest Spain. BirdStudy 44: 45–52.

Pfister, C., Harington, B.A. & Lavine, M. 1992. The impact ofhuman disturbance on shorebirds at a migration stagingarea. Biol. Conserv. 60: 115–126.

Piersma, T., Hoekstra, R., Dekinga, A., Koolhaas, A., Wolf, P.,Battley, P. & Wiersma, P. 1993. Scale and intensity of inter-tidal habitat use by Knots Calidris canutus in the westernWadden Sea in relation to food, friends and foes. Neth. J. SeaRes. 31: 331–357.

Rehfisch, M.M., Clark, N.A., Langston, R.H.W. & Greenwood, J.J.D.1996. A guide to the provision of refuges for waders: an ana-lysis of 30 years of ringing data from the Wash, England.J. Appl. Ecol. 33: 673–687.

Roberts, G. 1995. A real-time response of vigilance behaviourto changes in group size. Anim. Behav. 50: 1371–1374.

Rogers, D.I. 2003. High-tide roost choice by coastal waders.Wader Study Group Bull. 100: 73–79.

Rufino, R., Araújo, A., Pina, J.P. & Miranda, P.S. 1984. Theuse of salinas by waders in the Algarve, South Portugal.Wader Study Group Bull. 42: 41–42.

Santos, C.D., Granadeiro, J.P. & Palmeirim, J.M. in press.Feeding ecology of Dunlin Calidris alpina in a southern Euro-pean estuary. Ardeola in press.

Scheiffarth, G. 2001. The diet of Bar-tailed Godwits Limosalapponica in the Wadden Sea: combining visual observationsand faeces analysis. Ardea 89: 481–494.

Venables, W.N. & Ripley, B.D. 2002. Modern Applied Statisticswith S, 4th edn. New York: Springer-Verlag.

Whitfield, D.P. 1985. Raptor predation on wintering waders insoutheast Scotland. Ibis 127: 544–588.

Worral, D.H. 1984. Diet of Dunlin Calidris alpina in the SevernEstuary. Bird Study 31: 203–212.

Zwarts, L., Blomert, A.M., Spaak, T. & Vries, B. 1994. Feedingradius, burying depth and siphon size of Macoma balthicaand Scrobicularia plana. J. Exp. Mar. Biol. Ecol. 183: 193–212.

Zwarts, L. & Wanink, J.H. 1993. How the food supply harvest-able by waders in the Wadden Sea depends on the variationin energy density, body weight biomass, burying depth andbehaviour of tidal-flat invertebrates. Neth. J. Sea Res. 31:441–476.

Received 13 January 2005; revision accepted 22 September 2005.

high water roost selection by waders: maximizing feeding opportunities or avoiding predation?

Documents