On the cost of reproduction in long-lived birds: the influence of environmental variability

Kjell Einar Erikstad


A central issue in life history theory is how animals balance their investment in young against their own chances to survive and reproduce in the future (e.g., Roff 1992, Stearns 1992). This life history trade-off is referred to as the cost of reproduction and was originally proposed by Williams (1966), based on clutch size studies on birds. If the reproductive effort in one year leads to a loss in future reproductive output (through decreased adult survival or reduced fecundity), then the optimal effort in the current season is less than the effort that would maximize the number of offspring produced in that season (Williams 1966, Charnov and Krebs 1974).

A common design used to examine the cost of reproduction in birds is to increase or decrease the effort of parents by manipulating the number of young in the nest. Such studies generally show that parents are able to raise more young than they actually produce in a single year, but that the increased effort involved leads to lower adult survival or lower future fecundity (see Linden and Moller 1989, Dijkstra et al. 1990, Daan et al. 1996). To date, however, most studies on the regulation of avian parental effort have been carried out on short-lived passerines (Linden and Moller 1989). In these species, the probability of survival to future reproduction is often so low that an increased investment in current reproduction, at the expense of parent survival, would be expected (Charlesworth 1980). Linden and Moller (1989) emphasized the need to examine the reproductive cost in long-lived species such as seabirds and geese. One would expect long-lived species to be more restrictive in the degree to which they exhibit increased effort, because even a small reduction in adult survival would reduce the number of subsequent breeding attempts, thereby greatly lowering lifetime reproductive success (Charlesworth 1980, Curio 1988, Wooller et al. 1992). Thus, it has been suggested that parental effort in long-lived species is regulated to a fixed schedule, independent of offspring need (Ricklefs 1987, 1992), in order to maximize the survival of adults (Saether et al. 1993). There is evidence supporting this hypothesis from studies of the Leach’s Storm-Petrel Oceanodroma leucorhoa (Ricklefs and Minor 1991, Hamer and Hill 1994, Bolton 1995, Mauck and Grubb 1995) and the Antarctic Petrel Thalassoica antarctica (Andersen et al. 1993, 1995, Saether et al. 1993). However, recent studies on geese and long-lived seabirds show that long-lived birds increase their breeding effort after experimental manipulation (Johnsen et al. 1994, Jacobsen et al. 1995, Tombre and Erikstad 1996, Erikstad et al. 1997), even at the expense of their own survival (Reid 1987, Jacobsen et al. 1995). A recent study of the Atlantic Puffin Fratercula arctica (Erikstad et al. 1997) even suggests that parents adjust their investment in the single chick according to both their own body condition and the size of the chick; the latter may indicate the chick’s prospect of survival and, hence, recruitment to the population.

The idea that long-lived birds should not invest in young at the expense of their own survival is based on the general assumption that lifetime reproductive success depends primarily on survival, rather than seasonal fecundity (Clutton-Brock 1988, Newton 1989). However, what has not been considered in this context is that long-lived birds often live in stochastic environments, where the possibilities of raising young in some years are low. For example, seabirds sometimes face food shortages during the breeding season, resulting in very low reproductive success (e.g., Monaghan et al. 1989, 1992, Vader et al. 1990, Anker-Nilssen 1992, Chastel et al. 1993, Hatch et al. 1993, Weimerskirch 1992). The large variation in reproductive output observed in a number of long-lived species might indicate that reproduction in good breeding seasons has a larger value with respect to fitness than reproduction in bad seasons. Large variations in breeding conditions may therefore favor a highly flexible reproductive effort (Erikstad et al. 1997).

This study is a theoretical exploration of the optimal response in reproductive effort by long-lived birds to varying breeding conditions. We examine the optimal phenotypic balance between reproduction and adult survival for given genotypes under variable breeding conditions. The model is applied to two genotypes differing in their potential fecundity. The first genotype refers to a long-lived seabird with a large clutch size (three eggs). The second genotype refers to a long-lived seabird with single-egg clutches. We assume that the phenotypic relationship between reproduction and survival is determined by breeding conditions, clutch size, and reproductive effort. We also assume that the birds are unable to predict breeding conditions later in the breeding cycle when the clutch is produced. Furthermore, we assume that the genotypes, in order to maximize fitness, respond to variations in breeding conditions by adjusting their reproductive effort. By increasing the clutch size in the model, we examine the optimal response to a phenotypic brood size manipulation.


To explore the functional relationships among reproduction, survival, and clutch size, we start by assuming constant breeding conditions. Furthermore, we assume annual reproduction and a constant population size. The rate of increase of a genotype in one breeding season is then given by (e.g., Schaffer 1974):

[Lambda] = R + S (1)

where [Lambda] is rate of increase; R is the product of the number of offspring produced and pre-reproductive survival, hereby referred to as reproductive output; and S is annual adult survival. For iteroparity, the constraint curve relating S to R should be concave (Schaffer 1974). Ricklefs (1977) suggested the following general power function relating S to R for birds:

S = A[1 – [(R/B).sup.z]] for 0 [less than or equal to] R [less than or equal to] B (2)

where A is adult survival when reproduction is zero; B is maximum reproductive output when adult survival is zero; and Z determines the curvature between S and R. Because the parameter A determines maximum adult survival and the parameter B determines maximum reproductive output, we define A as the trait longevity and B as the trait fecundity for the genotype.

Reduced survival (S [less than] A) is the consequence of all costly activities associated with reproduction. Accordingly, we introduce the variable reproductive effort (E) and define it in units of R. We propose that the constraint function (Eq. 2) describes the special case in which an individual achieves the maximum reproductive output for all possible values of reproductive effort such that R = E and S = A(1 – [(E/B).sup.z]), for 0 [less than or equal to] E [less than or equal to] B.

However, an individual might obtain a reduced reproductive output for a given reproductive effort (R [less than] E), through a mismatch between the effort spent and clutch size. We measure clutch size (c) in units of E, letting c = E define the line for which clutch size matches reproductive effort and, consequently, R = E. Furthermore, we define a function f(c, E), which gives the fraction of reproductive output realized. The expressions of R and S then become:

[Mathematical Expression Omitted]. (3)

Now, let u be the proportion between clutch size and reproductive effort, such that u = c/E. For u [less than] 1, more young could have been reared, because reproductive effort is too large relative to clutch size, resulting in a reduced reproductive output with respect to effort spent (R [less than] E). For u = 1, reproductive effort matches clutch size, and R = E. For u [greater than] 1, reproductive effort is too small relative to clutch size, resulting in an increased mortality of young and, consequently, a reduced R. Accordingly, we assume that the function f(c, E) is of the form: f(c, E) = [Alpha](c/E) – [Beta][(c/E).sup.n], where [Alpha] and [Beta] are constants and n determines the skewness of the function. By definition, when u = 1, then f(c, E) = f(u) = 1 and df(u)/du = 0. Solving with respect to [Alpha] and [Beta] yields: [Alpha] = n/(n – 1) and [Beta] = 1/(n – 1). Furthermore, because reproduction cannot be negative, f(c, E) [greater than or equal to] 0, and we define f(c, E) = 0 for E [less than] [cn.sup.-1/(n – 1)]. Thus we have:

f(c, E) = 0 for 0 [less than or equal to] E [less than] [cn.sup.-1/(n – 1)]

f(c, E) = n/n -1 (c/E) – 1/n – 1 [(c/E).sup.n] for [cn.sup.-1/(n – 1)] [less than or equal to] E [less than or equal to] B (4)

where c [greater than or equal to] 0 and n [greater than or equal to] 2. See Fig. 1 for an illustration of the relationships among clutch size (c), reproductive effort (E), and reproductive output (R), as given by Eqs. 3 and 4.

For a constant c, the rate of increase and optimal reproductive effort ([E.sup.*]) is found by combining Eqs. 1, 3, and 4 and differentiating with respect to E. This yields:

[[Lambda].sub.c] (E) = A[1 – [(E/B).sup.z]]

[[E.sub.1].sup.*] = 0 for 0 [less than or equal to] E [less than] [cn.sup.-1/(n – 1)]


[Mathematical Expression Omitted]. (5)

Note that we have two local optima of [Lambda]: [Lambda]([[E.sub.1].sup.*]) where reproductive effort is zero, and [Lambda]([[E.sub.2].sup.*]) where reproductive effort is positive. An individual with a given clutch size, optimizing reproductive effort, should therefore choose [[E.sub.1].sup.*] or [[E.sub.2].sup.*], depending on which gives the higher rate of increase. If [Lambda]([[E.sub.1].sup.*]) [greater than] [Lambda]([[E.sub.2].sup.*]), then maximum fitness is achieved by not reproducing; if [Lambda]([[E.sub.1].sup.*]) [less than] [Lambda]([[E.sub.2].sup.*]), then maximum fitness is achieved through reproduction.

In order to examine the optimal reproductive effort, reproductive output, and survival for a given genotype in a variable environment, we introduce stochasticity in breeding conditions to the model. Assume a stochastic, independent (not autocorrelated) variation in breeding conditions between seasons. The breeding condition (x) is determined by variables such as territory size, territory quality, weather conditions, food supply, predator density, etc. We introduce phenotypic plasticity in the fecundity trait (B) and in the longevity trait (A), such that B and A become functions of breeding conditions. We let the reaction norm of B be given by the function b(x), and the reaction norm of A be given by the function a(x). We assume that Z is constant for all environments.

Under “infinitely” good breeding conditions, the maximum number of fledged chicks will be determined by the physiological limits of the parents, and not by a further improvement in breeding conditions (e.g., Drent and Daan 1980). We therefore assume an upper threshold (K) in the fecundity trait during favorable breeding conditions, such that b(x) approaches K asymptotically for increasing x. We let x = 0 define the minimum breeding condition for which breeding is possible. Accordingly, we propose the following reaction norm in the fecundity trait:

b(x) = 0 for [less than or equal to] 0

b(x) = K(I – [e.sup.rx]) for x [greater than] 0 (6)

where r determines the rate of increase toward K.

The reaction norm in the longevity trait (a(x)) will only partly depend on breeding conditions, because a(x) is the average of maximum survival in the breeding period and survival outside the breeding season. We assume an upper threshold (L) in a(x) during good breeding conditions, and, furthermore, that a(x) can take only positive values. Accordingly, we propose the following relationship:

a(x) = 0 for x [less than] – 1/r ln(1/m)

a(x) = L(1 – [me.sup.-rx]) for x [greater than or equal to] – 1/r 1n(1/m) (7)

where m determines the relationship between a(x) and b(x) for x [greater than] 0. The defined reaction norms in the two life history traits b(x) and a(x) produce different tradeoff functions between reproductive effort E and adult survival S (Eq. 3) for different environments. Figure 2 illustrates the resulting reaction norm in the constraint function relating S to E.

For a given breeding condition (x) and clutch size (c), we assume that the rate of increase in a given breeding season ([Lambda]) is determined by the reproductive effort (E) spent by the parents. As before, we assume that [Lambda] is the sum of R and S, such that the rate of increase and optimal reproductive effort is found by replacing B by b(x) and A by a(x) in Eq. 5, which yields:

[[Lambda].sub.c,x](E) = a(x){1 – [[E/b(x)].sup.z]}

[[E.sub.1].sup.*] = 0 for 0 [less than or equal to] E [less than] [cn.sup.-1/(n – 1)]


[[Lambda].sub.c,x](E) = c/n – 1[n – [(c/E).sup.n-1]] + a(x){1 – [[E/b(x)].sup.z]}

[[E.sub.2].sup.*] = [[b[(x).sup.z][c.sup.n]/a(x)Z].sup.1/Z+n-1)] for [cn.sup.-1/(n-1)] [less than or equal to] E [less than or equal to] b (x). (8)

The maximization of [[Lambda].sub.c,x] in each season is only a valid criteria when population size is constant (e.g., Stearns 1992). Population sizes vary in a variable environment, and a more appropriate measure of fitness would be the mean stochastic rate of increase proposed by Orzack and Tuljapurkar (1989). However, we assume that the mean rate of increase is set to 1, and we argue, for simplicity, that the optimization criteria given in Eq. 8 are most appropriate in this context.

We assume that breeding conditions (x) are normally distributed, with mean [[Mu].sub.x] and standard deviation [[Sigma].sub.x]. We assume that the female at the time of egg-laying is unable to predict the breeding condition later in the season. Accordingly clutch size (c) is independent of later breeding conditions, and we use a constant clutch size in the model. In order to maximize [[Lambda].sub.c,x], the individual adjusts its reproductive effort (E) optimally according to the given x and c in each season (Eq. 8). Clutch size is set equal to the c that maximizes the mean [Lambda], depending on the probability density function of x. Furthermore, because we assume a stable population size, we find the mean breeding condition ([[Mu].sub.x]) such that mean rate of increase is equal to 1. [[Sigma].sub.x] is held constant.

We apply the model on two distinct long-lived genotypes. The genotype [G.sub.1] represents a long-lived species with high potential fecundity and the genotype [G.sub.2] represents a long-lived species with low potential fecundity. Accordingly, we give the two genotypes equal upper thresholds in the longevity trait (L), but unequal upper thresholds in the fecundity trait (K). In [G.sub.1], we assume a maximum number of three fledglings per nest; in [G.sub.2], we assume a maximum number of one fledgling equal to one per nest. We assume a survival from fledging until first breeding equal to 0.5 for both genotypes, and a sex ratio of 0.5. The upper threshold in the fecundity trait (K) is then equal to 3 x 0.5 x 0.5 = 0.75 and 1 x 0.5 x 0.5 = 0.25 for [G.sub.1] and [G.sub.2], respectively. The upper threshold in the longevity trait (L) is set to 0.95 for both genotypes. The parameters used are based on demographic studies on the Kittiwake (Coulson and White 1959, Jacobsen et al. 1995) for [G.sub.1], and studies on alcid species (Hudson 1985) for [G.sub.2]. We suggest that the parameter in (Eq. 7), which determines the slope of the regression between a(x) and b(x), is small. In other words, we assume that breeding conditions only have minor influences on the longevity trait compared to the fecundity trait. Such an assumption seems appropriate, because resources needed for reproduction generally differ from those needed for subsequent survival. For example, many seabirds rely on a local food supply during breeding, yet, once the breeding season is over, they leave the colony areas and disperse over large areas to locate food. We use m = 0.05 for both genotypes. The parameter Z, which determines the curvature of the constraint function relating adult survival to reproductive effort (Eq. 3), is set equal to 6, the mean value found by Ricklefs (1977). The resulting reaction norm of the constraint function relating reproductive effort (E) to (S) for different breeding condition, is shown in Fig. 2 for [G.sub.1]. We use n = 6 (Eq. 4), indicating a large skewness in the function f(c, E) [ILLUSTRATION FOR FIGURE 1 OMITTED] and, thus, a large increase in chick mortality for reduced reproductive effort and increased clutch size. In order to predict an optimal response to an experimentally increased brood size, we increase the optimal clutch size for the two genotypes by 50%.

The reaction norms with respect to total rate of increase, adult survival, and reproductive output for increasing breeding conditions are shown for [G.sub.1] (genotype with high potential reproductive output) and [G.sub.2] (genotype with low potential reproductive output) in Fig. 3A. The estimated distribution of breeding conditions giving a mean rate of increase equal to 1 is shown as a solid line. Clutch size is set constant and equal to the optimal clutch size for the estimated distributions of breeding conditions. During poor breeding conditions, fitness achieved through reproductive investment is less than fitness achieved through abandoning the brood: [[Lambda].sub.c,x]([[E.sub.1].sup.*]) [less than] [[Lambda].sub.c,x]([[E.sub.2].sup.*]) (see Eq. 8). Thus, reproduction is zero and fitness is equal to adult survival. At a certain threshold in breeding conditions, [[Lambda].sub.c,x]([[E.sub.2].sup.*]) becomes larger than [[Lambda].sub.c,x]([[E.sub.1].sup.*]) and reproduction is profitable. At this threshold, survival decreases and reproduction increases abruptly. The threshold is a consequence of the following assumptions: (1) a constant clutch size; (2) an increased mortality of eggs and chicks for an increased mismatch between reproductive effort and clutch size; and (3) an increased survival cost of a given reproductive effort for decreasing breeding conditions. The general concave form of the reaction norms reflects the upper thresholds of fecundity and survival. The threshold for breeding is found at better breeding conditions for [G.sub.1] than for [G.sub.2]. This result is a consequence of the difference in potential fecundity. In order to determine the distribution of environments giving a constant population size, mean breeding conditions are found at a larger value and closer to the upper limits of reproduction and survival for [G.sub.2] than for [G.sub.1]. As a consequence of the concave form of the reaction norms, the variation in life history traits is smaller in [G.sub.2] than in [G.sub.1]. In seasons when reproduction is profitable, the model predicts that the relative value of reproduction in terms of fitness is higher in [G.sub. 1] than in [G.sub.2]. Thus, in general, we find a low probability but a high gain of reproduction in [G.sub.1], and a high probability but a low gain of reproduction in [G.sub.2].

The optimal response to a 50% increase in clutch size is given in Fig. 3B, C for the two genotypes. The environmental threshold for breeding is found at better breeding conditions among enlarged broods than among controls. Thus, under normal conditions, a large number of birds in the enlarged group will abandon the brood. However, when the threshold for breeding is reached, the parents with enlarged broods invest considerably more than the parents of the control group. The maximum difference in adult survival (19% in [G.sub.1] and 7% in [G.sub.2]), is reached at the breeding threshold. Such a response to the experiment is a consequence of (1) an increased mortality of eggs and chicks for an increased mismatch between reproductive effort and clutch size; (2) an increased survival cost of a given reproductive effort for decreasing breeding conditions; and (3) ability of the parents to assess a change in the demands of the brood. In other words, when the clutch size is increased under good breeding conditions, the parents should respond to the increased demands of the brood by increasing their reproductive effort, thereby reducing the mortality of young. They would, however, do this at the cost of their own subsequent survival. Under poor breeding conditions, they should desert their brood rather than respond to the increased demands, thereby maximizing their own survival. When breeding conditions are above the threshold, the value of the brood is larger in [G.sub.1] than in [G.sub.2]. Therefore, [G.sub.1] parents with enlarged broods will increase their effort to a higher cost (with respect to adult survival) than [G.sub.2] parents.


The model presented describes the relationships among parental effort, reproductive output, and adult survival in long-lived birds breeding in a stochastic environment. We assume that parents put in a flexible breeding effort and adjust it according to the likelihood that the young will survive and according to their own chances of breeding in the future. As a consequence of the assumptions made, there is a threshold in environmental conditions where breeding is profitable and where breeding effort increases abruptly. For long-lived species with a high potential fecundity, the threshold for breeding is higher than for long-lived birds with a low potential fecundity. When the brood size is experimentally enlarged, the threshold for breeding will be higher among enlarged than among control broods. Thus, during normal breeding conditions, most parents with enlarged broods will abandon their chicks. However, in years when the breeding conditions are good and the threshold for breeding is reached, parents with enlarged broods increase their breeding effort above that of controls and suffer from an increased mortality.

There are at least three important assumptions in the model that need to be considered. First, we assume that individuals are unable to adjust their clutch size at the time of egg-laying in response to the conditions they meet later in the reproduction cycle. Any mechanism that increases the female’s ability to assess breeding conditions at the time when the clutch is laid will reduce the effect of the predicted breeding threshold, i.e., that parents start to breed at poorer breeding conditions. Numerous studies have shown that females adjust their clutch size at the time of egg-laying to factors such as current body condition (e.g., Ankney and MacInnes 1978) and territory size (Hogtsedt 1980), and that laying terminates when body reserves reach some lower critical level (Ankney and MacInnes 1978). However, such proximate mechanisms of clutch size determination do not give information about ultimate mechanisms and, thereby, how females adjust their clutch to future requirements of the young. So far, the hypothesis has only been examined experimentally in the Pied Flycatcher Ficedula hypoleuca (Sanz and Moreno 1995). For seabirds, the most common pattern observed is that there is much larger variation in reproductive output than in clutch size. For example, in a study of the Black-legged Kittiwake Rissa tridactyla, Hatch and Hatch (1990) found no relationship between the variation in clutch size and the variation in reproductive output, suggesting that, at the time of egg-laying, the female has small possibilities of regulating clutch size according to future breeding conditions. Similar results were obtained by Jacobsen et al. (1995) in a clutch size manipulation study of the Black-legged Kittiwake, in which the number of chicks produced was found to be independent of both the original and experimental clutch size. Therefore, we argue that some of the variation in breeding conditions cannot be predicted and optimally adjusted for at the time of egg formation. Furthermore, we assume an increase in the mortality of eggs and chicks for an increased mismatch between reproductive effort and clutch size. In bad seasons, the clutch size is too large relative to the optimal reproductive effort, and the parents have to adjust the brood size accordingly. Such adjustments can be made through brood reduction (reviewed in Amundsen and Slagsvoid 1996) or flexible chick growth (e.g., Barrett and Rikardsen 1992), for instance. However, such mechanisms can never compensate completely for an “erroneous” clutch size (Apricario 1993) and, presumably, there will be an associated reproductive cost. In the model, we therefore use two time steps in the acquisition of information about breeding condition. First, the clutch size is determined by the probability distribution of the breeding conditions at the time of egg-laying, and second, the reproductive effort is adjusted according to the actual breeding conditions experienced.

A second crucial assumption in the model is that there is a cost, with respect to adult survival, for an increase in reproductive effort, and furthermore, that the parents are able to adjust their effort according to offspring need. Although widely accepted in the literature (e.g., Roff 1992, Stearns 1992), the concept of reproductive costs has been questioned for long-lived seabirds. It has been proposed that parents regulate their feeding effort according to a fixed schedule, independent of the chick’s need in order to maximize their own survival. Evidence supporting this hypothesis has been found in studies of the Leach’s Storm-Petrel Oceanodroma leucorhoa (Ricklefs and Minot 1991) and the Antarctic Petrel Thalassoica antarctica (Salther et al. 1993). However, many recent studies have shown that long-lived seabirds have a flexible investment in their young (Johnsen et al. 1994, Bolton 1995, Jacobsen et al. 1995). For example, a recent study of the Atlantic Puffin (Erikstad et al. 1997) has shown that parental effort is regulated by an interaction between the parents’ body condition and the chicks’ need and prospects. Furthermore, studies of the Glaucous-winged Gull Larus glaucescens and the Black-legged Kittiwake have showed that females rearing enlarged broods increased their effort, lost more mass, and had lower survival to the next breeding season (Reid 1987, Jacobsen et al. 1995). Thus, reproductive costs also seem to exist among long-lived species.

The third crucial assumption in the model is that, during poor breeding conditions, most birds should either not breed or abandon their eggs or chicks, and that such decisions should be made according to female body mass (prospects of survival), quality of the breeding season, and potential fitness of the brood. The model also predicts that seabirds with small clutch sizes have a lower threshold for breeding than seabirds with large clutch sizes.

It has been proposed that seabirds should refrain from breeding in bad seasons (e.g., Drent and Daan 1980) and that they need to reach an upper threshold in body condition before they can start to breed. When seabirds face food shortages, they may respond either by not breeding or by deserting their eggs or chicks during the breeding season (see Chastel et al. 1995a, b). A high frequency of nonbreeding has been described in a number of studies (e.g., Coulson 1984, Weimerskirch 1992, Hatch et al. 1993, Chastel et al. 1995a). The frequency of nonbreeding is very high in some years. Coulson (1984) noted that as many as 60-80% of experienced female Common Eiders Somateria mollissima did not breed in some seasons, and a similar level of nonbreeding ([approximately]50%) has been described in the Blue Petrel Halobaena caerulea in poor breeding seasons (Chastel et al. 1995b). In the latter study, nonbreeding was related to low body mass of females at the start of the breeding season.

Desertion of eggs and chicks when the adult body mass reaches some lower threshold is also a common pattern among seabirds. Studies of both the Blue Petrel (Chaurand and Weimerskirch 1994) and the Antarctic Petrel (Tveraa et al., in press) during the incubation period showed that parents deserted the egg when their body masses reached some critical lower threshold. Similar patterns have also been described for the Canada Goose Branta canadensis (Alderich and Raveling 1983) and for the Common Eider, in which the likelihood of egg or chick desertion is determined by both the female body mass and the size of the clutch (Erikstad et al. 1993, Erikstad and Tveraa 1995).

Desertion of chicks has also recently been demonstrated in an experimental study of the Atlantic Puffin Fratercula arctica (Johnsen et al. 1994). The decision whether or not to desert the chick was taken in response not only to the parents’ body condition, but also to the size of the chick, which may reflect its status and prospects for recruitment to the population (Erikstad et al. 1997). A non-experimental, long-term study of the Atlantic Puffin, also in northern Norway, gives further support to the threshold idea of investment in relation to breeding season quality (Anker-Nilssen 1992). Puffins studied by Anker-Nilssen relied on young herring Clupea harengus for food during the breeding seasons. In years when the spawning of herring was very low, Puffins did not produce eggs at all; in other years, they fed their chick for a week or so and then suddenly abandoned the nest and left the colony. There was, however, no evidence of a decrease in adult survival during such events (Anker-Nilssen 1992).

The most important consequence of the model assumptions is that, in some years, long-lived birds should also risk some of their own survival probability in order to produce young. Such a decrease in adult survival would occur in years when a certain threshold in environmental condition is reached. Long-lived birds with a high fecundity would risk their survival at a lower threshold in breeding conditions than those with a low fecundity. Only a few manipulation studies of parental effort in long-lived birds have measured the effect on adult survival (Lessells 1986, Reid 1987, Jacobsen et al. 1995, Tombre and Erikstad 1996). As predicted from the model, results of these studies vary. Glaucous-winged Gull (Reid 1987) and Black-legged Kittiwake (Jacobsen et al. 1995) studies documented a reduced survival of parents that increased their effort, whereas studies of Canada Geese and (Branta canadensis) and Barnacle Geese Branta leucopsis (Lessells 1989, Tombre and Erikstad 1996) did not document such an effect. However, such an effect is predicted from the model, given that the breeding season is good when the parental effort is increased (Tombre and Erikstad 1996).

In conclusion, we have demonstrated that the predicted response to an experimental manipulation of brood size may be a result of optimal decisions made by the parents in relation to the breeding conditions, the fitness of the brood, and their own prospects of survival. This hypothesis should be tested by experimental manipulation of brood size (or parental effort) over several years. Such experiments will give new insight into how long-lived birds optimize their lifetime reproduction in a stochastic environment.


The study was supported by the Research Council of Norway and The Norwegian Foundation for Nature Research and Cultural Heritage Research. We acknowledge Rolf A. Ims, Rob Barrett, and two anonymous reviewers for helpful comments on the manuscript.


Alderich, T. W., and D. G. Ravling. 1983. Effects of experience and body weight on incubation behavior of Canada Geese. Auk 199:670-679.

Amundsen, T., and T. Slagsvoid. 1996. Lack’s brood reduction hypothesis and avian hatching asynchrony: what next? Oikos 76:613-620.

Andersen, R., B.-E. Saether, and H. C. Pedersen. 1993. Resource limitation in a long-lived seabird, the Antarctic Petrel Thalassoica antarctica: a twinning experiment. Fauna Norwegica Serie C, Cinclus 16:15-18.

Andersen, R., B.-E. Saethen and H. C. Pedersen. 1995. Regulation of parental investment in the Antarctic Petrel Thalassoica antarctica: an exchange experiment. Polar Biology 15:65-68.

Anker-Nilssen, T. 1992. Food supply as a determinant of reproduction and population development in Norwegian Puffins Fratercula arctica. Dissertation. University of Trondheim, Trondheim, Norway.

Ankney, C. D., and C. D. MacInnes. 1978. Nutrient reserves and reproductive performance of female Lesser Snow Geese. Auk 95:459-471.

Apricario, J. M. 1993. The effect of clutch-size error on fitness: a hypothesis. Oikos 68:186-190.

Barrett, R. T, and F. Rikardsen. 1992. Chick growth, fledging periods, and adult mass loss of Atlantic Puffin Fratercula arctica during years of prolonged food stress. Colonial Waterbirds 15:24-32.

Bolton, M. 1995. Experimental evidence for regulation of food delivery to Storm Petrel, Hydrobates pelagicus, nestlings: the role of chick body condition. Animal Behavior 50:231-236.

Charlesworth, B. 1980. The evolution in age-structured populations. Cambridge University Press, Cambridge, UK.

Charnov, E. L., and J. R. Krebs. 1974. On clutch-size and fitness. Ibis 116:217-219.

Chastel, O., H. Weimerskirch, and P. Jouventin. 1993. High annual variability in reproductive success and survival of an Antarctic seabird, the Snow Petrel Pagodroma nivea: a 27-year study. Oecologia 94:278-285.

Chastel, O., H. Weimerskirch, and P. Jouventin. 1995a. Body condition and seabird productive performance: a study of three petrel species. Ecology 76:2240-2246.

Chastel, O., H. Weimerskirch, and P. Jouventin. 1995b. Influence of body condition on reproductive decision and reproductive success in the Blue Petrel. Auk 112:964-972.

Chaurand, T., and H. Weimerskirch. 1994. The regular alternation of short and long foraging trips in the Blue Petrel Halobaena caerulea: a previously undescribed strategy of food provisioning in a pelagic seabird. Journal of Animal Ecology 63:275-282.

Clutton-Brock, T. H., editor. 1988. Reproductive success. University of Chicago Press, Chicago, Illinois, USA.

Coulson, J. C. 1984. The population dynamics of the Eider Duck Somateria mollissima and evidence of extensive nonbreeding by adult ducks Ibis 126:525-543.

Coulson, J. C., and E. White. 1959. The postfledging mortality of the Kittiwake. Bird Study 6:97-102.

Curio, E. 1988. Relative and realized lifespan and delayed cost of parental care. American Naturalist 131:825-836.

Daan, S., C. Deerenberg, and C. Dijkstra. 1996. Increased daily work precipitates natural death in the Kestrel. Journal of Animal Ecology 65:539-544.

Dijkstra, C., A. Bult, S. Bijlsma, S. Daan, T Meijer, and M. Zijlstra. 1990. Brood size manipulations in the Kestrel Falco tinnunculus. Journal of Animal Ecology 59:269-285.

Drent, R., and S. Daan. 1980. The prudent parent: energetic adjustment in avian breeding. Ardea 68:225-252.

Erikstad, K. E., M. Asheim, P. Fauchald, L. Dalhaug, and T Tveraa. 1997. Adjustment of parental effort in the Puffin; the roles of adult body condition and chick size. Behavioral Ecology and Sociobiology 40:95-100.

Erikstad, K. E., J. O. Bustnes, and T Moum. 1993. Clutch-size determination in precocial birds: a study of the Common Eider. Auk 110:623-628.

Erikstad, K. E., and T. Tveraa. 1995. Does the cost of incubation set limits to clutch size in Common Eider Somateria mollissima? Oecologia 103:270-274.

Hamer, K. C., and J. K. Hill. 1994. The regulation of food delivery to nestling Cory’s Shearwaters Calonectris diomedea: the roles of parent and offspring Journal of Arian Biology 25:198-204.

Hatch, S. A., and M. Hatch. 1990. Components of breeding productivity in a marine bird community: key factors and concordance. Canadian Journal of Zoology 68:1680-1690.

Hatch, S. A., B. Roberts, and B. S. Fadly. 1993. Adult survival of Black-legged Kittiwakes Rissa tridactyla in a Pacific colony. Ibis 135:247-254.

Hogstedt, G. 1980. Evolution of clutch-size in birds: adaptive variation in relation to territory quality. Science 210: 1148-1150.

Hudson, P. J. 1985. Population parameters for the Atlantic Alcidae. Pages 233-261 in D. N. Nettleship and T. R. Birkhead, editors. The Atlantic Alcidae. Academic Press, London, UK.

Jacobsen, K. O., K. E. Erikstad, and B.-E. Saether. 1995. An experimental study of the costs of reproduction in the Kittiwake Rissa tridactyla. Ecology 76:1636-1642.

Johnsen, I., K. E. Erikstad, and B.-E. Saether. 1994. Regulation of parental investment in a long-lived seabird, the Puffin Fratercula arctica. Oikos 71:273-278.

Lessells, C. M. 1986. Brood size in Canada Geese: a manipulation experiment. Journal of Animal Ecology 55:669-690.

Linden, M., and A. P. Moller. 1989. Cost of reproduction and covariation of life history traits in birds Trends in Ecology and Evolution 4:367-371.

Mauck, R. A., and T. C. Grubb. 1995. Petrel parents shunt all experimental increased reproductive costs to their offspring Animal Behaviour 49:999-1008.

Monaghan, P., J. D. Uttley, and M.D. Burns. 1992. Effect of changes in food availability on reproductive effort in Arctic Tern Sterna paradisea. Ardea 80:71-81.

Monaghan, P., J. D. Uttley, M.D. Burns, C. Thaine, and J. Blackwood. 1989. The relationship between food supply, reproductive effort and breeding success in Arctic Terns Sterna paradisea. Journal of Animal Ecology 58:261-274.

Newton, I., editor. 1989. Lifetime reproduction in birds. Academic Press, London, UK.

Orzack, S. H., and S. Tuljapurkar. 1989. Population dynamics in variable environments. VII. The demography and evolution iteroparity. American Naturalist 133:901-923.

Reid, W. V. 1987. The cost of reproduction in the Glaucouswinged Gull. Oecologia 74:458-467.

Ricklefs, R. E. 1977. On the evolution of reproductive strategies in birds: reproductive effort. American Naturalist 111:453-478.

—– 1987. Response of adult Leach’s Storm-Petrels to increased food demand at the nest. Auk 104:750-756.

—– 1992. The roles of parent and chick in determining feeding rates in Leach’s Storm Petrels. Animal Behaviour 43:895-906.

Ricklefs, R. E., and E. O. Minor. 1991. Concluding remarks: parent-offspring interactions in the evolution of avian life histories Acta Congressus Internationalis Ornithologici 20: 992-997.

Roff, D. A. 1992. The evolution of life histories. Chapman and Hall, New York, New York, USA.

Saether, B.-E., R. Andersen, and H. C. Pedersen. 1993. Regulation of parental effort in a long-lived seabird: an experimental manipulation of the cost of reproduction in the Antarctic Petrel Thalassoica antarctica. Behavioural Ecology and Sociobiology 33:147-150.

Sanz, J. J., and J. Moreno. 1995. Experimental induced clutch-size enlargements affect reproductive success in the Pied Flycatcher. Oecologia 103:358-364.

Schaffer, W. M. 1974. Optimal reproductive effort in fluctuating environments. American Naturalist 108:783-790.

Stearns, S.C. 1992. The evolution of life histories Oxford University Press, Oxford, UK.

Tombre, I., and K. E. Erikstad. 1996. Regulation of parental effort: an experimental study of the high arctic Barnacle Geese Branta leucopsis. Journal of Animal Ecology 65: 325-331.

Tveraa, T., B.-E. Saether, and S. H. Lorentsen. 1997. Regulation of foraging trips and the cost of incubation shifts in the Antarctic Petrel Thalassoica antarctica: an experiment. Behavioral Ecology 8:465-469

Vader, V., R. T. Barrett, K. E. Erikstad, and K. B. Strann. 1990. Differential responses of Common and Thick-billed Murres to a crash in the capelin stock in the Barents Sea. Studies in Avian Biology 14:175-180

Weimerskirch, H. 1992. Reproductive effort in long-lived birds: age-specific patterns of condition, reproduction and survival in the Wandering Albatross. Oikos 64:464-473.

Williams, G. C. 1966. Natural selection, the cost of reproduction, and a refinement of Lack’s principle American Naturalist 100:687-690

Wooller, R. D., J. S. Bradley, and J.P. Croxall. 1992. Long-term population studies of seabirds. Trends in Ecology and Evolution 7:111-114.

COPYRIGHT 1998 Ecological Society of America

COPYRIGHT 2000 Gale Group

You May Also Like

Line transect sampling under varying conditions with application to aerial surveys

Line transect sampling under varying conditions with application to aerial surveys Pham X. Quang Line transect sampling is a method …

Molecular Ecology and Evolution: Approaches and Applications.

Molecular Ecology and Evolution: Approaches and Applications. – book reviews Thomas S. Kantz The rapid development of molecular tech…

Impact of water calcium on the phenotypic diversity of alpine populations of Gammarus fossarum

Impact of water calcium on the phenotypic diversity of alpine populations of Gammarus fossarum Jean-Claude Meyran INTRODUCTION <p…

Biodiversity and the Law.

Biodiversity and the Law. – book reviews Douglas A. Henderson As noted conservationist Aldo Leopold concluded in A Sand County alman…