Track distributions of snowshoe hare (Lepus americanus), fox (Vulpes fulva) and lynx (Lynx canadensis) were related to distance from anthropogenic corridors (abandoned road, ditch and hydroline) and habitat types in a disturbed winter landscape near Cochrane, northeastern Ontario. More hare tracks were found in mature conifer habitat than expected. There were fewer hare tracks in the corridor, in immature conifer habitat and in mature mixedwood habitat than expected. The rest of the habitat types contained hare tracks close to expected values. Fifty-four percent of all hare tracks were found 10 to 30 metres from the centre of the corridor. Far fewer hare tracks were found 0 to 10 metres from the centre of the corridor than expected. More tracks were found 10 to 30 metres from the corridor centre than expected. Distances farther than thirty metres from the corridor centre contained hare tracks close to expected values. Loglinear analysis indicated that distance from the corridor had more influence on the distribution of hare tracks than did habitat type. The interaction of distance and habitat had an effect as well, but much less so than distance alone. Information analysis also demonstrated that distance from the corridor was more significant to the distribution of hare tracks than habitat type. The distance/habitat interaction was strong as well, with specific distance interval/habitat type combinations contributing much to the observed pattern. The corridor habitat at 0 to 10 metres, mature conifer at 10 to 20 metres, mature hardwood habitat at 10 to 30 and 40 to 50 metres, and mature mixedwood habitat at 20 to 50 metres are specific habitat/distance combinations contributing the most to the observed track pattern. These combinations represent patches which were either avoided or preferentially used by hare, with mature conifer at 10 to 20 metres being used preferentially while the other combinations are avoided. The sample sizes for fox and lynx were very small (n = 30 and 24 respectively). Statistical testing at p 0.05 could not reject the hypotheses that fox and lynx tracks were found in distance intervals and habitat types according to availability. At p 0.1, analysis showed only that less fox tracks were found in hardwood habitat than expected. However, when the fox and lynx datasets were combined into a single 'predator' dataset, goodness-of-fit testing rejected the hypothesis that tracks were found in distance intervals and habitat types according to availability. However, no variable classes showed significant chi-square values at the p 0.05 level. At the p 0.1 level, analysis showed that the combined fox and lynx tracks were found in mature conifer habitat more than expected. As well, the combined fox and lynx tracks were found 40-50 metres from the centre of the corridor in less than expected values.

Patterns of resource partitioning have been widely described, but the causes of these patterns are incompletely understood. To explore how the patterns are formed, I did a quantitative literature survey. In Chapter One, I (1) discuss current misconceptions in the literature, (2) document and describe five ways in which resource partitioning can occur, and (3) establish guidelines for predicting the two most common ways in which patterns of partitioning are formed. In Chapter One, the most common way in which patterns of resource partitioning were formed was through shared preference niche organization. Centrifugal organization extends shared preference organization from one resource gradient to multiple gradients. In Chapter Two, experimental evidence supporting the validity of the centrifugal organization model is reviewed and predictions of the model are presented. The centrifugal organization model predicts how the size of species pools changes along a resource gradient. Two other models from the literature make contradictory predictions. The three models were tested in Chapter Three using biomass/species composition data from 33 eastern North American wetlands. The resulting 640 quadrats produced a pattern of pool size consistent with a previously untested model, the species pool model. This model states that the pattern of pool size is the same as the pattern of alpha diversity along a biomass gradient. This suggests that (1) the more easily measured alpha diversity values can be used to predict where large species pools occur and (2) ecological processes that are associated with changes in alpha diversity may also influence the species pool.

In this thesis, I examine the factors that influence egg size, chick development, and the nest departure decision of young Thick-billed Murres, Uria lomvia. The study was conducted at a colony of 30,000 breeding pairs at Coats Island, N.W.T., Canada, in 1994 and 1995. Experienced birds initiated egg formation earlier in the season, and deposited more yolk per day, than younger birds. Among experienced pairs, egg size and laying date were unrelated; eggs laid by young birds at the same time as those of experienced pairs were similar. There was a positive relationship between egg size and the rate at which nestling Thick-billed Murres grew their wing feathers. Growth rate in mass was independent of egg size. Chicks with long wings left the nest to go to sea at a younger age than those with shorter wings in one of two years. The chicks of high quality, experimentally-delayed breeders grew as quickly, and departed at similar mass, as those of unmanipulated early-breeding birds. I conclude that the poor quality of late-breeding birds, more than a seasonally-deteriorating environment, causes the declines in growth of nestlings. I tested two predictions of the Ydenberg model of alcid nest departure for the Thick-billed Murre, which employs a nest departure strategy intermediate within the Alcidae. Among and within colonies, was little support for the model. I suggest that there are developmental and behavioural considerations in the nest departure decision not adequately explained by the Ydenberg model. I propose that the strategy centres around getting the chick from the cliff to the sea as quickly as possible, but with the mortality risk involved in making the transition minimized. (Abstract shortened by UMI.)

Ten experimental stream channels, fed from a Laurentian steam were used to examine the short-term impact of a pulse of Velpar L on stream periphyton and invertebrates. Velpar L was added to five alternating channels for 24 hours with the other five channels serving as controls. The mean concentration of hexazinone in the treatment channels varied over time from 145 $\mu$g L$\sp{-1}$ to 432 $\mu$g L$\sp{-1}$. Periphyton chlorophyll a-specific productivity was significantly reduced during the Velpar L treatment. Within 24 hours after the termination of the Velpar L treatment, productivity returned to control levels. Mean periphyton biomass, measured as the quantity of chlorophyll a per unit area, was not affected by the Velpar L treatment. In addition, the concentration of hexazinone that reduced periphyton chlorphyll a-specific productivity by 50% after a four hour exposure was determined. This value was only 3.6 $\mu$g L$\sp{-1}$ of hexazinone. Thus, low concentrations of Velpar L affect stream periphyton but the impact appears to be temporary. Velpar L did not cause an increase in the drift of insects or other macroinvertebrates. Mean density of benthic macroinvertebrates increased slightly with time in the Velpar L-treated channels and nearly doubled in the control channels. The mean length of individual macroinvertebrates decreased slightly over time in the control channels but not in the treated channels. However, the changes in mean density, biomass and length were not statistically significant. The variance of the invertebrate lengths increased over time in both the control and Velpar L-treated channels, which suggests that the size distribution of the invertebrate community varied with time. (Abstract shortened by UMI.)

The fate of the phototoxic phytochemical, $\alpha$-terthienyl ($\alpha$-T), which has been proposed as a novel mosquito larvicide was studied in target and non-target organisms. The lethality of this compound was determined in the laboratory by uniform acute toxicity bioassays using the target mosquito larvae, Aedes atropalpus, and two non-target species. Daphnia magna represented a potentially sensitive non-target organism and Limnephilus indivisus (caddisfly) larvae represented a potentially insensitive non-target species. The mosquito larvae had a lethal concentration for 50% mortality (LC$\sb{50})$ value determined to be 31.97 ppb. However, D. magna showed an order of magnitude more sensitivity towards $\alpha$-T with an LC$\sb{50}$ of 1.74 ppb while L. indivisus larvae showed considerably less sensitivity at 64.76 ppb. The bioaccumulation potential was investigated in the three species by toxicokinetic studies. Initial rates of accumulation by the first compartment were highest for A. atropalpus, followed by D. magna and L. indivisus. A. atropalpus reached a steady state in the second compartment. The accumulation of radiolabeled ($\sp3$H) $\alpha$-T did not plateau in the non-target species. The clearance rate constants were highest for D. magna and lowest for L. indivisus, indicating that the most sensitive species was eliminating $\alpha$-T the fastest, in contrast with previous studies. The results from the above studies, combined with previous field trials, predicted mortality at the applied concentrations. This did not translate into severe impacts in the fate study under natural conditions, probably due to volatilization, presence of dissolved organic matter in the water, and rapid photodegradation of $\alpha$-T. In addition, the microcosms were highly anoxic relative to the natural pools. (Abstract shortened by UMI.)

Concentrations of Cd, Cr, Cu, Fe, Hg, Ni, Pb, and Zn were measured in surficial sediments and in five aquatic macroinvertebrate taxa from shallow littoral areas of the St. Lawrence River, near Cornwall, Ontario. The principal objectives were to assess metal levels in littoral sediments and littoral macroinvertebrates, and to determine the influence of local point sources relative to upstream inputs. Hg concentrations were, however, greatly enriched at a site 1 km downstream of two historical point sources of Hg (Domtar Fine Papers/ICI Forest Products). Littoral sediments had lower metal levels than contaminated areas of the Great Lakes and deeper sites in the Cornwall area. Results of these comparisons, however, suggested a problem of Hg contamination in the Cornwall area. Metal levels in littoral macroinvertebrates were assessed for the amphipod Gammarus fasciatus, the decapod Orconectes sp., the gastropod Bythinia tentaculata, the Chironomidae (midges) and the Oligochaeta (worms). Metal concentrations in macroinvertebrates were found to differ among taxonomic groups, and these differences were metal-specific. When compared to concentrations in sediments, invertebrates had lower levels of Cd (except for B. tentaculata), Cr, Fe, Ni and Pb, and higher concentrations of Cu and Zn. Concentrations of Hg were similar to levels in the sediments, except for slightly higher concentrations in B. tentaculata. Concentrations of Cr, Fe, Ni and Zn in chironomids and oligochaetes were similar or higher than levels reported for contaminated areas of the Great Lakes and deeper sites in the Cornwall area, and this despite the relatively much lower concentrations in littoral sediments. (Abstract shortened by UMI.)

The objective of this study was to measure and compare several physical and chemical characteristics of two white sucker (Catostomus commersoni) populations from the St. Lawrence River, near Cornwall, Ontario and Massena, New York, that could be used to address environmental quality issues in this area. Seven hundred and sixty-two white suckers from the St. Lawrence River were collected upstream and downstream of the Moses-Saunders power dam near Cornwall, Ontario and Massena, New York during 1994 and 1995. Upstream white suckers were shorter and weighed less at older ages, had a lower average fecundity, a greater mean egg diameter, lower condition factor at older ages, greater overall mean age, and a higher incidence of lip and body papillomas than the downstream fish. In addition, a total of 80 fish (40 fish upstream and 40 fish downstream) were analysed for organic chemical contaminants. The mean total concentration of 34 polycyclic aromatic hydrocarbon compounds (PAH), analysed in muscle tissue using GC-MS-SIM techniques, were significantly higher in white suckers caught upstream of the dam (75.2 $\pm$ 30.4 ng/g dry wt.) than in those caught downstream (53.6 $\pm$ 29.7 ng/g dry wt.). Concentrations of 132 polychlorinated biphenyl (PCB) congeners and 31 organochlorine pesticides were also analysed in muscle tissue using GC electron-capture detectors (ECD). Detectable levels of PCB and organochlorine pesticides were found in nearly all fish sampled. However, there was no statistical difference between organochlorine contaminant concentrations of upstream and downstream fish, except for total aldrin (3.4 $\pm$ 4.0 ng/g dry wt. upstream, 1.9 $\pm$ 2.7 ng/g dry wt. downstream). Concentrations of most organochlorine compounds increased linearly with lipid content, but not relationship was found between contaminant concentration and age of the fish, except for DDT and mirex. (Abstract shortened by UMI.)

The Firth River is located in a sector of northern Yukon that has not been touch by the latest glaciations. The objective of the study is to determine if the Firth River naled ice has an impact on the surrounding soils and vegetation. The analysis of the soils and the observation of the vegetation along 3 transects, one longitudinal and 2 cross-sectional, has showned a differential evolution. The study and datation of an island in the center of the valley has permitted to establish that the river has cut in the valley after 16084 +/$-$ 340 B.P. The grain size distribution of the soils gives an estimation of the deposition modes which, in this case, is mainly fluvial but where the ice may play a role in its dynamic. The soils salt concentration show a relationship with the distance from the center of the valley. It is an indicator of the impact of the ice on the soils and act as a signature of the presence of the ice. The observation of the vegetation growing in the valley also show the influence of the ice: its destruction shows the presence of the ice although its development stages indicate the absence of the ice on a long or short term.

Many studies on forest or vegetation response to global warming have been done using the gap model or empirical models. Thus far, there is no good regional model allowing to predict forest change at an intermediate scale. In this study, we have developed a model of this type, called Knowledge Base Forest Model (KBFM), using an information analytical tool (P scEGASE) based on information theory. Using this model and data from the Canadian Climatic Centre general circulation model, we could predict the future distribution of forest types in the research area: the Province of Manitoba. The study shows that the KBFM may well be used to predict the future regional distribution of forest types. Its main advantages are: (1) environmental variables used as predictors can be qualitatives (e.g. soil texture) as well as quantitative (e.g. temperature); (2) the KBFM provides the possibility to account for the role of soil factors in the forest response to global warming; (3) the KBFM can predict forest type distribution using various climatic scenarios; (4) the KBFM can predict forest type distribution with greater details than empirical models.

The factors that regulate bioaccumulation of mercury in wetland systems and aquatic plants in particular have yet to be identified. Concentrations of total mercury were determined in water, sediments, and the yellow pond lily (Nuphar variegatum L.) for 22 wetlands from South-Central Ontario (Lat. 45$\sp\circ\ 11\sp\prime,$ Long. 78$\sp\circ\ 50\sp\prime$; Lat 46$\sp\circ\ 19\sp\prime$, Long. 80$\sp\circ\ 47\sp\prime)$ and for 23 wetlands from the St. Lawrence River (Lat. 45$\sp\circ\ 02\sp\prime,$ Long. 74$\sp\circ\ 44\sp\prime).$ The South-Central wetlands are remote from any industrial sources of mercury, whereas the wetlands from the St. Lawrence River come from an area where known Hg contamination exists. Positive correlations were found between the organic content of the sediments and the sediment Hg concentrations for both the South-Central and the St. Lawrence River wetlands. However, no relationships between water Hg and pH, alkalinity, and dissolved organic carbon (DOC) were found in either wetland group. The yellow pond lily is a common floating leaved aquatic plant which is the nutrient and energy source for a variety of wildlife. Thus, the importance of determining if the yellow pond lily accumulates toxic levels of mercury is evident. The results indicated that Nuphar variegatum does not accumulate mercury in relation to pH, alkalinity, DOC, or the organic content of the sediment, despite a wide range in these variables from the two wetland groups. Nuphar variegatum is an "excluder" or "non-indicator" of mercury. Remote South-Central wetlands had significantly higher concentrations of mercury in the sediments and the water compared to the St. Lawrence River wetlands. The South-Central wetlands also had significantly higher amounts of organic matter in the sediments and DOC in the water, both of which are known to bind mercury. (Abstract shortened by UMI.)