The importance of stopover habitat for developing effective conservation strategies for migratory animals


Although stopover habitats are used by many species as refuelling stations during migration and can be critical for survival and successful reproduction, they are rarely incorporated in year-round population models and conservation strategies. We incorporate stopover habitat into a density-dependent population model and then use this model to examine how optimizing one-time land purchase strategies for a migratory population is influenced by variation in the quality and the strength of density-dependence in a stopover habitat used for both fall and spring migration. As the strength of the density-dependence in the stopover habitat increases, the optimal amount of stopover habitat purchased increases while the amount of habitat during the stationary periods of the annual cycle (breeding and wintering) decreases. Any change in the cost of purchasing stopover habitat affects investment strategies in all three periods of the annual cycle. When the quality of the stopover habitat is high, the optimal strategy is to invest in low-quality habitat during breeding and wintering and when the stopover habitat quality is low, the optimal strategy switches to investing in high-quality habitat during the stationary periods. We apply this model to a threatened warbler population and demonstrate how purchase decisions to conserve stopover habitat that are not coordinated with conservation actions on the breeding and wintering grounds can potentially result in a lower population carrying-capacity compared to considering habitat in all three periods of the annual cycle simultaneously. Our model provides potential guidelines for developing conservation strategies for animals that rely on refueling habitats between the stationary breeding and non-breeding periods of the migratory cycle.


Although evidence suggests that migratory animal populations are influenced by the interaction of events throughout the annual cycle (Fretwell 1972; Bêty et al. 2003; Bearhop et al. 2004; Norris et al. 2004; Norris and Taylor 2006), conservation strategies for species typically focus on a single period of the year (e.g., Robbins et al. 1992; Basili and Temple 1999; Dankel et al. 2008; but see Klaassen et al. 2008). Stopover sites are used by many species as key refueling stations during migration and have a large influence on the rate of mass gain (Bairlein 1985; Kelly et al. 2002; Delingat et al. 2006; Schaub et al. 2008), which can, in turn, have consequences for the timing (e.g., Smith and Moore 2003) and success (e.g., Sandberg and Moore 1996; Bêty et al. 2003; Drent et al. 2003) of reproduction, as well as survival during the stationary non-breeding season (e.g., Haramis et al. 1986; Pfister et al. 1998). Possible mechanisms by which stopover habitat can influence migratory populations include density-dependent intra-specific competition for limited resources (e.g., Russell et al. 1992; Kelly et al. 2002) or variation in habitat quality via weather (e.g., Schaub and Jenni 2001). Thus, stopover sites should be a critical component of conservation plans.

Here, we use a three-season, state-structured population model to optimize the carrying-capacity of a migratory population that occupies multiple breeding and wintering habitats and a single stopover site. The model is designed to predict how to optimally allocate funds to purchase habitat across these three periods of the annual cycle given the relative strength of density-dependence and habitat cost, differences in habitat quality within a season, density-dependence on the stopover site, general life history, and total budget size. We use our model to explore how the optimal amount of breeding, wintering, and stopover habitat is affected by the density-dependence, habitat cost, and quality of the stopover habitat, as well as by the general life history of the species. We then apply the model to a threatened migratory warbler population to examine how stopover sites influence optimal conservation decisions.

Population model

In a two-season model, the annual cycle begins in the wintering (or non-breeding) season, W, and ends after the breeding season B. The population size, N, during the breeding season, N B, can be represented as the population size during the previous wintering season, N W, multiplied by the per capita reproductive output during the breeding season (b t):

$$ N_{{B_{t} }} = N_{{W_{t} }} (1 + b_{t} ) $$

The population size on the wintering grounds, N W, can be represented as the population size at the end of the breeding season in the previous year multiplied by the probability of an individual surviving the wintering season (d t):

$$ N_{{W_{t} }} = N_{{B_{t - 1} }} (d_{t} ) $$

The per capita reproductive output (b t) and the wintering survival (d t) can be represented as linear functions:

$$ b_{t} = b - b^{'} N_{{{\text{W}}_{t} }} $$
$$ d_{t} = d - d^{'} N_{{B_{t - 1} }} $$

where b and d are density-independent parameters (intrinsic habitat quality as the population approaches zero) for the breeding and non-breeding periods, respectively, and b′ and d′ are density-dependent parameters for the same periods. The values of b and d can be influenced by the overall habitat quality, with higher values associated with higher quality breeding and wintering habitat, respectively. The values of b′ and d′ are affected by changes in population size or amount of habitat. Although the values of b′ and d′ are small (e.g., b′ = 0.00005; Sutherland 1996), they can have a significant impact on population abundance. Since an individual’s mass from the stopover habitat may affect reproductive success (breeding grounds) or survival (wintering grounds), b and d can then be multiplied by a fitness function for the stopover site, (f ss = ω − ωN), where ω and ω′ represent the density-independence (intrinsic quality) and density-dependence at the stopover site, respectively. The equations for N B, and N W can now be modified such that:

$$ N_{{B_{t} }} = N_{{W_{t} }} (1 + (b(\omega - \omega^{'} N_{{W_{t} }} ) - b^{'} N_{{W_{t} }} )) $$
$$ N_{{W_{t} }} = N_{{B_{t - 1} }} (d(\omega - \omega^{'} N_{{B_{t - 1} }} ) + d^{'} N_{{B_{t - 1} }} ) $$

The value of ω varies between 0 and 1, with 1 being the quality of the stopover site in which there is no effect on b or d (i.e., highest quality). For ω, a value of 0 implies that density-dependence during stopover has no effect on b or d. The stopover habitat may appear to act in a similar fashion as a carry-over effect (i.e., a residual effect in one season that can carry-over to influence individual success in the following season) from one stationary period to another (Norris 2005; Norris and Taylor 2006), but there are important differences. Where the carry-over effect acts in a single direction (e.g., from winter to summer), the stopover habitat function influences both b and d over the course of an annual cycle. Also, carry-over effects outlined by Norris (2005) and Norris and Taylor (2006) only affect the habitat quality parameter of the stationary period, while the stopover parameters incorporated here influence both quality and density-dependent parameters during stationary periods.

When the population is at equilibrium, the population size at the end of both breeding and wintering seasons are equal to the sizes during the previous year \( (N_{{W_{t} }} = N_{{W_{t - 1} }} ,N_{{B_{t} }} = N_{{B_{t - 1} }} ) \). We define the carrying-capacity, K, as the population size at the end of the wintering period, N W, at equilibrium because it captures both birth and death processes over the course of a single annual cycle and is, therefore, always the lowest population estimate for any given period of the year. Defining K as the population size at the end of the breeding period risks developing ‘optimal’ conservation plans for populations that could potentially go extinct by the end of the following winter season due to high mortality rates.

Conservation model

Sheehy et al. (2010) showed that the relative density-dependence between the breeding (b′) and wintering (d′) habitats, along with the relative habitat costs (C B and C W) and density-independence (b and d), can be used to predict the optimal proportion of each habitat to purchase for a migratory population that occupies a single breeding and wintering habitat. Assuming that when habitat is lost the population will occupy the remainder of the habitat, such that the new density will equal the previous density multiplied by the inverse of the proportion of habitat remaining, N B and N W can be written as:

$$ N_{{B_{t} }} = N_{{W_{t} }} \left( {1 + \left( {b\left( {\omega_{spring} - {\frac{{\omega^{'}_{spring} }}{x}} N_{{W_{t} }} } \right) - b^{'} N_{{W_{t} }} } \right)} \right) $$
$$ N_{{W_{t} }} = N_{{B_{t - 1} }} \left( {d\left( {\omega_{fall} - {\frac{{\omega_{fall}^{'} }}{x}} N_{{B_{t - 1} }} } \right) - b^{'} N_{{B_{t - 1} }} } \right) $$

where p, q, and x are the proportion of breeding, wintering, and stopover habitat purchased, respectively (all vary between 0 and 1). To incorporate two different quality habitats in the breeding and wintering season, we assume an equal area within each habitat and that there is a higher cost associated with the high-quality habitat. A difference in habitat quality is represented by variation in the density-independent parameter (d high, d low for wintering, b high, b low for breeding). Thus, d and b are the weighted averages of these parameters. For example, d is:

$$ d = {\frac{{q_{\text{low}} (d_{\text{low}} ) + q_{\text{high}} (d_{\text{high}} )}}{{q_{\text{low}} + q_{\text{high}} }}} $$

A similar equation applies in a two-quality breeding habitat model, and in both cases higher values are associated with higher qualities. For both breeding and wintering habitats, the proportion purchased is the average of the two different quality habitats purchased. For example, p is:

$$ p = {\frac{{p_{\text{high}} + p_{\text{low}} }}{2}} $$

Cost constraints

CI is the cost of purchasing habitat during the given season and is equal to the amount of habitat available, LI (in hectares), multiplied by the cost per unit of habitat, PI (in dollars/hectare), where I = B (breeding), W (wintering), or S (stopover site). We assume that the total budget, Ct, is fixed and is always less that the cost of purchasing all of the winter, breeding, or stopover habitat. Thus, the optimal strategy will always entail spending the entire budget, such that:

$$ C_{t} = (p_{\text{low}} )(C_{{B_{\text{low}} }} ) + (p_{\text{high}} )(C_{{B_{\text{high}} }} ) + (q_{\text{low}} )(C_{{W_{\text{low}} }} ) + (q_{\text{high}} )(C_{{W_{\text{high}} }} ) + (x)(C_{S} ) $$

Dividing C B, C W and C S by C t gives:

$$ {\text{l}} = (p_{\text{low}} )(C_{{B_{\text{low}} }}^{*} ) + (p_{\text{high}} )(C_{{B_{\text{high}} }}^{*} ) + (q_{\text{low}} )(C_{{W_{\text{low}} }}^{*} ) + (q_{\text{high}} )(C_{{W_{\text{high}} }}^{*} ) + (x)(C_{S}^{*} ) $$

where C *B , C *W and C *S are the ratios of the costs needed to purchase all L B, L W and L S in relation to the total budget, respectively.


We ran simulations to maximize K (population size at the end of the wintering period) in the following way. For a given set of parameter values, we varied the amount of each habitat and/or habitat quality purchased between 0 and 1 in increments of 0.001 to find the strategy that resulted in the highest K and also met the constraint of being less than or equal to C t. All simulations were run using PELLES C (ver. 5.0). We used population parameter estimates from Eurasian Oystercatchers (Haematopus ostralegus; Sutherland 1996) and obtained approximate land cost values from the MBCC 2008 Report (Migratory Bird Conservation Commission 2008). These parameter estimates were used to establish realistic values for the simulations and not to investigate conservation strategies specifically for the Oystercatcher.

Model analysis and results

Effect of including stopover habitat on population model

Because the abundance parameters (N W or N B) in the reproductive output and survival equations are multiplied by the density-dependent and density-independent stopover parameters [Eqs. (5, 6)], an increase in the intrinsic quality of the breeding or wintering habitat (b or d) increases the density-dependence on the stopover habitat during the following migration. This, in turn, buffers the positive effect of an increase in the breeding or wintering habitat quality on population size. In other words, there is a relatively smaller response of K to changes in b or d compared to models that do not incorporate stopover habitat because of the two-way compensatory response at the stopover site (Fig. 1).

Fig. 1

The percentage decrease in K (population size at the end of the wintering period) in relation to the percentage decrease of b (intrinsic growth rate for breeding period) for a population model which does not include stopover habitat [Eqs. (3) and (4)] and a model that includes stopover habitat [Eqs. (5) and (6)]. For both models, d = 0.95, b′ = 0.00005, and d′ = 0.00011. For the model that incorporates stopover habitat, ω = 1 and ω′ = 0.0000275

Effect of density-dependence and cost of stopover habitat on purchase decisions

The between-season purchase decisions are determined by the relative density-dependence (Fig. A1A in supplementary material) and cost [Electronic Supplementary Material (ESM) Fig. A1B] between the wintering, breeding, and stopover habitats (Fig. 2a). As the strength of the density-dependence on the stopover site (ω′) increases, the optimal amount of stopover site purchased increases and the optimal amount of the breeding and wintering habitat decreases (Fig. 2a). More generally, a change in the cost of habitat in any of the three periods influences the optimal purchase strategy for all three periods. For example, as the cost of purchasing stopover habitat (C S) increases, the amount of habitat to purchase in all three periods of the annual cycle should be reduced. However, the greatest reduction in purchase should occur during the period when the cost of habitat has decreased (results not shown). Lastly, because the density-dependence for the stopover habitat is multiplied by the intrinsic growth rates (b and d), any increase in these values (through purchasing high-quality habitat) may also increase the amount of stopover habitat purchased. This has the effect of actually decreasing the amount of breeding and wintering habitat purchased (results not shown).

Fig. 2

The optimal proportion of habitat purchased on the breeding, wintering, and stopover habitat in relation to the mean strength of density-dependence on the stopover site (ω′) (a) and the strength of density-independence (intrinsic quality) on the stopover site (ω) (b). For a, low-quality breeding and wintering habitats are not shown as they closely follow the general trends displayed in the high-quality habitats. The optimal proportion of habitat is equal to the amount purchased divided by the amount available. For a, ωspring varies and ω = 1. For b, ωspring = 0.0000125 and ω varies. For b, the dotted line represents the value of ω at which there is no effect on the intrinsic growth rates (b and d). For a and b, b′ = 0.00005, d′ = 0.00011, b high = 0.4, b low = 0.35, d high = 0.95, d low = 0.94, \( C_{{B_{\text{high}} }}^{*} \) = 1.85, \( C_{{B_{\text{low}} }}^{*} \) = 1.4, \( C_{{W_{\text{high}} }}^{*} \) = 2.6, \( C_{{W_{\text{low}} }}^{*} \) = 2.15, \( C_{S}^{*} \) = 2, and \( \omega_{\text{fall}}^{'} \) = 0.0000275

Effect of stopover habitat quality on purchase decisions

Which quality of breeding or wintering habitat to invest in depends on the quality of the stopover site (ω). If ω = 1 (high-quality stopover site) and the optimal purchase strategy is to invest in low-quality habitat (in either breeding or wintering seasons), then there is a threshold value of ω in which the strategy changes to investing in high-quality habitat. Otherwise, ω will not have an effect on the quality of habitat invested in during either stationary period of the annual cycle. Based on the parameter values we used (refer to Fig. 2 legend), ω typically has an effect on the quality of wintering habitat purchased and does not affect the quality of the breeding habitat (Fig. 2b). However, it is possible for ω to only affect the investment in the type of breeding habitat, both breeding and wintering habitat, simultaneously (ESM Fig. A2A), or neither (ESM Fig. A2B). This depends on the relative qualities and costs within a season.

Effect of life history on purchase decisions

When the ratio of b:(1 − d) is small (low breeding output relative to mortality, akin to K-selected species), a higher overall quality of stopover habitat is required to maintain a positive population growth rate compared to when the b:(1 − d) ratio is large (akin to r-selected species; ESM Fig. A2A). Furthermore, consistent with Sheehy et al. (2010), the optimal purchase strategy typically involves purchasing high-quality wintering habitat for K-selected species and low-quality wintering habitat for r-selected species. Thus, variation in the quality of the stopover habitat will typically not affect purchase strategies for K-selected species (ESM Fig. A2B) but will affect strategies for r-selected species (ESM Fig. A2A).

Application of the model

The Hooded Warbler (Wilsonia citrina) is a small (approx 10 g) long-distance migratory passerine that is listed as threatened under the Canadian Species at Risk Act (COSEWIC 2000), with the only Canadian breeding population located in southwestern Ontario. Based on stable isotope studies, evidence suggests that the Ontario population uses stopover sites in the gulf coast of the USA during the spring and fall migration (Langin et al. 2009). Populations over-winter from east-central Mexico to Belize (Ogden and Stutchbury 1994). We used habitat quality and density-dependent estimates from this species and closely related wood-warblers (Black-throated Blue Warbler Dendroica caerulescens, American Redstart Setophaga ruticilla, and Wilson’s Warbler Wilsonia pusilla; see Appendix and Table A1 in the ESM) and assumed that the size of the stopover habitat was equal to half of the available breeding/wintering habitat.

Sheehy et al. (2010) used land cost estimates from the breeding grounds in southern Ontario and the wintering grounds in Belize to estimate the optimal amount of habitat to be purchased between these two periods of the annual cycle. With a total budget equal to the combined costs of these two parcels of habitat, the optimal strategy was to purchase 164 ha of high-quality breeding habitat and 95 ha of high-quality wintering habitat, despite the fact that breeding habitat was over tenfold more expensive. Here, we ask what the optimal purchase decision would be if we added a single stopover site and used a state-structured model. In 2008, the Cat Island National Wildlife Refuge in Louisiana, which Hooded Warblers commonly use as stopover habitat during both the spring and fall migration, was expanded by 345 ha at a cost of USD $1.755 million (MBCC 2009). Thus, we added $1.755 million to the budget (total = $4.055 million), assumed that the stopover habitat was not yet purchased, and then calculated the optimal combination of habitat purchases when considering all three periods of the annual cycle simultaneously. If these extra funds were included in the model from Sheehy et al. (2010) with no stopover habitat incorporated, the optimal strategy would be to purchase 288 ha of high-quality breeding habitat and 166 ha of high-quality wintering. Interestingly, when we included stopover habitat into our stage-structured model, we found that the optimal strategy was to not spend most of the additional budget on stopover habitat, but to purchase less (266 ha) high-quality breeding habitat in southern Ontario, more (420 ha) high-quality wintering habitat in Belize, and only 12 ha of stopover habitat in Louisiana (see Appendix in ESM for calculations). Thus, for this particular strategy, funds that would have been allocated towards the breeding habitat are now directed towards the wintering and stopover habitat. This purchase strategy resulted in a K of 69 individuals, which was higher than that when all of the extra funds was invested solely on the stopover site (345 ha; K = 57). This strategy also yielded a higher K than that obtained when the combination of land purchase decisions was based solely on what would be for sale in the current market (29 ha of breeding habitat, the 2,500 ha of the wintering habitat, and the 345 ha of stopover habitat; K = 0; see Appendix in the ESM for calculations), which is not a formal coordinated effort to protect this species across these three periods of the annual cycle. Of course, these results apply when there is limited funding and do not address how to maximize long-term population persistence.

We also performed a sensitivity analysis and found that the optimal amounts of each habitat purchased were most sensitive to changes in the breeding habitat parameters and least sensitive to changes in the stopover habitat parameters (for both density-dependence and habitat cost; ESM Table A2). For example, a 25% decrease in the strength of breeding density-dependence resulted in purchasing 4 ha less high-quality breeding habitat (−1.4%), 49 ha more high-quality wintering habitat (+11.7%), and 1 ha more stopover habitat (+7.7%). In comparison, a 25% decrease in the strength of density-dependence in the stopover habitat (during the fall migration) resulted in purchasing 0.94 ha more high-quality breeding habitat (+0.4%), 6.6 ha more high-quality wintering habitat (−1.6%), and 0.95 ha less stopover habitat (−7.7%). However, the strategy remained to invest primarily in high-quality wintering habitat and purchase high-quality breeding habitat and small amounts of stopover habitat (ESM Table A2).


We show how a model with stopover parameters will generally lead to a more stable carrying-capacity, meaning that changes in the intrinsic quality of the habitat during the stationary periods of the annual cycle will have less of an impact on K when a stopover site is included. This is a general result that should be widely applicable to most systems regardless of the structure of the model. Since density-dependent and -independent effects from stopover habitats have been shown to influence individual success in migratory species (e.g., Russell et al. 1992; Schaub and Jenni 2001; Kelly et al. 2002), a model incorporating stopover sites should also have a better ability to predict future changes in the population size of migratory species.

Our results also demonstrate the importance of including stopover habitat into optimal conservation strategies. In our model, the density-dependence, cost, and quality of the stopover habitat had a significant impact, not only on the amount of habitat purchased at the stopover site but also on habitat purchased on the breeding and wintering grounds. The strength of density-dependence and the cost of the stopover habitat had similar effects on the optimal decisions as analogous parameters on the stationary breeding and non-breeding periods of the year (Sheehy et al. 2010). As the cost of the stopover habitat increases, it is optimal to invest less in all three periods of the migratory cycle. As the strength of density-dependence at the stopover site increases, the relative importance of the stopover site for maximizing K increases, and it therefore becomes optimal to invest more in stopover habitat. Again, these results should be applicable to a wide variety of models and parameter values.

We also found that the mean quality of the stopover habitat influences whether to purchase high- or low-quality habitat during the wintering or breeding period. When the quality of the stopover habitat (ω) is low, it is optimal to invest in high-quality habitat during the stationary period to compensate for the poor condition of the individuals arriving from the stopover site. If only low-quality habitat is purchased, then the survival rate is too low to maintain the population. Conversely, when the quality of the stopover habitat is high, it becomes optimal to invest in low-quality wintering and/or breeding habitat, which capitalizes on the low cost of the low-quality habitat and results in a larger amount of habitat that is conserved and a higher K.

Surprisingly, we found that the quality of the stopover habitat has the opposite effect on the purchase strategies compared to the breeding or wintering habitat quality. When habitat quality increases on either the breeding or wintering grounds, the optimal strategy is to purchase smaller amounts of high-quality habitat during all other periods of the annual cycle (Sheehy et al. 2010). However, because we have shown here that it is optimal to purchase low-quality winter habitat when the quality of the stopover is high, the amount of funding for purchasing additional habitat is also relatively high. As a consequence, the optimal strategy involves purchasing additional habitat in all three periods of the migratory cycle. Thus, somewhat counter intuitively, the two general purchase strategies involve either a large amount of high-quality stopover habitat or a small amount of low-quality habitat. This result reinforces the importance of estimating the quality of the stopover habitat prior to the formation of any conservation strategy.

Our model can be applied to developing conservation strategies for migratory songbirds. Using the Hooded Warbler as an example, we have shown that current efforts to purchase these habitats (that are not coordinated) would lead to a lower carrying-capacity compared to the results predicted from our optimization model. The model also predicted that an optimal K would be achieved by purchasing only a small fraction of the stopover habitat that was available (12 of 345 ha) and investing more in the wintering habitat. This is likely due to the fact that stopover habitat in this example had a much smaller density-dependence to cost ratio than habitats in the other periods of the annual cycle.

For Hooded Warblers, we also found that the optimal habitat purchase strategy is most sensitive to changes in breeding habitat parameters, likely due to higher density-dependence on the breeding grounds relative to other seasons of the annual cycle. When the density-dependence on the breeding grounds is decreased, the relative importance of the wintering and stopover habitat increases, and the amounts of both of these habitats increase by a higher percentage due to their lower habitat costs. Conversely, stopover habitat parameters have a very small effect on the results due to the small density-dependence to cost ratio.

Our model framework can be applied to any species that relies on stopover sites during migration. For example, Sockeye salmon (Oncorhynchus nerka) typically gather at the mouths of major river outlets to wait for optimal conditions prior to migrating upstream to spawn (Levy and Cadenhead 1995). During southward migration, Monarch butterflies (Danaus plexippus) often congregate in the thousands at discrete coastal locations before crossing large water bodies (Mackenzie and Friis 2006). Our model can also be used to design conservation strategies for trans-Saharan migratory birds, who breed in Europe and Asia and use stopover sites in the Mediterranean and northern Africa. One interesting case is the Aquatic Warbler (Acrocephalus paludicola), which is one of the most threatened long-distance migratory passerines in the world. A significant portion of the breeding and stopover habitat used by this species has been identified and protected (Heredia et al. 1996; Atienza et al. 2001). However, large portions of wintering habitat are not protected and presumably under threat (Schaffer et al. 2006). Applying our models to this species would provide an estimate of how much wintering habitat would be necessary to maximize population size and how much, if any, additional breeding and stopover habitats need to be conserved. It is important to note, however, that applying our model to these examples requires detailed information on the habitat quality, the cost of conserving habitats, and how these habitats influence subsequent reproduction or survival. Given this information, it is plausible that, in some situations, the optimal strategy would be to divert funding towards the protection of stopover sites in order to maximize the global carrying-capacity. How much habitat to protect would depend on the strength of the density-dependence and the cost.

In the model presented here, we assumed that any habitat lost on the stopover habitat was of average quality, which best applies to small decreases in habitat size (Sutherland 1996). However, it is relatively easy to incorporate multiple stopover habitats into an optimization model like the one presented here. The challenge will be to track the degree of habitat loss and identify different quality stopover sites in migratory species. For simplicity, we only included a single stopover habitat, whereas most migratory species typically use a series of stopover locations between the breeding and wintering habitats (e.g., Shimazaki et al. 2004; Lehnen and Krementz 2005), and some species will even use different sets of stopover locations depending on the season (e.g., Spear and Ainley 1999). Useful extensions of this model would be to incorporate multiple stopover sites in a spatially and temporally explicit framework and to incorporate multiple stopover habitats over the course of the annual cycle.

We also assumed that, when habitat was lost, the remainder of the individuals would crowd into the remaining habitat [Eqs. 7, 8)]. This implies that individuals cannot exclude others from the remaining habitat, a situation that may or may not hold for some species. However, two observations suggest that this assumption may be quite robust. First, most habitat that is lost in a given year is relatively small compared to the total habitat available. Thus, even in territorial species, individuals that formerly occupied lost habitat are likely to find space in the remaining habitat. Second, when large amounts of habitat are lost, then the strength of the density-dependence increases significantly in the crowded population. Compensatory effects of increased mortality and decreased birth rates in the remaining habitat likely produce similar patterns compared to cases when mortality increases simply due to exclusion.

Our model demonstrates that designing effective conservation strategies for migratory animals requires accurate estimates of density-dependence, habitat quality, and the costs of performing conservation actions at stopover habitats. Although obtaining such parameter estimates will be challenging for many species, such information is essential for conserving migratory populations that rely on stopover habitats during migration.


  1. Atienza JC, Pinilla J, Justribo JH (2001) Migration and conservation of the aquatic warbler Acrocephalus paludicola in Spain. Ardeola 48:197–208

    Google Scholar 

  2. Bairlein F (1985) Body weights and fat deposition of Palaearctic passerine migrants in the central Sahara. Oecologia 66:141–146

    Article  Google Scholar 

  3. Basili GD, Temple SA (1999) Dickcissels and crop damage in Venezuela: defining the problem with ecological models. Ecology 9:732–739

    Google Scholar 

  4. Bearhop S, Hilton GM, Votier SC, Waldron S (2004) Stable isotope ratios indicate that body condition in migrating passerines is influenced by winter habitat. Proc R Soc Lond B Biol Sci 271:S215–S218

    Google Scholar 

  5. Bêty J, Gauthier G, Giroux J-F (2003) Body condition, migration and timing of reproduction in Snow Geese, a test of the condition-dependent model of optimal clutch size. Am Nat 162:110–121

    Article  Google Scholar 

  6. Committee on the Status of Endangered Wildlife in Canada (2000) COSEWIC assessment and update status report on the Hooded warbler Wilsonia citrina in Canada. Committee on the Status of Endangered Wildlife in Canada, Ottawa

  7. Dankel DJ, Skagen DW, Ulltang O (2008) Fisheries management in practice: review of 13 commercially important fish stocks. Rev Fish Biol Fish 18:201–233

    Article  Google Scholar 

  8. Delingat J, Dierschke V, Schmaljohann H, Mendel B, Bairlein F (2006) Daily stopovers as optimal migration strategy in a long-distance migrating passerine: the northern wheatear Oenanthe oenanthe. Ardea 94:593–605

    Google Scholar 

  9. Drent R, Both C, Green M, Madsen J, Piersma T (2003) Pay-offs and penalties of competing migratory schedules. Oikos 103:274–292

    Article  Google Scholar 

  10. Fretwell SD (1972) Populations in a seasonal environment. Princeton University Press, Princeton

    Google Scholar 

  11. Haramis GM, Nichols JB, Pollock KH, Hines JE (1986) The relationship between body mass and survival of wintering canvasbacks. Auk 103:506–514

    Article  Google Scholar 

  12. Heredia B, Rose L, Painter M (1996) Globally threatened birds in Europe, action plans. Council of Europe Publishing, Strasbourg

    Google Scholar 

  13. Kelly JF, DeLay LS, Finch DM (2002) Density-dependent mass gain by wilson’s warblers during stopover. Auk 119:210–213

    Article  Google Scholar 

  14. Klaassen M, Bauer S, Madsen J, Possingham HP (2008) Optimal management of a goose flyway: migrant management at minimum cost. J Appl Ecol 45:1446–1452

    Article  Google Scholar 

  15. Langin KM, Marra PP, Németh Z, Moore FR, Kyser TK, Ratcliffe LM (2009) Breeding latitude and timing of spring migration in songbirds crossing the Gulf of Mexico. J Avian Biol 40:309–316

    Article  Google Scholar 

  16. Lehnen SE, Krementz DG (2005) Turnover rates of fall-migrating pectoral sandpipers in the lower mississippi alluvial valley. J Wildl Manage 69:671–680

    Article  Google Scholar 

  17. Levy DA, Cadenhead AD (1995) Selective tidal stream transport of adult sockeye salmon (Oncorhynchus nerka) in the Fraser River estuary. Can J Fish Aquat Sci 52:1–12

    Article  Google Scholar 

  18. Mackenzie SA, Friis CA (2006) Long Point bird observatory 2005 field operations report. Long Point bird observatory, Long Point (Port Rowan)

  19. Migratory Bird Conservation Commission (2008) In: 2007 Annual Report—Migratory Bird Conservation Commission. U.S. Fish and Wildlife Service, Arlington, pp 1–50

  20. Migratory Bird Conservation Commission (2009) In: 2008 Annual Report—Migratory Bird Conservation Commission. U.S. Fish and Wildlife Service, Arlington, pp 1–53

  21. Norris DR (2005) Carry-over effects and habitat quality in migratory populations. Oikos 109:178–186

    Article  Google Scholar 

  22. Norris DR, Taylor CM (2006) Predicting the consequences of carry-over effects in migratory populations. Biol Lett 2:148–151

    Article  Google Scholar 

  23. Norris DR, Marra PP, Kyser TK, Sherry TW, Ratcliffe LM (2004) Tropical winter habitat limits reproductive success on the temperate breeding grounds in a migratory bird. Proc R Soc B Biol Sci 271:59–64

    Article  Google Scholar 

  24. Ogden LJ, Stutchbury BJ (1994) Hooded Warbler (Wilsonia citrina). In: Poole A, Gill F (eds) The birds of North America. Academy of Natural Sciences/American Ornithologists’ Union, Philadelphia/Washington D.C.

  25. Pfister C, Kasprzyk MJ, Harrington BA (1998) Body fat levels and annual return in migrating semipalmated sandpipers. Auk 115:904–915

    Article  Google Scholar 

  26. Robbins CS, Fitzpatrick JW, Hamel PB (1992) A warbler in trouble: Dendroica cerulea. In: Hagan JM, Johnston DW (eds) Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington D.C., pp 549–562

    Google Scholar 

  27. Russell RW, Carpenter FL, Hixon MA, Paton DC (1992) The impact of variation in stopover habitat quality on migrant rufuous hummingbirds. Conserv Biol 8:483–490

    Article  Google Scholar 

  28. Sandberg R, Moore FR (1996) Fat stores and arrival on the breeding grounds: reproductive consequences for passerine migrants. Oikos 77:577–581

    Article  Google Scholar 

  29. Schaffer N, Walther BA, Gutteridge K, Rahbek C (2006) The African migration and wintering grounds of the aquatic warbler Acrocephalus paludicola. Bird Conserv Int 16:33–56

    Article  Google Scholar 

  30. Schaub M, Jenni L (2001) Variation of fuelling rates among sites, days and individuals in migrating passerine birds. Funct Ecol 15:584–594

    Article  Google Scholar 

  31. Schaub M, Jenni L, Bairlein F (2008) Fuel stores, fuel accumulation, and the decision to depart from a migration stopover site. Behav Ecol 19:657–666

    Article  Google Scholar 

  32. Sheehy J, Taylor C, Norris DR (2010) Optimal conservation planning for migratory animals: integrating demographic information across seasons. Conserv Lett 3:192–202

    Article  Google Scholar 

  33. Shimazaki H, Tamura M, Higuchi H (2004) Migration routes and important stopover sites of endangered oriental white storks (Ciconia boyciana) as revealed by satellite tracking. Mem Natl Inst Polar Res 58:162–178

    Google Scholar 

  34. Smith RJ, Moore FR (2003) Arrival fat and reproductive performance in a long-distance passerine migrant. Oecologia 134:325–331

    Article  Google Scholar 

  35. Spear LB, Ainley DG (1999) Migration routes of sooty shearwaters in the pacific ocean. Condor 101:205–218

    Article  Google Scholar 

  36. Sutherland WJ (1996) Predicting the consequences of habitat loss for migratory populations. Proc R Soc B Biol Sci 263:1325–1327

    Article  Google Scholar 

Download references


DRN was supported by grants from the Natural Sciences and Engineering Research Council of Canada, and an Early Researcher Award from the Ontario Ministry of Innovation and Training. Kevin McCann provided valuable comments on earlier drafts of the manuscript.

Author information



Corresponding author

Correspondence to Justin Sheehy.

Additional information

Communicated by F. Bairlein.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 90609 kb)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Sheehy, J., Taylor, C.M. & Norris, D.R. The importance of stopover habitat for developing effective conservation strategies for migratory animals. J Ornithol 152, 161–168 (2011).

Download citation


  • Conservation models
  • Population dynamics
  • Migratory birds
  • Stopover sites