Habitat loss and population decline: a meta-analysis of the patch size effect

Ecology, March, 1998 by Darren J. Bender, Thomas A. Contreras, Lenore Fahrig

INTRODUCTION

Human activities, such as forest clear-cutting or the expansion of agricultural land, have exacerbated the natural fragmentation of landscapes. For many organisms, new habitat patches are interspersed in an unfamiliar or hostile environment, which leads to the heterogeneous distribution of populations at different scales (Weins 1989a, Gilpin and Hanski 1991). There has been considerable interest in modeling the spatial dynamics of populations in patchy or fragmented landscapes so that we may better understand how the spatial pattern of the landscape influences population processes (reviewed by Levin 1976, Hastings 1990, Kareiva 1990, Dunning et al. 1995). Recently, there has been a growing interest in applying these models in conservation strategies (Fahrig and Merriam 1994), particularly for determining the consequences of habitat fragmentation on resident populations (e.g., Murphy and Noon 1992, Lamberson et al. 1994, Baz and Garcia-Boyero 1995, Turner et al. 1995).

Habitat fragmentation is a term that has been used in many different ways. Habitat fragmentation, by definition, is an event that creates a greater number of habitat patches that are smaller in size than the original contiguous tract(s) of habitat. Yet, the term commonly is used to describe human practices that destroy habitat. This usage is misleading because there are situations in which habitat can be removed without fragmenting the landscape whatsoever. To avoid confusion, we define our terms here. We use the term habitat destruction to refer to processes, particularly anthropogenic, that remove habitat cover. Habitat destruction can then be pictured as having two distinct components: habitat loss per se and habitat fragmentation (Fahrig 1997). The effect of losing habitat is obvious: when habitat is lost from the landscape, the animals that are subsequently displaced may also be lost, producing a population decline. The effects of fragmentation are less obvious, but they are very important; fragmentation effects can potentially compound the effects of pure habitat loss, often producing an even greater population decline.

We examined the most notable effect of habitat destruction: the reduction in average patch size. In patchy or fragmented landscapes, patch size effects should be relatively simple to detect. Consider the following illustration. Habitat loss is expected to produce a proportional decline in the number of animals living in a particular landscape. For example, if a piece of forest habitat supports a large population of some animal and 50% of that forest is removed, then one might expect a decline in animal abundance of 50%. However, it often has been found that species abundance declines beyond that predicted by habitat loss alone. This difference stems from the effects of reduced mean patch size and decreased connectivity in the landscape (i.e., a reduction in the rate of successful dispersal, sensu Merriam 1984; see also Venier and Fahrig 1996). A common way to test for patch size effects is to compare the relative densities of organisms within different-sized patches. If patch size effects account for any additional decline in abundance, then we would expect density to be positively related to patch size. Therefore, the strength of the relationship between patch size and density can be used to index the strength of the patch size effect.

Our first goal in this study was to evaluate the conditions under which patch size effects are important determinants of local population size for animals living in patchy landscapes. To accomplish this, we quantitatively reviewed results of published studies that tested for the presence of patch size effects on population density. Using these results, we generalized about the conditions under which patch size effects influence population size. Because habitat fragmentation creates patchy landscapes, such generalizations are useful for generating predictions about situations in which patch size effects will emerge when a landscape undergoes loss and fragmentation. Thus, our second goal was to formulate testable predictions about how patch size effects are expected to affect the population size of animals following habitat loss and fragmentation.

METHODS

We adopted a procedure that is relatively new to ecological research (Arnqvist and Wooster 1995), known as meta-analysis. Meta-analysis is a powerful, quantitative form of analysis used for summarizing and analyzing multiple independent studies. In this form of analysis, the statistical results (as opposed to the raw data) of numerous studies are analyzed to determine whether studies share a common statistical relationship, and whether any "general" relationships are influenced by one or more predictor variables. Although the benefits and criticisms of using meta-analysis in ecological research can be found elsewhere (Gurevitch et al. 1992, Gurevitch and Hedges 1993, Arnqvist and Wooster 1995), we wish to emphasize that meta-analysis is particularly suited to studies that examine large-scale phenomena. Large-scale studies are often impractical to conduct because of the time, effort, and money that must be devoted to a single project. For example, it is very difficult to experimentally test hypotheses about habitat fragmentation because the units of observation are landscapes. However, a large body of literature deals with research conducted on single landscapes. Meta-analysis is a more practical approach to studying large-scale questions, because it allows one to accumulate information from these independent studies and to view each as replicates or observations in subsequent statistical analysis. Using meta-analysis, one can test for relationships that occur between the characteristics of a study (e.g., landscape attributes) and their outcomes.