Date of Award
Master of Science
Natural Resource Ecology and Management
Stephen J. Dinsmore
CHAPTER 2. A COMPARISON OF TWO METHODS TO ESTIMATE BUTTERFLY DENSITY
The Pollard-Yates transect is a widely used method for sampling butterflies. Data from these traditional transects are analyzed to produce density estimates, which are then used to make inferences about population status or trends. A key assumption of the Pollard-Yates transect is that detection probability is 1.0 out to a fixed distance (generally 2.5 m on either side of a transect line), although this assumption has not been well tested with field data. Our objectives were to (1) estimate detection probabilities for butterflies with differing natural histories, (2) compare butterfly density estimates from Pollard-Yates line transects to those that incorporate distance sampling, and (3) offer advice on future butterfly sampling techniques to estimate population density. We conducted Pollard-Yates transects and distance-sampling transects at four grassland sites in central Iowa in 2014. For comparison to densities derived from Pollard-Yates transects (number of detections/total area sampled), we used Program DISTANCE to model detection probability (p) and estimate density (D) for eight butterfly species representing a range of life-history attributes. Although we noted little variation in detectability within 2.5 m, we found that detection probability among species varied widely beyond 2.5 m, with considerable variation apparent even within 5 m of the line. Such variation corresponded with size and natural-history attributes of each butterfly species. Program DISTANCE provided more robust density estimates at these greater distances, often considerably <1.0, particularly for smaller, cryptic species. Estimated detection probabilities ranged from 0.53 to 0.79 across the eight species. Thus, we recommend that researchers integrate distance sampling into butterfly sampling and monitoring, particularly for studies utilizing survey transects >5 m wide. Narrower transects are more likely to meet the assumption of perfect detection, but this assumption was clearly violated for many species when transect widths exceeded 10 m.
CHAPTER 3: LANDSCAPE EFFECTS ON WOODLAND, SAVANNA, AND SCRUB-SHRUB BIRDS AT MULTIPLE SCALES IN IOWA
North American birds across multiple guilds continue to experience widespread declines. As a result, there has been increasing interest in defining habitat associations and the impact of landscape configuration on avian population dynamics, especially in increasingly fragmented landscapes. Using bird-survey data collected at 476 sites across Iowa during the Multiple Species Inventory and Monitoring (MSIM) program from 2007-2014, we estimated rates of occupancy, colonization, and extinction for 10 bird species of forest, savanna, and scrub-shrub habitats. We used presence-absence data in robust design occupancy models to estimate the effects of landscape-habitat metrics (edge density and percentage of landscape in grassland, agriculture, and woodland) at multiple scales (200 m, 500 m, 1000 m, and 5000 m). We also modeled the impact of weather covariates (temperature, cloud cover, and wind speed) on detection probability. Occupancy probability ranged from 0.021 for Pileated Woodpecker to 0.610 for Eastern Wood-Pewee. Colonization probability (mean = 0.159) was estimated for all ten species and ranged from 0.021 for Kentucky Warbler to 0.508 for Eastern Wood-Pewee while extinction probability (mean = 0.239) ranged from 0.069 for Eastern Wood-Pewee to 0.398 for Wood Thrush. Our results suggested that habitat associations and dynamics of site colonization and extinction varied by species and spatial scale. The contrasting positive impacts of woodland and negative effect of row-crop agriculture in the landscape were evident for multiple forest birds, particularly area-sensitive species, whereas savanna and scrub-shrub birds showed more species-specific results. Wind speed had a negative impact on detectability across guilds. Our results provide further insight into the habitat associations of birds spanning multiple habitats and serve as a means to enhance management and conservation of birds in highly fragmented landscapes.
CHAPTER 4: LOCAL AND LANDSCAPE EFFECTS ON BUTTERFLIES AT MULTIPLE SCALES IN IOWA
Butterflies are declining around the globe, and they are particularly at risk in increasingly fragmented landscapes, such as those found in the Midwestern USA. Growing concern over these taxa, which are often regarded as proxies for other insects and the environment as a whole, has intensified interest in understanding the impacts of local and broad-scale features on butterfly population dynamics. Using butterfly-survey data collected at 314 sites across Iowa during the Multiple Species Inventory and Monitoring (MSIM) program 2007-2014, we estimated occupancy, colonization, and extinction probabilities for four butterfly species: giant swallowtail (Papilio cresphontes), summer azure (Celastrina neglecta), viceroy (Limenitis archippus), and pearl crescent (Phyciodes tharos). We used presence-absence data in robust design occupancy models to estimate the effects of microhabitat characteristics (litter depth, canopy cover, visual obstruction, and presence/absence of larval food plant), as well as landscape-habitat metrics (edge density and percentage of landscape in grassland, agriculture, and woodland) at multiple scales (100 m, 200 m, 500 m, 1000 m, and 5000 m). We also modeled the impact of weather covariates (temperature, cloud cover, and wind speed) on detection probability. Occupancy probability ranged from 0.16 for P. cresphontes to 0.73 for P. tharos. Likewise, colonization probability spanned from 0.083 for P. cresphontes to 0.409 for P. tharos and extinction probability ranged from 0.141 for L. archippus to 0.437 for C. neglecta. Landscape covariates surfaced in the top model on occupancy for all four species and on colonization rates for three species. Microhabitat variables impacted colonization for one species and extinction for two species; the direction of these effects were positive in each case. Woodland and grassland at differing spatial scales were each more prevalent than agriculture in the best models. Wind showed a strong negative impact on detection for two species, while percent cloud cover (negative effect) and temperature (positive effect) moderately impacted one species each. In what may be the first robust design occupancy study of butterflies in North America, we hope our findings will provide new insights into the life histories and conservation of butterflies in Iowa and beyond.
Shane Shaffer Patterson
Patterson, Shane Shaffer, "Local and landscape effects on population dynamics of birds and butterflies in Iowa" (2016). Graduate Theses and Dissertations. 15992.