The continental population of lesser scaup (Aythya affinis [Eyton]; hereafter scaup) has been in decline since the 1980s, leading researchers to speculate that reduced feeding efficiency and subsequent lowered body condition during spring migration may have adverse impacts on reproductive success. Although scaup-habitat associations have been studied in the past, lack of long-term data concerning the quantity and quality of suitable wetland habitat has hindered the evaluation of historic trends in relation to habitat use and more importantly scaup population decline. In this study, I assessed scaup-habitat relationships to determine the variables that best described scaup use of wetlands in South Dakota’s Prairie Pothole Region (PPR). Empirical data were used to develop a temporally dynamic spatial model to describe historic habitat suitability. The model was then applied to other parts of the PPR used by migrating scaup. The PPR spans from northern Iowa northward to Canada’s boreal forest region where about 70% of the scaup population breeds. In 2004, a census survey of spring migrating scaup was coupled with chemical, physical and biotic variables collected from wetland habitats. Site-specific wetland use was positively related with wetland size, amphipod density and the proportion of coarse sediments (≥2000 µm diameter). Scaup use was negatively associated with chloride, potassium, and nitrate concentrations in the water and the proportion of fine sediment (≤ 150 µm diameter). Site-specific evaluations revealed negative associations between amphipod abundance and water quality (chlorides, potassium and ammonim). Higher concentrations of chlorides, potassium and ammonium were associated with roads and development (i.e., housing, boat docks, sewage treatment, etc.). Moreover, modeling efforts showed that amphipod abundance and submerged aquatic vegetation (SAV) explained an appreciable amount of variation (64%) in scaup use. At the landscape scale, scaup wetland use was positively associated with landscapes that had high proportions of semipermanent wetlands and negatively associated with landscapes dominated by temporary wetlands. This would explain their affiliation to the Prairie Coteau, which has the highest density of semipermanent wetlands out of all the physiographic regions in eastern South Dakota. However, wetland resources in the PPR are not static. Apart from the expected expansion of wetlands in wet periods, a comparison of wetland resources between average (1979-1986) and above-average (1995-1999) water condition years revealed that spatial configuration patterns such as proximity, clumpedness and connectivity may explain “losses” of smaller temporary and seasonal wetlands by merging with larger semipermanent wetlands.
Using data on scaup-habitat associations, I developed a spatial habitat suitability index model comprised of three sub-models: (1) a population and rule-based sub-model describing potential amphipod production (AMPROD) in response to seasonal climate, (2) a landscape sub-model describing changes in total wetland area at broad scales and (3) a sub-model accounting for changes in open water area. Indices for each of the three sub-models were derived using WETSIM (Version 3.1), a process oriented wetland model that simulates the hydrodynamic and vegetation changes of a semipermanent PPR wetland from climate data. To my knowledge, this is the first attempt to model the population dynamics of an aquatic invertebrate in response to local climate. The model predicted high spatial and temporal variation between stations in response to local climate, however, stations that were closer together tended to be more similar. This reflected the variability observed in field data collected over three years. Based on climate alone, average long-term habitat suitability (1950-2000) was predicted to be highest in Iowa, followed by Minnesota, South Dakota and then North Dakota, which means that local weather patterns in the eastern portions of the PPR produce more favorable habitat than the west and habitat suitability got poorer as we went northward. However, continued deterioration in wetland quality and quantity by agricultural activities in the eastern portions of the PPR could place severe strains on migrant scaup, particularly in dry periods when habitats in the west are unusable. At most stations, the model predicts a decline in average habitat suitability from the 1950s to the 1960s followed by an increase in the 1970s. Habitat suitability in South Dakota then improved from the 1970s to the 1990s but steadily declined in Minnesota and North Dakota. Results from one station with more recent data (Brookings, SD 1950-2006) suggest that harsh winter conditions in 2001 resulted in a steep decline in habitat quality to levels below those experienced 30 years ago, which matches the low frequency of occurrence of amphipods in scaup diets collected from wetlands in close proximity. Although the model does not predict uniform trends across the PPR, it emphasizes the danger that habitat losses in the east pose to migrant scaup as these are located in the most favorable climates for production. The model does not fully support the spring condition hypothesis (SCH), which postulates that migration habitat suitability in the upper-Midwest has declined since the 1980s. However, it predicts that migration habitat suitability is cyclic. Interestingly, habitat suitability was predicted to have declined in those areas where the SCH was formulated. Furthermore, the model raises questions relating to the metapopulation dynamics of amphipods in the western PPR, which are presumed to experience higher local extinction rates due to higher drought frequency.