Behavioral Ecology doi:10.1093/beheco/arl080 Advance Access publication 23 November 2006 Capital or income breeding? A theoretical model of female reproductive strategies Alasdair I. Houston,a Philip A. Stephens,b Ian L. Boyd,c Karin C. Harding,d and John M. McNamarab School of Biological Sciences, University of Bristol, Woodland Road, Bristol, BS8 1UG, UK, b Department of Mathematics, University of Bristol, University Walk, Bristol, BS8 1TW, UK, cSea Mammal Research Unit, Gatty Marine Laboratory, University of St Andrews, St Andrews, Fife, KY16 8LB, UK, and d Department of Marine Ecology, Göteborg University, Box 461, 405 30 Göteborg, Sweden a Energy storage is an important component of life-history variation. A distinction is recognized between species that provision offspring using energy gained concurrently (income breeders) and those that provision offspring using energy stores accumulated at an earlier time (capital breeders). Although this distinction has been recognized for some time, surprisingly little attention has been paid to the general adaptive value of the 2 strategies. Here, we present a simple, general framework for modeling female reproductive strategies. We show that our framework can be applicable either to annual breeders that aim to maximize the energy delivered to their offspring before independence, or to species with shorter reproductive cycles that aim to maximize reproductive rate, given that their offspring must build up a given level of reserves before independence. For both scenarios, we show that the costs of accumulating capital can lead to pure income breeding, pure capital breeding, or a mixture of the 2 strategies. Our model allows the effects of a variety of parameters to be assessed. Length of gestation, offspring metabolism, efficiency of energy transfer from mother to offspring, and the relative rates of energy gain by females with and without offspring are all important factors. The cost associated with accumulated capital is a particularly critical determinant of the strategy adopted. More detailed approaches to specific systems may provide a greater understanding of the factors promoting different maternal strategies for offspring provisioning. Key words: blubber, energy storage, lactation, mass-dependent costs, optimal behavior. [Behav Ecol 18:241–250 (2007)] ife-history theory is concerned with how an animal should allocate resources to reproduction over its life in order to maximize fitness (e.g., Roff 1992; Stearns 1992; Roff 2002). The storage of energy represents an important component of life-history variation. One aspect of energy storage that has generated particular interest is the use of stored energy for reproduction. A distinction is recognized between income breeders, in which reproduction is financed using current energetic income, and capital breeders, in which compensatory feeding takes place in advance of breeding, so that reproduction may be financed from stored energetic capital (e.g., Thomas 1988; Stearns 1992). In practice, a range of strategies is possible, from pure income breeding at one extreme to pure capital breeding at the other extreme (Thomas 1988). The potential costs and benefits of income and capital breeding have been reviewed from 2 different perspectives ( Jönsson 1997; Bonnet et al. 1998). There is a tendency to view income breeding as the preferred strategy in endotherms and capital breeding as a necessary adaptation to adverse conditions. This is because of the costs of maintaining and converting stores, as well as the dangers of error in prior estimates of required capital (Jönsson 1997). By contrast, many ectotherms are predisposed toward storing energetic capital and so, for these species, capital breeding strategies may be more energetically efficient than income breeding strategies (Bonnet et al. 1998). Although this dichotomy has general support, there is clearly a great deal of variation among both endotherms and ectotherms. Differing modes of transport and var- L Address correspondence to P.A. Stephens. E-mail: philip.stephens@ bristol.ac.uk. Received 27 March 2006; revised 10 October 2006; accepted 16 October 2006. The Author 2006. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: [email protected] iation in the thermal stress of environments can both affect whether capital or income breeding is more economical in either type of organism. In low productivity environments, food availability can also limit the feasibility of income breeding if potential rates of energy delivery to offspring are low. Although there have been a number of taxon-specific treatments of the behavior and physiology of various capital breeders (e.g., Festa-Bianchet et al. 1998; Bonnet et al. 1999; Beck, Bowen, Iverson 2003; Beck, Bowen, McMillan, Iverson 2003), there have been far fewer general attempts to examine the adaptive value of capital and income breeding strategies. To date, all such attempts have been limited to the order Pinnipedia (Boyd 1998, 2000; Schulz and Bowen 2005; Trillmich and Weissing 2006). Pinniped species show several strikingly different strategies for offspring provisioning, broadly divided along phylogenetic lines (but see further in Discussion) (Bonner 1984; Oftedal et al. 1987; Boness and Bowen 1996). Among these, female true seals (family Phocidae) tend to be relatively large, with mothers fasting throughout the weaning period and relying entirely on stored capital to provision their pups; thus, female phocids are often cited as classic examples of capital breeding. By contrast, female fur seals and sea lions (family Otariidae) tend to be smaller, and employ a foraging cycles strategy, in which mothers intersperse bouts of lactation with long foraging trips, thus provisioning pups from current income; as a result, female otariids are often cited as classic income breeders. Previous models (Boyd 1998; Trillmich and Weissing 2006) and correlative approaches (Boyd 2000) have shown support for the hypothesis that body mass is likely to have a strong influence on female strategies. However, the focus on body mass alone cannot explain the pinniped dichotomy, owing to overlap in size between capital breeding phocids and income breeding otariids (Schulz and Bowen 2005; Trillmich and Weissing 2006). Consequently, it seems likely that a more general approach, Behavioral Ecology 242 focusing less on body mass, will allow other factors to be identified for further investigation (see also Schulz and Bowen 2005). Here, we present a simple model framework that allows for different rates of energetic gain depending on whether or not the female is caring for offspring. Differences may arise because the presence of offspring reduces foraging efficiency or increases energetic demands or because it is necessary to raise offspring in an area of relatively low food availability. For clarity of description, our model is framed in terms of mammals, with a gestation period and a period of lactation. We return to this point in the Discussion, where we consider its potential application to, and implications for, other taxa. Previous approaches have used energy delivery rate to offspring as the sole criterion for optimization (Boyd 1998; Trillmich and Weissing 2006). Although this is a key aspect of the model that we present, our framework allows for different implications arising from different ultimate optimization criteria. Here, we focus on 2 different criteria: maximization of the rate of offspring production and maximization of the quality of offspring produced (see further below). Using either criterion, we show that relative rates of maternal energy gain before and after giving birth can predict pure income breeding, pure capital breeding, or strategies intermediate between the 2. Our formulation also allows the effects of other life-history parameters to be assessed. THE MODEL Basic assumptions and structure Let t be time in years and t ¼ 0 be the time at which the previous offspring achieves independence (all model notation is summarized in Table 1). For simplicity, we consider the case where a female cannot become pregnant until her previous offspring is independent and where energy gain rate and metabolism are independent of t. Beyond this, the model rests on 5 basic assumptions as follows. (i) Females will seek to maximize the speed at which offspring energy reserves increase. This will be true regardless of whether the female aims to minimize the time taken to produce offspring with a given level of reserves or to maximize the level of her offspring’s reserves within a given time interval. (ii) The rate at which a female can accumulate energy reserves will be affected by her current level of reserves. This will be true where current reserves (e.g., fat stores) have an impact on metabolism, behavior, or foraging ability. Specifically, where current reserves (above some baseline, viable, nonpregnant level) are denoted x, we consider the case where the net rate of energy gain, G(x), is a decreasing function of current reserves (Figure 1A). It is assumed that G(x) is a decreasing function of x because carrying reserves becomes increasingly costly. We refer to these costs (of carrying capital reserves) as mass-dependent costs. Their principal outcome is that G#(x) , 0. We denote the time taken to raise a given level of reserves, x, as R x dy t̂ðxÞ ¼ y¼0 GðyÞ : (iii) Immediately after birth, all available capital is transferred to the offspring. Transfer of energy from mother to offspring need not be completely efficient, so is modeled as occurring with efficiency a. For simplicity, transfer of capital is assumed to be instantaneous (we return to this point in the Discussion). (iv) Female net income between birth and offspring independence is not dependent on reserves. This is because all available capital is transferred to the offspring immediately after birth and, thus, the female returns to her baseline state of reserves (x ¼ 0) at that time. She maintains this level of reserves by transferring all income (in excess of her own metabolic needs) Table 1 Summary of notation used Notation Description t x t̂ðxÞ G(x) Time in years (t ¼ 0 is assumed to be the time at which a female is free to begin a reproductive attempt). Female energetic reserves, above a viable, lean, nonpregnant level. Time taken to raise reserves from zero to x. A female’s net rate of energy gain during periods when she is not provisioning offspring and can, potentially, accumulate reserves. This is assumed to be a decreasing function of current reserves (see model assumption [ii]). A female’s net rate of energetic income when foraging as an income breeder. This is assumed to be independent of current reserves as the female is assumed to maintain herself at a baseline level (x ¼ 0) while income breeding (see model assumption [iv]). Efficiency with which female reserves can be transferred to her offspring. Offspring metabolic rate. The point at which net rate of energy transfer to offspring that results from capital accumulation (by the mother) before birth is equal to the rate achieved by income breeding. The time taken to raise x* reserves (i.e., t * ¼ t̂ ðx *Þ). The minimum reserves required by the female at the time of birth, comprising the energy content of the offspring at birth, together with any nonretrievable energy in tissues used to support the pregnancy. The time taken to raise x0 reserves (i.e., t0 ¼ t̂ ðx0 Þ). The fixed duration of gestation. Required offspring energy reserves at independence (in the rate maximization model). A female’s initial net rate of energy gain during periods when she is not provisioning offspring; cF ¼ G(0). The proportion by which central place foraging (when caring for offspring) reduces the surplus rate of energy gain in females with offspring relative to cF. Parameter describing the strength of mass-dependent costs. The degree of capital breeding, here defined as K ¼ C/(C 1 I), where C is the increase in offspring energy as a result of transfer from maternal capital and I is the increase in offspring energy as a result of provisioning from maternal income. cL a c x* t* x0 t0 T E cF q k K Houston et al. • Modeling capital and income breeding Figure 1 Optimal accumulation of reserves in the absence of other constraints. (A) Increasing reserves lead to decreases in the rate of energy accumulation, so female reserves, x, show a declining rate of increase with time (as shown by the black line). If the female gives birth at any time, t, she will pass her stored capital to her offspring with efficiency, a. Consequently, the value of capital (in terms of energy available for the offspring) must be discounted by a (as indicated by the gray line). We term this discounted value of stored capital ‘‘effective offspring reserves.’’ (B) If the female pursued an income breeding strategy from the outset (giving birth at time t ¼ 0), offspring energy reserves would increase as shown, at the rate acL c. (C) When female reserves are x*, the rate at which effective offspring reserves will increase as a result of delaying birth (and continuing to accumulate capital) is equivalent to the rate at which offspring reserves will increase as a result of provisioning from income; that is, aG(x*) ¼ acL c, so the gradients of the effective capital curve (gray line) and effective income function (black line) are equal. Ignoring other constraints, this is the point at which the female should give birth, in order to maximize the speed at which offspring energy reserves increase. The time taken to raise this optimal level of reserves is denoted t*. to her offspring. We denote this net rate of income, cL. Transfers from income are also assumed to occur with efficiency a. The rate at which offspring reserves increase between birth and independence is thus acL c, where c is the offspring’s metabolic rate (Figure 1B). (v) The female can adjust the time at which she gives birth, for example, by mating at a given time or by delaying implantation. These 5 assumptions dictate that a female will adjust the timing of her offspring’s birth in order to maximize offspring energy reserves at any time between birth and independence. To illustrate this point, consider Figure 1. The gray line in 243 Figure 1A shows the increase in effective offspring reserves that results from a female accumulating capital until any time, t. The initial gradient of this line is aG(0), where G(0) is the initial rate of surplus accumulation for a female with no capital reserves. Figure 1B shows the rate at which offspring reserves increase as a result of giving birth and provisioning from income. This function has gradient acL c [see model assumption (iv)]. Given that the constraints imposed by caring for offspring are likely to have a negative effect (if any) on the efficiency with which the female forages, it follows that cL G(0). This suggests that the initial rate at which energy accumulated as capital can be translated into offspring reserves will be higher than the rate at which surplus energetic income can be translated into offspring reserves, that is, aG(0) . acL c. As the female accumulates capital, so her rate of capital accumulation decreases (i.e., G#(x) , 0). Owing to the asymptotic nature of the capital accumulation curve, G(x) decreases to zero. It follows, therefore, that there will be some x at which aG(x) , acL c. The consequence of this is that aG(x), the function that dictates how energetic capital can be translated into offspring reserves at birth, has an initial rate of increase that is greater than, but a subsequent rate of increase that is less than, the rate at which offspring reserves increase from income provisioning. The crucial point then is that there will be an intermediate value of reserves, x*, at which the rate of increase of effective offspring reserves resulting from the female accumulating further capital before birth is equivalent to the rate of increase of offspring reserves if the female gives birth and switches to an income breeding strategy; i.e., aG(x*) ¼ acL c. The time taken to raise x* reserves is t̂ðx *Þ ¼ t *; and this is the point at which income provisioning becomes more efficient than further accumulation of capital. Ignoring any other constraints, this will be the optimal point at which the female should give birth (Figure 1C). Temporal constraints on optimality Clearly, the basic model is highly simplified. In particular, it is extremely unlikely that a female will be free to give birth at any time from immediately after her previous offspring has attained independence. In light of this, we introduce 2 further temporal requirements. First, we assume that a female must have reserves of at least x0 at term, otherwise her offspring dies. At the time of birth, the female loses x0 reserves (x0 comprises the energy content of the offspring at birth, together with any nonretrievable energy in tissues used to support the pregnancy). The time taken to raise x0 reserves is t̂ðx0 Þ ¼ t0 : Second, we assume that there is some fixed gestation time, T, during which the female can accumulate reserves xT. When T depends on aspects of fetal development, only weakly related to maternal energy availability (e.g., skeletal or neuronal development), then T may be unrelated to t0. Timing of birth, timing of implantation (ti), and maternal strategy for increasing offspring reserves can all be shown to depend on the relative magnitudes of the 3 time periods: T, t0, and t* (Figure 2). Optimization criteria and maternal strategy Although maximizing the rate of increase of offspring energy reserves is a key component of the model, we consider 2 ultimate optimization criteria that both provide a context for that aim but which have subtly different implications for the degree of capital and income breeding. In the first, which we term ‘‘rate maximization,’’ offspring must accumulate some fixed threshold level of reserves, E, prior to gaining independence. We assume that females will seek to minimize the time Behavioral Ecology 244 taken to raise their offspring to independence, either: 1) because in a seasonal environment, they should make the most of a limited breeding season and produce offspring as early as possible during the period of relative food abundance or 2) because in an aseasonal environment, minimizing the length of the breeding cycle will have the effect of maximizing their rate of reproduction and, hence, their fitness. The first of these would also apply to situations in which multiple breeding attempts were possible within the breeding season. The alternative criterion that we explore is one of ‘‘offspring energy maximization.’’ Here, we assume that females are constrained to annual breeding, with the production of independent offspring necessarily occurring at a given time of year. We assume that the value of independent offspring depends on their condition and that females will seek to maximize the amount of reserves of their offspring at the time of independence (t ¼ 0 in the following year). Both the optimization criteria considered are bound by the constraints introduced in the previous section and, thus, can be assessed in light of the 3 regions of parameter space summarized in Figure 2. However, both also introduce an additional constraint. In the rate maximization model, the additional constraint is that the offspring’s total reserves at independence must be E. We assume that E . 0 and is a measure of the increase in offspring reserves after birth (over and above x0). In the energy maximization model, the additional constraint is that the length of the breeding cycle is limited to exactly t ¼ 1. Here, we assume that t0, T , 1. We can assess the consequences for maternal strategies of these additional constraints by defining a measure of the degree of capital breeding, K. Within the context of the current framework, a useful metric is K ¼ C/(C 1 I ), where C is the increase in offspring energy as a result of transfer from maternal capital and I is the increase in offspring energy as a result of provisioning from maternal income. Using this metric, the maternal strategies arising under each optimization criterion are summarized in Table 2. Clearly, the outcomes are broadly similar, but the additional constraints affect the situations under which pure capital or mixed strategies would be expected when t* . T, t0, and also affect the calculation of K. The reasoning underlying Table 2 is provided in greater detail in the Appendix. Effects of parameters In spite of the simplicity of this model, it is clear that a number of aspects of biology (including metabolic rates, food availability, and efficiency of energy transfers from mother to offspring) can have an important bearing on the model outcomes. Here, we examine the key parameters of the model to assess how changes in those parameters will affect maternal strategies. In particular, we examine the effects of offspring metabolism, efficiency of energy transfer, surplus energy accumulation rate by a mother with offspring, and different aspects of the relationship between surplus accumulation with and without offspring. Parameters of the basic model may have a variety of effects, depending on which of the regions of parameter space delineated in Table 2 applies. Here, we consider the unconstrained region of parameter space, where t* . T, t0 (e.g., Figure 2A). In this region, we can examine the effects of parameters using the relationship G(x*) ¼ cL c/a. First, setting u ¼ c/a allows us to examine the effects of c and a because x* is a function of this parameter: Gðx *ðuÞÞ ¼ cL u ð1Þ Figure 2 Optimal strategies under different time constraints. In each case, x* and t* are the same, as are a, cL, and c. Only the relative magnitudes of t0 and T are varied. Under either optimization criterion (rate maximization or offspring energy maximization), the timing of birth (and implantation) is dictated by the larger of the 3 time periods, t*, t0, and T. (A) t* . T, t0. As in Figure 1, the optimal time to give birth is t* [except where E , a(x* x0), see further in Table 2 and Appendix]. Implantation is delayed until ti and capital in excess of the basic demands of parturition (x* x0) is transferred to the offspring after birth. (B) t0 . T, t*. The minimum time until birth is now dictated by the minimum time taken to raise the basic demands of parturition. Implantation is again delayed but, at birth, there is no excess of capital and hence no capital transfer after birth. (C) T . t*, t0. A longer minimum gestation period determines the timing of birth. Consequently, implantation is immediate, and capital (in excess of the basic demands of parturition) is available for transfer immediately after birth. In all scenarios, the heavy black line shows the trajectory of offspring energy reserves after birth, once the female begins provisioning from income. In scenario B, the female must pursue a pure income strategy. Only in scenarios A and C will the female pursue a pure capital or mixed capital and income breeding strategy. Whether a pure or a mixed strategy is more likely depends on the ultimate optimization criterion, as shown in Table 2. Differentiating Equation 1 with respect to u gives dx * 1 ¼ : du G#ðx *ðuÞÞ Because G(x) is a decreasing function, x* increases as u increases. Increasing c for constant a will mean that G(x) will remain greater than cL c/a for higher x, and offspring energy gain from capital, C, will increase (Figure 3A). Houston et al. • Modeling capital and income breeding 245 Table 2 Optimal strategies under different optimization criteria Conditions Rate maximization Offspring energy maximization t* . T, t0 If a(x* x0) , E Optimal implantation time ¼ t* T; a mixed strategy results, where K ¼ a(x* x0)/E If a(x* x0) E Optimal implantation time ¼ t̂ðx0 1E=aÞ T ; pure capital breeding results (K ¼ 1). If t* , 1 Optimal implantation time ¼ t* T; a mixed strategy results, where K ¼ a(x* x0)/[a(x* x0)1 (1 t*)(acL c)] If t* 1 Optimal implantation time ¼ 1 T; pure capital breeding results (K ¼ 1). t0 . T, t* Optimal implantation time ¼ t0 T Pure income breeding results (K ¼ 0). Optimal implantation time ¼ t0 T Pure income breeding results (K ¼ 0). T . t*, t0 Immediate implantation (at t ¼ 0) is optimal. If a(xT x0) , E A mixed strategy results with K ¼ a(xT x0)/E If a(xT x0) E Pure capital breeding results (K ¼ 1). Immediate implantation (at t ¼ 0) is optimal. A mixed strategy results with K ¼ a(xT x0)/[a(xT x0) 1(1 T)(acL c)]. Similarly, the delay between previous offspring independence and birth, t*, will also increase (Figure 3C). Under both optimization criteria, the probability of the female pursuing a pure capital breeding strategy will be increased (because increasing x* will increase the likelihood that a(x* x0) E in the rate maximization model and that t* 1 in the energy maximization model). Even where mixed strategies are followed, increasing c will increase the overall degree of capital breeding (K; see Table 2, Figures 3A,C). Decreasing a will have similar effects on x* to increasing c. However, decreasing a will also decrease the relative transfer from capital, reducing C in either model (Figure 3A). Reducing C in the rate maximization model will also reduce K, but in the energy maximization model, effects on K will depend on the time remaining for income breeding (which will also reduce because t* increases with decreasing a, Figure 3C), together with the total amount of energy gained by the offspring during that time. Differentiating Equation 1 with respect to cL gives dx * 1 ¼ ; dcL G#ðx *ðuÞÞ so that x* decreases as cL increases. Consequently, higher intake rates during the care period will tend to reduce the proportion of the costs of the breeding effort that are financed by capital (Figures 3B,D), potentially switching the female to a pure income breeding strategy. Following model assumption (ii), we have specified that G(x) is a decreasing function. It is possible to decompose that function into 2 components. These are the initial surplus rate of gain when no offspring are present, G(0), which is affected by foraging ability and food abundance, and the mass-dependent costs, which affect how rapidly G(x) declines below G(0) as x increases. We can consider the importance of these 2 aspects separately. First, we consider the initial surplus rate of gain; later, we return to the mass-dependent costs. For ease of notation, we define cF ¼ G(0). To emphasize that accumulation of capital depends on cF , we use the notation GðxÞ ¼ G̃ðx; cF Þ: Notice that now, the gain rate may differ in situations where the mass-dependent costs are identical, solely because the initial rate of gain is different (perhaps because food is less abundant). It follows that G̃ ð0; cF Þ ¼ cF and from assumption (ii), G̃ ðx; cF Þ , cF for x . 0: To determine how x* varies with cF , we can follow the same logic as used above to determine how x* varies with other parameters. Thus, we now write Equation 1 as G̃ ðx *ðu; cF Þ; cF Þ ¼ cL u: ð2Þ Differentiating Equation 2 with respect to cF gives dx * @ G̃=@cF ¼ dcF @ G̃=@x which is positive. Note that although C increases with increasing cF (Figure 3B), t* may not necessarily increase (Figure 3D). In the offspring energy maximization model, a reduction in t* will lead to increases in the duration of the income breeding phase and, potentially, increases in the proportion of offspring reserves that are derived from income. In the previous example, cF was assumed to vary independently of cL. However, unless the female is migratory (switching environments between the prebirth and postbirth periods), it is likely that energy gain rates with and without offspring will be linked. One way that the surplus rate of accumulation when caring for offspring could be linked to the initial surplus rate of accumulation in the absence of offspring is if the need to return to her offspring means that the mother is a central place forager (Orians and Pearson 1979). For example, a female without offspring may be able to remain in the vicinity of a food source, but a female that is caring for offspring must return to her offspring at regular intervals. In this case, if the amount of food in the environment changes, then both cL and cF will change. For convenience, we assume that this change will be proportional, such that the surplus rate of accumulation for mothers with offspring is a constant fraction of the initial rate of surplus accumulation for females without offspring, that is, cL ¼ qcF ; where 0 q 1. In this situation, we can write G̃ ðx *ðu; cF Þ; cF Þ ¼ qcF u: Now, we can examine the consequences of varying food availability, given that changing cF will have a proportional impact Behavioral Ecology 246 Figure 3 Effects of parameters on model outcomes. (A, B) Effects of changes in parameters on the amount of capital transferred after birth: (A) effects of changing intrinsic factors as indicated by the legend in (C) (including k, the parameter controlling the magnitude of costs associated with accumulated reserves; c, offspring metabolic rate; and a, the efficiency of energy transfer from mother to offspring); (B) effects of changing extrinsic factors as indicated by the legend in (D) (including cL, the surplus accumulation rate for mothers with offspring; cF , the initial surplus accumulation rate for mothers without offspring; cF1, both accumulation rates when mass-dependent costs are an additive function of accumulated capital; and cF3, both accumulation rates when mass-dependent costs are a multiplicative function of accumulated capital; see text for further details). (C, D) Effects of changes in parameter values on t*, the time taken to raise x* reserves: (C) effects of changing intrinsic factors, as before; (D) effects of changing extrinsic factors, as before. Note that all comparisons are relative to the situation shown in Figure 2A. Note, also, that changes leading to an increase in transfer from capital do not necessarily lead to an increase in t* (e.g., effects of changing a in [A] and [C]). In the rate maximization model, increases in a(x* x0) will always lead to increases in the degree of capital breeding. In the energy maximization model, this is less clear and depends on the time remaining within the annual cycle. In this case, if a(x* x0) increases but t* decreases (e.g., cF1 in [B] and [D]), it is possible that the degree of capital breeding might also decrease. on cL. In this case, optimal strategies depend on the interaction between intake rates and foraging cost due to increased reserves. The consequences of this interaction may depend critically on the form of the function that specifies how surplus accumulation declines with increasing x. For example, accumulated reserves may have a straightforward, additive effect on accumulation rate. This may happen, for instance, if accumulated reserves affect only one aspect of energy balance, such as the metabolic rate (or cost of foraging). In this case, G̃ ðx; cF Þ ¼ cF hðxÞ; where h#ðxÞ.0: For example, let G̃ ðx; cF Þ ¼ cF kx 2 : Then, h#ðx *Þ dx * ¼ 1 q; dcF so that dx * . 0: dcF As previously, x* will increase with cF (Figure 3B), but t* may decrease (Figure 3D), leading to uncertain impacts on K. It is also possible to envisage situations where the accumulation of reserves affects several components of energy balance simultaneously, potentially leading to multiplicative impacts on the rate of surplus accumulation. Here, it may be that G̃ ðx; cF Þ ¼ cF H ðxÞ; where H #ðxÞ,0: For example, G̃ ðx; cF Þ ¼ cF =ð11kx 2 Þ: Then, H ðx *Þ ¼ q u=cF ; and hence x* decreases as cF increases (Figure 3B). This will tend to reduce K as cF increases and may thus have the opposite effect to the case where the costs of accumulated reserves were additive. This demonstrates the critical importance of understanding the functional form of the costs of carrying capital. Finally, we note that in either of the cases above, increasing k will increase the strength of mass-dependent costs, causing net energy gain rate to decline more rapidly with increasing accumulation of capital. In the absence of changes in other parameters, this will always have the effect of decreasing both x* and t* (e.g., see Figures 3A,C for the additive case), making a pure income breeding strategy more likely and reducing K in both the rate maximization and energy maximization models. DISCUSSION The simple model illustrated above demonstrates that females may adopt a range of breeding strategies, including pure income breeding, pure capital breeding, and a mixture of the 2 strategies. The model can be applied to annual breeders and to species with shorter reproductive cycles. By maintaining a general structure, we have shown how different aspects of ecology and physiology might be expected to affect the selected strategy. Previous models (e.g., Boyd 1998; Trillmich Houston et al. • Modeling capital and income breeding and Weissing 2006) have tended to focus on body mass as the major determinant of the adaptive value of capital and income strategies. By contrast, our model enables us to pose a number of general hypotheses regarding other differences that may have led to divergent strategies among related species. These include that capital breeding would be favored among species: 1) with relatively low neonate mass or relatively long fixed terms of gestation; 2) with offspring that have relatively high metabolic rates; 3) where female foraging is relatively constrained while caring for offspring; and 4) where relatively low costs are incurred by carrying stored capital. The model also indicates ways in which efficiency of energy transfer from mother to offspring and environmental richness (or food availability) might be expected to influence maternal strategies. These impacts are less clear cut, however, and will depend on specific circumstances, including the values of other model parameters. Finally, our model also identifies the complex relationships between rates of energy gain for females with and without offspring and the specific form of mass-dependent cost functions. Clearly, the costs of carrying high-energy reserves (which might be additive or multiplicative and can be affected by a range of physiological and locomotory factors, as well as absolute body size) must be understood if predictions about maternal strategies are to be made regarding the consequences of variation in food supply. Our findings can be interpreted in light of other general assessments of the adaptive value of maternal provisioning strategies. For example, it is argued that large-bodied pinnipeds are more likely to be capital breeders because they seldom have access to rich local food supplies at the breeding site, whereas small bodied pinnipeds tend to be income breeders because they cannot support large reserves (Boyd 1998). Within our framework, these findings are equivalent to relatively high costs of capital storage among smaller pinnipeds and, potentially, a higher difference between cF and cL in larger species (perhaps because of the constraints imposed by the need for central place foraging). By identifying the importance of the costs of accumulated reserves, we have to some extent formalized arguments regarding which taxa might be predisposed toward capital breeding due to lower costs of carrying capital (Bonnet et al. 1998). Our model goes beyond previous efforts in 3 ways. First, it is more general and therefore applicable to a greater range of species than previous models that have been constructed and parameterized specifically for pinnipeds (Boyd 1998; Trillmich and Weissing 2006). Indeed, in contrast to these previous models, our framework has no requirement for central place foraging, showing that capital breeding can still arise under a variety of conditions. Second, our model is novel in that it allows for a mixture of strategies, focusing attention on the degree of capital or income breeding, rather than viewing these as dichotomous strategies. Smaller phocids are known to employ mixed strategies (e.g., Bowen et al. 1999) and our model might suggest explanations for these intermediate situations. Third, rather than focusing on body mass as the sole potential driver of different strategies, our framework allows an assessment of the effects of a variety of other parameters. Although there is a general tendency for capital breeding phocids to be larger than income breeding otariids (Costa 1991), exceptions do exist (e.g., the Steller sea lion, Eumetopias jubatus) (Pitcher and Calkins 1981), confounding attempts to explain maternal strategies solely on the basis of maternal body size (Trillmich and Weissing 2006). Although many of the parameters in our model are likely to show some relationship with lean body mass, some parameters may clearly be affected by other broad differences in life history. In particular, differing modes of insulation, foraging, and locomotion among phocids and otariids might be 247 expected to lead to different mass-dependent costs in the 2 families, whereas physiological constraints (such as mammary gland function, e.g., Lang et al. 2005) might also affect the choice of strategy. We have deliberately maintained a general model structure, and so it is unlikely that our model will apply precisely to any particular system. Nevertheless, it is worth considering the realism of the major underlying assumptions. Perhaps most critical to the model is the assumption of a negative relationship between the quantity of stored capital and the rate of capital accumulation (i.e., the assumption of an asymptotic maximum level of stores). Although the relative costs of reserve accumulation among small and large-bodied animals has been implicit in previous models of capital and income breeding, our model draws explicit attention to the critical importance of these costs in dictating maternal breeding strategies. Loss of mobility (Kullberg et al. 2005) and resultant vulnerability to predation are widely cited as costs of stored reserves in birds (e.g., Hedenström 1992; Witter and Cuthill 1993; Gosler et al. 1995), but these and other costs and benefits in other taxa are less well researched. Among air-breathing divers (many of which are constrained while tending by the need to raise offspring on land but forage at sea), emerging evidence suggests that efficiency of locomotion (Hind and Gurney 1997; Fish 2000), diving behavior (Lovvorn and Jones 1991; Beck et al. 2000; Sato et al. 2003; Miller et al. 2004), and thermoregulation (Kvadsheim et al. 1997; Costa and Gales 2003; Harding et al. 2005) are all affected by fat reserves. However, these relationships are rarely demonstrated in terms of their effects on optimal fat loading or rates of energetic gain. Similarly, the longer term health costs of increased work rates owing to increased body mass are, as yet, poorly understood (e.g., Nilsson 2002). Overall, it seems reasonable to assume that all motile organisms will have a maximum energy storage capacity and that most will show some decline in the rate of energy accumulation before they reach that maximum. Even among sedentary organisms, energy stores will require supporting tissues, the manufacture and maintenance of which will be associated with costs. Large size (resulting from large stores of capital) is also likely to be associated with higher threats from predation and parasitism, both of which may incur costly avoidance strategies. Consequently, our assumption that net energy gain will be a declining function of stored capital is likely to be valid, at least at high levels of stored energy. One way in which many species might be expected to differ from the types of energy accumulation functions that we present is where low reserves also carry a cost. For example, if limited energy stores are associated with poor insulation, rapid energy loss and, thus, high metabolism, then a sigmoid gain function might be more reasonable than the monotonic functions that we have discussed. These might be expected to favor income breeding if they lead to a long delay in accumulating the capital required for birth. However, the outcome of such situations is not intuitively obvious as low initial rates of capital accumulation also imply low rates of surplus energy gain for income breeding females. Furthermore, given the dangers (in a stochastic environment) of maintaining a lean body when capital can only be accumulated slowly, we might expect individuals in such a scenario to maintain an intermediate level of energy stores as their baseline (thus, omitting the lower part of the sigmoid curve and conforming to the monotonic functions that we have discussed). A second assumption of our model is that of the instantaneous transfer of capital from mother to offspring after birth. Although some capital breeders achieve very high-speed transfers of energy to their young after birth (e.g., the hooded seal, Cystophora cristata) (Bowen et al. 1985), instantaneous transfers 248 are not an accurate reflection of the average capital breeding system. However, the purpose of this model is merely to explore the circumstances that favor capital breeding or income breeding. Lower metabolic overheads during the provisioning phase are often cited as an advantage of capital breeding (e.g., Schulz and Bowen 2004, 2005), a factor exaggerated by our assumption of an instantaneous capital phase. Were this assumption relaxed to allow for longer capital transfer phases, we would still expect the same qualitative conclusions. Similar arguments may be advanced for 2 other assumptions, including female maintenance of reserves during the income breeding phase and female ability to manipulate timing of birth. Although neither of these are likely to be exactly accurate (e.g., see Boyd et al. 1997 for an example of where the former assumption is violated), both are plausible and sufficiently accurate for the purposes of a heuristic model. Finally, our model also assumes that whether they breed annually or within shorter cycles, females will seek to maximize the speed at which they transfer energy to their offspring. This is not necessarily the case. Decisions regarding offspring condition and the speed at which offspring reserves should be increased can only properly be viewed in light of optimization over a full annual routine (e.g., Houston and McNamara 1999). Trillmich and Weissing (2006) are concerned that annual routine models which consider longer term aspects of fitness would require such species-specific parameterization that they would be vulnerable to accusations of adaptive storytelling (cf. Gould and Lewontin 1979). We disagree that this is necessarily the case. For example, models of generalized pinniped life histories can examine the influence of one or a few parameters on reproductive strategies, without the requirement to tailor these to specific species. Such approaches would also permit an examination of other aspects of food supply, such as seasonality and predictability (to which our analytical model is not well suited). There is some debate about how the costs of reproduction should be measured in the context of capital breeding (Bonnet et al. 1999; Lewis and Kappeler 2005), and it is often far from clear which components of these costs are financed from stored reserves and which are financed from concurrent intake (Casas et al. 2005). Consequently, the best metric to indicate the degree of capital or income breeding is also unclear. We have suggested one metric here (total energy from capital as a proportion of total offspring reserves at independence) because it is easily envisaged and applied within the framework of our model. However, we do not suggest that this is the only metric or, indeed, the best. Other measures may be more easily quantified in empirical studies or more appropriate to other taxa. For situations where our model predicts mixed strategies, the influence of parameters on the degree of capital breeding is likely to depend on the metric used. Our model has been formulated and presented in terms most relevant to mammalian life history. However, there is no reason why it should not be adaptable to other taxa. Although it is framed in terms of a female producing a single offspring, it applies equally well to multiple offspring. Similarly, we describe a female that expends substantial reserves at the time of birth and, thereafter, transfers energy to her offspring, for example, by lactation. Again, however, there is no reason why the initial expense should not apply to the production of eggs, while subsequent transfers of resources may represent a combination of heat transfers (during incubation) and direct provisioning thereafter. Such adaptations would require careful thought regarding the meaning of capital and income breeding. For example, some authors have suggested that for birds, the source of energy used for egg production (rather than energy used for subsequent pro- Behavioral Ecology visioning) defines the capital-income split (Klaassen et al. 2001). Such a view would be hard to reconcile with our framework. Even in cases where the provisioning of offspring is clearly the focus of strategic differences, the model that we present would require substantial modifications before it could be applied to species in which males play an active role in offspring provisioning. There are clear parallels between the offspring provisioning strategies that we consider and maternal reproductive strategies that have been the focus of recent interest in several other taxa. For example, Trexler and DeAngelis (2003) used modeling approaches to consider how matrotrophy, the nourishment of developing embryos by a source other than yolk, has evolved from lecithotrophy (in which developing embryos are provisioned only from resources stored in the egg prior to fertilization) in fish. In lecithotrophic systems, all the resources required for development to independence must be supplied at the outset (analogous to capital breeding), whereas matrotrophic systems allow for continued maternal inputs as the embryo develops (akin to income breeding). Johnson (2006) used an experimental approach to assess the factors underlying foraging by ant queens between founding a colony and producing their first brood. Claustral queens store reproductive energy prior to mating flights and rear their first broods from stored reserves (i.e., they are capital breeders), whereas semi-claustral queens must forage prior to producing their first brood (and so are viewed as income breeders) ( Johnson 2006). In both of these studies, the authors found compelling evidence to suggest that environmental variability and predictability were important factors that would be difficult to incorporate into our structure (requiring, instead, a dynamic programming or simulation approach). Nevertheless, our simple model might have implications for other factors that could influence the transition between lecithotrophic and matrotrophic fish or claustral and semi-claustral ants. In particular, neither Trexler and DeAngelis (2003) nor Johnson (2006) considered explicitly the initial costs of raising capital that must be invested at the outset, which our model suggests could have important implications for strategic optima. A final limitation of our model is that it focuses on energy, making no allowance for other types of stores accumulated in advance of reproduction. For birds, there is evidence that, for egg development, protein may be a more important factor than energy (Meijer and Drent 1999). Similarly, in gray seals (Halichoerus grypus), maternal protein stores after parturition may be more important than fat stores in determining pup weaning weight (Mellish et al. 1999). Storage and catabolism of protein and fat may be associated with very different costs and benefits, but a model considering different types of stores would, necessarily, be considerably more complex than the one that we have presented (e.g., Noren and Mangel 2004). Indeed, the costs of mobilizing long-term and short-term stores of different types clearly vary greatly among taxa (Jönsson 1997; Bonnet et al. 1998), and our single parameter for controlling transfer of efficiency is very crude in this regard. To keep our model simple and thereby maximize its generality and heuristic value, we have deliberately ignored many behaviors that might reasonably be incorporated. The model could be extended to include more explicit costs of travel, in utero investment and the costs of pregnancy, spatial variation in food availability, the demands of mate encounters, and risks of predation. In spite of these and other limitations discussed above, we believe that our model should stimulate further consideration of the factors underlying maternal strategies for offspring provisioning. More detailed annual routine models of specific systems and, in particular, more attention to the form of mass-dependent cost functions could reveal much about the adaptive significance of income and capital Houston et al. • Modeling capital and income breeding breeding and the relative frequency of these and intermediate strategies in different taxa. APPENDIX For both optimization criteria, we consider 3 regions of parameter space. Here we clarify the optimal decisions in each case, showing how the degree of capital breeding, K (as given in Table 2), is derived in each case. Our reasoning rests on the fact that, without additional constraints, the female will favor giving birth (and switching to income provisioning) at t*. This is the point at which income provisioning becomes more efficient than further accumulation of capital. If t* is longer than t0 and T, then the female will usually be free to follow this strategy. If either t0 or T is the longer period, however, the female will not be free to give birth at t*, and other strategies will be favored. 249 capital after birth) will be a(xT x0). This presents 2 possibilities as follows: 1. If a(xT x0) , E, then the female cannot raise sufficient capital in T for her offspring to reach independence through the transfer of capital alone. Remaining reserves required must come from income provisioning, so the proportion of offspring reserves at independence that come from maternal capital will be K ¼ a(xT x0)/E. 2. If a(xT x0) E, then capital to supply all the energy that the offspring requires can be raised in t T. Either the female will limit her accumulation of capital or she will only transfer a proportion of her reserves as capital following birth. In either case, she will transfer E/a reserves as capital, so that all offspring reserves prior to independence come from capital, that is, K ¼ 1. The offspring energy-maximization model The rate-maximization model t* . T, t0 The female is free to delay giving birth until t*. Female capital reserves at t* are, by definition, x*. Thus, energy available to her offspring (transferred as capital after birth at t*) would be a(x* x0). This presents 2 possibilities as follows. 1. If a(x* x0) , E, then the female cannot raise sufficient capital in t* for her offspring to reach independence through the transfer of capital alone. She will delay giving birth until t ¼ t* and will delay implantation until ti ¼ t* T. Offspring energy reserves will increase (following birth) by a(x* x0) as a result of transfer from capital. Remaining reserves required must come from income breeding, so the proportion of offspring reserves at independence that come from maternal capital will be K ¼ a(x* x0)/E. 2. If a(x* x0) E, then capital to supply all the energy that the offspring requires can be raised in t t*. Specifically, the offspring will require E reserves, so the female must raise x0 1 E/a reserves before giving birth. The time taken to raise this quantity of reserves is t̂ðx0 1E=aÞ: In order to give birth at this time, the female must delay implantation until ti ¼ t̂ðx0 1E=aÞ T : Birth will occur at ti ¼ t̂ðx0 1E=aÞ: In birth, the female will lose x0 reserves and so will transfer E/a reserves to her offspring with efficiency a. The offspring requires no additional reserves for independence, and so all offspring reserves prior to independence come from capital, that is, K ¼ 1. t0 . T, t* Here, the time taken to raise the minimum reserves needed for birth is the largest of the 3 time periods. Consequently, the female will give birth at t0 (implanting at ti ¼ t0 T). Notice that this is after the point at which income provisioning becomes more efficient than further accumulation of capital. By the time of birth, therefore, the female will only have accumulated x0 reserves, all of which will be lost in birth. As a result, she will have no additional reserves to pass to her offspring. All offspring reserves must therefore come from income provisioning, so the degree of capital breeding is K ¼ 0. T . t *, t 0 In this case, the gestation period dictates the time until birth, so there will be immediate implantation. By definition, the total reserves accumulated by the female by T will be xT and the amount of energy available to her offspring (transferred as t* . T, t0 As in the rate-maximization model, if the female delays giving birth until t*, the amount of energy available to her offspring (transferred as capital after birth) will be a(x* x0). Again, this presents 2 possibilities as follows: 1. If t* , 1, then the female will have time remaining after giving birth (at t*) in which to provision from income. The amount of time remaining will be 1 t*, during which time she will be able to supply her offspring with (1 t*)(acL c) from income provisioning. She will therefore follow a mixed strategy with K ¼ a(x* x0)/ [ a(x* x0) 1 (1 t*)(acL c)]. 2. If t* 1, then offspring energy reserves will be maximized when the female spends the whole year accumulating capital, giving birth (and transferring all available capital to her offspring) at t ¼ 1. She will not benefit from giving birth before this, as (by definition) the point at which income provisioning becomes the more efficient mode of maximizing offspring energy reserves (t*) will not be reached within the annual cycle. Consequently, pure capital breeding results (K ¼ 1) and implantation occurs at ti ¼ 1 T. t0 . T, t* Here, the time taken to raise the minimum reserves needed for birth dictates the timing of birth. As in the rate-maximization model, the female will give birth at t0 (implanting at ti ¼ t0 T). By the time of birth, the female will only have accumulated x0 reserves, all of which will be lost in birth. As a result, she will have no additional reserves to pass to her offspring. All offspring reserves must come from income provisioning, so the degree of capital breeding, K ¼ 0. T . t*, t0 In this case, the gestation period dictates the time until birth, so there will be immediate implantation. By definition, the total reserves accumulated by the female by T will be xT and the amount of energy available to her offspring (transferred as capital after birth) will be a(xT x0). By definition, T , 1 and, consequently, some time will remain for income provisioning. The female will follow a mixed strategy, where K ¼ a(xT x0)/ [a(xT x0) 1 (1 T)(acL c)]. This work was funded by the Natural Environment Research Council (grant number 2003/00616). A.I.H. and J.M.M. were supported by Leverhulme Trust fellowships. We thank the editorial team and 2 anonymous referees for helpful comments on an earlier draft. 250 REFERENCES Beck CA, Bowen WD, Iverson SJ. 2000. Seasonal changes in buoyancy and diving behaviour of adult grey seals. J Exp Biol. 203:2323–2330. Beck CA, Bowen WD, Iverson SJ. 2003. Sex differences in the seasonal patterns of energy storage and expenditure in a phocid seal. J Anim Ecol. 72:280–291. Beck CA, Bowen WD, McMillan JI, Iverson SJ. 2003. Sex differences in the diving behaviour of a size-dimorphic capital breeder: the grey seal. Anim Behav. 66:777–789. Boness DJ, Bowen WD. 1996. The evolution of maternal care in Pinnipeds. Bioscience. 46:645–654. Bonner WN. 1984. Lactation strategies in pinnipeds: problems for a marine mammalian group. 51:253–272. Bonnet X, Bradshaw D, Shine R. 1998. Capital versus income breeding: an ectothermic perspective. Oikos. 83:333–342. Bonnet X, Naulleau G, Shine R, Lourdais O. 1999. What is the appropriate timescale for measuring costs of reproduction in a ‘capital breeder’ such as the aspic viper? Evol Ecol. 13:485–497. Bowen WD, Boness DJ, Iverson SJ. 1999. Diving behaviour of lactating harbour seals and their pups during maternal foraging trips. Can J Zool. 77:978–988. Bowen WD, Oftedal OT, Boness DJ. 1985. Birth to weaning in 4 days: remarkable growth in the hooded seal, Cystophora cristata. Can J Zool. 63:2841–2846. Boyd IL. 1998. Time and energy constraints in pinniped lactation. Am Nat. 152:717–728. Boyd IL. 2000. State-dependent fertility in pinnipeds: contrasting capital and income breeders. Funct Ecol. 14:623–630. Boyd IL, McCafferty DJ, Walker TR. 1997. Variation in foraging effort by lactating Antarctic fur seals: response to simulated increased foraging costs. Behav Ecol Sociobiol. 40:135–144. Casas J, Pincebourde S, Mandon N, Vannier F, Poujol R, Giron D. 2005. Lifetime nutrient dynamics reveal simultaneous capital and income breeding in a parasitoid. Ecology. 86:545–554. Costa DP. 1991. Reproductive and foraging energetics of high latitude penguins, albatrosses and pinnipeds: implications for life history patterns. Am Zool. 31:111–130. Costa DP, Gales NJ. 2003. Energetics of a benthic diver: seasonal foraging ecology of the Australian sea lion, Neophoca cinerea. Ecol Monogr. 73:27–43. Festa-Bianchet M, Gaillard JM, Jorgenson JT. 1998. Mass- and densitydependent reproductive success and reproductive costs in a capital breeder. Am Nat. 152:367–379. Fish FE. 2000. Biomechanics and energetics in aquatic and semiaquatic mammals: platypus to whale. Physiol Biochem Zool. 73:683–698. Gosler AG, Greenwood JJD, Perrins C. 1995. Predation risk and the cost of being fat. Nature. 377:621–623. Gould SJ, Lewontin RC. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist program. Proc R Soc Lond B Biol Sci. 205:581–598. Harding KC, Fujiwara M, Axberg Y, Harkonen T. 2005. Massdependent energetics and survival in harbour seal pups. Funct Ecol. 19:129–135. Hedenström A. 1992. Flight performance in relation to fuel load in birds. 158:535–537. Hind AT, Gurney WSC. 1997. The metabolic cost of swimming in marine homeotherms. J Exp Biol. 200:531–542. Houston AI, McNamara JM. 1999. Models of adaptive behaviour: an approach based on state. Cambridge (UK): Cambridge University Press. Johnson RA. 2006. Capital and income breeding and the evolution of colony founding strategies in ants. 53:316–322. Jönsson KI. 1997. Capital and income breeding as alternative tactics of resource use in reproduction. Oikos. 78:57–66. Klaassen M, Lindstrom A, Meltofte H, Piersma T. 2001. Arctic waders are not capital breeders. Nature. 413:794. Kullberg C, Jakobsson S, Kaby U, Lind J. 2005. Impaired flight ability prior to egg-laying: a cost of being a capital breeder. Funct Ecol. 19:98–101. Kvadsheim PH, Gotaas ARL, Folkow LP, Blix AS. 1997. An experimental validation of heat loss models for marine mammals. J Theor Biol. 184:15–23. Lang SLC, Iverson SJ, Bowen WD. 2005. Individual variation in milk composition over lactation in harbour seals (Phoca vitulina) Behavioral Ecology and the potential consequences of intermittent attendance. 83: 1525–1531. Lewis RJ, Kappeler PM. 2005. Are Kirindy Sifaka capital or income breeders? It depends. Am J Primatol. 67:365–369. Lovvorn JR, Jones DR. 1991. Effects of body size, body fat and change in pressure with depth on buoyancy and costs of diving in ducks (Aythya spp.). Can J Zool. 69:2879–2887. Meijer T, Drent R. 1999. Re-examination of the capital and income dichotomy in breeding birds. Ibis. 141:399–414. Mellish JAE, Iverson SJ, Bowen WD. 1999. Variation in milk production and lactation performance in grey seals and consequences for pup growth and weaning characteristics. Physiol Biochem Zool. 72:677–690. Miller PJO, Johnson MP, Tyack PL, Terray EA. 2004. Swimming gaits, passive drag and buoyancy of diving sperm whales Physeter macrocephalus. J Exp Biol. 207:1953–1967. Nilsson JA. 2002. Metabolic consequences of hard work. Proc R Soc Lond B Biol Sci. 269:1735–1739. Noren DP, Mangel M. 2004. Energy reserve allocation in fasting Northern elephant seal pups: inter-relationships between body condition and fasting duration. Funct Ecol. 18:233–242. Oftedal OT, Boness DJ, Tedman RA. 1987. The behavior, physiology and anatomy of lactation in the pinnipedia. 1:175–245. Orians GH, Pearson NE. 1979. On the theory of central-place foraging. In: Horn DJ, Mitchell RD, Stairs GR, eds. Analysis of ecological systems. Columbus (OH): Ohio State University Press. p. 154–177. Pitcher KW, Calkins DG. 1981. Reproductive biology of Steller sea lions in the gulf of Alaska. J Mammal. 62:599–605. Roff DA. 1992. The evolution of life histories. Theory and analysis. New York: Chapman and Hall. Roff DA. 2002. Life history evolution. Sunderland (MA): Sinauer Associates. Sato K, Mitani Y, Cameron MF, Siniff DB, Naito Y. 2003. Factors affecting stroking patterns and body angle in diving Weddell seals under natural conditions. J Exp Biol. 206:1461–1470. Schulz TM, Bowen WD. 2004. Pinniped lactation strategies: evaluation of data on maternal and offspring life history traits. Mar Mamm Sci. 20:86–114. Schulz TM, Bowen WD. 2005. The evolution of lactation strategies in pinnipeds: a phylogenetic analysis. Ecol Monogr. 75:159–177. Stearns SC. 1992. The evolution of life histories. Oxford: Oxford University Press. Thomas VG. 1988. Body condition, ovarian hierarchies, and their relation to egg formation in Anseriform and Galliform species. In: Quellet H, ed. Acta XIX Congressus Internationalis Ornithologici. Ottawa (Canada): National Museum of Natural Science. p. 353–363. Trexler JC, DeAngelis DL. 2003. Resource allocation in offspring provisioning: an evaluation of the conditions favoring the evolution of matrotrophy. 162:574–585. Trillmich F, Weissing FJ. 2006. Lactation patterns of pinnipeds are not explained by optimization of maternal energy delivery rates. Behav Ecol Sociobiol. 60:137–149. Witter MS, Cuthill IC. 1993. The ecological costs of avian fat storage. Philos Trans R Soc Lond B Biol Sci. 340:73–92.
© Copyright 2026 Paperzz