Independent Effects of Forest Cover and Fragmentation on the Distribution of Forest Breeding Birds

The aims o f this study were ( 1 ) to determine the relative importance o f the independent effects o f forest cover and fragmentation on the distribution o f forest breeding birds, and ( 2 ) to test the hypothesis that the negative ef fect o f forest fragmentation on species distribution increases with decreasing forest cover, i.e., the negative interaction ef fect o f forest cover and fragmentation on distribution. The independent effects o f forest cover and forest fragmentation on the distribution o f forest breeding birds were studied in 94 landscapes, 10 X 10 k m each, ranging in forest cover from 2.5% to 55.8%. For each landscape, percent forest cover was measured, and a fragmentation index (independent o f forest cover) was generated using PCA from the measures o f mean forest patch size, number o f forest patches, and total forest edge. Presence o f 3 1 forest breeding bird species in each landscape was determined using Breeding Bird Atlas data. The effects o f forest cover and forest fragmentation on species presence were analyzed using multiple logistic regression. All responses o f individual species to forest cover were positive. Responses to forest fragmentation were weak and variable. There were only two species for which the interaction between cover and fragmentation was significant (one positive, one negative). W e found no evidence for the hypothesized negative interaction ef fect between forest cover and forest fragmentation on species distribution. W e conclude that ( 1 ) forest cover at the 10 X 10 k m (Universal Transverse Mercator [UTM]) scale has a positive ef fect on the distribution o f forest breeding birds, (2 ) forest fragmentation does not have a consistent negative ef fect on the distribution o f forest breeding birds, ( 3 ) the ef fect o f forest cover is greater than that o f forest fragmentation, ( 4 ) responses to forest fragmentation are generally weak and highly variable among species, and ( 5 ) the ef fect o f forest fragmentation on species distribution does not increase with decreasing forest cover. These results suggest that conservationists' primary focus should be on preventing a decrease in forest cover. They should not be misled by recent discussions o f "fragmentation effects" to think that the negative effects o f forest loss can somehow be countered by careful consideration o f the spatial pattern o f remaining forest. Our results indicate that they cannot. Key \i.ords: agricultural landscapes: birr1 distribution: Breeding Bird Atlas; forest cover: fragmentatiotz: habitat loss: landscape pcittern; multiple logistic regression: Ontario: Quebec: SPANS GIs; UTM squares. extinction (den Boer 1981); and ( 2 ) providing temporal Both habitat loss and fragmentation are usually and spatial refugia for prey (Huffaker 1958, Levin thought to negatively affect species persistence (Wil1976, Kareiva 1987, Roland 1993). cox and Murphy 1985, Wilcove et al. 1986, Burkey Because the effects o f habitat amount and fragmen1989, Herben et al. 1991, Noss 1991, Perry and Gontation are confounded in many studies, it is not clear zalez-Andujar 1993, Adler and Nuernberger 1994). Exwhich has the greater impact (Haila and Hanski 1984). planations for the negative impact o f fragmentation inSimulation results o f Fahrig (1997, 1998) predict that clude increased mortality o f individuals moving bethe effects o f habitat loss are much greater than the tween patches, lower recolonization rates o f empty effects o f habitat fragmentation on population extincpatches, and reduced local population sizes resulting tion. To date, the only empirical test o f the relative in increased susceptibility to extinction (Fahrig and importance o f habitat loss and fragmentation was conMerriam 1994). There is also some evidence that habducted by McGarigal and McComb (1995), who comitat fragmentation can have a positive ef fect on regional pared the relative effects o f forest cover and fragmenpopulation survival by ( 1 ) desynchronizing the dynamtation on bird species abundances in late-sera1 forest ics o f local populations, thereby spreading the risk o f in Oregon. They found that the ef fect o f habitat amount was much greater than the ef fect o f habitat pattern. However, their study was conducted in a predominantly Manuscript received 29 September 1997; accepted 29 April 1998: final version received 26 May 1998. forested area, and they urged comparison o f their rePresent address: Department of Biology, Dalhousie Unisults with studies conducted in predominantly agriculversity. Halifax, Nova Scotia B3H 451, Canada. tural landscapes. The first objective o f this study was May 1999 FOREST COVER AND FRAGMENTATION 5 87 FIG. 1 . Locations of 94 10 X 10 km landscapes to determine the relative importance of the independent effects of forest cover and forest fragmentation on the distribution of forest breeding birds in a predominantly agricultural area. A few studies have suggested that fragmentation has the strongest effect when the amount of suitable habitat is low (McLellen et al. 1986, AndrCn 1994, Fahrig 1998), and that habitat fragmentation has an insignificant effect at high amounts of habitat. In other words, they predict a negative interaction effect between habitat fragmentation and habitat amount. The implication of this prediction is that when suitable habitat is abundant, the degree of fragmentation does not affect population persistence. The second objective of this study was to test the hypothesis that the negative effect of forest fragmentation on species distribution increases with decreasing forest cover. Forest loss and fragmentation occur in a landscape over time. If we begin with a completely forested landscape and gradually remove forest, fragmentation begins when the forest is broken into more than one patch. With further forest removal, fragmentation continues as the number of forest patches increases. At very low levels of forest cover, removal of whole forest patches will actually decrease forest fragmentation. Therefore, a study of a single landscape over time will inevitably confound the effects of forest cover and fragmentation. In this study, we compared many landscapes over a large spatial domain and a short time to separate the effects of forest cover and fragmentation. The landscape pattern at the time of study resulted from the concurrent processes of forest loss and fragmentation, and each landscape sampled in this study represents one possible outcome. We quantified forest loss and fragmentation separately for each landscape, thus allowing us to test for the separate effects of forest cover and fragmentation on species distribution, while retaining the ability to evaluate their relative impacts and test for their interaction.