Below are brief summaries for 11 scientific articles relating to wildlife habitat analysis with a focus on seabird habitat and forest nesting habitat.
Augutis, D., and S. Sinkevičius. 2005. Application of Geographic Information System (GIS) Technologies in Identification of Potential Nesting Habitats of Black Stork (Ciconia nigra). Acta Zoologica Lithuanica. 15: 1392-1657.
The study focuses on nesting habitat of the Black Stork in forests of central and eastern Lithuania. Data on nesting locations were used from previous studies. The study had 3 goals: 1) analysis of factors affecting nesting site selection, 2) identify suitable nesting habitat in other forest districts, 3) comparison of suitable nesting habitats with nesting sites not used in the analysis. 1) nesting sites and random point were analyzed using 16 GIS layers: water bodies, stagnant waters, flowing waters, rivers width >10m, river width <10m, forest boundaries, forests, forest habitats, settlement boundaries, infrastructure, farmstead boundaries, highways, country roads, regional roads, local roads, and railways. The Distance Matrix extension from ArcView GIS was used to measure the distance from each point (nest and random) to the nearest location for each attribute type. The resulting data was analyzed using Statistica-6. 2) They created a new GIS layer of suitable nesting habitat based on the data from objective 1. Results: significant differences were found between nest and random points for 6 of 16 GIS layers (stagnant waters, settlement boundaries, infrastructure, regional roads, local roads, and railways). The important habitat characteristics (GIS layers) were used to predict suitable forest nesting habitat for the other forest districts. The multiple layer analysis is a good example of a simple forest habitat analysis.
Earnst, S. L., R. Platte, and L. Bond. 2006. A landscape-scale model of yellow-billed loon (Gavia adamsii) habitat preferences in northern Alaska. Hydrobiologia. 567: 227-236.
Researchers used aerial surveys to detect presence and absence of loon species then used GIS applications to determine landscape scale habitat preferences. Landscape scale parameters measured were: lake area, maximum lake depth (shallow, medium, or deep), proportion of the shoreline in aquatic vegetation, an index of shoreline complexity, hydrological connectivity (lakes perimeter within 100m of a stream), distance from the lake’s centroid to a river, and distance from the lake’s centroid to the coast of the Beaufort Sea. Most distance measurements were taken from paper quad maps with other variables calculated from digital maps. Statistical analysis was done using the SAS extension GLIMMIX since the surveyed lakes were not independent. Small planes surveyed 757 lakes in a study area of 23,500 km² documenting yellow-billed loons on 15% (n = 115) and Pacific loons on 42% (n = 321). Yellow-billed loon presence was positively and significantly related to increasing lake size, depth, proportion of shoreline in aquatic vegetation, shoreline complexity, and connectivity. Lake depth, size, and connectivity had the largest effect on yellow-billed loon presence. Also, Pacific loon presence was a significant predictor (inverse relationship) of yellow-billed loon presence. This study does not use heavy GIS analysis but most of the measurements derived and analyzed were digitally measured using a GIS application. The study likely could have benefited from the use (if available) of vegetation and water layers to analyze landscape scale habitat features.
Hébert, P. N., and R. T. Golightly. 2008. At-sea distribution and movements of nesting and non-nesting Marbled Murrelets Brachyramphus marmoratus in northern California. 36: 99-105.
Researchers used radio-telemetry to track the movements of murrelets along the coast of northern California and southern Oregon. GIS applications were utilized to summarize the movements of the murrelets. They used the Animal Movement extension for ArcView 3.3 to analyze locations, range of movements, and home range size (minimum convex polygon estimator- MCP) for radio-marked murrelets. Each bird had the following variables calculated: average distance to nearest shore, maximum extent of alongshore movements (distance between the two locations farthest apart in the north-south direction), mean distance each bird traveled north or south of where Redwood Creek enters the ocean (this was adjacent to nest sites), and home range size (km², MCP). For each home range, the farthest 5% of outliers and any land mass were removed from each polygon (MCP). They used SPSS software for statistical analysis and performed ANOVAs comparing all variables between male and female as well as nesting and non-nesting murrelets. Results: distance from shoreline did not differ between years, sexes or breeding status; mean standardized maximum alongshore distance traveled by nesting murrelets was significantly less than non-nesting murrelets, the standardized maximum alongshore distance traveled by non-nesting males tended to be greater than that of nesting males, but the standardized maximum alongshore distance traveled by non-nesting females was similar to that of nesting females; home range size was similar between years, was significantly larger for males opposed to females, non-nesting murrelets opposed to nesting murrelets and non-nesting males opposed to non-nesting females.
Meyer, C. B., and S. L. Miller. 2000. Use of Fragmented Landscapes by Marbled Murrelets for Nesting in Southern Oregon. Conservation Biology. 16: 755-766.
Researchers used logistic regression models on landscapes in southern Oregon. They addressed two central questions: whether old-growth forest fragmentation was associated with use of an area by murrelets and whether proximity to certain marine features was associated with use of forest fragments by murrelets. A GIS vegetation map, including marine and topographic features, was derived from Landsat thematic mapper imagery. Circular plots of 400-, 800-, 1600-, and 3200-m radius were placed over surveyed inland areas occupied or unoccupied by murrelets. Old growth and other cover types were classified at a fine resolution (25-m pixel) using ArcInfo. U.S. Forest Service inland surveys for murrelets were used to classify a stand as “occupied” or “unoccupied.” Within the landscape-sized plots, fragmentation variables were calculated for each circular plot using FRAGSTATS (raster version). Results: All but one of the 46 occupied plots was located within the fog-influenced western hemlock zone. Statistical models identified the following variables associated with murrelet occupation: lower elevations, closer to ocean and major bay, less fragmented old-growth forest, closer to river mouths, higher percentage of old-growth forest. Compared with topographic and marine variables, old-growth variables moderately contributed to the accuracy of the best models, and the contribution was strongest for models based on the largest plots. GIS applications were the backbone for this study and all models and statistical analysis were based on measurements done in ArcView and FRAGSTATS.
Olson, G. S., E. M. Glenn, R. G. Anthony, E. D. Forsman, J. A. Reid, P. J. Loschi, and W. J. Ripple. 2004. Modeling Demographic Performance of Northern Spotted Owls Relative to Forest Habitat in Oregon. J. of Wildlife Management. 68: 1039-1053.
Northern Spotted Owls are a federally threatened species that are associated with older-aged forests of the Pacific Northwest. Studying the demography of Spotted Owls in the wild is expensive and involves the capture and handling of many birds each year. The authors investigate using statistical models to predict owl occurrence and demography based on remote sensing habitat data layers. A forest cover-type map based on aerial photographs was used to measure habitat characteristics around owl territories. Demographic parameters were derived from owl monitoring surveys dating back to the mid-1980s. Forest cover types were identified from 1:12,000 scale aerial photographs and included 8 cover types based on age of dominant over-story trees. The original 8 cover types were combined into 4 categories (non-forest, mid-seral conifer, broad-leaf, and late-seral conifer) for habitat composition analysis and 3 categories (non-forest, mid-seral conifer, and late-seral conifer) for habitat pattern analysis. Landscape composition was measured around activity centers for each territory at 3 levels: 600m radius for nest core area, 1500m for home range, and 2400m for winter foraging areas. ArcInfo GIS was used to measure the area for each cover type at all landscape levels. FRAGSTATS was used to evaluate 6 landscape pattern variables at the 1500m radius level. Results from models developed showed that increases in late-seral forest had a positive effect on survival while increases in early seral and non-forest had a negative effect. The best model indicated a nonlinear relationship between mid- and late-seral forest and survival where high levels of this cover type showed a slight decrease in survival. A mixture of all habitat types produced the highest levels of productivity and survival. Survival was highest near the mean mid- and late-seral levels while productivity increased linearly as the amount of edge between these forest types and other habitats increased.
Ripple, William J., S. Kim Nelson, and Elizabeth M. Glenn. 2003. Forest Landscape Patterns around Marbled Murrelet Nest Sites in the Oregon Coast Range. NW Naturalist. 84: 80-89.
Researchers located 41 Marbled Murrelet nest sites using dawn surveys and tree climbing in Oregon between 1990 and 1998. They analyzed habitat types around nest sites and random points (n = 41) with a photogrammetric method using a scanning stereoscope on aerial photographs at two scales (1:31,000 and 1:24,000). Nest sites were analyzed at the 0.5 km radius and the 1.0 km radius levels. They determined the composition of each circular plot around nest and random points using a grid-based GIS system (ERDAS Imagine). Vegetation was classified into 6 habitat classes and mature-old growth patch density, core habitat, and perimeter density were determined. Nest site habitat composition was compared to random sites using uni- and multivariate regression models. The models showed that murrelets selected for greater amounts of pole-young and mature old-growth forests, less edge, and more cohesive nest patch shape.
Ro¨nka¨, M., H. Tolvanenb, E. Lehikoinena, M. von Numersc, and M. Rautkarid. 2008. Breeding habitat preferences of 15 bird species on
south-western Finnish archipelago coast: Applicability of
digital spatial data archives to habitat assessment. Biological Conservation. 141: 402-416.
Authors “tested the applicability of GIS and digital data archives for the analysis of coastal bird habitats by conducting a multivariate analysis on the relationship between physical island characteristics and the breeding site selection of 15 species of ducks, waders, larids and alcids in 2001–2005 in the fragmented archipelago coast of south-western
Finland.”
Bird surveys detected the presence of breeding birds on islands.
Models for shoreline data, elevation, and bathymetry were used to calculate five physical variables for each island: land area (ha), maximum elevation (m), total land area of adjacent islands within 200m (ha), average water depth within 200m (m), and mean fetch (km).
GIS analysis used a shoreline data, a DEM, and a 200 m buffer around Results: island area, water depth, mean fetch and island elevation affected the occurrence of all target species.
A suitable habitat model was created for 14 of 15 target species except for the Tufted Duck.
The following is a summary of important factors for groups of birds: waterfowl- large island area, shorebirds- large island area and shallow water depth, gulls/terns- large island area and deep water depth, black guillemot- deep water depth and open shorelines (fetch).
The combination of GIS and digital data archives provides a quantitative and cost-effective assessment of coastal breeding habitats of birds.
Stralberg, D., K. E. Fehring, L. Y. Pomara, N. Nur, D. B. Adams, D. Hatch, G. R. Geupela, and S. Allen. 2009. Modeling nest-site occurrence for the Northern Spotted Owl at its southern range limit in central California. Elsevier. 90:76-85.
Predictive spatial models of species-habitat relationships and predicted occurrence or species/habitat distribution models are becoming prevalent in wildlife management. Large scale models have many uses but are not sufficient for small scale local land management decisions. The study evaluates the value of small scale occurrence models for Northern Spotted Owls. Researchers used breeding owl survey data from 1998–2001 and GIS layers representing topographic, anthropogenic, and vegetation-based landscape characteristics to build logistic regression models of owl nest-site occurrence. A total of 44 occupied nesting areas were identified and 88 random forest locations were derived using ArcView 3.2a. Point and landscape metrics were measured for each nest-site and random point. Point metrics included aspect and distance to nearest stream and road. Landscape measurements were done at 200 m, 400 m, and 800 m radii. Landscape metrics included mean slope, mean elevation, elevational position in watershed ((nest elevation—minimum elevation within a 400-m radius)/total elevation range within a 400-m radius), forest cover proportion, proportion of vegetation cover types (conifer versus hardwood, and dominant tree species), forest stand size class (a proxy for age), and total forest edge. Two vegetation data layers were used: layer derived from measurements collected at all 132 locations and another layer was created from various satellite imagery databases. Resulting models suggested that nest sites were more likely to occur at south-facing sites that were lower in the watershed; in lower in mean elevation (400m); a higher proportion of woodland within (400m); a lower proportion of bishop pine (800m); a lower proportion of urban development (200m); and less woodland edge (200m).
Vandegraft, D. L. 2005. Cartographic and GIS activities in the U.S. Fish and Wildlife Service. Federal Register (http://pubs.usgs.gov/of/2005/1428/pdf/vandegraft.pdf). Digital Mapping Techniques: 49-53.
The U.S. Fish and Wildlife Service is making use of modern GIS applications to effectively manage the National Wildlife Refuge System and other land entrusted to them. They have been using computers for refuge mapping, land management, and habitat analysis since the 1980s. GIS is used at the local, regional, and national levels with the refuge system divided in to 7 regions. The Geographic Information Systems and Spatial Data website (http://www.fws.gov/data/gishome.html) is devoted specifically to sharing spatial data with researchers and the public. The website uses ArcIMS software from ESRI and data is compatible with most GIS applications. The first fully digital map of the National Wildlife Refuge System was completed in 2002. Also in 2002, a new digital cartographic format was developed. The standards use USGS digital orthophoto quads and digital raster graphs as a base with color shading to represent the various land status categories. In 2004, they established standards to store digital boundaries and land status in ArcGIS format to view and analyze ownership boundaries and their associated land attributes.
Wanless, S., P. J. Bacon, M. P. Harris, and A. D. Webb. 2002. Evaluating the coastal environment for marine birds. Journal of Coastal Conservation 8: 17-24.
The authors describe a preliminary Geographic Information System (GIS) in which spatial data on environmental variables (seabird colony locations, sea depth and seabed sediments) are integrated with realistic energy constraints faced by marine birds during the breeding season along the Scottish coast. The European shag (Phalacrocorax aristotelis) is an important avian predator in the inshore marine community and was selected as the focal species for the model. To demonstrate the GIS approach (large scale) we selected a study area including the Firths of Forth and Tay in Southeast Scotland. Data on the bathymetry and seabed sediments from British Geological Survey were combined with data on the sizes and locations of seabird breeding colonies then digitized and stored in a GIS. The bathymetry (bands of 0 - 10, 11 - 20, 21 - 30 m etc.) and sediment type polygons were intersected and then superimposed on a 1 km x 1 km grid of the area to produce small (< 1 km2) polygons of known depth, sediment type, and location. The final layer in the GIS model involved the likely distribution of a preferred food source (lesser sandeels) and the associated seabed substrates (mostly sandy sediments). The daily feeding times of a marine bird (shag) were derived using a daily energetic requirement estimate and an estimate of the foraging success within varying levels of seabed substrate quality. They also incorporated potential affects of a reduced food supply and a large oil spill on the foraging habits of seabirds. The authors conclude that “the use of a GIS-based modeling approach can not only indicate which sea areas are likely to be important but can also be used to explore the consequences of changing patterns of human activity and management on marine species.”
Yen, P.P.W., F. Huettmann, and F. Cooke. 2004. A large-scale model for the at-sea distribution and abundance of Marbled Murrelets (Brachyramphus marmoratus) during the breeding season in coastal British Columbia, Canada. Ecological Modeling. 171: 395–413.
Researchers developed an ecological modeling system that allowed for the comparison of biotic and abiotic factors in relation to at-sea Marbled Murrelet densities. The study is the first to provide a model-based link between the murrelet’s marine and terrestrial environment. Data on the marine distribution of murrelets from count surveys, density surveys, and pertinent environmental variables were combined and analyzed in a GIS database. Murrelet density layers in the GIS database were density (birds/km²) and count (birds/survey). Abiotic environmental layers were current tidal speed, sea surface temperature, estuaries, and glaciers. Biotic environmental layers were Herring spawns, sandy shorelines and old-growth forest. Databases were combined and analyzed using ArcView 3.2. They measured the distance in meters from every point (evenly spaced and surveyed points) to the nearest feature for some environmental themes (glacier, sandy substrate, estuaries), and prey type theme (herring spawn). For the same points, sea surface temperature and tidal current speed overlays were extracted. These distances were compared using multivariate analysis. They identified that Marbled Murrelets show preference for cooler sea surface temperatures and are found closer to estuaries and sandy bottoms. Negative correlations with proximity to glaciers and high herring spawn index were found. They then used predictive models to map out predicted murrelet density based on relation to environmental factors for the entire coast of British Columbia.