This study examined female mate choice in the House Wren (Troglodytes aedon). The study was divided into two parts. The first part examined which territory or male traits contribute to female fitness. The second part examined whether females mate non-randomly with respect to those traits. Female fitness (nesting success, number of fledglings, nestling weight) was not related to any territory or male trait. Females mated randomly with respect to all traits except one: they preferred males who arrived early from migration to males who arrived late. However, the date of male arrival was not related to female fitness.

Abstract not available.

Warning (aposematic) coloration generally involves highly conspicuous colors, which are usually displayed by different prey species in characteristic patterns. The key function of warning coloration is to advertise prey unpalatability to potential predators as well as to facilitate avoidance learning. Many studies have examined the effect of conspicuousness of prey compared to cryptic prey on avoidance learning by predators. However, no previous studies examined the effect of warning coloration of prey compared to conspicuous and cryptic coloration of prey on avoidance learning in wild-ranging birds. Therefore, this problem was the central theme of my experimental study in which I used the American crow as a predator and chicken eggs of the three coloration patterns with palatable or unpalatable contents as prey. I hypothesized that avoidance of unpalatable eggs would be acquired through learning rather than being genetically determined. Further, I predicted that crows would learn faster to avoid eggs with warning coloration than those with either conspicuous or cryptic coloration. Finally, I predicted that avoidance of originally unpalatable eggs should persist longer during the subsequent control trials (contents of all eggs made palatable) for eggs with warning coloration compared to eggs with conspicuous or cryptic coloration patterns. To test these predictions, I trained crows in several series of trials to avoid potential food by offering them an opportunity to learn to associate unpalatability of prey (eggs) with selected coloration patterns (conspicuous, cryptic, and warning coloration). My results indicated that the American crow did not exhibit an innate avoidance of eggs with warning coloration. I also found that crows did not learn faster to avoid unpalatable prey with warning coloration compared to prey with conspicuous or cryptic coloration. (Abstract shortened by UMI.)

Human impacts such as habitat loss, species invasions and climate change are thought to be leading to biodiversity losses. However. little is known about the relative magnitudes of these effects. The purpose of this study is to determine to what extent climate change may have been responsible for observed changes in avian species' abundances. I investigated this question using data from the North American Breeding Bird Survey, one of the largest and most comprehensive biodiversity databases available. I found that 40% of the bird populations studied show significant linear changes in abundance from 1970 to 1999. Over the same period, temperature shows significant linear chances at 13% of the survey routes studied, whereas precipitation does not show significant trends. I analyzed the spatial relationships between abundance and climate and found that they are not strong (median R 2 = 0.119) but are generally consistent through time. Temporal trends in abundance were only very weakly related to trends in either temperature or temperature sub-optimality from 1970 to 1999. I found that temperature change is spatially autocorrelated over very long distances, whereas abundance change is autocorrelated over much shorter distances. Observed abundance changes in bird abundance are more likely to reflect local processes such as habitat modification.

The expansion of agriculture throughout the world has precipitated serious biodiversity losses. Countries with relatively extensive natural habitats, such as Canada, continue to intensify agricultural land uses, threatening to expand the scope of the present mass extinction. This thesis tests likely mechanisms of butterfly species decline in agricultural landscapes of varying intensities in the most biologically diverse region of Canada. I measured site variables such as the richness of plants suitable for larval development, patch area and shape index, and landscape variables such as heterogeneity and land-cover within buffers that varied in size from 100m to 1000m. Several rare species were not present in the agricultural sites, whereas a few species typically associated with disturbed habitat were present and abundant among most study sites. I found that agricultural intensification acts across landscapes to reduce butterfly species richness by reducing the proportion of natural habitat available. This effect was especially pronounced at 100m to 200m distances, suggesting a threshold effect beyond which the proportion of natural land is less critical to butterfly biodiversity. Within-site factors, such as plant species richness and habitat area were not affected by agricultural intensity, nor did they affect butterfly species richness and abundance. These results suggest that the mechanism linking butterfly species decline to agriculture, at least for butterflies in this region, is the loss of small habitat remnants.

The accumulation of persistent organic pollutants (POPs) in mountain regions was investigated by collecting trout from eight lakes spanning an elevation gradient of 760 to 2360 m.a.s.l. in the Canadian Rocky Mountains. All lakes were located within (or close to) national parks in Alberta and British Columbia. Concentrations of several organochlorine compounds increased significantly with lake elevation. The compounds, which increased the most with elevation, were the less volatile organochlorine pesticides (OCPs) (e.g. dieldrin, DDTs). The relationship was not as strong for the more volatile organochlorines (e.g. HCHs, HCB). Biological factors such as growth dilution of contaminants appears to be a major determinant of contaminant concentrations of organochlorine compounds in alpine lakes. Principal component analysis (PCA) showed that POP composition in trout relate strongly to octanol/water partition co-efficient (Kow) and lake elevation.

The incubation shift length of the Ancient Murrelet ( Synthliboramphus antiquus), an exceptionally long and varied for an auk. I studied colonies of this species at East Limestone Island (1993-1995, 2002-2003: data collected by Laskeek Bay Conservation Society) and Reef Island (2007-2008), Haida Gwaii, British Columbia. Incubation shift length was correlated between pair members and my results show that maintaining incubation schedule was an essential component of reproductive success. Incubation shift length varied in response to prevailing local weather and sea conditions (e.g. wind speed, wave height), perhaps as a consequence of reduced foraging efficiency. Incubation shift length was longer in years when sea surface temperature in April was high. In years with longer shift, birds had lower reproductive success and chicks departed the nest with a lower body mass. My results explained if we assume that multi-day incubation shifts in Ancient Murrelets are the adaptively preferred strategy, through reduction in predation risk, but that actual shift lengths are modified by immediate weather and foraging constraints.

Les repercussions de l'activite humaine sur la productivite des ecosystemes sont rarement mesurees en relation a la productivite attendue des aires intactes. J'ai utilise des observations de productivite nette de l'ecosysteme durant la saison de croissance (gNEP) pour les aires strictement protegees afin de predire la gNEP attendue dans l'absence d'activite humaine a travers d'un aire d'etude Canadienne, sans influence de l'autocorrelation spatiale. Comparant la gNEP predite avec la gNEP actuelle entre 1998 et 2005, j'ai mesure les deviations pertinents au point de vue biologique positifs et negatifs causes par l'activite humaine, sur 4% et 7.5% du territoire canadien, respectivement. Les activites humaines a long terme etaient lies typiquement aux deviations negatifs, e.g., urbanisation, tandis que les deviations positifs avaient trait a l'alteration temporaire du territoire, e.g., deforestation. Les deviations de gNEP changent a un taux de +/-1.6%·annee-1 avec quelques differences entres utilisations du terrain. Les deviations positifs croissaient en aire presque trois fois plus rapidement que les deviations negatifs. Les deviations inattendues de gNEP refletaient fortement les anomalies a court terme de la meteo, ce qui suggere des impacts des changements climatiques acceleres.

The Blanding's turtle, a semi-aquatic freshwater species found in North America, is listed as a species at risk in 17 of the 18 state/provincial jurisdictions across its range (NatureServe, 2009). Furthermore, approximately 20% of the global range of the Blanding's turtle is contained in Ontario (COSEWIC, 2005). The purpose of this study was to characterise the spatial ecology of Blanding's turtles in Ontario at multiple spatial scales, from a single island population to Ontario-wide habitat suitability modeling. I followed 38 Blanding's turtles (20 males, 13 gravid females, and 5 non-gravid females) in 2008 and 2009 on Grenadier Island in the St. Lawrence River, Ontario via radio-telemetry. Furthermore, I built habitat suitability models using historical Blanding's turtle records in Ontario and two machine learning algorithms: maximum entropy modeling (MAXENT) and boosted regression trees (BRTs). At the local scale, Blanding's turtles selected colder water with more submerged and floating vegetation and avoided open water. Reproductive class and month did not have a significant effect on the mean daily movement of turtles in May, July, and August. In June, however, gravid females moved significantly more than males. Gravid females also had significantly larger home ranges than both males and non-gravid females. At the landscape level, Blanding's turtle habitat suitability was best predicted by air temperature, wetland area, open water area, cropland area, and road density. Habitat suitability increased with increasing air temperature, wetland area, forested area, alvar area, bedrock outcrop area, and decreased with increasing cropland area, pastures and field area, precipitations, terrain ruggedness, and settlements and developed land. Mean area under the operating curve (AUC) values for habitat suitability models tested on independent data ranged from 0.878 to 0.912. My results highlight the importance of stratifying spatial analyses by reproductive class and time and of including terrestrial habitat in management plans for Blanding's turtles.

The combination of climate change and anthropogenic land use changes are having a substantial effect on species' distributions worldwide. While climate change has led to range expansion in many areas, habitat loss and degradation are leading causes of extinction. It is therefore important to consider how these two forces are acting together to alter species distributions. Here, I have tested whether responses of butterfly species to climate change can be predicted based on range size and whether the level of human impact differs in areas where ranges have expanded and where they have collapsed. I found that the number of widespread species that have shifted their ranges north is not statistically different from the number of range-restricted species that have also shifted north. I also found that human impacts are greater in areas where ranges had collapsed than where they had expanded, and that this pattern was stronger for range-restricted species than for widespread ones. Butterfly species are not tracking changing climate in any predictable way with respect to range size, and are likely limited in their ability to do so by habitat losses.