Surveying vernal pools in Halifax County, Nova Scotia to identify landscape correlates of pool features and pilot a survey methodology

Abstract

Vernal pools are small, ephemeral wetlands that provide critical breeding habitat for amphibian populations, increase biodiversity on a landscape scale, and influence local and regional hydrography. They have unique annual hydroperiods, typically going from full inundation to full desiccation within a single growing season. This project addresses the current lack of research on vernal pools in specifically Nova Scotian landscapes. A cross-sectional, systematically random pool survey has been conducted within Halifax County, NS. Differences in pool features and dominant land-cover were compared among three ecoregions (Atlantic Coastal Ecoregion (800), Western Ecoregion (700), Eastern Ecoregion (400)), and between two ecodistricts within Ecoregion 400 (Eastern Granite Uplands Ecodistrict (430), Eastern Interior Ecodistrict (440)). Differences in pool features were additionally compared among well-sampled ecosections within each ecoregion, and between well-sampled ecosections in each ecodistrict. Systematic sampling at the ecosection level was not possible due to logistical constraints. This project aims to provide baseline data on the distribution and characteristics of vernal pools within Halifax County, to investigate possible landscape influences on pool features among ecoregions and between ecodistricts, and to evaluate the efficiency and reliability of the survey methods in the context of future provincial vernal pool surveys. A total of 156 pools were surveyed in the field. The percent canopy per pool, pool surface area, and pool density per site (# pools/site) were recorded for survey sites in each ecoregion and ecodistrict. Pool pH levels were also recorded within Ecodistricts 430 and 440. The total means for all recorded values were: percent canopy = 46.9, surface area = 46.1m2, density = 1.3, and pH = 5.1. The dominant land-cover (forest, wetland, open) was also recorded for each site. Pool features and land-cover were tested in Minitab (Minitab 16) for significant differences among 9 ecoregions (ANOVA) and between ecodistricts (2 sample t-test). Pairwise comparisons were conducted (Tukey HSD) among ecoregions for pool features that differed significantly. Among ecoregions there were significant differences in percent canopy levels (p=0.014), forest land cover (p=0.033), and open-land cover (p=0.000). Between ecodistricts there was a significant difference in pH (p=0.000). The type of ecosection each survey site occurred in was determined retroactively in a GIS (ArcGIS 10.1) through visually analyzing the overlay of survey site locations (input from a Garmin eTrex Legend GPS) on the Halifax County ecosection layer. Pool features were tested for significant differences among the dominant ecosections within each ecoregion (ANOVA) and between the dominant ecosections of each ecodistrict (2 sample t-test). All results were insignificant. The total mean values of recorded pool features were compared to relevant literature, and it was found that all means fell within expected ranges. Pool density per km2 was relatively high when compared with a small amount of literature. Both the barren heath/shrub ecosystem typical of Ecoregion 800 and the high abundance of forest cover in Ecoregions 700 and 400 contributed to significant differences in percent canopy levels, forest land-cover, and open land-cover among ecoregions. Differences in wetland abundance, exposed granite bedrock, and possibly pool substrate contributed to a significant difference in pool pH between ecodistricts. Differences in pool surface area and pool density were insignificant at all geographic scales, indicating that a) variation in vernal pool site density and surface area are not significantly affected by variation in geographic landscape features at the ecosection, ecodistrict, or ecoregion scale or b) that the sample size of this study was not large enough to reflect significant variation in pool site density and surface area among distinct landscape settings. There were no significant differences in pool density or pool surface area at 10 the ecosection, ecodistrict or ecoregion scales. Tentative relationships between landscape factors and Ecodistrict 440’s high pool density and Ecoregion 700’s large pool surface were suggested, however the significance of these relationships will remain in question until an increased survey sample size is obtained. Data trends in pool features on the ecosection scale mimicked trends on the ecodistrict scale but not the ecoregion scale. This indicates that sites within well-sampled ecosections provide similar depictions of landscape variability in pool features as those stratified within ecodistricts, and that sites surveyed in well-sampled ecosections may allow for the determination of landscape trends that are not apparent at the ecoregion scale. However, if a pool survey obtained adequate sample sizes from all ecosections within an ecodistrict or ecoregion, the recorded pool features are likely to reflect a finer level of local landscape variation than that attained on both ecodistrict and ecoregion scales. The survey methods were expected to be efficient and reliable. However, designating area as the control variable for each transect decreased efficiency, as the uniform transect route was inflexible to variations in terrain difficulty. The uniform transect route was also inflexible to changes in visibility levels. This increased the potential of false negatives in areas with low visibility and diminished the overall reliability of the survey. An alternative survey method has been suggested that uses survey time as the control variable for each transect, which increases flexibility in response to difficult terrain, changes in visibility, and habitat variability. The benefits of incorporating remote sensing into the survey methods depends on the quality of the images, the training of the analyst, and the time of year the images were taken, however benefits remain minimal if the survey area is heavily forested. 11 Finding relatively high density of vernal pools in Halifax County suggests that other counties in NS may have high densities of vernal pools as well. This provides scientific rationale for the initiation of a provincial vernal pool assessment project, as well as the development of a vernal pool conservation plan within Halifax County. This project has provided baseline data on the characteristics and features of vernal pools in Halifax County, which lays the groundwork for future research in other geographic regions in NS. Ultimately, it provides some first steps towards protecting vernal pool ecosystems from anthropogenic impact within NS.