University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Faculty Publications in the Biological Sciences Papers in the Biological Sciences

2004 Juvenile and Adult Survival in the Sociable Weaver (Philetairus Socius), a Southern-Temperate Colonial Cooperative Breeder in Africa Rita Covas University of Cape Town, [email protected]

Charles R. Brown University of Tulsa, [email protected]

Mark D. Anderson Department of Tourism, Environment and Conservation

Mary Bomberger Brown University of Nebraska-Lincoln, [email protected]

Follow this and additional works at: http://digitalcommons.unl.edu/bioscifacpub Part of the Biology Commons, and the Ornithology Commons

Covas, Rita; Brown, Charles R.; Anderson, Mark D.; and Brown, Mary Bomberger, "Juvenile and Adult Survival in the Sociable Weaver (Philetairus Socius), a Southern-Temperate Colonial Cooperative Breeder in Africa" (2004). Faculty Publications in the Biological Sciences. 467. http://digitalcommons.unl.edu/bioscifacpub/467

This Article is brought to you for free and open access by the Papers in the Biological Sciences at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Faculty Publications in the Biological Sciences by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. JUVENILE AND ADULT SURVIVAL IN THE SOCIABLE WEAVER (PHILETAIRUS SOCIUS), A SOUTHERN-TEMPERATE COLONIAL COOPERATIVE BREEDER IN AFRICA Author(s): Rita Covas, Charles R. Brown, Mark D. Anderson, and Mary Bomberger Brown Source: The Auk, 121(4):1199-1207. Published By: The American Ornithologists' Union DOI: http://dx.doi.org/10.1642/0004-8038(2004)121[1199:JAASIT]2.0.CO;2 URL: http://www.bioone.org/doi/full/10.1642/0004-8038%282004%29121%5B1199%3AJAASIT %5D2.0.CO%3B2

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. The Auk 121(4):1199–1207, 2004 JUVENILE AND ADULT SURVIVAL IN THE SOCIABLE WEAVER (PHILETAIRUS SOCIUS), A SOUTHERN-TEMPERATE COLONIAL COOPERATIVE BREEDER IN AFRICA R C,1,4 Cmq R. B|s,2 Mp D. Ass,3 s M Bgl B|s2 1Percy Fitzpatrick Institute, University of Cape Town, Rondebosch 7701, South Africa; 2Department of Biological Sciences, University of Tulsa, Tulsa, Oklahoma 74104, USA; and 3Department of Tourism, Environment and Conservation, Private Bag X5018, Kimberley 8300, Northern Cape, South Africa

Agh.―We used capture–recapture analyses to describe juvenile and adult survival from 1993 to 2001 in a population of Sociable Weavers (Philetairus socius), a colonial, cooperatively breeding passerine of southern Africa. We examined temporal variation in survival and the role that the breeding season’s length and environmental factors play in determining survival pa erns in the population. Annual survival probability (mean ± SE) was 0.66 ± 0.02. In contrast to most passerines, juveniles and adults had similar survival probabilities; survival rates did not vary signifi cantly between years. We found no relationship among temperature, rainfall, and survival. Relatively high survival rates in Sociable Weavers probably result from a benign climate and easy access to food in winter. Juvenile survival may also be enhanced by prolonged parental care and delayed dispersal. Received 1 April 2003, accepted 24 June 2004.

Rs.—Utilizamos análisis de captura y recaptura para describir la supervivencia de juveniles y adultos entre 1993 y 2001 en una población de Philetairus socius, una especie de ave paserina colonial con cría cooperativa del sur de África. Examinamos la variación temporal en la supervivencia, y la importancia de la longitud del período de apareamiento y de los factores ambientales para la supervivencia de los individuos de la población. La probabilidad de supervivencia anual (media ± EE) fue de 0.66 ± 0.02. En contraste con la mayoría de los Passeriformes, los individuos juveniles y adultos mostraron probabilidades de supervivencia similares, y las tasas de supervivencia no variaron signifi cativamente entre años. No encontramos relación alguna entre la temperatura, la cantidad de lluvia y la supervivencia. La relativamente alta probabilidad de supervivencia observada se encuentra probablemente relacionada con un clima benigno y fácil acceso al alimento durante el invierno. La supervivencia de los juveniles puede además estar infl uenciada positivamente por un período de cuidado parental prolongado y por la dispersión retardada de las crías.

C life-history evolution their northern-temperate counterparts (Rowley requires knowing how survival varies with age and Russell 1991, Martin 1996, Ghalambor and and environmental conditions (Stearns 1992, Martin 2001), parents may invest more energy Charlesworth 1994). Studying survival, how- in fewer young (Williams 1966, Martin 1996, ever, is o en diffi cult, because individuals must Russell 2000). As a result, juvenile (i.e. post- be marked and monitored for multiple years. In fl edging) survival may be higher at southern birds, we know especially li le about annual latitudes than in more temperate locations survival rates of populations in the tropics and (Martin 1996, Martin et al. 2000, Russell 2000). the Southern Hemisphere, where relatively Recent work has shown that adult survival in few long-term population studies have been some southern populations is higher than in sim- conducted. Because birds in southern regions ilar taxa in northern-temperate regions (Johnston generally have smaller clutches, longer develop- et al. 1997, Ghalambor and Martin 2001, Peach et mental periods, and higher adult survival than al. 2001). How survival at southern latitudes var- ies between years and how it may be aff ected by environmental factors is largely unknown. 4Present address: ICAPB, University of Edinburgh, Passerine survival can be aff ected by food avail- King’s Buildings, West Mains Road, Edinburgh EH9 ability (Jansson et al. 1981, Newton 1998) and 3JT, United Kingdom. E-mail: [email protected] winter severity (McNamara and Houston 1990,

1199 1200 C q. [Auk, Vol. 121

Cuthill and Houston 1997). However, most tropi- typically high, with hot days and cool–warm nights cal and southern-temperate regions experience from October to March (approximately 8–40°C) mild winters—which probably support relatively and mild days and very cold nights from May to high food levels (Oatley 1982, Rowley and Russell September (approximately –8 to 25°C) Study species.—The Sociable Weaver is endemic 1991)—and less extreme weather conditions than to southern Africa, with its distribution centered more northerly areas. Knowing how survival of in the Northern Cape and Namibia in strong asso- southern-temperate species is aff ected by food ciation with southern Kalahari vegetation (Maclean availability and weather conditions would help 1973a; Mendelsohn and Anderson 1997). Sociable in understanding the mechanisms that regulate Weavers weigh between 26 and 32 g, and sexes are survival at southern latitudes. indistinguishable in the fi eld. Sociable Weavers feed Here, we investigate juvenile and adult sur- predominantly on insects, but also on seeds and other vival in the Sociable Weaver (Philetairus socius), plant products (Maclean 1973d). Colonial cooperative a colonial cooperative passerine endemic to breeders, they show great variation in colony size the semi-arid savannas of southern Africa. The (2–500 individuals per colony; Maclean 1973b) and in number of helpers a ending a brood (up to nine latitudinal trend in avian life histories predicts helpers per brood; Maclean 1973c). Sociable Weavers higher adult survival in Sociable Weavers build a huge nest (1–4 m wide) with separate cham- than in most northern-temperate passerines. bers in which a pair (with their off spring or other High juvenile survival should also occur birds) roost and breed. The nest is built on a variety because—besides benefi ting from factors that of sturdy structures, from Acacia trees to telephone might increase adult survival—young in this poles (Maclean 1973b, Mendelsohn and Anderson sedentary species receive extended parental 1997). At Benfontein, nests are constructed on Acacia care and remain in the natal colony for at erioloba trees or, occasionally, on A. tortilis. The colony least four months (Covas 2002, R. Covas and can remain active for several decades, being occupied C. Doutrelant unpubl. data). Sociable Weavers by successive generations of birds, which continually add to the structure (Maclean 1973b). are suitable for study of environmental factors Sociable Weavers breed aseasonally in response to and survival because they inhabit a highly rainfall (Maclean 1973c). Rainfall is also a major deter- fl uctuating environment, where rainfall, which minant of duration of the breeding period, number of is variable and o en low, is the main determi- broods (1–8), and clutch size (2–6; Maclean 1973c). In nant of food availability and reproductive eff ort our study area, Sociable Weavers usually start breeding (Maclean 1973c, Lloyd 1999). In our study area, in September or October, and the breeding season can winters are usually sunny, with mild daytime last between three and nine months (Covas et al. 2002). temperatures. However, night-time tempera- Field methods.―From August 1993 to November tures can drop to several degrees below zero, 2000, we conducted a capture–mark–recapture study representing a potential cost in terms of ther- at Benfontein. The study area contained 25 Sociable Weaver colonies. At 16–18 of those colonies, residents moregulation (White et al. 1975, du Plessis were captured twice a year. During the fi rst fi ve and Williams 1994). The present study aims to years, the capture eff ort was distributed throughout (1) estimate juvenile and adult survival in the the year (capturing the birds in 1–2 colonies each Sociable Weaver and (2) investigate temporal month). Beginning in mid-1998, we concentrated our variation in survival—in particular, how sur- capture eff orts during 1–2 consecutive months at the vival is aff ected by winter temperature, rainfall, beginning and end of the breeding period. Birds were and duration of the breeding season. captured with mist nets placed around the nesting tree before dawn. Before sunrise, we approached the Mm colony and fl ushed the birds into the nets. A small number of birds usually managed to escape (by fl y- Study site.―The study was conducted at Benfontein ing over the nets), but most individuals present were Game Farm, situated ∼6 km southeast of Kimberley, in caught. Birds were removed from the nets, placed the Northern Cape Province, South Africa (∼28°53’S, individually in linen bags, and processed. The dura- 24°89’E). Vegetation consists of open savanna and is tion of the procedure depended on the size of the dominated by Stipagrostis grasses and camelthorn tree colony but usually lasted 1–3 h. Birds were marked (Acacia erioloba). Study area is semi-arid, experiencing with a unique numbered ring from the South African low and unpredictable rainfall (average 431 ± 127 mm Bird Ringing Unit (SAFRING) and, in recent years, year–1; Weather Bureau, Pretoria), with most of the with an additional individual color-combination. precipitation falling during the summer months from Juvenile birds molt into adult plumage when they are September to April. Daily temperature ranges are approximately four months old; before that, their age October 2004] Survival in Sociable Weavers 1201

can be estimated through the development of a black fi t (judging by AICc values and weights) than one patch on the face and throat (Maclean 1973c, R. Covas without it, survival was considered to diff er among unpubl. data). the groups. We devised a candidate set of models a During the fi rst fi ve years of the study, the breed- priori that tested legitimate survival hypotheses, and ing activity of the birds was not directly monitored. those are the ones presented in the tables. Number of To determine when the breeding season took place, estimable parameters in our models was provided by we used development of the black patch on captured MARK on the basis of the chosen model structure and juveniles to estimate month of hatching. Presence or statistical information provided by our data. absence of a brood patch also provided an indication Before comparing the fi t of the candidate set of of the breeding season. In total, we considered eight models, we performed a goodness-of-fi t test for each breeding periods (there was no breeding activity in dataset, using RELEASE (Burnham et al. 1987). That the summer of 1994–1995): August 1993 (beginning of program evaluated how well the data met variance the study), October 1993 to April 1994, October 1995 assumptions inherent in the binomial distribution to April 1996, October 1996 to March 1997, June 1997 used in mark–recapture analysis. When a data set to May 1998 (no capture during that period), October does not meet the assumptions, it is usually because 1998 to January 1999, September 1999 to May 2000, the data are overdispersed, refl ecting lack of indepen- and September 2000 to January 2001. dence or some heterogeneity among observations and Statistical methods.—We estimated annual survival o en brought about by the presence of transients or probabilities and tested hypotheses using the general trap-dependence. We assessed goodness-of-fi t by fi rst methods of Lebreton et al. (1992) and Burnham and calculating a combined chi-square value on the basis Anderson (2002). We used MARK (Cooch and White of tests 3m, 2ct, and 2cl in RELEASE. That subset of 1998, White and Burnham 1999) to generate maxi- tests can incorporate age-dependence in determining mum-likelihood estimates of survival and recapture goodness-of-fi t (Pradel et al. 2004); our most highly probabilities. A so ware package that computes parameterized models (for which goodness-of-fi t was survival and recapture parameters using encounter assessed) all contained age-dependence. The total histories of individually marked birds, MARK com- chi-square value allowed estimation of a variance pares and tests the fi t of diff erent statistical models, infl ation factor, ĉ, as chi-square divided by degrees allowing one to evaluate the plausibility of diff erent of freedom (χ2/df). The ĉ value was used in MARK biological hypotheses (Lebreton et al. 1992). to adjust the AICc through quasi-likelihood, result-

We use the general notation of Lebreton et al. (1992), ing in a QAICc whenever ĉ departed from 1.0. In our in which annual survival probability is denoted as φ case, model selection and parameter estimation was and recapture probability as p. Subscripts indicate based on the model with the lowest QAICc value as whether parameters in a model are time-dependent described above for AICc. That variance infl ation

(e.g. φtime, ptime), vary with age (and if so, with how adjustment allowed use of data that departed from many age classes; for example, φage2, page2 for two age the assumptions of the binomial distribution. In such classes), are constant over time (e.g. φconstant, pconstant), cases, a er adjustment, maximum likelihood can still are group-specifi c (e.g. φgroup, pgroup), or describe an provide optimal point estimators of model parameters interaction between group and time (e.g. φgroup×time, (Wedderburn 1974, Burnham and Anderson 2002). p group×time). Because Sociable Weavers are unpredictable, asea- We compared the fi t of competing models with sonal breeders, time intervals in-between our capture the Akaike Information Criterion (AIC; Akaike 1973, occasions were not constant (see above). Each interval Lebreton et al. 1992, Burnham and Anderson 2002) was defi ned as time from the start of the nonbreeding corrected for sample size (AICc) as provided by period until the end of the successive breeding period. MARK. In theory, the model with the lowest value Thus, as in most mark–recapture studies, the breeding of AICc is the so-called “best” model. The AICc is season was considered to take place in an instant in a formal criterion used to select among competing time, with zero mortality during that breeding sea- models; the currently accepted convention (Burnham son. For the eight capture occasions, corresponding and Anderson 2002) is that models with AICc that dif- to the eight breeding periods that occurred between fer by <2 are indistinguishable in terms of their fi t to July 1993 and January 2001, the intervening intervals the data. We also present normalized AICc weights, a consisted of 0.67, 2, 0.92, 1.2, 0.58, 1.4, and 0.67 years. measure of a model’s relative probability of being the However, we standardized all survival estimates to best model for the data as compared with alternative intervals of 1 year, using the set-time-intervals util- models (Burnham and Anderson 2002). We assessed ity in MARK. That made all survival probabilities potential diff erences in survival among groups of comparable (and consistent with other studies), being birds (e.g. age classes) by comparing various models, estimated for a period of the same length. Program some of which modeled the eff ect of each group, MARK uses the actual time interval as an exponent whereas others considered the groups as a common of the estimated survival probability to correct for the pool. If a model with a group eff ect provided a be er length of the time interval. Although some birds were 1202 C q. [Auk, Vol. 121 caught during nonbreeding periods, those captures We tested for survival diff erences between juve- were not used in constructing individual encounter niles and adults in two ways. (1) We treated birds histories; a bird had to be caught during a breeding banded as adults and those banded as juveniles as period to be designated as surviving to that period. separate groups, with fi rst-year survival of juveniles Because of the variation in length of intervals constant but diff erent from all other cohorts; yearly between breeding periods, we checked to see if those survival of adults and juveniles in their later years diff erent intervals might have aff ected our survival was otherwise modeled as the same. That approach estimates. We compared the best-fi  ing time- also tested for other diff erences among birds marked dependent model (model 2; Table 1) with a struc- as adults versus those marked as juveniles, such as a turally similar model that had survival linearly greater proportion of transients in one age class. Such constrained on the length of the breeding season analysis used the entire data set, with all those banded (using the log-link function in MARK). The model as adults designated as one group and those banded with an eff ect of breeding-season length had a QAICc as juveniles designated as the second group (models 3 that was only 0.25 less than that for model 2, which and 8; Table 1). (2) We then repeated the age analysis indicates that it did not provide a be er fi t to our using only juveniles (known-age birds) and modeled data. For that reason, and also because we found no fi rst-year versus later-year survival for that set of strong eff ect of time in our analyses (below), we con- birds (models 11 and 13; Table 2). clude that the diff ering breeding-season lengths did Total summer rainfall (October–March) and mini- not aff ect our results or conclusions. mum and average winter temperatures (May–August) Juveniles were birds younger than four months were modeled as linear constraints on survival, using upon initial capture, and adults those with a fully the log-link function in MARK. Winter temperature developed black throat-patch (older than four was chosen because passerines are o en vulnerable to months). Juveniles were caught at diff erent ages; food shortage brought about by climatic severity during we initially assigned each individual to one of four winter (e.g. Newton 1998). Weather data were obtained classes based on its approximate age upon fi rst cap- from the Weather Bureau, Pretoria, South Africa. ture: 24–40 days old, 40–60 days old, 60–90 days old, and 90–120 days old. To determine if survival may Rq have varied across birds in the diff erent age classes, we compared a model that treated each age class as Goodness-of-fi t.―The variance infl ation factor, the same (φ , p ; model 10; Table 2) with one constant time ĉ, was calculated as 2.41 for the full data set that treated each age class as a separate group (φ , group4 and 1.69 for the subset of birds fi rst banded p ). The one treating all juveniles the same regard- time as juveniles. We thus used quasi-likeli- less of age at fi rst capture was a much be er fi t to our hood (e.g. QAICc) for survival estimation data (QAICc = 576.94, AICc weight = 0.9347) than the one treating survival among the four age classes as and model fi  ing. The lack of fi t was caused principally by trap-dependence among birds diff erent (AICc = 582.26, AICc weight = 0.0652). Thus, all juveniles were pooled for analysis, regardless of banded as adults (RELEASE test 2ct; χ2 = age at fi rst capture. 24.56, df = 5, P < 0.001). There may also have

Tgq 1. Models to assess eff ects of period (time), minimum (mintemp) and average (avgtemp) winter temperatures,

and rainfall (rain) on survival and recapture probabilities in Sociable Weavers, using all data (n = 1,486 birds).

Birds fi rst banded as adults and as juveniles were treated as separate groups (group2) in some models; fi rst-year

survival of juveniles were treated as separate (age2) in others. All models, except (9), had full time-dependence in recapture probability.

Number of

Model QAICc ∆QAICc QAICc weight estimable parameters

(1) φconstant , ptime 1226.9 0.0 0.4608 8 a (2) φtime,modifi ed, ptime 1228.9 2.0 0.1760 10

(3) φgroup2age2, ptime 1229.0 2.1 0.1693 9

(4) φmintemp, ptime 1229.9 3.0 0.1022 11

(5) φavgtemp, ptime 1232.0 5.1 0.0371 12 a (6) φtime, ptime 1232.2 5.3 0.0367 12

(7) φrain, ptime 1233.2 6.3 0.0199 11

(8) φtime×group2age2, ptime 1255.8 28.9 0.0000 13

(9) φconstant, pconstant 1443.2 216.2 0.0000 2 a In model (6), survival varied with each period (full time-dependence); whereas model (2) treated the fi rst, fourth, and fi h breeding periods as the same, and the remaining ones as separate. October 2004] Survival in Sociable Weavers 1203

Tgq 2. Models to assess eff ects of age on survival and recapture probabilities in Sociable Weavers using only the subset of birds fi rst banded as juveniles (n = 473 birds). The two age classes used were the fi rst year and

all older ages (age2), and time-dependent models (time) treated all periods as separate. All models with an eff ect

of time (time) for either survival or recapture were fully time-dependent. Number of

Model QAICc ∆QAICc QAICc weight estimable parameters

(10) φconstant, ptime 576.9 0.0 0.8566 8

(11) φage2, ptime 581.1 4.2 0.1089 9

(12) φtime, ptime 583.4 6.5 0.0343 12

(13) φtime×age2, ptime 593.1 16.2 0.0003 17

(14) φtime, pconstant 633.5 56.6 0.0000 8

(15) φconstant, pconstant 662.0 85.1 0.0000 2 been some trap-dependence among birds of 0.662 ± 0.019 for Sociable Weavers in the marked as juveniles (test 2ct; χ2 = 8.14, df = Kimberley population. Recapture probabilities 4, P = 0.09). (mean ± SE) estimated from that model were Eff ect of time.―Our data set consisted of 1,013 0.600 ± 0.057, 0.476 ± 0.064, 0.144 ± 0.030, 0, adults and 473 juveniles. Of those, 348 adults 0.106 ± 0.035, 0.827 ± 0.073, and 0.596 ± 0.047, (34.4%) and 173 juveniles (36.6%) were recap- respectively, for the seven intervals between the tured during at least one subsequent breeding observed breeding periods. period. There was no strong eff ect of time (i.e. Eff ects of winter temperature and rainfall.― interval between breeding periods) on survival There was li le evidence that Sociable Weaver probabilities in Sociable Weavers (Table 1). A survival was aff ected by either temperature fully time-dependent model (model 6; Table 1) or rainfall in winter. A model with minimum was a poorer fi t to our data than one with sur- winter temperature as a linear constraint on vival constant across all time periods (model 1; survival (model 4; Table 1) did not fi t as well Table 1). A model in which survival was treated as one without an eff ect of temperature (model as the same for the fi rst, fourth, and fi h breed- 1). Similarly, a model using average winter tem- ing periods and separately for the remaining perature as a constraint on survival (model 5; ones (model 2; Table 1) was a be er fi t than the Table 1) was far less plausible than one without fully time-dependent one (model 6; Table 1); but an eff ect, and a model with an eff ect of rainfall even the modifi ed time-dependent model did (model 7; Table 1) was even less likely. not provide a be er fi t than one with constant Eff ects of age.―We found li le evidence that survival (model 1; Table 1). The time-constant survival diff ered between adult and juvenile model was 12.5× more likely than the fully Sociable Weavers. A model treating birds time-dependent one and 2.6× more likely than banded as adults and juveniles as separate the modifi ed time-dependent model, as judged groups, with fi rst-year survival of the juve- from QAICc weights (Table 1). nile group diff erent from that of older classes We found a similar result when confi ning the (model 3; Table 1), was less plausible than the analysis to only known-age birds (those fi rst more parsimonious model without an age eff ect banded as juveniles). A model without an eff ect that treated the groups the same (model 1). of time (model 10; Table 2) was 25× more likely Similarly, using only the subset of known-age than a similar model with time-dependence juvenile birds (n = 473), a model with constant (model 12). Recapture probabilities did vary survival across time and age (model 10; Table with time, however (Tables 1 and 2); all mod- 2) was the best fi t. In the case of the full data set els with constant recapture probabilities (e.g. (Table 1) and the known-age subsample (Table models 9, 14, and 15) had substantially worse 2), the best-fi  ing model without an age eff ect fi t. That is consistent with the fact that our fi eld was 2.7× and 7.9× more likely, respectively, than eff ort varied between diff erent periods. the next best one that included an age eff ect,

Model 1 (Table 1) was used to derive the based on the QAICc weights. Using only the average annual-survival probability using the known-age birds, model φconstant, ptime (model total data set (n = 1,486 birds), which yielded 10; Table 2) estimated average (mean ± SE) an annual survival probability (mean ± SE) annual-survival probability of Sociable Weavers 1204 C q. [Auk, Vol. 121 as 0.659 ± 0.028, which is similar to the estimate wrens; Rabenold 1990), fi rst-year survival esti- from the total data set. mates overlapped with adult estimates, which suggests that our result could be found in other Dhs species. The relatively high juvenile survival reported Adult and juvenile survival.―We found rela- here could be related to prolonged parental care tively high adult and juvenile survival in the and delayed dispersal. Young Sociable Weavers Sociable Weaver. The annual survival estimate can be fed for up to six weeks a er fl edging, obtained for our population (0.66) was higher which may enhance survival by reducing risks than estimates for European passerine popula- taken during development of foraging skills tions, which are mostly around or below 0.5 (Langen 2000, Russell 2000; cf. Sullivan 1989). (Sæther 1989, Peach et al. 2001); but it was lower Delayed dispersal was shown to promote juve- than some estimates obtained for other south- nile survival in Siberian Jays (Perisoreus infaus- ern African passerines by Peach et al. (2001). tus; Ekman et al. 2000) and Brown Thornbills However, our estimate may be an underesti- (Acanthiza pusilla; Green and Cockburn 2001). mate, if there is any permanent dispersal away In a study conducted over three breeding sea- from the study area. As in virtually any survival sons and involving intensive color-banding of study of an open population using recaptures nestlings, young Sociable Weavers never le the or resightings, permanent emigration is con- natal colony in their fi rst four months of life and founded with mortality, leading to underesti- seldom did so in their fi rst year (Covas 2002). mates of true survival (Cilimburg et al. 2002). Spending that period in the company of parents That might be particularly true for some weaver or helpers and communally roosting in the nest species in southern Africa, which were found to chamber is likely to provide extra protection have higher tendency for dispersal than other from potential causes of mortality, such as pred- passerines (Peach et al. 2001). Thus, our results ators and cold nights. Still, it is possible that the are be er termed “apparent” or “local” survival. apparently higher juvenile survival for species Still, Sociable Weavers in and around Kimberley with delayed dispersal is a spurious result, appear to be highly sedentary. Of 164 individu- because in these species local survival estimates als ringed in the nest and recaptured one year approach the true fi gure, which would not hap- later, only 6% had moved to another colony pen in species with higher juvenile dispersal. (Covas 2002). The trend is similar when also The relatively high juvenile survival in considering the adult population: only 6.9% of Sociable Weavers could also appear to have the birds ringed (n = 2,094) were recaptured at resulted because we measured survival for other colonies (Covas et al. 2002). Moreover, we some juveniles that, when fi rst marked, were up monitored 18 colonies out of 25; thus, many of to 120 days old. That could lead to infl ated sur- the birds that dispersed were recaptured. vival estimates if we had missed a period just The most unusual result was the fi nding that a er fl edging when mortality could have been juvenile survival was the same as that of adults. higher. That seems unlikely, however, because Juvenile survival in small birds is o en thought we found no diff erences in survival among to be about half the adult survival rate (e.g. Gill juveniles fi rst marked at diff erent ages, from 1995; cf. Baillie and McCulloch 1993), though 24 to 120 days. The youngest birds were ones that generalization is based mostly on studies that had just fl edged, and if there was a period of northern-temperate species and relatively of unusually high mortality just a er fl edging, few studies have estimated juvenile survival we should have detected diff erences in survival in songbirds using mark–recapture statistics. among the juvenile age classes. Postfl edging survival at southern latitudes is Interannual survival.―Our data indicated not well studied, but work on cooperatively constant survival during the study period. That breeding passerine species from Australia and was unexpected, because arid environments are South America (though not based on mark– considered to be highly fl uctuating. Moreover, recapture statistics) reported fi rst year survival several studies have reported temporal varia- probabilities ranging from 0.35 to 0.76 (Stacey tion in survival of passerines (e.g. Newton 1998), and Koenig 1990, Rowley and Russell 1991). though again that is mainly based on the study of In some of those studies (e.g. Campylorhyncus northern-temperate species. Temporal variations October 2004] Survival in Sociable Weavers 1205 in survival are most commonly a ributed to energetic demands for thermoregulation. Still, environmental fl uctuations aff ecting food levels, the Sociable Weaver’s nest mass and its habit exposure to bad weather, or density-dependent of communal roosting help the birds cope with mechanisms. It seems, therefore, that the environ- cold night-time temperatures by reducing the mental variation experienced by our population metabolic cost of thermoregulation (White et does not place any obvious constraints on sur- al. 1975). vival. In addition, survival in Sociable Weavers That rainfall did not aff ect survival in the was not aff ected by variation in reproductive population was surprising, because rainfall, eff ort, as measured by duration of the breeding through its eff ect on insect abundance and season (which varied from one year when no production of seeds, is believed to be the main breeding activity was detected to a continuous determinant of food availability in semi-arid nine-month breeding season). That result is regions (Maclean 1973c, Harrison et al. 1997, interesting, because it is contrary to the predomi- Lloyd 1999, Dean and Milton 2001), including nant view of avian life histories, which suggests our study area (M. Picker pers. comm.). Thus, that survival is mainly a consequence of repro- our results contrast with studies on northern- ductive eff ort through the reproduction–survival temperate birds, in which food availability dur- trade-off (e.g. Lack 1968, Martin 1987, Stearns ing winter has been shown to infl uence survival 1992). Hence, contrary to our fi ndings, survival (Jansson et al. 1981, Bri ingham and Temple should have decreased a er years of prolonged 1988, Newton 1998). The absence of a rainfall breeding activity. Of course, a problem with cor- eff ect suggests that, though food levels can vary relative studies such as the present one is that the greatly in the region, they do not represent an probability of detecting reproductive costs based important constraint on survival. Relatively on natural variation is low, if individuals adjust high food levels outside the breeding season reproductive eff ort to environmental conditions. have also been suggested as being responsible Therefore, further work is needed to establish the for high adult survival of birds in southern real relationship between reproductive eff ort and Africa (Peach et al. 2001) and Australia (Ford et survival in this species. al. 1988, Rowley and Russell 1991). Environmental factors.―Survival in Sociable That we did not fi nd signifi cant eff ects of age, Weavers did not seem to be aff ected strongly time, or environmental conditions on survival by winter temperature. Winter severity is o en could indicate that our data set either was too thought to be the main survival constraint for small or had other characteristics (such as not birds (e.g. Newton 1998), because persistent bad enough recaptures) to enable a strong test of weather may decrease foraging opportunities those eff ects. In classical statistical terms, this (Cuthill and Houston 1997) and food availabil- could be a case in which failure to reject a null ity (Jansson et al. 1981, Bri ingham and Temple hypothesis might not necessarily mean that the 1988, Newton 1998) or increase energetic null hypothesis was supported. Whether this is a demands for thermoregulation (McNamara serious issue—both for our study and others—is and Houston 1990, Cuthill and Houston 1997). unknown, because there is no sort of power One of the main food sources for Sociable analysis available for mark–recapture modeling. Weavers, the harvester termite (Hodotermes However, we note that our total sample size and mossambicus; Maclean 1973d), increases its percentage of individuals recaptured at least daily activity in winter when temperatures once were relatively large for studies of passerine are cooler (Richardson 1985, R. Adam pers. survival, which suggests that the present study comm.). Moreover, Sociable Weavers inhabit an was at least as likely as most others to detect any area where winter days are usually sunny, with eff ects on survival if they existed. temperatures >15°C. Therefore, an eff ect of bad weather on foraging opportunities may be neg- Ahps|qls ligible. Hence, it is possible that variations in Field assistance was provided by W. Sinclair, E. winter temperature do not play a signifi cant role Oosthuysen, R. Becker, E. MacFarlane, T. A. Anderson, in limiting food availability in our study area. R. A. Anderson, and B. Wilson-Aitchison from 1993 to However, cold winter nights (that can reach 1997; and C. Doutrelant, R. Visagie, M. P. Melo, A. Pierce, –8°C), could aff ect survival through increased H. Wright, T. Pontynen, and many other volunteers 1206 C q. [Auk, Vol. 121 from 1998 to 2001. We thank M. A. du Plessis, P. Lloyd, C, R. 2002. Life history evolution and coop- and R. Pradel for helpful discussion and advice; and erative breeding in the Sociable Weaver. E. Cam, D. Diefenbach, C. Doutrelant, W. P. Peach, Ph.D. dissertation, Percy FitzPatrick Institute, T. E. Martin, M. P. Melo, and W. Thompson for useful University of Cape Town, Cape Town, South comments on a previous version of the manuscript. Africa. T. E. Martin kindly gave us access to unpublished C, R., C. R. B|s, M. D. Ass, s material. De Beers Consolidated Mines provided access M. B. B|s. 2002. Stabilizing selection on to Benfontein and logistic support; further logistic sup- body mass in the Sociable Weaver Philetairus port was obtained from the Department of Tourism, socius. Proceedings of the Royal Society of Environment and Conservation (South Africa). London, Series B 269:1905–1909. The project was supported by the Open Research Cmqq, I. C., s A. I. Hs. 1997. Managing Programme of the South African National Research time and energy. Pages 97–120 in Behavioural Foundation through a grant to M. A. du Plessis and Ecology: An Evolutionary Approach (J. R. with funds from the Sandton Bird Club and Wildlife Krebs and N. B. Davies, Eds.). Blackwell, and Environment Society (Northern Cape Region). Cambridge, United Kingdom. R.C. was supported by Fundação para a Ciência e Ds, W. R. J., s S. J. Mqs. 2001. Responses Tecnologia, Portugal (Praxis XXI, BD11497/97). C.R.B. of birds to rainfall and seed abundance in the and M.B.B. were supported by the National Science southern Karoo, South Africa. Journal of Arid Foundation (DEB-9613638, DEB-0075199) during the Environments 47:101–121. preparation of this manuscript. We thank J. C. Ruiz for Pq, M. A., s J. B. Wqq. 1994. translating the abstract into Spanish. Communal cavity roosting in Green Woodhoopoes: Consequences for energy L C expenditure and the seasonal pa ern of mor- tality. Auk 111:292–299. App, H. 1973. Information theory and an exten- Eps, J., A. Bqs, s H. Tlq¾. 2000. sion of the maximum likelihood principle. Parental nepotism enhances survival of Pages 267–281 in International Symposium on retained off spring in the Siberian Jay. Information Theory (B. N. Petran and F. Csaki, Behavioral Ecology 11:416–420. Eds.). Akademiai Kiadi, Budapest, Hungary. F, H. A., H. Bqq, R. N, s R. Np. 1988. Bqq, S. R., s N. MhCqqhm. 1993. Modelling The relationship between ecology and the inci- the survival rates of passerines ringed during dence of cooperative breeding in Australian the breeding season from national ringing birds. Behavioral Ecology and Sociobiology and recovery data. Pages 123–140 in Marked 22:239–249. Individuals in the Study of Bird Population Gmqg, C. K., s T. E. Ms. 2001. (J.-D. Lebreton and P. M. North, Eds.). Fecundity–survival trade-off s and parental Birkhäuser Verlag, Basel, Switzerland. risk-taking in birds. Science 292:494–497. Bslm, M. C., s S. A. Tq. 1988. Impacts Gqq, F. B. 1995. Ornithology. W. H. Freeman, New of supplemental feeding on survival rates of York. Black-capped Chickadees. Ecology 69:581–589. Gs, D. J., s A. Chpgs. 2001. Post-fl edging Bsm, K. P., s D. R. Ass. 2002. care, philopatry and recruitment in Brown Model Selection and Inference: A Practical Thornbills. Journal of Animal Ecology 70: Information-theorectic Approach, 2nd ed. 505–514. Springer-Verlag, New York. Hs, J. A., D. G. Aqqs, L. G. Usmqq, M. Bsm, K. P., D. R. Ass, G. C. Wm, C. Hs, A. J. T, V. Pp, s C. J. B|s, s K. H. Pqqhp. 1987. Design B|s. 1997. The Atlas of Southern African and Analysis Methods for Fish Survival Birds. BirdLife South Africa, Johannesburg. Experiments Based on Release–Recapture. Jss, C., J. Eps, s A. V. B¾s. 1981. American Fisheries Society Monograph, no. 5. Winter mortality and food supply in Parus spp. Cmq|m, B. 1994. Evolution in Age- Oikos 37:313–322. structured Populations. Cambridge University Jmss, J. P., W. J. Phm, R. D. Gl, s Press, Cambridge, United Kingdom. S. A. Wm. 1997. Survival rates of tropical and Cqgl, A. B., M. S. Lsgl, J. J. T|pg, temperate passerines: A Trinidadian perspec- s S. J. Hoq. 2002. Eff ects of dispersal on tive. American Naturalist 150:771–789. survival probability of adult Yellow Warblers Lhp, D. 1968. Ecological Adaptations for Breeding (Dendroica petechia). Auk 119:778–789. in Birds. Methuen, London. Chm, E., s G. Wm. 1998. Program MARK: Lsls, T. A. 2000. Prolonged off spring depen- A gentle introduction. [Online.] Available at dence and cooperative breeding in birds. phidot.org/so ware/mark/docs/book. Behavioral Ecology 11:367–377. October 2004] Survival in Sociable Weavers 1207

Lgs, J.-D., K. P. Bsm, J. Cqg, s Pq, R., O. Gs, s J.-D. Lgs. D. R. Ass. 1992. Modelling survival and 2004. Principles and interest of GOF tests for testing biological hypotheses using marked multistate capture–recapture models. Animal animals: A unifi ed approach with case studies. Biodiversity and Conservation. 27: in press. Ecological Monographs 62:67–118. Rgsq, K. N. 1990. Campylorhyncus wrens: The Lq, P. 1999. Rainfall as a breeding stimulus and ecology of delayed dispersal and cooperation clutch size determinant in South African arid- in the Venezuelan savanna. Pages 159–196 in zone birds. Ibis 141:637–643. Cooperative Breeding in Birds (P. B. Stacey and Mhqs, G. L. 1973a. The Sociable Weaver, part W. D. Koenig, Eds.). Cambridge University 1: Description, distribution, dispersion and Press, Cambridge, United Kingdom. populations. Ostrich 44:176–190. Rhms, P. R. K. 1985. The social behaviour Mhqs, G. L. 1973b. The Sociable Weaver, part of the aardwolf, Proteles cristatus (Sparrman, 2: Nest architecture and social organisation. 1783) in relation to its food resources. Ph.D. Ostrich 44:191–218. dissertation, University of Oxford, Oxford. Mhqs, G. L. 1973c. The Sociable Weaver, part R|q, I., s E. Rqq. 1991. Demography of pas- 3: Breeding biology and moult. Ostrich 44: serines in the temperate Southern Hemisphere. 219–240. Pages 22–44 in Bird Population Studies: Mhqs, G. L. 1973d. The Sociable Weaver, part 5: Relevance to Conservation and Management (C. Food, feeding and general behaviour. Ostrich M. Perrins, J.-D. Lebreton, and G. J. M. Hirons, 44:254–261. Eds.). Oxford University Press, Oxford. Ms, T. E. 1987. Food as a limit on breed- Rqq, E. M. 2000. Avian life histories: Is ing birds: A life-history perspective. Annual extended parental care the southern secret? Review of Ecology and Systematics 18: Emu 100:377–399. 453–487. SŠm, B.-E. 1989. Survival rates in relation Ms, T. E. 1996. Life history evolution in tropi- to body weight in European birds. Ornis cal and south temperate birds: What do we Scandinavica 20:13–21. really know? Journal of Avian Biology 27: Sh, P. B., s W. D. Ksl. 1990. Cooperative 263–272. Breeding in Birds. Cambridge University Ms, T. E., P. R. Ms, C. R. Oqs, B. J. Press, Cambridge, United Kingdom. Hsl, s J. J. Fss. 2000. Parental Ss, S. C. 1992. The Evolution of Life Histories. care and clutch sizes in North and South Oxford University Press, Oxford. American birds. Science 287:1482–1485. Sqqs, K. A. 1989. Predation and starvation: MhN, J. M., s A. I. Hs. 1990. The Age-specifi c mortality in juvenile juncos (Junco value of fat reserves and the trade-off between phaenotus). Journal of Animal Ecology 58: starvation and predation. Acta Biotheoretica 275–286. 38:37–61. Wgs, R. W. M. 1974. Quasi-likelihood Msqms, J. M., s M. D. Ass. 1997. functions, generalized linear models, and Sociable Weaver Philetairus socius. Pages the Gauss-Newton method. Biometrika 61: 534–535 in The Atlas of Southern African Birds 439–447. (J. A. Harrison, D. G. Allan, L. G. Underhill, Wm, F. N., G. A. Bmq|, s T. R. M. Herremans, A. J. Tree, V. Parker, and H|qq. 1975. The thermal signifi cance of the C. J. Brown, Eds.). BirdLife South Africa, nest of the Sociable Weaver Philetairus socius: Johannesburg. Winter observations. Ibis 117:171–179. N|s, I. 1998. Population Limitation in Birds. Wm, G. C., s K .P. Bsm. 1999. Program Academic Press, London. MARK: Survival estimation from popula- Oq, T. B. 1982. The Starred Robin in Natal, part tions of marked animals. Bird Study 46 2: Annual cycles and feeding ecology. Ostrich (Supplement):120–139. 53:193–205. Wqq, G. C. 1966. Natural selection, the cost of Phm, W. P., D. B. Hs, s T. B. Oq. 2001. reproduction, and a refi nement of Lack’s prin- Do southern African songbirds live longer ciple. American Naturalist 100:687–690. than their European counterparts? Oikos 93: 235–249. Associate Editor: M. Bri ingham