Genetically modified (GM) canola (Brassica napus L.) has been widely adopted in Canada since its commercial release in 1995 and now represents over 85% of the canola grown in western Canada. Concurrently, GM canola volunteers have become an increasing management problem in cultivated fields and are ubiquitous in adjacent ruderal (non-cropped disturbed) habitats. However, systematic post-release monitoring is lacking and the ecological and agronomic impacts of escaped GM canola are poorly understood. In this dissertation, I characterize the escape, demography and distribution of GM canola in ruderal habitats in southern Manitoba, at multiple spatial and temporal scales. I characterized GM herbicide tolerance traits in 16 escaped canola populations. The progeny of 129 plants were tested in herbicide trials; 74% of plants produced glyphosate-tolerant progeny, 63% produced glufosinate-tolerant progeny, and 34% produced multiple herbicide-tolerant progeny as a result of gene flow between escaped plants. At the population-scale, four escaped GM canola populations were monitored and periodic matrix models were constructed to describe the dynamics and persistence of flowering plants. Escaped populations were observed to flower in synchrony with adjacent crops and were projected to persist for 2 to 5 years, confirming the potential for gene flow between escaped and cultivated canola populations. At the landscape-scale, the distribution of escaped canola was surveyed in three agricultural regions. Regional factors were important determinants of distribution; escaped canola density was positively correlated with canola cropping intensity and with traffic intensity, and was negatively correlated with distance to grain distribution centres. Local seed dispersal had negligible impact on distribution compared to landscape-scale anthropogenic seed inputs resulting from agricultural transport. These findings suggest that escaped canola persists as a metapopulation, where long-distance dispersal and colonization compensate for frequent extinction of local populations. Escaped populations play an important role in the persistence and spread of GM traits at large spatial scales, with substantial implications for the coexistence of GM and non-GM crops, and especially for organic and reduced-tillage farming operations. Landscape-scale management approaches, designed and implemented collaboratively by multiple stakeholders, are necessary to mitigate the risks of contamination resulting from GM trait escape. Regulation and ongoing monitoring of GM crops must acknowledge and address the dynamic regional nature of seed- and pollen-mediated gene flow.

Anopheles melas is a brackish water mosquito found along the coast of West-Africa where it can be the dominant malaria vector locally. In order to facilitate genetic studies of this species and to examine the usefulness of microsatellite markers when used in a sibling species, 45 microsatellite loci originally developed for Anopheles gambiae were sequenced in An. melas. These loci were evaluated on their suitability as polymorphic markers based on repeat structure, length, and polymorphism in wild An. melas populations. Of the 45 loci, 18 were not considered promising markers in An. melas. A total of 48 out of 90 An. gambiae primers contained at least one mismatch with the An. melas annealing site. An. melas-specific primers were designed for 27 loci, and their variability was examined in two wild populations from Equatorial Guinea. Based on a low level of polymorphism, Hardy-Weinberg disequilibrium, or poor amplification, a further 12 loci were excluded. The remaining fifteen loci were screened in four additional wild populations from a wider geographic region including Equatorial Guinea, Cameroon, The Gambia, and Guinea Bissau. These loci showed an average heterozygosity ranging from 0.18 to 0.79, with 2.5 to 15 average alleles per locus, yielding 13 highly polymorphic markers and two loci with more limited variability in a wide geographic region. To examine the effects of cross species amplification, five of the original An. gambiae markers were also amplified in the An. melas populations. Null alleles were found for one of these An. gambiae markers. We discuss the pitfalls of using microsatellite loci even in a very closely related species, and conclude that in addition to the well-known problem of null alleles associated with this practice, many loci may prove to be of very limited use as polymorphic markers even when used in a sibling species. Fifteen An. melas-specific markers were subsequently amplified and analyzed in 11 wild An. melas populations from throughout the range of this species, including Bioko Island, Equatorial Guinea. We analyzed pair-wise population differentiation between all populations, and found that all but two comparisons were significant (p-val.<0.05), and populations clustered into three distinct groups representing Bioko Island, Central Africa, and West Africa populations. A Bayesian clustering analysis found little, if any, evidence for migration from mainland to Bioko Island populations, although there was evidence of migration from Bioko Island to the West population cluster, and from the Central to the West population cluster. Simulations of historical gene followed these same patterns and further support our predictions of unidirectional gene flow. Comparison of 1161 nucleotides amplified and sequenced from the ND4 and ND5 regions of the mtDNA showed that differentiation between An. melas population clusters is on par with levels of differentiation between member species of the An. gambiae complex, with low support for internal nodes in a maximum likelihood tree, which suggests that observed An. melas clusters are not monophyletic. From this we hypothesize that Bioko Island An. melas populations are derived from Tiko, Cameroon, and that these populations became isolated from one another when sea levels rose after the last glaciation period (?10,000-11,000 years ago), cutting off Bioko Island populations from the mainland and significantly reducing migration. Our conclusions have implications for vector control within the region, as Bioko Island is the subject of an intensive malaria control campaign, and the lack of migration from mainland West Africa to Bioko Island make it unlikely that eradicated populations of this malaria vector will be repopulated by mainland immigrants.

The overall aim of the study was to assess the risk of gene flow from Brassica crops by insectmediated pollen transport. I measured the viability of pollen in Brassica flowers throughout crop development and compared this with the viability of pollen transported by insects inside and outside one early- and one late-season crop. In order to evaluate the relative importance of different species in pollen transport, I measured abundance of flower visitors during crop development, and measured the foraging behaviour of five key pollinator species throughout the growing season, in relation to variation in microclimate, crop phenology and the relative abundance of other pollinator species competing for flower resources. Flower visiting insects of Brassica rapa crops were highly diverse, and their abundance and diversity changed with crop phenology. I found similar abundances at the family level for both crops studied, although capture rates were greater in the early- than in the late-season crop. Across flowering development, the greatest numbers of insects were captured at the peak of flowering for both crops. During the flowering period, Diptera was the most abundant order collected, followed by Hymenoptera. The most abundant family in Hymenoptera was Apidae which tracked crop development in both fields, with greater numbers of insects captured inside than outside the field. Standardized-count pollen loads were smaller in Diptera than in Hymenoptera. Of the five key pollinator species sampled, Lasioglossum sordidum (Hymenoptera: Halictidae), Apis mellifera (Hymenoptera: Apidae) and Bombus terrestris (Hymenoptera: Apidae) transported similar pollen loads, which were much greater than those carried by Eristalis tenax (Diptera: Syrphidae) and Melangyna novae-zealandiae (Diptera: Syrphidae). The numbers of insects captured outside of the crop were 10% and 33% of the totals captured inside for the early- and the late-season crop, respectively. The proportion of insects entering versus leaving the crop varied considerably across species, crops and trap location (i.e., whether traps were inside or 50 m outside the border of the crop). However, it is worth noting that not uncommonly more insects were attracted into the crop early in the season, staying there rather than leaving, and then when flowers started to disappear there was a massive escape of insects leaving. This research provides evidence for the influence of crop age on the foraging behaviour of key pollinators and for species-specific variation in the foraging behaviour of Brassica visitors with crop development. Temporal variation in the rate and variability of movement between flowers, and the duration and variability in time spent on each flower, throughout the growing season differed markedly between pollinator species. Flower density, plant density, and the abundance of other insects contributed to the observed variation in pollinator behavioural activity for A. mellifera, E. tenax, M. novae-zelandiae and L. sordidum. Bombus terrestris had the greatest rates and variability of movement, and the greatest rates of flower visitation among all key pollinators studied. Therefore B. terrestris might contribute to gene flow to a greater extent than other key pollinators. Additionally B. terrestris had the greatest variability in the rate of movement, increasing the risk of pollen movement over long distances. In summary, I found that (i) insect abundance and diversity changed with crop phenology and Diptera was the most abundant order collected, (ii) flower density, plant density, and the abundance of other insect pollinators were important factors explaining pollinator behaviour for all key pollinators, except B. terrestris, (iii) B. terrestris might contribute to gene flow to a greater extent than other key pollinators, because it has a greater rate of flower visitation and a greater flight distance between flowers than other pollinators, and (iv) pollen viability tended to decrease with crop development and declined sharply even just 50 m outside the edge of the crop.

Genetically modified (GM) canola (Brassica napus L.) has been widely adopted in Canada since its commercial release in 1995 and now represents over 85% of the canola grown in western Canada. Concurrently, GM canola volunteers have become an increasing management problem in cultivated fields and are ubiquitous in adjacent ruderal (non-cropped disturbed) habitats. However, systematic post-release monitoring is lacking and the ecological and agronomic impacts of escaped GM canola are poorly understood. In this dissertation, I characterize the escape, demography and distribution of GM canola in ruderal habitats in southern Manitoba, at multiple spatial and temporal scales. I characterized GM herbicide tolerance traits in 16 escaped canola populations. The progeny of 129 plants were tested in herbicide trials; 74% of plants produced glyphosate-tolerant progeny, 63% produced glufosinate-tolerant progeny, and 34% produced multiple herbicide-tolerant progeny as a result of gene flow between escaped plants. At the population-scale, four escaped GM canola populations were monitored and periodic matrix models were constructed to describe the dynamics and persistence of flowering plants. Escaped populations were observed to flower in synchrony with adjacent crops and were projected to persist for 2 to 5 years, confirming the potential for gene flow between escaped and cultivated canola populations. At the landscape-scale, the distribution of escaped canola was surveyed in three agricultural regions. Regional factors were important determinants of distribution; escaped canola density was positively correlated with canola cropping intensity and with traffic intensity, and was negatively correlated with distance to grain distribution centres. Local seed dispersal had negligible impact on distribution compared to landscape-scale anthropogenic seed inputs resulting from agricultural transport. These findings suggest that escaped canola persists as a metapopulation, where long-distance dispersal and colonization compensate for frequent extinction of local populations. Escaped populations play an important role in the persistence and spread of GM traits at large spatial scales, with substantial implications for the coexistence of GM and non-GM crops, and especially for organic and reduced-tillage farming operations. Landscape-scale management approaches, designed and implemented collaboratively by multiple stakeholders, are necessary to mitigate the risks of contamination resulting from GM trait escape. Regulation and ongoing monitoring of GM crops must acknowledge and address the dynamic regional nature of seed- and pollen-mediated gene flow.

Population structure, gene flow and dispersal are some of the most commonly estimated population parameters in population genetics, evolutionary biology and conservation genetics. The primary aim of thesis is to test the precision and accuracy of genetic estimates of population structure, gene flow and dispersal. The controlled replicated Drosophila melanogaster experiments of known effective population size (Ne = 14.3) and dispersal rate (m = 0.0025 - 0.04) all adhered to Wright??s demographic island model. Three statistical approaches were empirically tested: 1) the conversion of population structure to gene flow using FST, RST, SHUA and PhiST ; 2) the private alleles method to estimate gene flow; 3) a Bayesian assignment method to estimate dispersal (BAYESASS 1.2). Even in the best-case scenario, almost all current methods except SHUA significantly underestimate population structure, and consequently overestimate gene flow and dispersal when applied to real populations. It was crucial to ensure that the manipulated rate of gene flow was correctly defined. This led to three supporting investigations of hybrid performance, inversion polymorphisms and effective population size. The hybrid performance investigation demonstrated that the manipulated rate of gene flow had not been unexpectedly inflated by hybrid vigour or reduced by breakdown. These experiments also demonstrated that close inbreeding is not a necessary precondition for hybrid vigour or breakdown, which is important for conservation strategies involving induced dispersal. The investigation of inversion polymorphisms ensured that the manipulated rate of gene flow was not affected by selection on inverted regions. The effective population size investigation used a temporal estimation method to confirm that the Ne was accurately predicted by an N:Ne ratio of 0.286. Additionally this experiment showed that the single-sample estimation methods implemented by ONeSAMP or LDNE resulted in downwardly biased estimates of Ne in structured populations. In conclusion these results call into question the confidence that biologists may have in some of the most widely used molecular tools in conservation biology.

Asymmetrical dispersal is when dispersal rates differ in opposite directions. This is expected to be common in natural populations. This work aims to study the symmetrical and asymmetrical dispersal through the use of a simulation program, simuPOP. The main questions were (i) "what are the differences between asymmetrical and symmetrical dispersal in relation to genetic differentiation and equilibrium?" and (ii) "Is it possible to identify asymmetrical dispersal structure from observed patterns of genetic differentiation between populations, and variation within populations?". To address these questions, simulations were conducted with two and three subpopulations subject by three different dispersal rate contrasts and several spatial patterns of dispersal. Variables were estimated at drift-dispersal equilibrium included genetic differentiation between subpopulations (θ) and heterozygosity. With pairwise θ for three subpopulations it was possible to determine whether the metapopulations were subject to symmetrical or asymmetrical dispersal and sometimes to identify the structure of dispersal. Equilibrium heterozygosities did not aid diagnosis of asymmetrical dispersal patterns. I also checked the applicability of two predictions originally made for symmetrical dispersal: Wright's expectations for θ at equilibrium, and Whitlock's expectations fro time to half of equilibrium θ. In most cases these expectations were not applicable. Study of asymmetrical dispersal on living organisms is strongly encouraged.

The overall aim of the study was to assess the risk of gene flow from Brassica crops by insectmediated pollen transport. I measured the viability of pollen in Brassica flowers throughout crop development and compared this with the viability of pollen transported by insects inside and outside one early- and one late-season crop. In order to evaluate the relative importance of different species in pollen transport, I measured abundance of flower visitors during crop development, and measured the foraging behaviour of five key pollinator species throughout the growing season, in relation to variation in microclimate, crop phenology and the relative abundance of other pollinator species competing for flower resources. Flower visiting insects of Brassica rapa crops were highly diverse, and their abundance and diversity changed with crop phenology. I found similar abundances at the family level for both crops studied, although capture rates were greater in the early- than in the late-season crop. Across flowering development, the greatest numbers of insects were captured at the peak of flowering for both crops. During the flowering period, Diptera was the most abundant order collected, followed by Hymenoptera. The most abundant family in Hymenoptera was Apidae which tracked crop development in both fields, with greater numbers of insects captured inside than outside the field. Standardized-count pollen loads were smaller in Diptera than in Hymenoptera. Of the five key pollinator species sampled, Lasioglossum sordidum (Hymenoptera: Halictidae), Apis mellifera (Hymenoptera: Apidae) and Bombus terrestris (Hymenoptera: Apidae) transported similar pollen loads, which were much greater than those carried by Eristalis tenax (Diptera: Syrphidae) and Melangyna novae-zealandiae (Diptera: Syrphidae). The numbers of insects captured outside of the crop were 10% and 33% of the totals captured inside for the early- and the late-season crop, respectively. The proportion of insects entering versus leaving the crop varied considerably across species, crops and trap location (i.e., whether traps were inside or 50 m outside the border of the crop). However, it is worth noting that not uncommonly more insects were attracted into the crop early in the season, staying there rather than leaving, and then when flowers started to disappear there was a massive escape of insects leaving. This research provides evidence for the influence of crop age on the foraging behaviour of key pollinators and for species-specific variation in the foraging behaviour of Brassica visitors with crop development. Temporal variation in the rate and variability of movement between flowers, and the duration and variability in time spent on each flower, throughout the growing season differed markedly between pollinator species. Flower density, plant density, and the abundance of other insects contributed to the observed variation in pollinator behavioural activity for A. mellifera, E. tenax, M. novae-zelandiae and L. sordidum. Bombus terrestris had the greatest rates and variability of movement, and the greatest rates of flower visitation among all key pollinators studied. Therefore B. terrestris might contribute to gene flow to a greater extent than other key pollinators. Additionally B. terrestris had the greatest variability in the rate of movement, increasing the risk of pollen movement over long distances. In summary, I found that (i) insect abundance and diversity changed with crop phenology and Diptera was the most abundant order collected, (ii) flower density, plant density, and the abundance of other insect pollinators were important factors explaining pollinator behaviour for all key pollinators, except B. terrestris, (iii) B. terrestris might contribute to gene flow to a greater extent than other key pollinators, because it has a greater rate of flower visitation and a greater flight distance between flowers than other pollinators, and (iv) pollen viability tended to decrease with crop development and declined sharply even just 50 m outside the edge of the crop.

In this study different molecular techniques are contrasted (RAPD's, allozyme, sequencing mtDNA, sequencing ribosomal spacers) and appropriate analytical methods (allelic and infinite-sites approaches; inbreeding and coalescent models) used for estimating population genetic parameters in parasites. A range of population genetic questions at different scales were chosen to emphasise the importance of tailoring techniques and analytical methods to the particular question being investigated. The realisation that each question formulated has a particular scale means the appropriate technique and markers must be useful at that scale to attempt to answer the question. The useful scale of a technique depends several factors including the region of DNA examined, the density of sampling of the technique, and the mode of evolution of the markers. Each technique will produce a useful range of variability. Below the lower limit there is no variation, above the upper limit the variation is too high to produce useful comparisons. Parasites are of interest for many reasons, primarily because they can cause disease and thus impact on their host's population dynamics. They are often closely associated with their hosts and may undergo co-evolution, as well as causing an ongoing immunological "arms race" with their hosts. The parasitic mode of live is found throughout nearly all taxonomic groupings and thus classical models of population genetics based on sexual, diploid vertebrates do not fit well with the entire diversity of parasite groups. Genetic diversity within and among populations of Echinococcus granulosus was examined contrasting a RAPD dataset with an allozyme dataset. Two models of variation in Echinococcus have been proposed, those of Smyth and Rausch, and the expected genetic structure from each was compared to the observed genetic structure. The premise of Smyth’s model, predominant self-fertilisation, was supported, but the resultant pattern of genetic variation followed Rausch’s model. RAPD data, being dominant, present challenges to analysis. An approach to overcome this dominance problem and allow standard allelic frequency analysis is described using the selfing rate estimated from allozyme data. The RAPD data were also analysed using both band-sharing and nucleotide diversity approaches. A population genetic study of Ostertagia ostertagi in the USA was extended to two different scales: within an Australian state and between the USA and Australian continents. Three alternative explanations for the observed discrepancy between genetic structure and differentiation in an important biological trait, hypobiosis, were explored. A number of programs and analyses were compared including coalescent geneflow estimates. Variation among multiple copies of two spacer regions of rDNA was examined within individuals of Ostertagia ostertagi. Both the intergenic spacer and internal transcribed spacer 1 regions were found to include repeat regions, with different numbers of repeats creating length differences in clones from the same worm. Multi-copy genes present extra challenges in analysis to ensure that only homologous copies are being compared. Many studies fail to look for variation within populations or within individuals. The two major conclusions from these examples are that: 1). The study of variation necessarily involves an implicit scale, and markers must be chosen that are appropriate to the question being explored. 2). Using several methods of analysis of genetic data allows contrasts to be made, and if different methods produce similar results gives much more confidence in the conclusions drawn. Incongruence in results leads to new questions and reexamination of the assumptions of each analysis.

Asymmetrical dispersal is when dispersal rates differ in opposite directions. This is expected to be common in natural populations. This work aims to study the symmetrical and asymmetrical dispersal through the use of a simulation program, simuPOP. The main questions were (i) "what are the differences between asymmetrical and symmetrical dispersal in relation to genetic differentiation and equilibrium?" and (ii) "Is it possible to identify asymmetrical dispersal structure from observed patterns of genetic differentiation between populations, and variation within populations?". To address these questions, simulations were conducted with two and three subpopulations subject by three different dispersal rate contrasts and several spatial patterns of dispersal. Variables were estimated at drift-dispersal equilibrium included genetic differentiation between subpopulations (θ) and heterozygosity. With pairwise θ for three subpopulations it was possible to determine whether the metapopulations were subject to symmetrical or asymmetrical dispersal and sometimes to identify the structure of dispersal. Equilibrium heterozygosities did not aid diagnosis of asymmetrical dispersal patterns. I also checked the applicability of two predictions originally made for symmetrical dispersal: Wright's expectations for θ at equilibrium, and Whitlock's expectations fro time to half of equilibrium θ. In most cases these expectations were not applicable. Study of asymmetrical dispersal on living organisms is strongly encouraged.

Investigating secondary contact of historically isolated lineages can provide insight into how selection and drift influence genomic divergence and admixture. Here, we studied the genomic landscape of divergence and introgression following secondary contact between lineages of the Western Diamondback Rattlesnake (Crotalus atrox) to determine whether genomic regions under selection in allopatry also contribute to reproductive isolation during introgression. We used thousands of nuclear loci to study genomic differentiation between two lineages that have experienced recent secondary contact following isolation, and incorporated sampling from a zone of secondary contact to identify loci that are resistant to gene flow in hybrids. Comparisons of patterns of divergence and introgression revealed a positive relationship between allelic differentiation and resistance to introgression across the genome, and greater-than-expected overlap between genes linked to lineage-specific divergence and loci that resist introgression. Genes linked to putatively selected markers were related to prominent aspects of rattlesnake biology that differ between populations of Western Diamondback rattlesnakes (i.e., venom and reproductive phenotypes). We also found evidence for selection against introgression of genes that may contribute to cytonuclear incompatibility, consistent with previously observed biased patterns of nuclear and mitochondrial alleles suggestive of partial reproductive isolation due to cytonuclear incompatibilities. Our results provide a genome-scale perspective on the relationships between divergence and introgression in secondary contact that is relevant for understanding the roles of selection in maintaining partial isolation of lineages, causing admixing lineages to not completely homogenize.