Fragmentation due to the creation of linear features such as roads and trails can cause changes in habitat structure and resource levels near forest edges (Chen et al. 1992). These changes may negatively impact songbirds which are adapted to forest interior habitats, and may cause them to experience increased nest predation, brood parasitism, and competition near forest edges (Ambuel and Temple 1983; Donovan et al. 1997). Conversely, some species which are generalists or edge specialists may be positively affected by fragmentation due to decreased patch isolation, increased habitat diversity, or favourable responses to increased edge (Fahrig 2003). As a result of these two contrasting effects, linear feature creation may lead to significant shifts in the composition of forest songbird communities.
Figure 2. Examples of a forest interior specialist (boreal chickadee) and a forest edge specialist (chipping sparrow).
Linear features may have a particularly significant effect on forest bird populations in northeastern Alberta. This area is one of the most productive breeding zones for songbirds in North America (Schmiegelow and Monkkonen 2002), but it has also been experiencing large increases in energy sector activity (Figures 3, 4). At present, anthropogenic edge density is approximately 1.8 km/km2, but is projected to increase to 8km/km2 over the next 100 years (Schneider et al. 2003). Most of this increase is expected to be due to the creation of seismic lines, which are 8-10 m wide linear features created by bulldozers during oil and gas exploration.
Bird abundances are commonly estimated using point count surveys, in which all birds heard within a specified time period are recorded. This technique is based on the assumption that the number of individuals detected represents a constant proportion of the actual number present across space and time (Thompson 2002). However, this assumption is rarely met because detection probabilities are not constant between species, nor across habitats. As a result, bird counts must be adjusted for differences in detectability prior to analysis in order to avoid making incorrect conclusions. This type of correction is becoming increasingly common in univariate analyses, but does not appear to have been applied in a multivariate context.
1. Determine the effects of seismic line density on the composition of songbird communities in trembling aspen (Populus tremuloides) and black spruce (Picea glauca) stands.
The creation of seismic lines results in changes to habitats which are expected to lead to shifts in songbird communities, such that forest interior specialists decrease and edge specialists or generalists increase. However, due to the narrow width of seismic lines, it is also possible that they may have little effect on bird community composition.
2. Assess whether adjusting for differences in detectability affects the results of ordinations.
Detectability of bird songs varies depending on species and habitat type, and has been found to significantly affect results in univariate tests. Consequently, it may also be important to account for differences in detectability when analyzing multivariate data.
Seismic line density |
Figure 5. Expected shift in bird community composition as seismic line density increases. Interior species (black-coloured birds) are expected to decrease and edge species (red-coloured birds) to increase in abundance. |
Detectability
Correcting for differences in detectability between species is expected to have only a small effect on the relationship between seismic line density and the composition of bird communities (Figure 6). Larger effects might be observed if detectability was also influenced by the seismic line treatment (Figure 7). Simulated data in which detectability is altered between treatments will therefore be used to examine how such differences could influence results.
Figure 6. a) Mean relative abundance of species A, B, and C (counts not adjusted for differences in detectability between species); b) Estimated mean density (number per 1 hectare) of the species A, B, and C after correcting for detectability. Species C is harder to detect and thus is not recognized as the dominant species in the community until the counts are corrected.
Figure 7. a) Mean relative abundance of one species at locations with no seismic lines (none), a single seismic line (single), and two intersecting seismic lines (double). b) Estimated mean density (number per 1 hectare) at each treatment type after correcting for differences in detectability between treatments. In this hypothetical example, detectability improves as seismic line density increases and a negative correlation between bird density and seismic line treatment becomes evident after this difference in detectability is taken into account.