Department of Biology, University of the South, 735 University Avenue, Sewanee, TN 37383 email dhaskell@Sewanee.edu
(Abstract of Paper)
Introduction
The effect of edges created by habitat fragmentation is a prominent and recurring theme in many studies of conservation biology. When roads cut through otherwise undisturbed habitats, they create long, thin edges resulting in fragmentation. Despite their ubiquity, however, our understanding of the ecological effects of roads has lagged behind studies of edges such as those created by timber harvesting. Studies that have investigated the ecological effects of roads show that roads may negatively affect the ecosystems around them. For example, roads and road-construction activities are significant sources of sediment that can affect local water quality or water quality in downstream marine ecosystems. Roads can also serve as conduits for invasive, edge-adapted species, and road traffic can be a significant source of mortality for native fauna.
I investigated whether roads in forests affect the abundance and richness of the macroinvertebrate fauna of the soil leaf litter. Macroinvertebrates are an essential component of the nutrient- and energy-processing abilities of the soil. In addition, these animals are significant components of the diets of many vertebrates, especially birds and amphibians. My study focused on the network of one and two lane unpaved roads in a U.S. National Forest in the Southern Appalachian Mountains. The Southern Appalachians have been identified as an internationally important center of biodiversity and have been subject to locally intense road development. This region is therefore an important test case for investigating the ecological effects of roads.
Methods
I conducted the study in the southern portion of the Cherokee National Forest, Tennessee, in June and early July of 1998. I chose sample points on roads by haphazardly marking points on a map of all unpaved roads within the forest. This process was performed without prior knowledge of the site conditions at each point. I excluded some of these points from the study because they marked the edges of timber harvests. Thus, all roads included in the sample traversed forest that was otherwise unaffected by human-caused edges.
At each sampling point I measured the width of the road and the percent canopy cover above the road. I measured canopy cover by visually examining the canopy from the center of the road through a cone held 1.8 m above the surface of the road.
I ran one, 100 m transect perpendicular to the road at each sampling point and took samples at 1, 5, 15, 35, 60. and 100 m along this transect using a corer at each of the points to remove three soil samples. I pushed the corer into the soil just far enough to reach the mineral soil layer; the sample thus contained the entire litter layer of the soil including the fermentation and humus or humification layers.
I used visual searches of sifted soil samples because alternative methods such as Berlese-Tullgren funnels are less successful at sampling the macroinvertebrates that are the focus of my study (Coleman & Crosslev 1996). I identified all macroinvertebrates (defined here as organisms >1 mm long) to the level of order except for Chilopoda and Diplopoda, which I identified to class. I used the total number of taxa in each sample as an estimate of taxonomic richness.
The macroinvertebrate soil fauna was significantly less abundant and less diverse close to the forest roads in this study. Leaf-litter depth was also reduced close to roads. Visual inspection of the data suggests that the effects of roads on faunal abundance and leaf-litter depth may persist up to 100 m into the forest, whereas the effect on faunal richness persists to 15 m. Roads that had steep declines in macroinvertebrate abundance and richness tended to be wide and to have open canopies, but neither of these correlations was significant, however.
The methods I used have three potential biases. First, because I quantified the soil fauna by visual searches, the results are likely biased toward large organisms. Second, because I used timed searches, large samples are likely to have been less thoroughly searched than were small samples. Third, I measured taxonomic richness in terms of numbers of higher taxa, not individual species.
I believe, however, these biases did not produce or contribute to the significant trends I report. The visual search bias applies equally at all distances along the transect, so is unlikely to produce a spurious correlation. Nevertheless, my study addresses only macro- invertebrates (>1 mm), and the effects of roads on soil microfauna are presently unknown. The second bias, arising from the use of timed searches, actually increases the conservatism of my study. Because there was a significant increase in litter depth as distance from the road increased, samples close to the road were less bulky and were likely searched more thoroughly, than those far from the road. But, there was a significant increase in invertebrate abundance away from roads despite this bias, not because of it. The third bias, the use of higher taxa as a measure of richness, also increases conservatism because many species can be lost from a sample without affecting the number of higher taxa (Gaston 1996). The fact that roads significantly affect richness measured at the level of orders and classes suggests that the effect at the level of species may be more severe.
Although reductions in the abundance and richness of macroinvertebrates near roads are cause for concern in themselves, these changes may have additional consequences for the functioning of the forest ecosystem and the biological diversity found within this system. One such consequence may be a change in the ability of the soil to process energy and nutrients. Changes or reductions in the abundance and richness of soil fauna have been shown to have important effects on these processing abilities. These changes affect both soil structure and nutrient availability, which may in turn affect the distribution and abundance of other organisms, particularly plants.
Wilke et al. (1993), for example, found significant differences in soil pH and plant-community composition associated with wind-caused differences in leaf-litter depth. The road-associated reductions in litter depth reported here suggest a mechanism by which forest roads may change the growth rates, abundance, and diversity of plants in a swath of forest on either side of a road. Studies in tropical forests have shown that forest fragmentation may negatively affect plant communities by altering the abundance and diversity of insect pollinators, predators, decomposers, and possibly herbivores. Future work in this area is needed to investigate the effects of forest roads on plant communities and other functional groups of invertebrates.
The vertebrate fauna of the forest may also be affected by changes in soil macroinvertebrates. Ground-foraging birds for example, rely heavily on litter dwelling macroinvertebrates as sources of food. Breeding birds in the Southern Appalachians which use soil macroinvertebrates as a source of food for adults and nestlings include Black-and-white Warblers, Worm-eating Warblers, Ovenbirds, Wood Thrushes, and Kentucky Warblers. All these birds are Neotropical migrants and, according to the North American Breeding Bird Survey, all are experiencing significant population declines in this region.
The availability of food during the breeding season can limit the productivity of bird populations in temperate forests, so roads may negatively affect bird populations in adjacent forests by depressing food availability. The effect reported here may be similar to the results of Burke and Nol (1998), who studied invertebrate abundance in forest fragments in an agricultural landscape in southern Ontario. They reported that small forest fragments had significantly lower numbers of invertebrates in the leaf litter than did large fragments. They also reported that female Ovenbirds avoided these small fragments, perhaps because of the dearth of food.
The data I present suggest that such effects are not confined to isolated forest fragments, but that roads through otherwise contiguous forest may also depress the abundance of the invertebrate prey of breeding birds. Rich et al. (1994) found that some forest-nesting Neotropical migrant birds in eastern North America were less abundant near roads and power lines. My data suggest this reduction in bird abundance may be caused partly by reduced food availability near roads. Road-associated reductions in bird density have also been reported in Europe, where traffic noise has been implicated as a cause of declines in numbers of birds (Reijen et al. 1995).
The Southern Appalachians are an internationally significant area for amphibian biodiversity (Petranka 1998). This region hosts a particularly diverse assemblage of salamanders, particularly "woodland" salamanders. These animals live in the leaf litter and prey exclusively on soil invertebrates. My study shows that forest roads depress both the depth of the leaf litter in which the salamanders live and reduce the availability of the salamanders' prey. The abundance of salamanders is thought in some cases to be influenced by both habitat availability and food abundance (Hairston 1980, 1981), so my results provide evidence that forest roads may have negative consequences for the salamanders of the Southern Appalachian mountains.
The dead organic matter that forms the litter layer is the source of habitat, energy, and nutrients for many of the invertebrates that live in this layer, so a reduction in litter depth may cause a reduction in invertebrate abundance and richness. What, then, causes the thinning of the leaf litter near roads? Roads may serve as conduits for wind and may break the canopy layer, allowing more solar energy to reach the ground. An increase in wind could affect leaf litter in two ways: physical displacement of leaf litter by wind has been shown to have a significant effect on the spatial distribution of litter in forests (Wilke et al. 1993), and wind dries the litter, which affects rates of decomposition (Clein & Schimel 1994; Bruhn et al. 1995 Didham 1998).
Increased solar energy can have a similar drying effect on leaf litter. I found that roads with more open canopies had significantly steeper declines in litter depth. This is consistent with the hypothesis that increased exposure to solar energy is partially responsible for the thinning of the litter layer.
The finding that anthropogenic changes in the landscape are associated with changes in leaf litter is not unique to this study. Didham (1998), for example, reported rapid soil turnover rates at some tropical forest edges, although his study investigated edges of forest fragments, not roads. Whatever the mechanism by which leaf litter is thinned near roads, it is clear that even relatively narrow roads through forests are associated with reduced leaf-litter depths, which in turn are associated with reductions in the abundance and richness of the leaf-litter fauna.
The effect of roads on forest ecosystems depends not only on their local effect on biological patterns and processes but also on the number and distribution of roads within the forest. Even if roads have large local effects, as suggested by my data, the landscape-level consequences of these effects might be negligible if roads are confined to small areas of the forest. If, on the other extreme, roads are spread evenly through the entire forest, the cumulative result of small, local effects may be large.
Roads in forests provide benefits to society such as opportunities for recreation and natural-resource extraction. These benefits must be weighed against the ecological and economic costs of roads. I have not addressed where the balance between these costs and benefits might lie, but I have identified a previously undescribed effect of forest roads on the surrounding ecosystem. If a goal of forest management is to maintain the function and diversity of forest ecosystems, then my results suggest that managers should minimize both the density of roads and the extent to which roads sprawl across the landscape.