Reciprocal Tradeoffs between Molt and Breeding in Albatrosses

Reciprocal Tradeoffs between Molt and Breeding in Albatrosses Author(s) :Sievert Rohwer, Anthony Viggiano and John M. Marzluff Source: The Condor, 113(1):61-70. 2011. Published By: Cooper Ornithological Society URL: http://www.bioone.org/doi/full/10.1525/cond.2011.100092 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 Condor 113(1):61 70 The Cooper Ornithological Society 2011 RECIPROCAL TRADEOFFS BETWEEN MOLT AND BREEDING IN ALBATROSSES SIEVERT ROHWER 1,3, ANTHONY VIGGIANO 2, AND JOHN M. MARZLUFF 2 1 Burke Museum and Department of Biology, University of Washington, Seattle, WA 98195 2 College of the Environment, University of Washington, Box 352100, Seattle, WA 98195 Abstract. Many large birds cannot replace all their flight feathers annually, creating potential life-history tradeoffs between breeding and molting. We show for the Black-footed Albatross (Phoebastria nigripes) that adults with overly worn primary flight feathers suffer reduced fledging success in the current breeding season and are likely to skip the next breeding season, even though they are still alive. If worn flight feathers affect breeding detrimentally, it follows that worn flight feathers accumulate because time for molt is limited. We show that individuals that spend more time breeding replace fewer flight feathers in the following molt. Thus in the Black-footed Albatross incomplete molts link more time invested in breeding to reduced feather quality, and feather quality predicts breeding success in the current season and breeding effort in the following year. Detrimental effects of flightfeather wear on breeding should be expected in other birds that cannot replace all their flight feathers annually. As in other albatrosses, the primaries of the Black-footed Albatross are replaced in an intrinsic molt cycle that in most but not all individuals results in most flight feathers being replaced at least every second year. Key words: nigripes. life- history tradeoffs, time constraints, worn feathers, molt, breeding success, large birds, Phoebastria Compromisos Recíprocos entre Muda y Anidación en los Albatros Resumen. Muchas aves de tamaño grande no pueden reemplazar todas sus plumas de vuelo cada año, creando potenciales soluciones de compromiso entre la anidación y la muda. Mostramos que para Phoebastria nigripes, los adultos con plumas primarias de vuelo demasiado usadas sufrieron una reducción en el éxito de emplumamiento en la presente estación de cría y tuvieron una alta probabilidad de saltarse la próxima estación de cría, aunque todavía están vivas. Si las plumas de vuelo usadas afectan de modo negativo la anidación, se deduce que las plumas de vuelo usadas se acumulan porque el tiempo de muda es limitado. Mostramos que los individuos que gastaron más tiempo anidando reemplazaron menos plumas de vuelo en la siguiente muda. Por lo tanto, en P. nigripes las mudas incompletas se vinculan con más tiempo invertido en la anidación para reducir la calidad de las plumas, y la calidad de las plumas predice el éxito de anidación en la estación actual y el esfuerzo reproductivo en el año siguiente. Los efectos negativos del uso de las plumas de vuelo en la anidación deberían esperarse en otras aves que no pueden reemplazar todas sus plumas de vuelo cada año. Como en otros albatros, las primarias de P. nigripes son reemplazadas en un ciclo de muda intrínseco que en la mayoría, pero no en todos, los individuos da como resultado que casi todas las plumas de vuelo son reemplazadas al menos cada año de por medio. INTRODUCTION Body size is an important determinant of life-history strategy across a wide range of plants and animals. In endotherms the pace of life generally slows as body size increases, with the result that time may become limiting in the lives of large birds and mammals (Calder 1984, Schmidt-Nielsen 1984, Brown 1995). In birds the time required to raise young increases with body size (Calder 1984), leaving large birds with less time in their annual cycle for other activities. This is an important conundrum if those other activities cannot be completed in the time away from breeding. Apart from breeding, no other component of the annual cycle of large birds demands more time than the regular replacement of flight feathers. Feathers wear out, and even the best-constructed flight feathers become severely degraded over several years of use. Thus flight feathers must be replaced regularly. Most birds minimize or completely avoid temporal overlap of flight-feather molt and breeding because missing flight feathers compromise flight performance (Tucker 1991, Swaddle and Witter 1997, Lind 2001). Birds that do overlap molt and breeding affirm the high cost of replacing flight feathers while breeding because their molts are particularly slow and protracted (Foster 1975, Rohwer et al. 2009b). Molt breeding tradeoffs are further exacerbated by the allometric relationships that describe the time required to replace flight feathers Manuscript received 7 May 2010; accepted 5 September 2010. 3 E-mail: rohwer@uw.edu. The Condor, Vol. 113, Number 1, pages 61 70. ISSN 0010-5422, electronic ISSN 1938-5422. 2011 by The Cooper Ornithological Society. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press s Rights and Permissions website, http://www.ucpressjournals.com/ reprintinfo.asp. DOI: 10.1525/cond.2011.100092 61

62 SIEVERT ROHWER ET AL. as a function of mass (M). The growth rate of the primary flight feathers scales as M 0.170, while primary length scales as M 0.314. Because large birds devote more time to breeding, they have less time available for molting than do small birds, yet because the growth rate of flight feathers increases with body size more slowly than does their length, large birds require disproportionately more time to renew their flight feathers than do small birds (Rohwer et al. 2009a). Large birds that breed seasonally face tradeoffs between molt and breeding when successful breeding and a complete replacement of the flight feathers cannot be accomplished in a year. For example, more time breeding implies less time for molt, which results in the accumulation of worn flight feathers. These worn feathers may affect future breeding detrimentally in two ways, either by reducing foraging success (and thus breeding success) in the next season or by forcing birds to invest so much more time in molt that they must skip the next breeding season. Thus more time spent rearing young means less time for molt, and, reciprocally, a longer molt that clears overly worn feathers from the wing may make breeding in the next season impossible. Ornithologists have long known that in large birds flightfeather molts are often incomplete (Stresemann and Stresemann 1966, Prevost 1983, Edelstam 1984, Prince et al. 1993). From a study of population samples Langston and Rohwer (1996) suggested that the accumulation of worn flight feathers, following a succession of incomplete molts, might force Laysan (Phoebastria immutabilis) and Black-footed (P. nigripes) albatrosses periodically to skip breeding to undertake a longer molt and clear worn feathers from their wings. This idea offered the first interpretation for why about 25% of successful breeders skip the following breeding season, despite the fact that most of these birds were known to be living and not to be breeding elsewhere (Rice and Kenyon 1962, Fisher 1976). Population assessments cannot critically evaluate the hypothesis that birds that skip breeding are doing so to clear overly worn feathers from their wings in a longer molt. Missing have been data assessing whether individual albatrosses with worn flight feathers have lower breeding success in the current season and whether individual albatrosses starting a current breeding season with worn flight feathers are more likely to skip the following breeding season. This study has three important results. First, we show that Black-footed Albatrosses have an intrinsic cycle of primary replacement, in which molt of most of the primaries is initiated in alternate years. Second, we show that a longer investment in breeding results in fewer primaries being replaced in the following molt. Finally, we show that individuals with two middle primaries in their third year of use have lower breeding success in the current season and are more likely to skip breeding the following season than are individuals in which these same two feathers are newer. DEFINITIONS We refer to primaries and secondaries, but not to rectrices, as flight feathers. In the albatrosses, the primaries are the 10 outermost (and longest) flight feathers of the wing. Proximal to the primaries are the secondaries, which in the long wings of the Black-footed Albatross vary in number from 28 to 31 (Edwards and Rohwer 2005). Primaries are numbered distally, starting with P1, and secondaries are numbered proximally, starting with S1, as illustrated in Langston and Rohwer (1995). The breeding season of the Black-footed Albatross crosses calendar years. Adults return to the Hawaiian Island breeding colonies in late October; the mean date of laying is 21 November, the mean date of hatching is 25 January, and chicks fledge about 140 days later in June (Rice and Kenyon 1962). To be precise about years we refer to the two seasons of our field work as the 1998/1999 and 1999/2000 seasons. The age of birds is important, and we start counting age in the year the chick hatched, not in the year its egg was laid. Thus a chick that hatched in 1994 would be called 5 years old in its 1998/1999 breeding season. NATURAL HISTORY OF MOLT IN ALBATROSSES Even though it is small compared to most albatrosses, the Black-footed Albatross has a long breeding cycle that leaves it little time to molt. All combined, the pre-laying period, incubation, and care of nestlings take about 245 days. Variation around this mean is likely high, as a sample of 54 Laysan Albatross chicks varied by 40 days in when they fledged (Rice and Kenyon 1962). Adults are away from their breeding colonies for only about 4 months of the year, which leaves them with little time for molt, depending on when they finish breeding and the length of the pre-laying phase. Males and females invest about equally in incubation and feeding chicks, and parental care ends when chicks go to sea (Rice and Kenyon 1962). Because albatrosses molt at sea during their nonbreeding period, the rules of feather replacement were largely unstudied in any albatross until the Burke Museum preserved extended wings from a collection of 191 Laysan and 117 Black-footed albatrosses salvaged from drift nets set for squid (Johnson et al. 1993). Of these specimens 91% were caught from June to October during the period of active molt. These species differ little in their rules of flight-feather replacement. The primaries are divided into two molt series (Langston and Rohwer 1995), the proximal of which also includes the four outer secondaries (Edwards and Rohwer 2005). The direction of feather replacement is proximal in the inner series and distal in the outer (Langston and Rohwer 1995). There is some doubt over whether the division between these molt series lies between P5/6 or between P6/7, partly because these primaries are replaced only every 2 to 3 years, resulting in little data on feathers in active growth, even with the large sample of molting wings in the Burke collection (Rohwer and Edwards 2006). In this paper we follow

RECIPROCAL TRADEOFFS BETWEEN MOLT AND BREEDING 63 Langston and Rohwer (1995) and consider the break to occur at P5/6, but this technicality has no effect on our results. Unlike other albatrosses (Prince et al. 1993, 1997), the Laysan and Black-footed replace P8 10 in virtually every molt (Langston and Rohwer 1995), apparently because these outer primaries drag on the sand of the atolls where the birds breed and become severely abraded as parents commute to incubate or feed young (Langston and Rohwer 1995). Edwards and Rohwer (2005) documented that the flight feathers of Laysan and Black-footed albatrosses are organized into four molt series, each requiring about the same amount of time to replace. The most proximal secondaries (but not all the feathers of this molt series) also become seriously abraded in a single breeding season and tend to be replaced every year (Edwards and Rohwer 2005). Intrinsic biennial patterns of flight-feather replacement have been documented in several albatrosses. Documenting an intrinsic molt pattern requires either access to birds of known age (as in this study) or following individuals through successive years to score feather ages (Prince et al. 1993, 1997). Unfortunately, the field work for this study preceded Edwards and Rohwer (2005) showing that the molt series containing the inner primaries extends through S4. Thus, in working only with the primaries, we failed to include all the feathers in the next-to-outermost molt series. METHODS Our study site was Tern Island, French Frigate Shoals (23 45 N, 166 15 W), a crescent-shaped atoll in the Hawaiian Islands National Wildlife Refuge located about 800 km northwest of the main Hawaiian Islands. Tern Island is the largest (14 ha) of ten sandy islets within the atoll (Amerson 1971), and approximately 1400 pairs of Black-footed Albatrosses breed there annually. Albatrosses were occasionally banded on Tern Island as early as 1963; since 1981 all chicks have been banded annually, and all unbanded adults have been banded since 1998 (U.S. Fish and Wildlife Service, unpubl. data). We studied actively incubating pairs during the 1998/1999 and 1999/2000 breeding seasons. To address potential intrinsic cycling in primary replacement, we studied only pairs in which at least one individual had been banded as a chick. We used pairs in which the age of both adults was known to assess whether Black-footed Albatrosses mated assortatively or disassortatively with respect to age. Nests were marked with plastic survey flags and checked daily until both adults were captured and scored for feather wear, after which time they were checked once a week. Adults were banded with a unique alphanumeric plastic color band. We treated the time invested in reproduction as small or large: small investors failed in incubation or early chick rearing; large investors fledged their chick or failed late in chick rearing. We used culmen length to establish sex because most males have a longer culmen and wider head than females (Frings and Frings 1961). By inspecting the cloaca shortly after egg laying we determined sex in 200 mated pairs. Within pairs, the male s culmen was larger than the female s 98.5% of the time. Following Furness (1988) we attempted to score primary wear into three year-classes: fresh feathers (score 1), replaced in the last molt, were in their first year of use; worn feathers (score 2), replaced in the molt before last, were in their second year of use; and very worn feathers (score 3), replaced three molts earlier, were in their third year of use. Fresh feathers were glossy and almost black, with little or no wear. Worn feathers were brown with moderate amounts of wear. Very worn feathers were light brown, chalky to the touch, heavily abraded, and typically had bleached spots. As explained below, P6 7 are the two primaries most likely to be used for 3 years, so we used just these two primaries to investigate the effect of feather wear on fledging success and on the probability of returning to breed the following season. Because our wear scores were assigned in December, during early incubation, they were little affected by the increased rate of wear in the outer primaries associated with chick rearing (Langston and Rohwer 1995). Furthermore, there could be no researcher-expectation bias because we scored feather wear before recording data on reproductive success. We used only the right wing in our analyses of feather wear; feather wear was similar in the two wings. STATISTICAL ANALYSES We used logistic regression to evaluate the effect of feather wear on breeding success in the current season and the effect of breeding effort, body condition, extent of primary replacement, and feather wear on the probability of returning to breed in the following season. We assume a normal distribution of errors and use standard maximum-likelihood estimation of parameters. We report Wald s statistic to appraise the significance of each factor in the logistic regression (formally testing the null hypothesis that 0). Because the intrinsic molt cycle largely explains feather-wear scores, we avoided problems of multicollinearity by not including data on the intrinsic molt cycle (odd or even years of age) and data on feather wear in the same model. All main-effect variables (breeding effort, body condition, and either feather wear or molt cycle) were entered into the model simultaneously. We entered interactions between main effects in our models individually and excluded them if the difference in log-likelihood statistics for models with and without the interaction term was nonsignificant (P 0.05; Hosmer and Lemeshow 1998). We used a general linear model to investigate how investment in the 1998/1999 breeding season affected the number of primaries replaced in the following molt. We included four factors in the model: level of investment in breeding, scored as small or large; hatch year, scored as odd or even (to account for the intrinsic molt cycle); body condition, measured as mass

64 SIEVERT ROHWER ET AL. divided by tarsus length; and sex. Green (2001) has cogently criticized this sort of measure of body condition, but we could not reanalyze our data in a more appropriate way. Our measure of body condition was nonsignificant in all of our analyses, so it did not affect other results, even though it is a poor index of condition. We present means with 1 standard error of the mean (SE) and ran all statistical analyses with SPSS (version 9). RESULTS BIENNIAL MOLT CYCLES Our access to many birds of known age enabled us to show that Black-footed Albatrosses molt their primaries in an intrinsic biennial cycle. Unlike other albatrosses (Prince et al. 1993, 1997), Black-footed Albatrosses replace their outer three primaries every year. But they still show a strong biennial cycle in the replacement of their inner seven primaries: in even years of age (birds 6, 8, or 10... years old) 229 adults replaced a mean of 4.93 ( 0.103) of their 10 primaries, while in odd years of age (birds 5, 7, or 9... years old) 219 adults replaced a mean of 8.62 ( 0.110) of their 10 primaries (Fig. 1). There is no suggestion that this biennial cycle is lost as birds age: it was as strong in 6- to 8-year-old birds as it was in 17- to 19-year-old birds. In even years when fewer primaries are replaced, P8 10 are replaced, but P7 and P6 often are not replaced; additionally, some primaries of the inner molt series may be replaced (Fig. 2). In odd years when more primaries are replaced, P8 10 are replaced again and P1 7 are usually replaced. P6 7 are the two primaries most likely to be carried for a third year (Fig. 2). There was no sex difference in the average number of primaries replaced annually. The 399 females scored in 1998/1999 replaced an average of 6.92 2.39 primaries, and FIGURE 1. Mean number of primaries Black-footed Albatrosses of known age had replaced in the molt preceding the 1998/1999 (n 440) and 1999/2000 (n 333) breeding seasons. P8 10 are replaced every year; in odd years the additional primaries replaced are usually among P1 5, while in even years they tend to be P6 7. FIGURE 2. The percent of primaries in the three categories of wear, by the birds years of age as even or odd, to show how the intrinsic molt cycle affects the condition of each of the ten primaries.

RECIPROCAL TRADEOFFS BETWEEN MOLT AND BREEDING 65 FIGURE 3. Age differences between members of breeding pairs of Black-footed Albatrosses; for age differences between 0 and 7 years there were 133 pairs. Pairs present in both seasons of study were counted only once. the 400 males replaced an average of 6.67 2.23 primaries. There was no sex difference in the number of primaries replaced (t 0.16, df 1386, P 0.87), and the number of primaries replaced did not differ significantly by year for either sex (females: t 0.45, df 289, P 0.65; males: t 1.85, df 298, P 0.07). In the 1999/2000 season 110 (17.7%) of the 623 birds scored had replaced all 10 primaries. The discovery of biennial molt patterns raises the question of whether the birds might mate assortatively with respect to odd or even years of age. If a Black-footed Albatross cannot always replace the requisite number of primaries to breed annually, assortative mating might help pairs synchronize their need to skip breeding seasons. Alternatively, disassortative mating might allow pair-mates to compensate for each other in foraging efficiency, the mate with better feathers foraging more. But we found that pairing was random with respect to the intrinsic biennial cycle of primary replacement (Fig. 3). Most pairs were the same age or one year different in age. Between mates differing in age by 0 to 7 years, the frequency of both individuals being odd or even in years of age was almost identical to the frequency of one being odd and the other even (sign test, P 0.50). In Figure 3 this is reflected in an absence of alternating peaks at odd or even differences in age between mates. BREEDING INVESTMENT, PRIMARY REPLACEMENT, AND BREEDING THE FOLLOWING YEAR In other albatrosses that breed annually, the time invested in breeding affects the number of primaries replaced in the following molt: successful breeders replace fewer feathers than breeders that fail early. When viewed in the context of the biennial molt cycle, this same pattern was apparent in the Blackfooted Albatross. In odd years of age, when more primaries FIGURE 4. Number of primaries replaced in odd and even years of life by breeding Black-footed Albatrosses, split by the amount of time invested in breeding. are replaced, birds that fledged a chick or failed in late chick rearing replaced an average of 0.90 fewer primaries than birds that failed during incubation or early chick rearing (Fig. 4; t 3.22, df 182, P 0.001). The contrast is in the same direction for small molts (even years of age), though the average difference of 0.45 primaries is only half as large and just misses significance (Fig. 4; t 1.91, df 207, P 0.06). The intrinsic molt cycle explained 39% of the variance in the number of primaries replaced (F 20,279 8.74; P 0.001), while breeding effort only explained 2% of this variance (F 1,279 7.09; P 0.008). Some individuals can fledge offspring and still renew all of their primaries before the following breeding season, something that Langston and Rohwer (1995) considered improbable on the basis of calculated time required to replace the entire outer molt series (P6 10), which comprises the longest feathers of the wing. Of the 623 birds that returned to breed in the 1999/2000 season, 110 had replaced all of their primaries, and 16% of these had successfully fledged a chick in the 1998/1999 season. Breeding investment also had a strong effect on a bird s likelihood of returning to breed the following year. Birds that were successful or that failed late in chick rearing were significantly less likely to return the following year (P returning 0.75 0.02; n 340) than birds that failed during incubation or early chick rearing (P returning 0.83 0.02; n 493; Wald s F 1 3.82, P 0.05). Given that successful breeders replace fewer primaries, this result suggests that successful breeders are sensitive to the need to invest more time in molt. EFFECTS OF WORN FEATHERS ON CURRENT AND FUTURE BREEDING Because foraging commutes during most of the chick-rearing period involve round trips of 1600 14 000 km (Fernandez et al. 2001), we evaluated the reproductive cost of flying with overly worn primaries. We restricted these analyses to

66 SIEVERT ROHWER ET AL. FIGURE 6. Probability of breeding in the 1999/2000 season compared to the summed scores of wear of primaries 6 and 7, scored at the beginning of the 1998/1999 season. Most birds that skip breeding are still alive. Wear scores of 2, 4, or 6 usually indicate that these two primaries are in their first, second, or third year of use, respectively. FIGURE 5. Probability of fledging compared to the summed wear scores for primaries 6 and 7. Wear scores of 2 indicate that both P6 and P7 were in their first year of use, while scores of 5 or more indicate that one or both was in its third year of use. P6 7 because these are the two primaries most likely to be in a third year of use and highly worn. High summed wear scores for P6 7 were significantly associated with a reduced probability of fledging young in both breeding seasons (Fig. 5). When P6 7 were both new (score of 2), or when one was new and one was in its second year of use (score of 3), breeding success was high. But when both were in their second year of use (score of 4), breeding success dropped, and when one or both was in its third year of use (score of 5 or more), the weighted likelihood of fledging a chick for the 2 years of this study fell to 60% below that of birds with P6 7 new or little worn. Did the degree of wear on P6 and P7, scored at the beginning of the 1998/1999 breeding season, affect the likelihood of birds returning to breed in the 1999/2000 season? About 20% of the birds that bred in 1998/1999 failed to breed in 1999/2000; 20% is considerably higher than the annual mortality rate, which is currently about 8% and surely was below 5% prior to the onset of fisheries-related mortality (Véran et al. 2007). Although our data were collected after mortality from fishing had become high, the difference between the rates of not breeding (20%) and annual mortality (8%) means we can still be confident that most of the 1999/2000 nonbreeders were alive but skipping a year of breeding. There appears to have been a threshold effect of primary wear on the probability of returning to breed (Fig. 6; Wald s statistic 3.82, P 0.05). Birds with summed wear scores for P6 7 ranging from 2 to 5 did not differ in their probability of returning to breed in the following season. For each of these four wear categories about 80% of the birds returned to breed the following season. But the likelihood of returning to breed dropped to about 70% when P6 and 7 were both in their third year of use (Fig. 6). Whether or not birds skip the next breeding season seems driven largely by females. Starting a season with P6 and 7 worn was a highly significant predictor of skipping the subsequent breeding season for females (Wald s statistic 7.48, P 0.006), but not for males (Wald s statistic 0.77, P 0.38). Does the phase of the intrinsic molt cycle that a bird is entering affect the likelihood of returning to breed? Birds in even or odd years of age had similar likelihoods of returning to breed in 1999/2000 (Wald s statistic 0.083, P 0.77). Furthermore, when we restricted the comparison to just those birds that had made a large investment in breeding, there was still no effect of the intrinsic molt cycle on the likelihood of returning to breed the following year ( 2 0.19, P 0.89). 1 There may be no reason to expect an effect of the intrinsic molt cycle on skipping breeding if the flight feathers are allocated into molt series such that about the same amount of time is required for all the feathers of each series to be replaced (Edwards and Rohwer 2005).

RECIPROCAL TRADEOFFS BETWEEN MOLT AND BREEDING 67 DISCUSSION As far as we are aware ours is the first demonstration of a reciprocal tradeoff between molt and breeding. In the Blackfooted Albatross worn flight feathers reduce fledging success in the current breeding season and increase the likelihood of skipping the following breeding season. Many lineages of large birds that fly while molting fail to replace all their primaries annually (Rohwer et al. 2009a). Thus, negative effects of feather wear on fledging success and the likelihood of breeding in the next season may be a common theme in the life-history strategies of large birds that cannot replace all their flight feathers annually. Although a diversity of large birds have incomplete flight-feather molts (Rohwer et al. 2009a), these results for the Black-footed Albatross apparently constitute only the fifth demonstration that more time invested in breeding leads to replacing fewer flight feathers in the following molt. This relationship has previously been documented for an owl (Pietiainen et al. 1984) and for four other albatrosses that breed annually (Harris 1973, Furness 1988, Prince et al. 1993, Cobley and Prince 1998). Because such data are the starting point for searches for reciprocal tradeoffs between molt and breeding, we very much need to know if other lineages of large birds with incomplete primary molts also replace fewer primaries following longer investments in breeding. The following sections discuss the four general results of this study in more detail. BIENNIAL MOLT CYCLES Biennial cycles of flight-feather replacement have now been documented in four species of albatrosses, representing two of the four major clades of the family, and two species each that breed annually and biennially (Furness 1988, Prince et al. 1993, Nunn et al. 1996, Prince et al. 1997). Because albatrosses are much too large to replace all their flight feathers annually in the time they have away from breeding (Rohwer et al. 2009a), we think the intrinsic biennial molt cycle evolved to insure that most primaries and secondaries are replaced at about equal frequency and about every 2 years. This complex pattern of molting the flight feathers requires that they be broken into multiple replacement series that can be activated in alternate years. Multiple replacement series and the pattern of primary replacement within these series have been documented only for the Laysan and Black-footed albatrosses, the only albatrosses in which large samples of birds in active feather replacement have been studied (Langston and Rohwer 1995, Edwards and Rohwer 2005). However, an alternating biennial pattern of replacement is well established for the primaries and secondaries of the Wandering (Diomedea exulans; Prince et al. 1997) and Black-browed (Thalassarche melanophris; Prince et al. 1993) and for the primaries (secondaries not reported) of the Gray-headed (T. chrysostoma; Prince et al. 1993) and Black-footed (this study). Excellent data on the frequency of feather replacement in the Wandering Albatross shows that in its first molts it replaces few flight feathers and that those replaced are the feathers of the wing that suffer the most from wear and ultraviolet light the outermost primaries and the innermost secondaries; by age 8, nonbreeders have begun to replace almost half of their primaries and secondaries each year (Prince et al. 1997). Wandering Albatrosses usually start to breed by age 10 (Croxall et al. 1990), and they are so large that they require more than a year to raise a chick, so they breed only every second year when they are successful. Failed breeders often attempt breeding annually, but failed breeders that attempt to breed the season after a failure must replace an average of six primaries to keep their primaries largely clear of feathers in their third year of use (Prince et al. 1997). The Laysan and Black-footed are the first albatrosses shown to replace their outer three primaries annually. Presumably they do so because these feathers become highly abraded after a single breeding season because their wing tips drag on the sand or water as birds return to their breeding atolls (Langston and Rohwer 1995). Breeding adults replace an average of almost eight primaries every 2 years (Fig. 1), so the intrinsic biennial molt cycle results in average adults replacing all primaries at least once every 2 years. The frequency of primary replacement we found for the Black-footed Albatross closely matches that reported by Langston and Rohwer (1995) for birds collected at sea, except that we found all adults to replace P8 10 every year. Our evidence that P8 is replaced every year is probably more reliable than Langston and Rohwer s (1995) conclusion that P8 is sometimes not replaced because our scores were assigned in early incubation, before seasonal variation in feather wear made the assignment of feather age difficult for Langston and Rohwer. The Short-tailed Albatross (Phoebastria albatrus) is another member of the clade containing the Laysan and Black-footed albatrosses (Nunn et al. 1996), and it would be interesting to know how its primaries are replaced. It currently breeds on grassy slopes (Hasegawa and DeGange 1980) that resemble those used by Southern Hemisphere species that replace their three outer primaries in alternate years. Take-off is easier from slopes, and grassy slopes are less abrasive than sand, so Short-tailed Albatrosses might not replace their three outer primaries annually. On the other hand, they may have bred on low-lying islands when their range included the North Atlantic (Olson and Hearty 2003). The flight feathers of Black-footed and Laysan albatrosses are organized into four molt series, illustrated by Edwards (2008), each of which takes about the same amount of time for complete replacement (Edwards and Rohwer 2005). In theory, this system makes it possible for them to activate molt in just two series per year and have all their primaries replaced every 2 years. But feather replacement seems more sophisticated because birds replace feathers in any of these

68 SIEVERT ROHWER ET AL. series in any year (Fig. 2; Edwards 2008); thus some mechanism may enable them to replace feathers that are damaged or unusually worn, regardless of the biennial cycle. In the Blackfooted Albatross molt is also so time constrained by reproduction that some individuals fail to replace all their flight feathers every 2 years; these individuals either skip breeding or breed with worn feathers and suffer reduced success. BREEDING AFFECTS MOLTING In small birds that undergo a complete annual molt, increased investment in breeding often delays the start of molt, especially for females, (Svensson and Nilsson 1997, Hemborg 1999). In large birds that cannot replace all of their flight feathers annually, increased investment in breeding reduces the number of primaries replaced (Harris 1973, Pietiainen et al. 1984, Furness 1988, Prince et al. 1993, Cobley and Prince 1998). Our data show the latter to be true for the Black-footed Albatross, but the effect is small and statistically significant only in even years of age when more primaries are replaced. In the five species of annually breeding albatrosses that have been examined, failed breeders replace only about one more primary than successful breeders. This small difference is likely due to the in albatrosses intrinsic molt cycle. In contrast, the Ural Owl (Strix uralensis), which is not known to have an intrinsic molt cycle, replaces 12.2 primaries and secondaries per wing after a failed attempt to breed but only 7.9 after a successful attempt (Pietiainen et al. 1984). In the Laysan and Black-footed albatrosses, P6 and P7 are replaced more frequently, on average, than are P1 5, raising the question why P6 7 are ever used for a third year (Fig. 2). The answer is that the outer series of primaries (P6 10) requires the most time to replace, and birds initiate molt at P7 or at P6 only if they have more time to molt (Langston and Rohwer 1996). Thus birds that initiate molt late tend not to replace one or both of these primaries. Because they are outer primaries, P6 7 wear rapidly, making them especially useful in assessing the effect of flight-feather wear on current and future breeding. In small passerines the currency through which greater investment in one breeding season reduces breeding success in the following season appears to be the quality of the feathers generated in the intervening molt. Pairs of the Blue Tit (Cyanistes caeruleus) that were experimentally forced to renest molted later, survived the winter poorly, and nested late the following season (Nilsson and Svensson 1996). Remarkably, their basal metabolic rate was very high in cold conditions, suggesting that their late molt had generated a plumage with poor insulating properties. Similarly, captive European Starlings (Sturnus vulgaris) forced to molt faster by experimental reduction of the photoperiod grew primaries of lower initial mass that wore out faster than those of controls (Dawson et al. 2000). While these results suggest that feather quality is likely to be the currency linking higher investment in breeding in one year with lower success in the following year, an effect of feather quality on breeding success has never been assessed experimentally so that the effects of individual quality and feather quality can be disentangled. PLUMAGE QUALITY AFFECTS BREEDING If feather quality is the currency linking tradeoffs between seasons in reproductive performance, then birds with worn feathers should have reduced breeding success. Apparently this has been addressed only once before, and the results were opposite of predicted: Collared Flycatchers (Ficedula albicollis) with worn primaries had higher, not lower, breeding success (Merila and Hemborg 2000). Because the authors scored feather wear near the end of the nestling period, this result is easily explained by phenotypic plasticity: birds in good condition may invest more in breeding, thus wearing their feathers more and also fledging more young. Laysan and Black-footed albatrosses that fail to replace all their primaries every 2 years usually fail to replace P6 or P6 and 7 (Langston and Rohwer 1996). Thus we related the condition of these two primaries at the beginning of a breeding season to success in that season and to the likelihood of attempting to breed in the following year. When the combined score of these feathers wear was 5 or more, indicating that one or both of these primaries was entering a third year of use, the probability of the chick fledging was about 60% lower than that for individuals in which these same primaries were less worn. Apparently, the cost of flying with worn feathers severely impairs parental care in this species, which regularly travels 1600 to 14 000 km to forage during chick rearing (Fernandez et al. 2001). However, without experimental abrading of primaries, the causal effect of feather quality cannot be distinguished from an individual s quality. Starting a current breeding season with primaries 6 and 7 highly worn greatly reduced the likelihood of breeding the following season. Again, experimental feather abrasion is needed to establish causality. This effect was strong for females but not for males, suggesting that males may to be forced to follow the lead of their females or risk losing their mates. In the closely related Laysan Albatross a larger percentage of males (67% of 168) than females (48% of 124) that lost their mates missed the following breeding season (Fisher 1976). This difference suggests an excess of males in the population, in turn suggesting that a male that fails to breed with his female, regardless of the state of his flight feathers, might lose her to another male. Langston and Rohwer (1996) speculated that pairs might be synchronized in their molts to overcome the problem of one but not the other member of a pair needing to skip a breeding season to replace overly worn flight feathers. Mating assortatively with regard to the intrinsic molt cycle in the primaries might help achieve this, but Black-footed Albatrosses mate randomly with respect to their intrinsic molt cycle (Fig. 3). If the function of the intrinsic molt cycle is to ensure that all the

RECIPROCAL TRADEOFFS BETWEEN MOLT AND BREEDING 69 flight feathers are replaced as they are wearing out, which is about biennially for most of the wing, then there is no reason to expect mating to be related to the intrinsic molt cycle. However, assortative mating by quality, which is the rule in most animals, should result in Black-footed Albatross pairs tending to be alike in whether or not they can replace enough flight feathers in the time available for molt to breed the next year. FEATHER WEAR AND BREEDING Interconnected lines of evidence suggest that Black-footed Albatrosses are very sensitive to worn primaries. First, just two primaries in their third year of use greatly reduced fledging success in the current breeding season and increased the likelihood of skipping the following season. Second, Black-footed Albatrosses that invested more in breeding were less likely to breed the following year. This skipping is likely driven by a need to avoid accumulating worn primaries because birds that initiate molt later are less likely to replace P6 and P7 (Langston and Rohwer 1996). If P6 7 were already in their second year of use in a year the bird breeds successfully, then replacing them before the next attempt to breed seems especially important because breeding success is low for birds that breed with these feathers in a third year of use. All this apparent sophistication suggests that albatrosses may be sensitive to the condition of individual feathers. In chickens the follicles of the primaries and secondaries are lined with slowly adapting mechanoreceptors that discharge when these flight feathers are manually displaced in any direction (Brown and Fedde 1993), suggesting they might be sensitive to wear-related fluttering. Slowly adapting receptors in the follicles of primaries and secondaries suggest a neurophysiological link making it possible for birds that cannot replace all their flight feathers annually to replace worn flight feathers preferentially in their next molt. Experiments are needed. Would albatrosses preferentially replace experimentally abraded feathers? And, would females with experimentally abraded feathers tend to skip the following breeding season? ACKNOWLEDGMENTS This research was inspired by the collection of albatross wings at the Burke Museum. The U.S. Fish and Wildlife Service, Hawaiian Islands National Wildlife Refuge, provided financial and logistic support, with Chuck Monnett, Beth Flint, and Dave Johnson being especially helpful. Staff of the Tern Island Field Station and assistants Laura Carsten, Frans Joula, Lisa DeAmatio, Brandan Courtot, and Karen Fischer were especially helpful in the field. Lab mates Roarke Donnelly and Stephanie Zador helped with discussion and insight, and two anonymous reviewers provided helpful comments. Thanks to all. LITERATURE CITED AMERSON, B. A. 1971. The natural history of French Frigate Shoals, northwestern Hawaiian Islands. Atoll Research Bulletin 150. BROWN, J. H. 1995. Macroecology. University of Chicago Press, Chicago. BROWN,R.E., AND M. R. FEDDE. 1993. Airflow sensors in the avian wing. Journal of Experimental Biology 179:13 30. CALDER,W.A.I. 1984. Size, function, and life history. Harvard University Press, Cambridge, MA. COBLEY, N.D., AND P. A. PRINCE. 1998. Factors affecting primary molt in the Gray-headed Albatross. Condor 100:8 17. CROXALL, J. P., P. ROTHERY, S. P. C. PICKERING, AND P. A. PRINCE. 1990. Reproductive performance, recruitment and survival of Wandering Albatrosses Diomedea exulans at Bird Island, South Georgia. Journal of Animal Ecology 59:775 796. DAWSON, A., S. A.HINSLEY, P. N. FERNS, R.H. C. BONSER, AND L. ECCLESTON. 2000. Rate of moult affects feather quality: a mechanism linking current reproductive effort to future survival. Proceedings of the Royal Society of London B 267:2093 2098. EDELSTAM, C. 1984. Patterns of moult in large birds of prey. Annales Zoologici Fennici 21:271 276. EDWARDS,A.E. 2008. Large-scale variation in flight feather molt as a mechanism enabling biennial breeding in albatrosses. Journal of Avian Biology 39:144 151. EDWARDS,A.E., AND S. ROHWER. 2005. Large-scale patterns of molt activation in the flight feathers of two albatross species. Condor 107:835 848. FERNANDEZ, P., D. J. ANDERSON, P. R. SIEVERT, AND K. P. HUYVAERT. 2001. Foraging destinations of three low-latitude albatross (Phoebastria) species. Journal of Zoology, London 254:391 404. FISHER, H.I. 1976. Some dynamics of a breeding colony of Laysan Albatrosses. Wilson Bulletin 88:121 142. FOSTER, M. S. 1975. The overlap of molting and breeding in some tropical birds. Condor 77:304 314. FRINGS, H., AND M. FRINGS. 1961. Some biometric studies on the albatrosses of Midway Atoll. Condor 63:304 312. FURNESS, R.W. 1988. Influence of status and recent breeding experience on the moult strategy of the Yellow-nosed Albatross Diomedea chlororhynchos. Journal of Zoology, London 215: 719 727. GREEN, A. J. 2001. Mass/length residuals: measures of body condition or generators of spurious results? Ecology 82:1473 1483. HARRIS, M. P. 1973. The biology of the Waved Albatross Diomedea irrorata of Hood Island, Galapagos. Ibis 115:483 510. HASEGAWA, H., AND A. R. DEGANGE. 1980. The Short-tailed Albatross, Diomedea albatrus: its status, distribution and natural history. American Birds 36:806 814. HEMBORG, C. 1999. Sexual differences in moult breeding overlap and female reproductive costs in Pied Flycatchers, Ficedula hypoleuca. Journal of Animal Ecology 68:429 436. HOSMER, D.W., AND S. LEMESHOW. 1998. Applied logistic regression. Wiley, New York. JOHNSON, D.H., T.L. SCHAFFER, AND P. J. GOULD. 1993. Incidental catch of marine birds in North Pacific high seas driftnet fisheries in 1990. International North Pacific Fisheries Commission Bulletin 53:473 483. LANGSTON,N.E., AND S. ROHWER. 1995. Unusual patterns of incomplete primary molt in Laysan and Black-footed Albatrosses. Condor 97:1 19. LANGSTON,N.E., AND S. ROHWER. 1996. Molt breeding tradeoffs in albatrosses: life history implications for big birds. Oikos 76:498 510. LIND, J. 2001. Escape flight in moulting Tree Sparrows (Passer montanus). Functional Ecology 15:29 35. MERILA, J., AND C. HEMBORG. 2000. Fitness and feather wear in the Collared Flycatcher Ficedula albicollis. Journal of Avian Biology 31:504 510. NILSSON, J.-A., AND E. SVENSSON. 1996. The cost of reproduction: a new link between current reproductive effort and future reproductive

70 SIEVERT ROHWER ET AL. success. Proceedings of the Royal Society of London B 263:711 714. NUNN, G. B., J. COOPER, P. JOUVENTIN, C. J. R. ROBERTSON, AND G. G. ROBERTSON. 1996. Evolutionary relationships among extant albatrosses (Procellariiformes: Diomedeidae) established from complete cytochrome-b gene sequences. Auk 113:784 801. OLSON, S. L., AND P. J. HEARTY. 2003. Possible extirpation of a breeding colony of Short-tailed Albatross (Phoebastria albatrus) on Bermuda by Pleistocene sea-level rise. Proceedings of the National Academy of Sciences 100:12825 12829. PIETIAINEN, H., P. SAUROL, AND H. KOLUNEN. 1984. The reproductive constraints on moult in the Ural Owl Strix uralensis. Annales Zoologici Fennici 21:277 281. PREVOST, Y. 1983. The moult of the Osprey Pandion haliaetus. Ardea 71:199 209. PRINCE, P. A., S. RODWELL, M. JONES, AND P. ROTHERY. 1993. Moult in Black-browed and Grey-headed Albatrosses Diomedea melanophris and D. chrysostoma. Ibis 135:121 131. PRINCE, P. A., H. WEIMERSKIRCH, N.HUIN, AND S. RODWELL. 1997. Molt, maturation of plumage and ageing in the Wandering Albatross. Condor 99:58 72. RICE,D.W., AND K. W. KENYON. 1962. Breeding cycles and behavior of Laysan and Black-footed Albatrosses. Auk 79: 517 567. ROHWER, S., AND A. E. EDWARDS. 2006. Reply to Howell on primary molt in albatrosses. Western Birds 37:245 248. ROHWER, S., R. E. RICKLEFS, V.G.ROHWER, AND M. M. COPPLE. 2009a. Allometry of the duration of flight feather molt in birds. PLoS Biology 7:e1000132. ROHWER,V.G., S. ROHWER, AND M. F. ORTIZ-RAMIREZ. 2009b. Molt biology of resident and migrant birds of the monsoon region of west Mexico. Ornitología Neotropical 20:565 584. SCHMIDT-NIELSEN, K. 1984. Scaling: why is animal size so important? Cambridge University Press, New York. STRESEMANN,E., AND V. STRESEMANN. 1966. Die Mauser der Vogel. Journal für Ornithologie 107:1 447. SVENSSON, E., AND J.-A. NILSSON. 1997. The tradeoff between molt and parental care: a sexual conflict in the Blue Tit? Behavioral Ecology 8:92 98. SWADDLE, J. P., AND M. S. WITTER. 1997. The effects of molt on the flight performance, body mass, and behavior of European Starlings (Sturnus vulgaris): an experimental approach. Canadian Journal of Zoology 75:1135 1146. TUCKER, V. A. 1991. The effect of molting on the gliding performance of a Harris Hawk (Parabuteo unicinctus). Auk 108:108 113. VÉRAN, S., O. GIMENEZ, E.FLINT, W.L.KENDALL, P. F. DOHERTY, AND J.-D. LEBRETON. 2007. Quantifying the impact of longline fisheries on adult survival in the Black-footed Albatross. Journal of Applied Ecology 44:942 952.