Rendón et al. 1 1 Assessing sex-related chick provisioning in greater flamingo 2 Phoenicopterus roseus parents using capture-recapture 3 models 4 5 6 Miguel A. Rendón* a, Araceli Garrido b, Manuel Rendón-Martos c, José M. 7 Ramírez b and Juan A. Amat a 8 9 a Department of Wetland Ecology, Estación Biológica de Doñana (EBD-CSIC), calle 10 Américo Vespucio s/n, 41092 Sevilla, Spain 11 b 12 Ambiente y Agua de Andalucía, Consejería de Agricultura, Pesca y Medio Ambiente, 13 Junta de Andalucía, Parque Comercial Málaga Nostrum, Edificio Galia Center, calle 14 Jaén 9-3ª, 29004 Málaga, Spain 15 c 16 Medio Ambiente, Junta de Andalucía, Apartado 1, 29520 Fuente de Piedra, Spain Programa de Actuaciones de Aves Acuáticas en Andalucía, Agencia de Medio Reserva Natural Laguna de Fuente de Piedra, Consejería de Agricultura, Pesca y 17 * 18 Corresponding author: ma_rendon@ebd.csic.es 19 20 Word count: 8596 21 22 23 Running head: Provisioning movements in greater flamingos Rendón et al. 2 24 Summary 25 26 1. In sexually dimorphic species, the parental effort of the smaller sex may be 27 reduced due to competitive exclusion in the feeding areas by the larger sex, 28 or physiological constraints. However, to determine gender effects on 29 provisioning patterns other intrinsic and extrinsic factors affecting parental 30 effort should be accounted for. 31 32 2. Greater flamingos (Phoenicopterus roseus) exhibit sexual size dimorphism. 33 In Fuente de Piedra colony the lake dries out almost completely during the 34 breeding season and both parents commute between breeding and foraging 35 sites distant >130 km during the chick rearing period. 36 37 3. Applying multistate capture-recapture models to daily observations of 38 marked parents, we determined the effects of sex, and their interactions with 39 other intrinsic and extrinsic factors, on the probability of chick desertion and 40 sojourn in the colony and feeding areas. Moreover, using stable isotopes in 41 the secretions that parents produce to feed their chicks we evaluated sex- 42 specific use of wetlands. 43 44 4. The probability of chick attendance (complementary to chick desertion) was 45 >0.98. Chick desertion was independent of parental sex, but decreased with 46 parental age. Females stayed in the feeding areas for shorter periods (mean: 47 7.5 [95% CI: 9.4–6.0] days) than males (9.2 [11.8–7.3] days). Isotopic 48 signatures of secretions did not show sex differences in δ13C, but males’ Rendón et al. 3 49 secretions were enriched in δ15N, suggesting they fed on prey of higher 50 trophic levels than females. Both parents spent approximately one day in the 51 colony, but females prolonged their mean stay when the lake dried out. 52 Females also allocated more time to foraging in the flooded areas remaining 53 in the colony, likely because they were energetically more stressed than 54 males. 55 56 5. The results indicate that sex-specific provisioning behaviour in greater 57 flamingo is related to differential effects of both intrinsic and extrinsic 58 factors. Males seem forage less efficiently than females, whereas females 59 body condition seem to be lower after fed the chick. Our methodology may 60 be extended to species that feed on distant food sources, and that do not visit 61 their offspring daily, to elucidate patterns of chick provisioning behaviour. 62 63 64 Key-words: Foraging ecology, multistate capture-recapture models, parental care, 65 sexual size dimorphism, sexual segregation, stable isotopes, waterbirds. 66 67 68 69 70 71 72 73 Rendón et al. 4 74 Introduction 75 Chick rearing entails high energetic expenditures (Drent & Daan 1980; Monaghan & 76 Nager 1997). The time and energy that adults allocate to obtain resources for their 77 offspring are limited by both adult and offspring requirements (Ydenberg et al. 1992). 78 Therefore, studying provisioning patterns is important for the understanding of the 79 factors that limit parental investment and variations in the life-history of species (Sæther 80 1994). 81 The frequency of parental provisioning is one of the most important factors affecting 82 chick body condition (Gray et al. 2005). The patterns of provisioning may be affected 83 by the interaction of both extrinsic (e.g. resources and time) and intrinsic (e.g. body 84 condition, age) factors (Tinbergen & Verhulst 2000; Weimerskirch & Lys 2000; 85 Zimmer et al. 2011). Adult sex, for instance, may restrict parental effort due to 86 differences in foraging efficiency or dominance in dimorphic species (González-Solís et 87 al. 2000; Lewis et al. 2005; but see Lewis et al. 2002), as well as physiological 88 restrictions (e.g. nutrient limitation, Nisbet 1997) or sex differences in energy efficiency 89 (Barbraud et al. 1999). In sexually dimorphic species, pair members may reduce sexual 90 competition by using different environments (González-Solís et al. 2000; Quillfeldt et 91 al. 2011) or feeding on different prey (Forero et al. 2002; Quillfeldt et al. 2011). These 92 strategies may affect foraging patterns (Lewis et al. 2005), and consequently 93 provisioning behaviour (Weimerskirch & Lys 2000; Kato et al. 2001). In addition, sex 94 differences in parental investment at different breeding stages (egg production, 95 incubation, etc.) together with other intrinsic (Nol & Smith 1987; McGraw et al. 2001) 96 and extrinsic (Dawson & Bortolotti 2003; Hamer et al. 2005) conditioning factors may 97 lead to male and female parents adopting different decisions. Rendón et al. 5 98 Besides parental sex, provisioning is influenced by age at breeding (Nol & Smith 1987; 99 Forest & Gaston 1996; González-Solís et al. 2004), probably due to older individuals 100 being more experienced/efficient in acquiring resources. Also, as chick requirements 101 increase during growth, so should increase parental effort. In long-lived birds it has 102 been shown both flexible (Weimerskirch et al. 1997a; Hamer et al. 2005) and fixed 103 levels of parental effort to satisfy chicks’ requirements (Sæther et al. 1993; Navarro & 104 González-Solís 2007). These two strategies may not be mutually exclusive, however, as 105 adults may switch between them depending on food availability (Erikstad et al. 1998; 106 Granadeiro et al. 1998; Weimerskirch et al. 2001). 107 Here we investigate the effect of sex on offspring desertion and parental provisioning 108 strategies in the greater flamingo Phoenicopterus roseus. The greater flamingo is a 109 colonial nester, long-lived and sexually dimorphic bird (males are 20% larger than 110 females), which lays a single egg that is incubated by both parents (Johnson & Cézilly 111 2007). The chicks are fed with a secretion produced by their parents (Lang 1963; 112 Ziswiler & Farner 1972). Once chicks are in crèches, the parents move to forage in 113 wetlands distant up to 400 km from the breeding site, remaining several days foraging 114 in them (Amat et al. 2005). Male greater flamingos exhibit a greater effort in nest 115 defence prior to egg laying (Johnson & Cézilly 2007), as well as in incubation (Rendón- 116 Martos et al. 2000), but females are more likely to desert the nest (Cézilly 1993). 117 During incubation, nest attendance also depends on flooding around the colonies 118 (Rendón-Martos 1996), and parental age (Johnson & Cézilly 2007; Schmaltz et al. 119 2011). In the few days after hatching, parental care equalise between sexes (Rendón- 120 Martos et al. 2000). However, both chick provisioning and the factors that control it 121 have been little studied in flamingos. Previous studies have determined that chick 122 survival do not depend on parental age (Johnson & Cézilly 2007; Schmaltz et al. 2011), Rendón et al. 6 123 but the effect of parental sex on provisioning patterns remains unknown. Furthermore, 124 as colonial birds are central place foragers during breeding, their provisioning tactics are 125 expected to vary with environmental conditions (Tremblay & Cherel 2005). This may 126 be especially critical when foraging on spatially and/or temporally unpredictable 127 resources, and the food sources are distant from the colony, as in the case of greater 128 flamingos (Amat et al. 2005). 129 We applied multistate capture-recapture models to resightings of individually marked 130 adults attending a colony to analyse the effect of the interaction of adult sex with other 131 intrinsic (laying date, parent age, chick age) and extrinsic (observation date) factors on 132 chick provisioning. The analysis of resighting data from marked individuals by means 133 of capture-recapture methods may be an alternative to the study of animal movements 134 (Kendall & Nichols 2004). In particular, the use of multistate models with unobservable 135 states (Kendall & Nichols 2002; Fujiwara & Caswell 2002; Schaub et al. 2004) allowed 136 determining periodicities of reproductive events in species with temporal emigrations. 137 Though these models have been mainly used to study interannual demographic patterns, 138 multistate models have also been applied to document variations in reproductive effort 139 within a breeding season (Schmaltz et al. 2011). 140 A comprehensive analysis of parental investment takes body condition changes besides 141 reproductive effort by the parents into account (Weimerskirch et al. 1997b; Granadeiro 142 et al. 1998; Weimerskirch & Lys 2000). For this reason we also studied the effect of 143 intrinsic and extrinsic factors on the foraging patterns of parents. In addition, we 144 investigated whether male and female parents used different habitats and/or prey during 145 chick provisioning, by analysing stable isotopes (δ13C and δ15N) in secretions of parents 146 (i.e. chick food). 147 Rendón et al. 7 148 149 Materials and methods 150 151 STUDY SITES AND BREEDING PHENOLOGY 152 153 The study was conducted at Fuente de Piedra lake (FP; 37º07’N, 4º46’W; Fig. 1), a 154 seasonal wetland in which the size of breeding colonies of greater flamingos varies 155 annually depending on rainfall (Rendón-Martos 1996). The main foraging areas (FA) of 156 adults during chick provisioning are in the wetlands of Doñana National Park and 157 nearby zones, mainly the seasonal marshes (Guadalquivir marshes, GM), and Veta la 158 Palma fish-farm (VP) (Amat et al. 2005). 159 Laying dates in 2001 spanned from late February until mid-May, and colony size was 160 17700 nests. There were three waves of breeders (cohorts: c) (Fig. 2A): 9th (fourth week 161 of February) to 11th weeks (c1), 13th-15th weeks (c2) and 17th-19th weeks (c3). The 162 number of flamingos in GM increased in line with chick hatching in FP, whereas in VP 163 remained more or less stable. However, as GM dried out, there was a greater use of VP, 164 which coincided with the maximum number of chicks at FP (Fig. 2B). 165 166 167 OBSERVATIONS OF BREEDING ADULTS AND CHICKS 168 169 Since 1977 in Camargue (France), and since 1986 in Fuente de Piedra, about 10% of 170 greater flamingo chicks have been ringed using individually-coded leg-rings (Johnson Rendón et al. 8 171 & Cézilly 2007). The breeding behaviour of marked adults was recorded using a 172 spotting scope located 250 m from the breeding colony. The sex of adults was 173 determined based on their relative size and sex-specific behaviour from multiple 174 observations. This sexing procedure was very accurate, as 96% of adults for which sex 175 was determined using molecular methods were correctly classified when sex was 176 assigned using phenotypic and behavioural characteristics (Appendix S1). The age of 177 adults (Aa) was known, as they were individually marked as chicks. To investigate 178 whether seasonal trends in joining the colony by breeding adults were predicted by their 179 age (c), we used a linear model that included the interaction with the sex of adults 180 (c*sex). 181 When a parent was attending a chick, the age of the latter was allocated as a function of 182 both its size relative to the adult, and the development of its plumage. When a parent 183 was only recorded incubating, we assumed that it was in mid-incubation (incubation 184 lasts 28-30 days; Johnson & Cézilly 2007). Thus, we estimated for each parent both 185 laying and hatching dates. Laying dates were grouped according to each one of the three 186 cohorts (c1, c2 and c3; see above). Chick age was estimated based on four categories 187 (ac4), depending on their ability to forage by themselves and to fly (Allen 1956; Zweers 188 et al. 1995): chicks that (1) cannot fly and are provisioned by their parents (1-6 weeks), 189 (2) may forage by themselves but cannot fly (7-12 weeks), (3) may perform short flights 190 and forage by themselves (13-16 weeks), and (4) are able to fly and forage by 191 themselves (>16 weeks) 192 193 194 PRESENCE OF ADULTS IN THE BREEDING COLONY Rendón et al. 9 195 196 After hatching, the presence of individually marked adults in FP was recorded daily 197 during two periods (p): p1, 30 May–23 June (25 days), and p2, 25 July–19 August (26 198 days). Observations were carried out from 08:00 h to 12:00 and from 18:00 h to 21:00 h 199 (GMT+2h), and average time of observation was 4.3 ± (SD) 1.8 hours/day, though 200 resighting effort (E) varied between days (range 2-8 h). We chose these two time- 201 periods in order to avoid the hottest hours, when individuals were resting far from the 202 lake shore (Rendón-Martos 1996). We carried out observations at the mouth of a stream 203 at which the adults congregate during diurnal hours (Fig. 1). A total of 1934 204 individually marked greater flamingos were identified, of which 367 (188 males and 205 179 females) were selected because they were recorded incubating or attending chicks. 206 207 208 PROBABILITIES OF CHICK DESERTION AND COMMUTING 209 210 The analysis of the probability of chick desertion and commuting between FP and FA 211 was considered as a case of temporary emigration, for which we used multistate 212 capture-recapture models with unobservable states (Fujiwara & Caswell 2002; Schaub 213 et al. 2004). Multistate capture-recapture models include three kinds of parameters: 214 recapture probability Pt, survival rate St, and conditional transition probability ψt 215 (Nichols et al.1994). When emigration processes dominate on mortality, St may be 216 interpreted as probability of leaving the study area (Pradel et al. 1997). Survival 217 probability in greater flamingos is >0.9 (Cézilly et al. 1996; Tavecchia et al. 2001), 97% 218 of the individuals in this study were re-sighted in subsequent years after the study, Rendón et al. 10 219 therefore, St was considered as the probability of chick attendance, which is 220 complementary to the probability of desertion. 221 From the transition matrices both the probability of an individual commuting from FP to 222 FA (ψtFPFA) and its probability of returning from FA to FP (ψtFAFP) were calculated 223 (see Appendix S2 for model description). For an equally spaced discrete-time Markov 224 chain, the mean residence or sojourn time ( ti ) for the chain in a state i is the length of 225 time it stays there during a single visit (Guttorp 1995). As the expected sojourn time in 226 stage i is geometrically distributed with parameter 1-ψi→i (probability of leaving the 227 estate), we transformed transition probabilities in sojourn time for FP and FA, using the 228 expressions tFP =1/(1-ψFP→FP) and tFA =1/(1-ψFA→FA), respectively. 229 In order to estimate parameters values recapture probabilities for FA were fixed to 0, 230 and survival probabilities were constrained to be the same for FP and FA (Kendall & 231 Nichols 2002). Thus, the starting model considered the additive effects of sex, p and E 232 on the probability of recapture at FP (Psex+p+E), the effect of the interaction of sex with 233 ac4, Aa, c and p on the probability of residence (Ssex*(ac4+Aa+c+p)), and the effect of the 234 interaction of sex with age of chick (excluding the category >16 weeks old: ac3), Aa, c 235 and p on the probability of commuting between FP and FA (ψFPFAsex*(ac3+Aa+c+p)) and 236 vice versa (ψFAFPsex*(ac3+Aa+c+p)). Given the chicks >16 weeks old are no longer fed by 237 their parents (Rendón-Martos et al. 2000), its effect on commuting probability was not 238 included in the models. Capture-recapture models were analysed using MARK (White 239 & Burnham 1999). 240 Because no formal test exists to check the goodness-of-fit (GOF) in multistate models 241 with unobservable states, the data were fitted to a Cormack–Jolly–Seber model 242 St*sexPt*sex (Lebreton et al. 1992). Selection of more parsimonious models was done Rendón et al. 11 243 using Akaike’s information criterium corrected for finite sample size (AICc). When the 244 difference in AICc values among several models was <2, the model with the lowest 245 number of significant parameters was retained as the final model. In Results only a 246 selection of models are shown. All generated models are presented in Table S1. 247 248 249 FORAGING BEHAVIOUR 250 251 Abdominal profile index (API), a semi-quantitative index of abdominal roundness, was 252 used as a surrogate of the time that adults allocated to forage. In the greater flamingo 253 there are considerable variations in the abdominal roundness, not only related to body 254 fat stores, but also to the time devoted to drinking and feeding. APIs were estimated 255 based on six categories (Fig. 5 in Rendón et al. 2009). APIs of individually marked 256 individuals (654 observations from 286 individuals) were recorded both at FP and VP at 257 the same time period the continuous observations were carried out at FP colony. Using 258 mixed ordinal logistic models API was related to the interaction of sex with ac3, Aa, p 259 and locality of observation (l) as fixed factors (sex*[ac3+Aa+p+l]), and bird identity and 260 observer as random factors. Analyses were conducted with R package ordinal2 261 (Christensen 2010). Generated models are presented in Table S2. 262 263 264 265 STABLE ISOTOPES IN SECRETIONS Rendón et al. 12 266 In 2006 and 2007, parental secretions were collected from chicks ringed at FP (n=100 267 and n=118 chicks, respectively). By softly massaging the crops, the chicks were induced 268 to regurgitate their content (i.e. the secreted food by its parent). Samples were then kept 269 in vials at 4ºC during transportation and until analysis. Blood samples (1 ml) from a 270 tarsal vein were taken and stored in 70% ethanol for sex determination of chicks. Cells 271 extracted from crop content were used for molecular sex determination. As these cells 272 may originate either from the parent that fed the chick or from the chick itself, parental 273 sex was assigned only when molecular sexing based on secretion cells were different 274 from chick sex based on blood sample. This resulted in unambiguous sex identification 275 of 37 (25 and 12) female and 29 (6 and 23) male parents (in 2006 and 2007, 276 respectively). 277 The analysis of stable isotopes of the feeding secretions was conducted at the Iso- 278 Analytical Ltd. Laboratory (Sercon Ltd., Crewe, U.K.), using mass spectrometry to 279 obtain isotope ratios for carbon (13C/12C) and nitrogen (15N/14N), both relative to 280 reference material, and expressed in delta notation (δ13C and δ15N) as 281 δX(‰)=[(Rsample/Rreference)-1]*1000, where X is the heavier isotope and R is the isotopic 282 proportion (13C/12C, 15N/14N) in the sample or reference material. 283 Variations between individuals in δ13C and δ15N values were related to parental sex and 284 tarsus length of chicks, indicative of chick age, using a mixed model in which year was 285 included as random factor, and parental sex, tarsus length of chicks and their interaction 286 as fixed factors. 287 288 289 Rendón et al. 13 290 Results 291 292 EFFECT OF ADULT AGE ON LAYING DATE 293 294 There was no significant effect of the interaction between sex and cohort on the mean 295 age of breeders (F2, 415=0.219; P=0.803). Sex had no additive effect on parental age (F1, 296 415=0.001; 297 joined the colony earlier (c1) were older (14±[SE] 0.54 years, n=45) than those that 298 joined later (c2: 12.2±0.23 years, n=242 and c3: 11.3±0.29 years, n=149) (Tukey- 299 Kramer test: P<0.05), but there was no difference in parental age between c2 and c3 300 (P>0.05). P=0.972) after controlling for cohort (F2, 415=37.9; P<0.001). Parents that 301 302 303 MULTISTATE MODELS 304 305 GOF tests 306 307 The 3.SR component was not significant neither for males (229=19.7, P=0.902) nor 308 females (231=7.5, P=1). The 2.CT component was not significant for males (246=22.0, 309 P=0.999), but it was significant for females (248=72.0, P=0.014) because the 310 observation of females in FP was more likely when they were recorded the previous day 311 (trap dependence: Z=-1.98, P=0.048). The variance factor was ĉ=0.87, indicating minor Rendón et al. 14 312 underdispersion, and therefore a value of ĉ=1 was applied (Burnham & Anderson 313 2002). 314 315 316 Probability of recapture 317 318 The best model of recapture probability included resighting effort (Table 1: M07). The 319 model that also included the additive effect of observation period (Table 1: M04) had 320 slightly higher relative AIC values (ΔAICc=0.5), however, the 95% CI of p included the 321 0 value (βp1=-1.272; [SE: 1.142; 95% CI: -3.511 – 0.966]). Model M07 indicates that the 322 probability of recapture increased with resighting effort (βE=0.444; [0.151; 0.148 – 323 0.741]) from 0.53 (0.36 – 0.70) to 0.90 (0.63 – 0.99) in the range between 2 and 8 hours 324 of observation. 325 326 327 Probability of chick attendance 328 329 The lowest AICc value was due to the additive effect of chick and parental age on the 330 probability of chick attendance (Table 2: M17). Models M13 and M16 had ΔAICc=1.8. 331 Nevertheless, the 95% CI of the coefficients of interactions between sex and both 332 parental age and observation period in model M13 included the 0 value (-0.838 − 0.086 333 and -0.088 − 2.486, respectively), as well as observation period coefficient in model 334 M16 (-0.531 − 0.321). Therefore, we selected model M17 as the best model. Probability Rendón et al. 15 335 of chick attendance was very high for parents attending chicks <16 weeks old (weeks 1- 336 6: 0.993 [95% CI: 0.990 − 0.995]; 7-12 weeks: 0.981 [0.972 – 0.987]; 13–16 weeks: 337 0.982 [0.966 – 0.991]), and decreased for chicks >16 weeks old (0.794 [0.643 – 0.892]) 338 (Fig. 3A). In addition, older parents were less prone to desert from chick care than 339 younger parents (Fig. 3B; βAa=0.197 [SE: 0.101]). 340 341 342 Commuting between the colony and foraging areas 343 344 The model estimating the probability of commuting from FP to FA with the lowest 345 AICc value (Table 3: M27) included the interaction of sex with observation period and 346 the additive effect of parental age. The AICc value of the model M25, that in addition 347 considers the interaction between parental sex and age, differs by 1.5 with respect to 348 model M27, but the interaction term was not significant (βsex*Aa=0.215; [SE: 0.278]; 349 [95% CI: -0.760 – 0.331]). Given than the probabilities of transition of males and 350 females in model M27 were rather similar in period p1 (0.958; [0.927 – 0.977] and 351 0.955; [0.923 – 0.974], respectively), we tested whether there were no significant 352 intersex differences in such transition for p1. The new model (M33) provided a better fit 353 to data (AICc=6348.5), and the probability of commuting from FP to FA during p1 was 354 0.956 (0.934 – 0.971), corresponding to mean sojourn time in FP of 1.1 (1.01 – 1.00) 355 days (Fig. 4A). During p2 the probability of parents moving from FP to FA was lower, 356 (Fig. 4A), and it was lower for females (0.675; [0.569 – 0.765]) than for males (0.885; 357 [0.814 – 0.948]), corresponding to mean sojourn time in FP of 1.5 (1.8 – 1.3) days for 358 females and 1.1 (1.2 – 1.1) days for males (Fig. 4B). Independently of the observation Rendón et al. 16 359 period, older parents spent less time than younger parents in FP between consecutive 360 visits to FA (βAa=0.322; [SE: 0.134]). 361 The probability of remaining in FA is best explained by a model including parental sex 362 as the only explanatory variable (Table 3: M44). A model that also included the additive 363 effect of the observation period (M43) had ΔAICc<2 compared to model M44, but the 364 coefficient of observation period term was not significant (95% CI: -0.139 – 0.331). 365 Daily probability of parents in FA moving to FP was greater in females (0.134 [95% CI: 366 0.106 – 0.168]) than in males (0.109 [0.085 – 0.765]), which implied a mean sojourn 367 time of 7.5 (9.4 – 6.0) and 9.2 (11.8 – 7.3) days in FA for females and males (Fig. 4C), 368 respectively. 369 370 371 FORAGING BEHAVIOUR OF ADULTS 372 373 Four models relating API values in FP and VP (Table S2: Mvii, Mx, Mxii y Mxiii) had 374 ΔAIC values <2, all being submodels of Mvii (sex*[Aa+l]+p). The interaction between 375 parental sex and age for this last model was not significant (z=1.577; P=0.115), neither 376 the additive effects of parental age (z=-0.310; P=0.757) and observation period 377 (z=1.033; P=0.302), so that model Mviii including the interaction between parental sex 378 and locality of observation (βsex[males]*l[VP]=1.032 [SE: 0.432]; P=0.017) was selected. 379 During the stay in FP (Fig. 5) males had lower API (API2: 67%) than females (API3: 380 54%), suggesting that females spent probably more time foraging in the colony. When 381 males and females were in VP, there were no sex differences in API, and both sexes 382 exhibited greater APIs than when they were in FP (API4: 63% and 64%, for males and Rendón et al. 17 383 females, respectively), likely because both sexes spent similar time foraging and spent 384 more time foraging in VP than in FP. 385 386 387 STABLE ISOTOPES IN SECRETIONS 388 389 There were no significant effects of the interaction of parental sex*chick tarsus length 390 (F1, 58.06=1.11; P=0.297), chick tarsus length (F1, 58.32=0.686; P=0.411), or parental sex 391 (F1, 35.34=0.06; P=0.804) on δ13C values. δ15N was related neither to the interaction of 392 adult sex*chick tarsus length (F1, 59=0.75; P=0.391), nor to chick tarsus length (F1, 393 59=0.04; 394 (15.11±SE: 0.54‰) than secretions of female parents (13.22±0.5‰; F1, 59=5.11; 395 P=0.028). 396 397 398 399 400 401 402 403 404 P=0.844) although secretions of male parents had higher levels of δ15N Rendón et al. 18 405 Discussion 406 407 Our results show that both intrinsic (sex, parental age, and chick age), and extrinsic 408 factors (observation date) affected chick provisioning effort by parents. In contrast to 409 most breeding colonies of greater flamingos in Europe, the Fuente de Piedra colony is 410 located in a temporal natural wetland that usually dries out in early summer, and 411 therefore breeding adults have to move to other wetlands for foraging. These conditions 412 must be similar to those faced by the greater flamingo in other natural wetlands, and 413 thus our study may be representative of the environmental conditions that have shaped 414 the provisioning strategies in the species. 415 416 417 CHICK ATTENDANCE BY PARENTS 418 419 Both incubation and chick rearing require considerable energy expenditure by parents 420 (Monaghan & Nager 1997). In the greater flamingo the main limiting of breeding 421 success is desertion during incubation (Rendón-Martos 1996). Indeed, Schmaltz et al. 422 (2011) showed that incubation acts as a bottleneck in the greater flamingo, eliminating 423 individuals of lower quality (i.e. younger breeders), while the remaining parents are 424 equally able to assume chick rearing. We have shown that the probability of chick 425 attendance is very high during chick provisioning (>0.98), suggesting that after chick 426 hatching, most parents could afford paying the costs of parental care. This is in line with 427 Johnson & Cézilly (2007) reporting that parental age does not affect chick survival. Rendón et al. 19 428 Despite having demonstrated a significant effect of parental age on chick desertion, 429 such an effect is minimal while chicks are dependent on parental care (<16 weeks), and 430 increases when chicks are able to fly, with an average occurrence of >10% for the rank 431 of parental ages that we have analysed (Fig. 3B). This result suggests that the 432 prolongation of parental care depends mainly on parental age, and that only older 433 individuals are able of assuming such an effort during chick rearing. Similar patterns in 434 prolongation of parental care by older adults have been reported in gulls (Pugesek 435 1990). The prolongation of parental care before the chicks leave the natal colony likely 436 to have important consequences on their body condition that may affect survival (Nur 437 1984; Tinbergen & Boerlijst 1990), as well as their dispersal capabilities (Barbraud et 438 al. 2003). In years when parental age of the FP colony is highly structured (older 439 individuals start breeding earlier than younger ones), chick survival before fledging is 440 lower in younger chicks, and this may be partly due to chick desertion by younger 441 parents who initiate breeding later than older parents (M.A. Rendón et al., unpubl.). 442 We did not detect seasonal changes, neither on the probability of chick desertion, nor on 443 the frequency of chick provisioning, once the remaining explanatory factors were 444 controlled for. Differences in parental effort of individuals breeding in different dates 445 would have been due to environmental factors related to seasonal changes (e.g. food 446 availability) and/or differences in quality/experience of breeders. 447 Neither did we found a sex-related effect on the probability of chick desertion by 448 parents. The main factors that affect sexual conflict over parental care in birds depend 449 on chick precocity and on new mating opportunities of adults during the same breeding 450 season (Olson et al. 2008). Despite mate switching have been observed within the 451 breeding season (Cézilly & Johnson 1995), adults abandoning at chick stage may have 452 low chances to successfully complete a new breeding attempt. There is age-related Rendón et al. 20 453 assortative mating in the greater flamingo (Cézilly et al. 1997), and this may further 454 limit the opportunities for a sexual conflict over chick care. It is unlikely that one parent 455 would increase its fitness by reducing its effort and thus inducing its mate to 456 compensate for loss (Pärt et al. 1992), given that the residual reproductive value of both 457 parents would be similar. Therefore, it is expected that chick desertion by one of the 458 parents may determine desertion by the other parent, determining a rather similar 459 investment in chick care by both pair members during the provisioning period. 460 461 462 PROVISIONING MOVEMENTS 463 464 Sojourn time in foraging areas 465 466 In general, in dimorphic species the smaller sex feeds in more distant sites and has a 467 lower foraging efficiency than the larger sex during chick provisioning, and as a 468 consequence the smaller sex exhibits a lower level of parental care (Wearmouth & Sims 469 2008, for a review on marine species). Contrary to this pattern, female greater flamingos 470 remained in FA shorter periods than males between consecutive visits to the colony (7.5 471 vs. 9.2 days). This would allow the females to feed the chicks more frequently than 472 males. Central place foraging theory predicts that the time that individuals spend in 473 foraging areas increases with the distance to foraging patches from the central site, or 474 when energy gain in the patch diminishes (Orians & Pearson 1979). Accordingly, we 475 would expect that females forage in wetlands closer to the colony, or exploit more 476 efficiently than males the trophic resources. Nevertheless, satellite tracking of some Rendón et al. 21 477 individuals did not support this (Amat et al. 2005). Neither do δ13C values in secretions 478 indicate sexual differences in wetland use during chick provisioning. δ15N values in 479 male secretions, however, suggest that males feed on prey of higher trophic levels than 480 those of females (e.g. Bearhop et al. 2006). 481 Food availability may affect provisioning movements, as food searching time may vary 482 and this may determine variations in the quantity of food received by chicks 483 (Granadeiro et al. 1998; Weimerskirch et al. 2001). In southern Spain there is a dramatic 484 seasonal decrease in wetland availability from late spring due to the drying out of 485 temporal marshes in the Doñana area, so that the number of flamingos in permanent 486 wetlands increased threefold during our study. However, we did not find a relationship 487 between sojourn times in FA and observation period, suggesting that resource 488 availability did not directly affect foraging efficiency of parents during chick rearing. 489 490 491 Sojourn time in the colony 492 493 In contrast to the pattern of sojourn time in FA, the duration of stays by parents in FP 494 increased throughout the chick rearing period. However, this was affected by gender 495 differences: during p1 both members remained one day in FP, but during p2 females 496 remained longer (1.5 days) than males (1.1 days) in FP between consecutive visits to 497 FA. Given that females remained for shorter periods in FA than males, foraged on prey 498 of apparently lower quality, and both parents provisioning their chick with similar meal 499 sizes (M.A. Rendón et al. unpubl., but see Cézilly et al. 1994), the amount of parental 500 effort might have decreased female body condition more than male body condition Rendón et al. 22 501 (with more expressed differences with advance of the season due to the cumulative 502 effort). These, in turn, might have resulted in females spending more time foraging than 503 males late in the chick provisioning period when visiting the breeding site. 504 Older parents remained for shorter periods in FP between consecutive visits to FA than 505 younger ones. This pattern could mean that older parent spent less time to regain body 506 condition before returning to FA. Due to higher residual reproductive value, young 507 parents should invest less in chick rearing than older ones if investment in reproduction 508 affects survival. However, female flamingos <7 years old breeding for the first time 509 have lower survival than those breeding later, although this pattern is not found in males 510 (Tavecchia et al. 2001). Alternatively, as older individuals could be more 511 experienced/efficient in acquiring food resources (Bildstein et al. 1991), sojourn time in 512 FP would decrease with parental age because older parent spend less time to restore 513 their body condition before leaving the colony. 514 515 516 IS THERE REGULATION OF CHICK PROVISIONING? 517 518 Chick age did not affect commuting behaviour by parents, at least for the range of chick 519 ages that we studied. This indicates that parents may not vary their effort in response to 520 chick requirements. However, there are other aspects of parental effort besides 521 frequency of visits to the colony, for instance, food quantity/quality (Weimerskirch et 522 al. 1997b; Weimerskirch & Lys 2000). Parent males of greater flamingos vary the 523 duration of their feedings according to chick age (Cézilly et al. 1994). On the other 524 hand, the number of feedings received daily by chicks also varies with chick age Rendón et al. 23 525 (Rendón et al. 2012). Therefore, additional studies are necessary to determine whether 526 parental effort is fixed or varies according to chick needs. 527 528 529 SEX-RELATED PROVISIONING PATTERNS 530 531 Sex-related differences in commuting patterns suggest that females make a greater 532 parental investment than males during chick rearing. Commuting periods are shorter for 533 females than males, in spite of females being smaller and foraging on prey of apparently 534 lower quality than males. An explanation to the greater effort of females in chick 535 provisioning may be due to different gender allocation of effort in different phases of a 536 reproductive attempt. Thus, the cumulative effects of such effort would affect the body 537 condition of individuals investing more in previous phases, and this would affect 538 parental care decisions in latter phases (Heaney & Monaghan 1996). However, despite 539 male's greater effort in the first phases of breeding, it is more likely that females desert 540 the nest, and, therefore, males should invest less than females in incubation (Cézilly 541 1993). 542 In the present study a significant variation in isotopic signatures between sexes has been 543 shown. The higher δ15N values in male's secretions suggest segregation between sexes 544 by their selection for different prey size/type (e.g. Forero et al. 2002; Bearhop et al 545 2006). Sex difference in provisioning patterns is probably not due to competitive 546 exclusion from feeding areas, because according to this hypothesis we would expect 547 longer permanency of females in FA to compensate for their lower foraging efficiency. 548 Alternatively, sojourn time at FA could be consequence of size-mediated differences in Rendón et al. 24 549 sex-specific foraging behaviour and microhabitat use. Sex differences in bill size 550 (Cramp & Simmons 1977) would determine males filtering larger prey items than 551 females because of their larger bill, and thus secretions had higher δ15N values. On the 552 other hand, sexual dimorphism can promote sex-specific niche segregation by 553 exploiting different feeding microhabitats, reducing intersexual-competition. Using 554 satellite telemetry, J.A. Amat et al. (unpubl.) found that females foraged in shallow 555 water near the shore of wetlands, whereas males were spread over the wetlands using 556 deeper, open waters. This spatial segregation between sexes must determine differences 557 in both foraging behaviour (e.g. stamp-feeding vs. walk-feeding, Johnson & Cézilly 558 2007) and prey selection. Varo et al. (2011) found that brine shrimp (Artemia 559 parthenogenetica), a potential prey for greater flamingo, are larger and occur at higher 560 density at the bottom of the water column. Therefore, if males exploit larger but more 561 dispersed prey than females, which are restricted to shallower water where exploit more 562 benthonic preys (e.g. Chironomus larvae), optimum foraging duration at FA could be 563 longer for males because more time is needed to maximise energy gained per unit time 564 (Stephen & Krebs 1986). 565 566 567 Conclusions 568 569 By analysing resightings of individually marked flamingos we were able to study 570 commuting patterns during chick rearing. Our study provides further empirical evidence 571 that capture-recapture models can be successfully applied to estimate probabilities of 572 desertion and commuting when parents do not visit their chicks daily and the marking Rendón et al. 25 573 of adults to track their movements (e.g. telemetry) is difficult. Our results indicate that 574 sex differences in provisioning behaviour in the greater flamingo may arise due to both 575 gender-specific foraging behaviour and energy constraints. In foraging areas, males 576 seem to be less efficient than females in replenish body reserves. However, when 577 parents return to fed the chick, females must restoring their body condition staying 578 longer time periods than males in the colony, before returning to adult foraging areas. 579 580 581 Acknowledgements 582 583 The Consejería de Medio Ambiente of Junta de Andalucía authorised our study and 584 provided many facilities. We thank Juan Rubio for assistance during field work. 585 Pesquerías Isla Mayor S.A. granted access to Veta la Palma fish-farm. Logistical 586 support was provided by Laboratorio de Ecología Molecular, Estación Biológica de 587 Doñana, CSIC (LEM-EBD), where sexing of cells extracted from crop content was 588 performed by Mónica Gutiérrez. Sex determination of chicks was carried out at the 589 laboratory of Centro de Análisis y Diagnóstico de la Fauna Silvestre, Consejería de 590 Medio Ambiente, Junta de Andalucía. Data on counts of greater flamingos and water 591 levels in the National Park of Doñana were provided by Equipo de Seguimiento de 592 Procesos Naturales-EBD. The authors would like to thank the anonymous referees for 593 their valuable comments. We were supported by research grants BOS2002-04695 and 594 CGL2005-01136BOS from Ministerio de Educación y Ciencia of Spain, both with EU- 595 ERDF support. 596 Rendón et al. 26 597 References 598 599 Allen, R. P. (1956) The Flamingos: Their Life History and Survival. Research Report 5 600 of the National Audubon Society. National Audubon Society, New York, NY, USA. 601 Amat, J.A., Rendón, M.A., Rendón-Martos, M., Garrido, A. & Ramírez, J.M. (2005) 602 Ranging behaviour of Greater Flamingos during the breeding and post-breeding 603 periods: linking connectivity to biological processes. Biological Conservation, 125, 604 183-192. 605 Barbraud, C., Weimerskirch, H., Robertson, G.G. & Jouventin, P. (1999) Size-related 606 life history traits: insights from a study of snow petrels (Pagodroma nivea). Journal of 607 Animal Ecology, 68, 1179-1192. 608 Barbraud, C., Johnson, A.R. & Bertault, G. (2003) Phenotypic correlates of post- 609 fledging dispersal in a population of greater flamingos: the importance of body 610 condition. Journal of Animal Ecology, 72, 246-257. 611 Bearhop, S., Phillips, R.A., McGill, R., Cherel, Y., Dawson, D.A. & Croxall, J.P. 612 (2006) Stable isotopes indicate sex-specific and long-term individual foraging 613 specialisation in diving seabirds. Marine Ecology Progress Series, 311, 157-164. 614 Bildstein, K.L., Fredrick, P.C. & Spalding M.G. (1991) Feeding patterns and aggressive 615 behavior in juvenile and adult American Flamingos. Condor, 93, 916-925. 616 Burnham, K.P. & Anderson, D.R. (2002) Model Selection and Multimodel Inference: a 617 Practical Information-Theoretic Approach. 2nd Edition. Springer-Verlag, New York. 618 Cézilly, F. (1993) Nest desertion in the greater flamingo, Phoenicopterus ruber roseus. 619 Animal Behaviour, 45, 1038-1040. Rendón et al. 27 620 Cézilly, F., Tourenq, C. & Johnson, A.R. (1994) Variation in parental care with 621 offspring age in the greater flamingo. Condor, 96, 809-812. 622 Cézilly, F. & Johnson, A.R. (1995) Re-mating between and within breeding seasons in 623 the Greater Flamingo Phoenicopterus ruber roseus. Ibis, 137, 543-546. 624 Cézilly, F., Viallefont, A., Boy, V. & Johnson, A.R. (1996) Annual variation in survival 625 and breeding probability in greater flamingos. Ecology, 77, 1143-1150. 626 Cézilly, F., Boy, V., Tourenq, C.J. & Johnson, A.R. (1997) Age-assortative pairing in 627 the greater flamingo Phoenicopterus rubber roseus. Ibis, 139, 331-336. 628 Christensen, R. (2010) ordinal2—Regression models for ordinal data. R package 629 version 2011.05-03. URL: http://www.cran.r-project.org/package=ordinal. 630 Cramp, S. & Simmons, K.E.L. (1977). The birds of the Western Palearctic, Vol. I. 631 Oxford University Press, Oxford. 632 Dawson, R.D. & Bortolotti, G.R. (2003) Parental effort of American kestrels: the role of 633 variation in brood size. Canadian Journal Zoology, 81, 852-860. 634 Drent, R.H. & Daan, S. (1980) The prudent parent: energetic adjustment in avian 635 breeding. Ardea, 68, 225-252. 636 Erikstad, K.E., Fauchald, P., Tveraa, T. & Steen, H. (1998) On the cost of reproduction 637 in long-lived birds: the influence of environmental variability. Ecology, 79, 1781-1788. 638 Forero, M.G., Hobson, K.A., Bortolotti, G.R., Donázar, J.A. & Bertellotti, M. (2002) 639 Food resource utilisation by the Magellanic penguin evaluated through stable-isotope 640 analysis: segregation by sex and age and influence on offspring quality. Marine Ecology 641 Progress Series, 234, 289-299. Rendón et al. 28 642 Forest, L.N. & Gaston, A.J. (1996). The effect of age on timing of breeding and 643 reproductive success in the thick-billed murre. Ecology, 77, 1501-1511. 644 Fujiwara, M. & Caswell, H. (2002) A general approach to temporary emigration in 645 mark–recapture analysis. Ecology, 83, 3266-3275. 646 González-Solís, J., Croxall, J.P. & Wood, A.G. (2000) Sexual dimorphism and sexual 647 segregation in foraging strategies of northern giant petrels, Macronectes halli, during 648 incubation. Oikos, 90, 390-398. 649 González-Solís, J., Becker, P.H., Jover, L & Ruiz, X. 2004. Individual changes underlie 650 age-specific pattern of laying date and egg-size in female common terns (Sterna 651 hirundo). Journal of Ornithology, 145, 129-136. 652 Granadeiro, J.P., Nunes, M., Silva, M.C. & Furness, R.W. (1998) Flexible foraging 653 strategy of Cory’s shearwater, Calonectris diomedea, during the chick-rearing period. 654 Animal Behaviour, 56, 1169-1176. 655 Gray, C.M., Brooke, M.L. & Hamer, K.C. (2005) Repeatability of chick growth and 656 food provisioning in Manx shearwaters Puffinus puffinus. Journal of Avian Biology, 36, 657 374-379. 658 Guttorp, P. (1995) Stochastic Modelling of Scientific Data. Chapman & Hall, New 659 York, NY, USA. 660 Hamer, K.C., Quillfeldt, P., Masello, J.F. & Fletcherd, K.L. (2005) Sex differences in 661 provisioning rules: responses of Manx shearwaters to supplementary chick feeding. 662 Behavioral Ecology, 17, 132-137. 663 Heaney, V. & Monaghan, P. (1996) Optimal allocation of effort between reproductive 664 phases: the trade-off between incubation costs and subsequent brood rearing capacity. 665 Proceedings of the Royal Society of London B, 263, 1719-1724. Rendón et al. 29 666 Johnson, A.R. & Cézilly, F. (2007) The Greater Flamingo. T & AD Poyser, London, 667 UK. 668 Kato, A., Watanuki, Y. & Naito, Y. (2001) Foraging and breeding performance of 669 Japanese cormorants in relation to prey type. Ecological Research, 16, 745-758. 670 Kendall, W.L. & Nichols, J.D. (2002) Estimating state-transition probabilities for 671 unobservable states using capture-recapture/resighting data. Ecology, 83, 3276-3284. 672 Kendall, W.L. & Nichols, J.D. (2004) On the estimation of dispersal and movement of 673 birds. Condor, 106, 720-731. 674 Lang, E.M. (1963) Flamingoes raise their young on a liquid containing blood. 675 Experientia, 19, 532-533. 676 Lebreton, J.D., Burnham, K.P., Clobert, J. & Anderson, D.R. (1992) Modelling survival 677 and testing biological hypotheses using marked animals: a unified approach with case 678 studies. Ecological Monographs, 62, 67-118. 679 Lewis; S., Benvenuti, S., Dall-Antonia, L., Griffiths, R., Money, L., Sherratt, T.N., 680 Wanless, S. & Hamer, K.C. (2002) Sex-specific foraging behaviour in a monomorphic 681 seabird. Proceedings of the Royal Society of London B, 269, 1687-1693. 682 Lewis, S., Schreiber, E.A., Daunt, F., Schenk, G.A., Orr, K., Adams, A., Wanless, S. & 683 Hamer, K.C. (2005) Sex-specific foraging behaviour in tropical boobies: does size 684 matter? Ibis, 147, 408-414. 685 McGraw, K.J., Nolan, P.M., Stoehr, A.M. & Hill, G.E. (2001) Intersexual differences in 686 age-specific parental effort in the house finch (Carpodacus mexicanus). Etologia, 9, 35- 687 41. Rendón et al. 30 688 Monaghan, P. & Nager, R.G. (1997) Why don’t birds lay more eggs? Trends in Ecology 689 and Evolution, 12, 270-74. 690 Navarro, J. & González-Solís, J. (2007) Experimental increase of flying costs in a 691 pelagic seabird: effects on foraging strategies, nutritional state and chick condition. 692 Oecologia, 151, 150-160. 693 Nichols, J.D., Hines, J.E., Pollock, K.H., Hinz, R.L. & Link, W.A. (1994) Estimating 694 breeding proportions and testing hypotheses about costs of reproduction with capture- 695 recapture data. Ecology, 75, 2052-2065. 696 Nisbet, I.C.T. (1997) Female common terns Sterna hirundo eating mollusc shells: 697 evidence for calcium deficits during egg laying. Ibis, 139, 400-401. 698 Nol, E. & Smith, J.N.M. (1987) Effects of age and breeding experience on seasonal 699 reproductive success in the song sparrow. Journal of Animal Ecology, 56, 301-313. 700 Nur, N. (1984) The consequences of brood size for breeding blue tits. II. Nestling 701 weight, offspring survival, and optimal brood size. Journal of Animal Ecology, 53, 497- 702 517. 703 Olson, V.A., Liker, A., Freckleton, R.P. & Székely, T. (2008) Parental conflict in birds: 704 comparative analyses of offspring development, ecology and mating opportunities. 705 Proceedings of the Royal Society B, 275, 301-307. 706 Orians, G.H., & Pearson, N.E. (1979) On the theory of central-place foraging. Analysis 707 of Ecological Systems (eds. D.J. Horn, R.D. Mitchell & G.R. Stairs), pp. 154-177. Ohio 708 State University Press, Columbus, OH, USA. 709 Pärt, T., Gustafsson, L. & Moreno, J. (1992) Terminal investment and a sexual conflict 710 in the collared flycatcher (Ficedula albicollis). American Naturalist, 140, 868-882. Rendón et al. 31 711 Pradel, R., Rioux, N., Tamisier, A. & Lebreton, J.D. (1997) Individual turnover among 712 wintering teal in Camargue: a mark-recapture study. Journal of Wildlife Management, 713 61, 816-821. 714 Pugesek, B. (1990) Parental effort in the California gull: test of parent-offspring conflict 715 theory. Behavioral Ecology and Sociobiology, 27, 211-215. 716 Quillfeldt, P., Schroff, S., Noordwijk, H.J., Michalik, A., Ludynia, K. & Masello, J.F. 717 (2011) Flexible foraging behaviour of a sexually dimorphic seabird: large males do not 718 always dive deep. Marine Ecology Progress Series, 428, 271-287. 719 Rendón, M.A., Garrido, A., Amat, J.A. & Rendón-Martos, M. (2009) Monitoring of 720 Greater Flamingo colonies: some proposals for measuring and interpreting results. 721 Flamingo, Special Publication, 1, 62-75. 722 Rendón, M.A., Garrido, A., Guerrero, J.C., Rendón-Martos, M. & Amat, J.A. (2012) 723 Crop size as an index of chick provisioning in the greater flamingo Phoenicopterus 724 roseus. Ibis, 154, 379-388. 725 Rendón-Martos, M. (1996) La laguna de Fuente de Piedra en la dinámica de la 726 población de flamencos (Phoenicopterus ruber roseus) del Mediterráneo occidental.. 727 PhD thesis, University of Málaga, Málaga. 728 Rendón-Martos, M., Vargas, J.M., Rendón, M.A., Garrido, A. & Ramírez, J.M. (2000) 729 Nocturnal movements of breeding greater flamingos in southern Spain. Waterbirds, 23 730 (Special Publication 1), 9-19. 731 Sæther, B.-E. (1994) Food provisioning in relation to reproductive strategy in altricial 732 birds: a comparison of two hypotheses. Evolution, 48, 1397-1406. 733 Sæther, B.E., Andersen, R., & Pedersen, H.C. (1993) Regulation of parental effort in a 734 long-lived seabird: an experimental manipulation of the cost of reproduction in the Rendón et al. 32 735 Antarctic petrel Thalassoica antarctica. Behavioral Ecology and Sociobiology, 33, 147- 736 150. 737 Schaub, M., Gimenez, O., Schmidt, B.R. & Pradel, R. (2004) Estimating survival and 738 temporary emigration in the multistate capture–recapture framework. Ecology, 85, 739 2107-2113. 740 Schmaltz, L., Cézilly, F. & Béchet, A. (2011) Using multistate recapture modelling to 741 assess age-specific bottlenecks in breeding success: a case study in the greater flamingo 742 Phoenicopterus roseus. Journal of Avian Biology, 42, 1-9. 743 Stephens, D.W. & Krebs, J.R. (1986) Foraging Theory. Princeton University Press, 744 Princeton, NJ. 745 Tavecchia, G., Pradel, R., Boy, V., Johnson, A.R. & Cézilly, F. (2001) Sex- and age- 746 related variation in survival and cost of first reproduction in greater flamingos. Ecology, 747 82, 165-172. 748 Tinbergen, J.M. & Boerlijst, M.C. (1990) Nestling weight and survival in individual 749 great tits (Parus major). Journal of Animal Ecology, 59, 1113-1127. 750 Tinbergen, J.M. & Verhulst, S. (2000) A fixed energetic ceiling to parental effort in the 751 great tit? Journal of Animal Ecology, 69, 323-334. 752 Tremblay, Y. & Cherel, Y. (2005) Spatial and temporal variation in the provisioning 753 behaviour of female rockhopper penguins Eudyptes chrysocome fiholi. Journal of Avian 754 Biology, 36, 135-145. 755 Varo, N., Green, A.J., Sánchez, M.I., Ramo, C., Gómez, J., Amat, J.A. (2001) 756 Behavioural and population responses to changing availability of Artemia prey by 757 moulting black-necked grebes, Podiceps nigricollis. Hydrobiologia, 664, 163-171. Rendón et al. 33 758 Wearmouth, V.J. & Sims, D.W. (2008) Sexual segregation in marine fish, reptiles, birds 759 and mammals: behaviour patterns, mechanisms and conservation implications. 760 Advances in Marine Biology, 54, 107-170. 761 Weimerskirch, H., Mougey, T. & Hindermeyer, X. (1997a) Foraging and provisioning 762 strategies of black-browed albatrosses in relation to the requirements of the chick: 763 natural variation and experimental study. Behavioral Ecology, 8, 635-643. 764 Weimerskirch, H., Cherel, Y., Cuenot-Chailet, F. & Ridoux, V. (1997b) Alternative 765 foraging strategies and resource allocation by male and female wandering albatrosses. 766 Ecology, 78, 2051-2063. 767 Weimerskirch, H. & Lys, P. (2000) Seasonal changes in the provisioning behaviour and 768 mass of male and female wandering albatrosses in relation to growth of their chick. 769 Polar Biology, 23, 733-744. 770 Weimerskirch, H., Zimmermann, L. & Prince, P. (2001) Influence of environmental 771 variability on breeding effort in a long-lived seabird, the yellow-nosed albatross. 772 Behavioral Ecology: 12, 22-30. 773 White, G.C. & Burnham, K.P. (1999) Program MARK: survival estimation from 774 populations of marked animals. Bird Study, 46, 120–138. 775 Ydenberg, R.C., Welham, C.V.J., Schmid-Hempel, R., Schmid-Hempel, P. & 776 Beauchamp, G. (1992) Time and energy constraints and the relationships between 777 currencies in foraging theory. Behavioral Ecology, 5, 28-34. 778 Zimmer, I., Ropert-Coudert, Y., Poulin, N., Kato, A. & Chiaradia, A. (2011) Evaluating 779 the relative importance of intrinsic and extrinsic factors on the foraging activity of top 780 predators: a case study on female little penguins. Marine Biology, 158, 715-722. Rendón et al. 34 781 Ziswiler, V. & Farner, D.S. 1972. Digestion and the digestive system. Avian Biology. 782 vol. II. (eds. D.S. Farner, J.R. King & K.C. Parkes), pp. 343-430. Academic Press, New 783 York, NY, USA. 784 Zweers, G., De Jong, F., Berkhoudt, H. & Vanden Berge, J.C. (1995) Filter feeding in 785 flamingos (Phoenicopterus ruber). Condor, 97, 297-324. 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 Rendón et al. 35 802 Supporting Information 803 The following Supporting Information is available for this article online. 804 Appendix S1. Phenotypic sexing of adult greater flamingos. 805 Appendix S2. Description of the multistate capture-recapture models. 806 Table S1. Results of the model fitting for the selection of the more parsimonious model 807 for probabilities of residence and movement between the breeding site and foraging 808 areas. 809 Table S2. Results of the mixed-effects ordinal logistic models predicting abdominal 810 profile index values in adults grater flamingos. 811 812 Rendón et al. 36 813 FIGURE LEGENDS 814 815 Figure 1. Location of the study area at Fuente de Piedra lake, and the two main feeding 816 areas for breeding flamingos during the chick-rearing period in Doñana National Park. 817 818 Figure 2. (A) Weekly variation (week 1 is the first week of January) of the numbers of 819 adult flamingos at the colony and dispersed along the lake, and water depth at FP in 820 2001. Three waves of breeding (c) were recorded in 2001. Daily resightings of marked 821 adults were conducted in two periods (p). (B) Monthly variation of greater flamingos in 822 GM and VP in 2001. Water depth in the GM is also shown. 823 824 Figure 3. Chick (A) and parent (B) age effects on the mean (±95% CI) daily probability 825 of chick attendance by parents (S) in FP according to the M17-model (table 2). For 826 chick age effect, only the predicted values for categories 7-12 and >16 weeks old are 827 shown. 828 829 Figure 4. Mean daily probabilities (±95% CI) of movements by parent greater flamingos 830 (ψ) form FP to FA and vice versa, predicted by M44-model (table 3). The right axis 831 indicates the mean sojourn time in FP and FA. (A) Observation period and sex effects 832 on movement probabilities from FP to FA. (B) Effects of parental age and (C) sex on 833 the probability of movements are also shown for foraging displacements from FA to FP. 834 835 Figure 5. Predicted probabilities for API categories in greater flamingos relative to sex 836 (males: black bars; females: white bars) and locality of observation (FP: upper graph; 837 VP: lower graph), resulting from the ordinal logistic model: sex*l. 838 Rendón et al. 37 839 840 FIGURE 1 841 842 Doñana National Park Fuente de Piedra Lake 0 100 km Guadalquivir Marshes Breeding Colony Surveyed Area Veta la Palma 0 843 844 10 km 0 1 km Rendón et al. 38 845 846 FIGURE 2 847 848 p1 A p2 16000 14000 0.5 12000 0.4 10000 8000 0.3 c2 6000 c3 4000 0.2 Depth (m) Nº of individuals Adults dispersed Adults at colony Chicks Breeding waves Depth 0.6 c1 0.1 2000 0 0 0.0 8 B 10 12 14 16 18 20 22 24 28 30 32 34 36 25000 1.0 GM VP Depth 0.9 0.8 0.7 15000 0.6 0.5 10000 0.4 0.3 0.2 5000 0.1 0.0 0 8 849 850 851 852 10 12 14 16 18 20 22 Weeks 24 28 30 32 34 36 Depth (m) Nº of individuals 20000 Rendón et al. 39 853 854 855 856 857 858 859 FIGURE 3 Rendón et al. 40 860 861 862 863 FIGURE 4 Rendón et al. 41 864 865 FIGURE 5 866 0,7 0.7 0,6 0.6 0,5 0.5 0,4 0.4 0,3 0.3 Probability 0,2 0.2 0,1 0.1 0,0 0.0 1 1 2 2 3 3 4 4 5 5 1 1 2 2 3 3 4 4 5 5 0.7 0,7 0.6 0,6 0.5 0,5 0.4 0,4 0.3 0,3 0.2 0,2 0.1 0,1 0.0 0,0 867 868 869 API Rendón et al. 42 870 871 Table 1. Model selection for daily recapture probability for greater flamingos at FP 872 colny. AICc, ΔAICc, number of parameters (k), and deviance are shown. Variables in 873 uppercase letters are continuous. Most parsimonious model is highlighted in bold. 874 Model M01: sex+p+E M02: sex+p M03: sex+E M04: p+E M05: sex M06: p M07: E M08: . 875 876 877 878 879 880 881 AICc ΔAICc k Deviance 6371.2 2.3 52 6262.6 6403.6 34.7 51 6273.5 6371.0 2.1 51 6264.6 6369.4 0.5 51 6263.0 6374.5 5.6 50 6270.2 6375.6 6.7 50 6271.4 6368.9 0 50 6264.6 6373.5 4.6 49 6271.4 Abbreviations sex: parental sex; p: resighting period (p1 and p2); E: resighting effort (hours); .: constant; +: additive effect. Rendón et al. 43 882 883 Table 2. Model selection for daily probability of chick attendance for greater flamingos 884 at FP colony. AICc, ΔAICc, number of parameters estimated (k), and deviance are 885 shown. Variables starting with uppercase letters are continuous. Most parsimonious 886 model is highlighted in bold. 887 M09: M10: M11: M12: M13: M14: M15: M16: M17: M18 M19 M20: 888 889 890 891 892 893 894 Model sex*(ac4+Aa+c+p) sex*(Aa+c+p)+ac4 sex*(Aa+c+p) sex*(Aa+p)+ac4+c sex*(Aa+p)+ac4 sex*(p)+ac4+Aa sex+ac4+Aa+p ac4+Aa+p ac4+Aa ac4 Aa . AICc ΔAICc k Deviance 6368.9 10.5 50 6264.6 6363.6 5.2 47 6265.8 6370.5 12.1 44 6279.1 6363.4 5.0 45 6269.9 6360.2 1.8 43 6271.0 6360.7 2.3 42 6273.7 6362.1 3.8 41 6277.3 6360.2 1.8 40 6277.4 6358.4 0.0 39 6277.8 6361.4 3.0 38 6282.9 6375.9 17.5 36 6301.7 6378.7 20.3 35 6306.6 Abbreviations sex: parental sex; ac4: chick age (1-6 weeks, 7-12 weeks, 13-16 weeks, and >16 weeks); Aa: age of breeders (years); c: waves of breeding (c1, c2, and c3); p: resighting period (p1 and p2); .: constant; +: additive effect; *: interaction among variables. Rendón et al. 44 895 896 Table 3. Model selection for daily movements between FP and FA (ψ FP→FA and ψ 897 FA→FP 898 Variables starting with uppercase letters are continuous. Most parsimonious models are 899 highlighted in bold. ). AICc, ΔAICc, number of parameters estimated (k), and deviance are shown. 900 901 902 903 904 905 906 907 908 Transition ψFP→FA M21 M22 M23 M24 M25 M26 M27 M28 M29 M30 M31 M32 M33 Model sex*(ac3+Aa+c+p) sex*(Aa+c+p)+ac3 sex*(Aa+c+p) sex*(Aa+p)+c sex*(Aa+p) sex*(Aa)+p sex*(p)+Aa sex*(p) sex+p sex p . p1(.), p2(sex)+Aa AICc ΔAICc k Desviación 6358.4 9.9 39 6277.8 6356.1 7.6 37 6279.8 6355.3 6.8 35 6283.2 6354.9 6.4 33 6287.0 6352.0 3.6 31 6288.4 6357.2 8.7 30 6295.7 6350.5 2.1 30 6289.0 6354.3 5.8 29 6294.9 6359.5 11.0 28 6302.1 6418.1 69.6 27 6362.9 6368.6 20.1 27 6313.4 6431.8 83.3 26 6378.6 6348.5 0.0 29 6289.0 ψFA→FP sex*(ac3+Aa+c+p) sex*(ac3+Aa+p)+c sex*(ac3+Aa+p) sex*(Aa+p)+ac3 sex*(Aa+p) sex*(Aa)+p sex*(p)+Aa sex*(Aa) sex*(p) sex+p sex p . 6348.5 6344.9 6342.7 6339.3 6340.3 6338.4 6338.4 6337.0 6336.8 6334.9 6333.5 6339.8 6338.1 M34 M35 M36 M37 M38 M39 M40 M41 M42 M43 M44 M45 M46 15.0 11.5 9.2 5.8 6.8 4.9 4.9 3.5 3.3 1.4 0.0 6.3 4.6 29 27 25 23 21 20 20 19 19 18 17 17 16 6289.0 6289.7 6291.6 6292.3 6297.5 6297.7 6297.7 6298.4 6298.2 6298.3 6299.0 6305.3 6305.6 The most parsimonious model for ψ was determined departing from the model: FA FAFP Sap4, Aa PEFP ψ FP sex*(ap3 Aac p) ψ sex*(ap3 Aac p) . In the first place we established the best model for the FP→FA transition and, once determined, we selected the best model for the FP→FA transition. Abbreviations as in the table 3.