1. Habitat and habitat selection at different spatial and temporal scales
2. Habitat selection: The behavior
3. Implications of habitat selection for basic and applied ecology
Separately and in combination, the terms habitat and selection mean different things to different audiences. This chapter focuses on habitat selection behavior at the level of individuals and considers how the processes that affect the choices made by organisms at different spatial scales affect the distributions at the population level. Because we initially focus on habitat selection at the level of individuals, habitat can be defined as a location in which a particular organism is able to conduct activities that contribute to survival and/or reproduction. That is, habitat is organism-specific rather than being determined by features that may be obvious to humans (e.g., vegetation type). Selection can be defined as a behavioral process by which an organism chooses a particular habitat in which to conduct specific activities. Hence, habitat selection implies that individual organisms have a choice of different types of habitat available to them and that they actively move into, remain in, and/or return to certain areas rather than others.
conspecific attraction. Attraction of individuals to conspecifics during the process of habitat selection
habitat selection. The process by which individuals choose areas in which they will conduct specific activities
heterospecific attraction. Attraction of individuals to other potentially competing species during the process of habitat selection
indirect cues. Stimuli that are produced by factors that are correlated with other factors with direct effects on intrinsic habitat quality
intrinsic habitat quality. The expected fitness of an individual when it uses or lives in a given habitat, after controlling for any effects of conspecifics on fitness
microhabitat. An area used for a specific type of activity (e.g., foraging, oviposition, nesting)
natal habitat preference induction. Exposure to cues in an individual’s natal habitat increases the attractiveness of those cues during habitat selection
Habitats and habitat selection can occur at several different spatial and temporal scales. At larger scales, habitat refers to areas that are required for the long-term survival and reproduction of the members of a given population. In this case, habitat includes all of the areas required by all of the life stages of the members of that population, including areas that allow dispersers to travel among different patches of suitable habitat. For instance, from the perspective of a migratory bird, habitat includes breeding habitat, wintering habitat, and migratory stopovers that connect these venues. Habitat selection at large spatial and temporal scales occurs when individuals choose localities or regions that might be capable of supporting them, their offspring, and their descendents for an extended period of time.
At intermediate spatial and temporal scales, habitat refers to an area capable of supporting an individual for a biologically significant, finite period of its lifetime. Examples of habitat at this spatial scale include the selection of a feeding territory by a juvenile salmonid or an area suitable for feeding and oviposition by a female butterfly. Habitat selection at this scale is particularly important for sessile organisms such as barnacles because in this case a decision made early in life affects an individual’s fitness for the rest of its life. In contrast, mobile organisms may select new habitats several times over the course of their lives as a result of changes in resource requirements, experience, or competitive ability during development, or as a consequence of seasonal movements from one area to another.
Finally, at even smaller spatial and temporal scales, habitat refers to an area in which an organism is able to conduct specific activities, such as foraging, resting, courtship, oviposition, or parental care. The term microhabitat is often used to refer to such areas. With the exception of sessile species, microhabitat selection typically occurs multiple times and involves many different types of habitats over the course of an individual’s lifetime.
Recently, it has become apparent that scale matters and that models that predict behavior and distributions at small spatial scales may do a less satisfactory job at predicting them at larger spatial scales. In order to appreciate how scale affects habitat selection, it is helpful to consider one of the most influential general models of habitat selection, the ideal free distribution (IFD) (Fretwell and Lucas, 1970). The IFD predicts the area that an animal will select under the assumption that animals have accurate estimates of the intrinsic quality of the different areas that are available to them. In turn, intrinsic quality indicates the net fitness payoff that an individual can expect when it is using an area, after controlling for any effects of conspecifics on fitness. This particular model also assumes that animals compete with one another when they are using a habitat, such that fitness is inversely related to population density. Finally, it assumes that animals incur no costs when they are searching for a habitat, so they are always free to choose the habitat that maximizes their expected fitness. Under this simplified scenario, the probability that an individual will select a given area will be positively related to the relative intrinsic quality of that area and inversely related to the density of other individuals in that area.
Considerable empirical support for the IFD has been obtained in studies of habitat selection at small spatial scales, e.g., in studies of foraging patch selection conducted in tanks in the laboratory, or restricted areas in the field. This is not surprising because these are situations in which the assumptions of the ideal free distribution are most likely to be satisfied. Empirical support at microhabitat scales has also been obtained for modified versions of the IFD, e.g., models that assume that individuals differ with respect to competitive ability and sort themselves among habitat patches based on their competitive ability relative to the other individuals with whom they interact on a regular basis.
In contrast, studies of habitat selection at larger spatial and temporal scales often yield results at odds with the predictions of the IFD. These discrepancies have drawn attention to assumptions of this model that may not apply when animals select regions, localities, neighborhoods, home ranges, or territories for long-term use. One of the first assumptions of the IFD to be reevaluated was that individual fitness is inversely related to conspecific density. An alternative possibility is an Allee distribution, in which individual fitness increases as a function of density at low to intermediate densities and then declines at intermediate to high densities. Thus far, Allee distributions in the context of habitat selection have been documented for a wide array of taxa, including territorial animals, as well as species that live in colonies or groups. Of course, if individuals benefit from the presence of conspecifics at low to intermediate densities, then one would expect different patterns of habitat choice than if interactions between conspecifics were entirely competitive. For example, whereas the IFD predicts that empty patches of suitable habitat would be more attractive than a comparable habitat containing a moderate number of conspecifics, habitat selection under an Allee distribution predicts the reverse: newcomers should avoid an empty patch in favor of a patch that already contains members of their species.
Indirect Cues and the Effects of Conspecifics and Heterospecifics on Habitat Selection
One of the assumptions of the IFD that is scale-dependent is that organisms have accurate estimates of the relative intrinsic quality of all of the habitats that are available to them. This assumption is most likely to be valid when individuals are able to extensively sample and evaluate many different habitats. For instance, an animal that has lived in a home range for an extended period of time probably has reasonable estimates of the relative quality of different foraging patches within its home range. However, extensive sampling of potential habitats is less feasible when animals are choosing habitats at larger spatial and temporal scales. In this situation, sampling can be constrained by many factors, including limits on the amount of time or energy that is available to search for and investigate novel habitats or elevated risk of mortality as animals explore unfamiliar areas. In addition, because large-scale habitat selection involves areas that individuals will use for extended periods of time, there is no guarantee that factors with major impacts on fitness will even be present when individuals choose a habitat. For instance, the larval dispersers of some benthic marine invertebrates settle in spring, when important attributes of the attachment site (e.g., exposure to hot temperatures in midsummer, or exposure to storm surges in winter) cannot be directly assessed.
When direct assessment of habitat quality is not an option, organisms may rely on indirect cues of intrinsic habitat quality. Indirect cues are stimuli that can be reliably detected when organisms are searching for habitats and that are correlated with biotic or abiotic factors that affect fitness after they have chosen a habitat. For many years, biologists have focused on indirect cues involving structural features of the habitat, e.g., shapes, colors, odors, sounds, or other cues that are likely to be correlated with other factors (food, predators, parasites, etc.) with direct impacts on fitness when animals are using a habitat. More recently, researchers have expanded the notion of indirect cues to include stimuli produced by conspecifics and hetero-specifics. Thus, the conspecific cuing hypothesis argues that the presence or number of conspecifics in an area can provide information about other factors (predators, food supplies, parasites, etc.) that affect the intrinsic quality of that area. Similarly, the conspecific performance hypothesis argues that cues related to the reproductive performance of conspecifics in an area may provide information about factors that affect breeding performance in that area. Even other species that are potential competitors of a focal species may provide indirect cues to habitat quality, as is outlined in the heterospecific cuing hypothesis. These hypotheses have broadened the notion of indirect cuing to include the possibility that conspecifics, successful conspecific breeders, and heterospecifics affect habitat selection behavior because they provide information about other factors with major impacts on intrinsic habitat quality.
Indirect Cues versus Direct Benefits in the Effects of Cues on Habitat Selection
The issue of the effects of conspecifics and heterospecifics on habitat selection raises an important general question, namely, whether organisms respond to particular cues when selecting habitats because those cues are correlated with other factors that affect intrinsic habitat quality (an indirect cue), or whether they respond to those cues because they are produced by factors that directly affect fitness after they choose a habitat (direct benefits). That is, individuals may be attracted to cues from conspecifics or heterospecifics because these cues are associated with other factors that affect habitat quality (conspecific cuing, heterospecific cuing) and/or because individuals directly benefit from the presence of conspecific or heterospecific neighbors after settling in a habitat. As a result, two mutually nonexclusive hypotheses predict that individuals will be attracted to conspecifics (conspecific attraction) or to heterospecifics (heterospecific attraction) when choosing a habitat. In fact, a growing number of empirical studies indicate that individuals are attracted to conspecifics, to successful conspecifics, or to heterospecifics when choosing a habitat. However, it is not yet clear whether indirect cues, direct benefits, or both contribute to positive effects of cues from conspecifics or heterospecifics on habitat choice. Fortunately, this is a question that is currently under study, and we can expect more information on this topic in the coming years.
The Development of Responses to Cues during Habitat Selection
Over the years, it has become apparent that animals have two potential sources of information about associations between cues and intrinsic habitat quality: information from previous generations, via genes and maternal effects, and information from an individual’s immediate past, via learned associations between cues and factors with direct effects on habitat quality. Information from the past is reflected by preexisting biases, which in the context of habitat selection are expressed when naive individuals with no previous exposure to a natural cue are attracted to that cue. For instance, in nature, larval anemonefish disperse from their natal habitat (a sea anemone of a particular species) and then attempt to locate and settle on a new host anemone. Naive larvae raised in the absence of sea anemones are more strongly attracted to the odor of their usual host anemone than to the odor of other sea anemones. Similar examples of differential attraction by naive individuals to cues produced by “ancestral” habitats have been reported in many different taxa.
An individual’s personal experience can modify or alter preexisting biases for the attractiveness of cues from natural habitats. For instance, exposure to cues in an animal’s natal habitat may increase the attractiveness of those cues to natal dispersers, a process termed “natal habitat preference induction” (NHPI). Thus far, NHPI has been reported for a number of taxa, including many insects and a scattering of vertebrates. The latter includes the anemonefish mentioned earlier, in which larvae raised in the presence of their typical host anemone are more strongly attracted to olfactory cues from their host anemone than are naive larvae. Preexisting biases and personal experience interact throughout development to affect preferences for cues for particular patches or types of habitat. As a result, nature and nurture both affect the ways that animals respond to indirect cues during the process of habitat selection.
By definition, habitat selection at larger spatial scales involves longer travel distances and higher travel costs than habitat selection at smaller spatial scales. Travel costs include increased risk of mortality from predators, accidents, infection, or other adverse conditions individuals face when traveling through unfamiliar, often inhospitable, terrain. These costs are compounded when individuals have difficulty detecting suitable habitats from a distance or when suitable habitat is sparsely distributed in the landscape. In such cases, individuals may travel considerable distances without finding any suitable habitat. Moreover, habitat selection at larger spatial scales is often time- or energy-limited. For instance, natal dispersal in brush mice is restricted to a 1- to 2-week period before sexual maturation, and natal dispersal times and distances in bark beetles are constrained by the fact that they do not feed en route and must rely on energy stored before they leave their natal habitat.
Long travel distances, high travel costs, and time- or energy-limited search constrain habitat selection behavior in ways that can often be safely ignored in studies of microhabitat selection. In the absence of these constraints, it is reasonable to assume that preference and choice will map onto one another and that organisms will select the habitat that they perceive (correctly or incorrectly) will yield the highest fitness. However, when organisms choose habitats at larger spatial scales, preference is only one of several factors that affect habitat choice. For example, when energy-limited individuals search for a habitat, theory predicts that individuals with higher energy reserves will be more selective than individuals with lower energy reserves. In that case, individuals in poor condition will be more likely to accept less-preferred habitats early in the search and less likely to end up settling in highly preferred habitats than individuals in good condition. This is an example of a “silver spoon effect” in which favorable conditions early in life increase the chances that an individual will be successful later in life. In fact, this type of silver spoon effect has been documented in studies of benthic marine invertebrates (bryozoans), in which larvae with large food reserves are more selective and more likely to settle in highly preferred habitats than larvae with lower food reserves.
Adding search to habitat selection highlights a number of other reasons why animals should accept less-preferred habitats, even though more-preferred habitats exist elsewhere in the same landscape. High travel costs, difficulty in detecting suitable habitats, long distances between suitable habitats, a shortage of time or energy available for search, and a scarcity of highquality habitats are all factors that will favor individuals who are relatively nonselective when they are searching for a new habitat. Nonselective individuals still prefer some habitats to others and can express these preferences if provided with a choice of habitats located directly next to one another. However, under natural conditions, nonselective individuals will be more likely to accept any suitable habitat rather than incur the added costs of continuing to search for a more-preferred habitat. In turn, reduced selectivity during search increases the proportion of individuals that end up choosing habitats in proportion to their availability. Hence, individuals that are highly selective at smaller spatial scales (e.g., when choosing foraging patches within a home range or territory) may be considerably less so when choosing a habitat at larger spatial scales (e.g., when selecting a region in which to settle, or establishing a home range or territory within that region).
Most ecologists and conservation biologists are not nearly as interested in the behavioral processes that generate habitat selection as they are in the effects of these processes on animal distributions and population viability. Indeed, in the ecological literature, the term habitat selection usually does not refer to the behavior of individual animals but rather to differential patterns of habitat use. In this literature, habitat selection is inferred when the density of individuals in a particular type of habitat is higher than predicted on the basis of a null model that assumes that individuals use different types of habitat in proportion to their availability. By extension, it is assumed that higher-than-predicted densities in a given type of habitat occur because organisms preferentially settle in, use, or remain in that type of habitat. However, active habitat choice is only one of several factors that can produce differential habitat use, so this assumption need not always be valid. For instance, newly settled larvae of benthic marine fish and invertebrates are often strongly associated with certain types of habitat. In the past, researchers assumed that differential patterns of habitat use by new recruits were a result of active habitat choice, but recent studies indicate that habitat-specific predation in the hours to days immediately following settlement contributes to these patterns. Because most researchers had assumed that the factors affecting the mortality of new arrivals were the same as the factors affecting the mortality of settled larvae, habitat-specific mortality early in the settlement period in larval recruits went undetected for many years.
A major goal of studies of habitat selection in the ecological literature is to identify the types of habitat that are most suitable for the members of a population or species. The notion that differential habitat use reflects differences in intrinsic habitat quality rests on yet another assumption, namely that organisms accurately estimate the relative intrinsic quality of different types of habitat, so that preference and performance are positively correlated across different types of habitat. If this assumption is valid, then the relative abundance of organisms in a given type of habitat may provide useful information about the quality of that type of habitat, relative to the quality of other types of habitat in the same area.
The assumptions outlined in the previous two paragraphs are actually quite similar to the underlying assumptions of the IFD. As a result, differential habitat use patterns are most likely to reflect habitat quality when the assumptions of the IFD are satisfied. Recall that the IFD assumes that individuals compete with one another while living in a habitat and that all of the individuals in a species are comparable with respect to their competitive ability. However, if individuals benefit from the presence of conspecifics after settling in a habitat, then the density of individuals in a given type of habitat need not reflect the relative quality of that type of habitat. For example, if individuals prefer to settle in the company of conspecifics, lower-quality patches that contain a moderate number of conspecifics may be more attractive to both local recruits and to potential immigrants than empty patches of higher-quality habitat. Alternatively, if individuals differ with respect to their competitive ability, then highly competitive individuals may be able to exclude less-competitive individuals from higher-quality habitats. In this case, there is no guarantee that population densities will be higher in habitats of higher intrinsic quality, even if these habitats are preferred by every member of the species.
The use of indirect cues in habitat selection can also disrupt relationships among population density, habitat preferences, and habitat quality. Even under the best of circumstances, the association between indirect cues and habitat quality is correlational rather than causal, so that organisms that rely on indirect cues will occasionally prefer lower-quality habitats by mistake. And because indirect cues provide only approximate estimates of habitat quality, organisms that rely on them are likely to have difficulty discriminating among habitats that do not differ very much with respect to habitat quality. Indeed, when organisms rely on indirect cues for habitat selection, ecologists with accurate estimates of habitat-specific mortality and reproductive rates probably have a better notion of the relative quality of different types of habitat than do the organisms that are selecting those habitats.
Although associations between indirect cues and habitat quality have always been imprecise, humans have contributed more than their share to the disruption of correlations between indirect cues and habitat quality. The recent literature on “ecological traps” considers cases in which a sudden environmental change (e.g., addition of a novel predator, altered habitat structure) has decoupled indirect cues from the true quality of the type of habitat that produces them. Most empirical studies of ecological traps have focused on situations in which humans are responsible for changing correlations between indirect cues and habitat quality. Examples include birds that preferentially settle in plantations of exotic trees rather than natural forests but that suffer lower nesting success in the former as a result of nest predation, or mayflies that prefer to lay their eggs on dry asphalt roads rather than ponds because asphalt reflects more of the polarized light that these animals use to choose oviposition sites. Hence, even if indirect cues used to be strongly correlated with habitat quality, there is no guarantee that this is still the case in today’s altered world.
Another situation in which indirect cues can encourage mismatches between habitat preferences and relative habitat quality occurs when the attractiveness of indirect cues increases after exposure to those cues in the natal environment (NHPI). This is because NHPI encourages animals to select new habitats that are comparable to their natal habitat, even if other types of higher-quality habitats are available in the same landscape. Thus, NHPI may help explain situations in which animals raised in degraded habitats are reluctant to recruit to nearby patches of restored, high-quality habitat, or in which captive-raised or translocated animals fail to settle in habitats that are known to be of high quality for the members of their species.
Even if we are willing to assume that organisms have perfect estimates of the intrinsic quality and the density of conspecifics at every habitat that is available to them, and that every individual prefers the same type of habitat, adding search to habitat selection further complicates relationships among habitat preferences, habitat quality, and population density. When organisms have to search for a habitat, preference is no longer the only factor affecting habitat choice. Instead, the optimal behavior for a given individual depends not only on the benefits of finding a high-quality habitat but also on the costs of searching for it. Thus, if patches of high-quality habitat are rare and sparsely distributed, and if search is time- or energy-limited, then theory suggests that most of the individuals in a population will be relatively nonselective and, hence, likely to settle in habitats in proportion to their availability. As a result, habitat fragmentation and habitat degradation will not only reduce the amount of habitat that is available to support a population but also shift behavior in a direction that discourages individual selectivity and encourages individuals to accept habitats in proportion to their availability in the landscape. When this happens, an analysis of habitat use in relation to habitat availability might conclude (correctly) that individuals were not exhibiting habitat selection. However, it might also conclude (incorrectly) that individuals do not prefer some types of habitat to others or that all of the available habitats were of comparable intrinsic quality.
A number of other factors that occur when animals search for habitats can affect relationships among habitat choice, habitat preference, and the distribution of individuals. For instance, species with small perceptual ranges may have difficulty detecting suitable habitats. If searching individuals run a strong risk of not finding any suitable habitat, and if low-quality habitats produce cues that can be detected at longer distances than the cues from high-quality habitats, then individuals should be differentially attracted to, and differentially settle in, low- rather than high-quality habitats. This scenario may help account for the fact that pest species such as aphids recruit to large expanses of agricultural crops rather than to isolated patches of their native host plants, even though those crops are less suitable for feeding and oviposition than the native hosts. The condition of the individuals who are selecting habitats may also affect relationships between preference and choice because, as was noted above, when time- or energy-limited animals are searching for habitats, individuals in poor condition are expected to be less selective during search than individuals in good condition. Hence, habitat degradation may not only reduce the survivorship and reproductive success of individuals who live in those lower-quality habitats but also produce individuals who lack the stamina or stored resources necessary to locate and settle in patches of higher-quality habitat.
In conclusion, habitat selection behavior is scale-dependent. Although simple habitat-selection models do a reasonable job of predicting individual behavior and spatial distributions involving habitat selection at smaller spatial scales, more complex models may be required to predict patterns of habitat selection at larger spatial scales. Recent theoretical and empirical studies of habitat selection at larger spatial scales have expanded traditional models to consider situations in which organisms benefit from the presence of conspecifics or heterospecifics after settlement, rely on indirect cues to assess habitat quality, or incur costs when searching for potentially suitable habitats. On the debit side, this recent body of work provides a number of reasons why differential patterns of habitat use at larger spatial scales may not provide reliable estimates of either habitat preferences or intrinsic habitat quality. On the positive side, these recent studies have generated a number of new hypotheses about habitat-selection behavior, some of which have already been supported in studies of habitat selection at larger spatial scales. This new body of work provides possible explanations for distribution patterns that have been observed in nature and offers suggestions that may help applied biologists manage the habitat selection behavior of species of concern to humans.
Flaxman, Samuel M., and Christina A. de Roos. 2007. Different modes of resource variation provide a critical test of ideal free distribution models. Behavioral Ecology and Sociobiology 61: 877–886. A recent study illustrating how the predictions of the IFD have been tested and validated for habitat selection behavior at small spatial scales.
Fretwell, Steven D., and H. L. Lucas. 1970. On territorial behavior and other factors influencing habitat distribution in birds. I. Theoretical development. Acta Biotheoretica 19: 16–36. The classic paper that presented the initial model.
Trengenza, T. 1995. Building on the ideal free distribution. Advances in Evolutionary Research 26: 253–307. A general review of the topic.
Jones, Jason. 2001. Habitat selection studies in avian ecology: A critical review. Auk 118: 557–562. A review of habitat selection studies involving birds.
Stamps, Judy A. 2001. Habitat selection by dispersers: Integrating proximate and ultimate approaches. In J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, eds., Dispersal. Oxford: Oxford University Press, 230–242. A review of factors affecting habitat selection behavior at larger spatial scales, with a focus on natal dispersers.
Sutherland, William J. 1996. From Individual Behaviour to Population Ecology. Oxford: Oxford University Press. A book that illustrates the ways that IFD models and modifications of these models can be used to study habitat selection at small to intermediate spatial scales.
Underwood, Anthony J., Gee M. Chapman, and Tasman P. Crowe. 2004. Identifying and understanding ecological preferences for habitat or prey. Journal of Experimental Marine Biology and Ecology 300: 161–187. A review of methods of studying habitat preferences with an emphasis on marine organisms.
Johnson, Matthew D. 2007. Measuring habitat quality: A review. The Condor 109: 489–504. A review of methods for estimating habitat quality with a focus on birds.
Robertson, Bruce A., and Richard L. Hutto. 2006. A framework for understanding ecological traps and an evaluation of existing evidence. Ecology 87: 1075–1085. A recent review of empirical studies of “ecological traps”: situations in which indirect cues are no longer strongly correlated with habitat quality.
Stamps, Judy A., and V. V. Krishnan. 2005. Nonintuitive cue use in habitat selection. Ecology 86: 2860–2867. An overview of the use of indirect cues in habitat selection, with an emphasis of the use of indirect cues for habitat selection at larger spatial scales.