Species distribution models must balance the need for model generality with that for precision and accuracy. This is critical when modelling range-expanding species such as invasive species. Given the increased use of species distribution models to study invasive species-landscape relationships, a better understanding of the effect of spatial scales, sampling biases, model transferability and discrepancies between different models’ future predictions is necessary. This dissertation addresses these knowledge gaps using mute swans (Cygnus olor) as a case study species. I specifically examine mute swan’s distributions in parts of their native range of Britain and their non-native range of Ontario, Canada. I first investigate which environmental variables at which spatial scales best explain mute swan’s distribution in its non-native range. Second, I perform a sample selection bias study to examine predictive accuracy when species distribution models are built using varying ranges of environmental variables and applied to broader spatial extents. Third, I examine the potential for, and limitations of model transferability between native and non-native regions. Finally, I use two different modelling approaches and three different climate change and land use change scenarios to predict future mute swan habitat suitability. The results indicate that (1) models with better predictive accuracy include environmental variables from multiple ecologically-meaningful scales and measured at spatial extents that include a broad range of environmental variable values; (2) models can exhibit asymmetrical transferability; (3) climate change will facilitate mute swan range expansion in the future more than land use change; and (4) mute swans are often found near urban waterbodies. When modelling invasive species distributions, I suggest that ecologists consider: (I) spatial scale of the underlying landscape processes and species’ use of the landscape; (II) variability and range of each environmental variable used for building models; and (III) stage of establishment of the invasive species.

Species distribution models must balance the need for model generality with that for precision and accuracy. This is critical when modelling range-expanding species such as invasive species. Given the increased use of species distribution models to study invasive species-landscape relationships, a better understanding of the effect of spatial scales, sampling biases, model transferability and discrepancies between different models’ future predictions is necessary. This dissertation addresses these knowledge gaps using mute swans (Cygnus olor) as a case study species. I specifically examine mute swan’s distributions in parts of their native range of Britain and their non-native range of Ontario, Canada. I first investigate which environmental variables at which spatial scales best explain mute swan’s distribution in its non-native range. Second, I perform a sample selection bias study to examine predictive accuracy when species distribution models are built using varying ranges of environmental variables and applied to broader spatial extents. Third, I examine the potential for, and limitations of model transferability between native and non-native regions. Finally, I use two different modelling approaches and three different climate change and land use change scenarios to predict future mute swan habitat suitability. The results indicate that (1) models with better predictive accuracy include environmental variables from multiple ecologically-meaningful scales and measured at spatial extents that include a broad range of environmental variable values; (2) models can exhibit asymmetrical transferability; (3) climate change will facilitate mute swan range expansion in the future more than land use change; and (4) mute swans are often found near urban waterbodies. When modelling invasive species distributions, I suggest that ecologists consider: (I) spatial scale of the underlying landscape processes and species’ use of the landscape; (II) variability and range of each environmental variable used for building models; and (III) stage of establishment of the invasive species.

No description available.

Aim MaxEnt, a very popular species distribution modelling technique, has been used extensively to relate species’ geographic distributions to environmental variables and to predict changes in species’ distributions in response to environmental change. Here, we test its predictive ability through time (rather than through space, as is commonly done) by modeling colonizations and extinctions. Location Continental U.S. and southern Canada. Time period 1979-2009 Major taxa studied Twenty-one species of passerine birds. Methods We used MaxEnt to relate species’ geographic distributions to the variation in environmental conditions across North America. We then modelled site-specific colonizations and extinctions between 1979 and 2009 as functions of MaxEnt-estimated previous habitat suitability and inter- annual change in habitat suitability and neighborhood occupancy. We evaluated whether the effects were in the expected direction, we partitioned model’s explained deviance, and we compared colonization and extinction model’s accuracy to MaxEnt’s AUC. Results IV Colonization and extinction probabilities both varied as functions of previous habitat suitability, change in habitat suitability, and neighborhood occupancy, in the expected direction. Change in habitat suitability explained very little deviance compared to other predictors. Neighborhood occupancy accounted for more explained deviance in colonization models than in extinction models. MaxEnt AUC correlates with extinction models’ predictive ability, but not with that of colonization models. Main conclusions MaxEnt appears to sometime capture a real effect of the environment on species’ distributions since a statistical effect of habitat suitability is detected through both time and space. However, change in habitat suitability (which is much smaller through time than through space) is a poor predictor of change in occupancy. Over short time scales, proximity of sites occupied by conspecifics predicts changes in occupancy just as well as MaxEnt. The ability of MaxEnt models to predict spatial variation in occupancy (as measured by AUC) gives little indication of transferability through time. Thus, the predictive value of species distribution models may be overestimated when evaluated through space only. Future prediction of species’ responses to climate change should make a distinction between colonization and extinction, recognizing that the two processes are not equally well predicted by SDMs.

A relação entre a paisagem e a distribuição da ocorrência e abundância das espécies no espaço é uma das questões centrais em ecologia, com importantes aplicações diretas em tempos de intenso uso da terra e mudanças climáticas por atividades humanas. Contudo, para inferirmos sobre essas relações, temos que descrever a paisagem da forma mais próxima possível de como as espécies realmente a percebem. Uma paisagem pode ser descrita em diferentes níveis hierárquicos de organização do ambiente (e.g. quantidade de um mineral no solo, número de plantas em uma parcela, área de cobertura de floresta...), e cada nível pode ser descrito em diferentes escalas (resolução e extensão de descrição). Os níveis e escalas com maior poder de previsão da ocorrência/abundância de uma espécie são chamados de nível de efeito e escala (extensão) de efeito. Nesta tese, utilizo armadilhas fotográficas e modelos hierárquicos para avaliar relações entre a paisagem e espécies de mamíferos e aves. Meus objetivos são: (1) avaliar se existe relação entre nível e extensão de efeito para prever a abundância de espécies e a área de vida ou massa das mesmas; (2) investigar características do solo que possam ser determinantes da distribuição de mamíferos e aves que consumem solo (geofagia); e (3) a partir de ralações espécie-paisagem, estimar a distribuição da densidade de uma espécie, o veado-catingueiro, para diferentes datas. Não há suporte para que a área de vida ou a massa de uma espécie sejam relevantes para o nível ou extensão em que uma paisagem deva ser descrita a fim de prever a abundância de uma espécie. Isso implica na importância de se avaliar diferentes níveis e extensões de uma paisagem quando na busca por relações espécie-paisagem. Fatores locais, como a argila ou minerais do solo, podem ser importantes para algumas espécies. Descobri que o veado-mateiro e o caititu selecionam solos para consumo com base na quantidade e tipo de argila. O caititu também seleciona solos com base na concentração de microminerais, assim como a juriti-azul, a arara-azul-grande, o mutum, o aracuã e a pomba-galega. Uma descrição da paisagem em nível de composição do solo pode ser relevante para avaliar a distribuição destas e outras espécies. Relações espécie-paisagem podem ser usadas para prever a abundância de espécies no espaço. Estimei a densidade do veado-catingueiro em 1992 (2,07 ind/km2) e em 2011 (7,31 ind/km2), para uma região de pecuária extensiva no nordeste do Pantanal onde foi criada uma reserva em 1997. A densidade desta espécie aumentou 3,5 vezes entre 1992 e 2011, com o fim da pecuária no local. Investigações multinível e multiescala de relações espécie-paisagem ainda são incipientes, embora importantes aplicações destas relações já venham sendo feitas há décadas. / The relationship between the landscape and the distribution of species’ occurrence and abundance is one of the main questions in ecology, with important applications for the current period of intense land use and climate change. However, to infer about these relationships, we have to describe the landscape as closely as possible to how species actually realize it. We can describe landscapes at different hierarchical levels of the environment (e.g. mineral amount in soil, number of plants in a plot, forest cover area…), and each level can be described at different scales (resolution and extent). The best levels and scales to predict species’ occurrence/abundance are known as level of effect and scale (extent) of effect. In this PhD thesis, I use camera traps and hierarchical models to assess relationships between the landscape and mammals or birds. My goals are: (1) to evaluate possible relationships between the level and extent of effect to predict species’ abundance and species’ home range or mass; (2) to investigate soil features important to the distribution of mammals and birds engaged on geophagy (soil consumption); and (3) from species-landscape relations, to estimate the density of a species, the Gray Brocket in space for different dates. There is no support for species’ home range or mass as relevant traits related to the level and scale that a landscape should be described in order to predict species’ abundance. This highlight the importance of evaluating different levels and scales of a landscape when searching for species-landscape relationships. Local factors such as clay or minerals may be important for some species. I found that the Red Brocket and Collared Peccary select soils for consumption based on clay amount and type. The Collared Peccary also selects soil based on the concentration of trace minerals, as well as Blue Ground-dove, Hyacinth Macaw, Bare-faced Curassow, Chaco Chachalaca, and Pale-vented Pigeon. Describing the landscape at soil composition level may be important to evaluate the distribution of these and another species. Species-landscape relationships can be used to predict the abundance of species in space at different dates. I estimated the density of the Gray Brocket in 1992 (2.07 ind/km2) and 2011 (7.31 ind/km2), in a livestock region in northeastern Brazilian Pantanal where a reserve was established in 1997. Gray Brocket density increased 3.5 times between 1992 and 2011, after livestock ban. Multi-level and multi-scale approaches to investigate species-landscape relationships are still emerging, though important applications of such relationships have been done for decades.

A relação entre a paisagem e a distribuição da ocorrência e abundância das espécies no espaço é uma das questões centrais em ecologia, com importantes aplicações diretas em tempos de intenso uso da terra e mudanças climáticas por atividades humanas. Contudo, para inferirmos sobre essas relações, temos que descrever a paisagem da forma mais próxima possível de como as espécies realmente a percebem. Uma paisagem pode ser descrita em diferentes níveis hierárquicos de organização do ambiente (e.g. quantidade de um mineral no solo, número de plantas em uma parcela, área de cobertura de floresta...), e cada nível pode ser descrito em diferentes escalas (resolução e extensão de descrição). Os níveis e escalas com maior poder de previsão da ocorrência/abundância de uma espécie são chamados de nível de efeito e escala (extensão) de efeito. Nesta tese, utilizo armadilhas fotográficas e modelos hierárquicos para avaliar relações entre a paisagem e espécies de mamíferos e aves. Meus objetivos são: (1) avaliar se existe relação entre nível e extensão de efeito para prever a abundância de espécies e a área de vida ou massa das mesmas; (2) investigar características do solo que possam ser determinantes da distribuição de mamíferos e aves que consumem solo (geofagia); e (3) a partir de ralações espécie-paisagem, estimar a distribuição da densidade de uma espécie, o veado-catingueiro, para diferentes datas. Não há suporte para que a área de vida ou a massa de uma espécie sejam relevantes para o nível ou extensão em que uma paisagem deva ser descrita a fim de prever a abundância de uma espécie. Isso implica na importância de se avaliar diferentes níveis e extensões de uma paisagem quando na busca por relações espécie-paisagem. Fatores locais, como a argila ou minerais do solo, podem ser importantes para algumas espécies. Descobri que o veado-mateiro e o caititu selecionam solos para consumo com base na quantidade e tipo de argila. O caititu também seleciona solos com base na concentração de microminerais, assim como a juriti-azul, a arara-azul-grande, o mutum, o aracuã e a pomba-galega. Uma descrição da paisagem em nível de composição do solo pode ser relevante para avaliar a distribuição destas e outras espécies. Relações espécie-paisagem podem ser usadas para prever a abundância de espécies no espaço. Estimei a densidade do veado-catingueiro em 1992 (2,07 ind/km2) e em 2011 (7,31 ind/km2), para uma região de pecuária extensiva no nordeste do Pantanal onde foi criada uma reserva em 1997. A densidade desta espécie aumentou 3,5 vezes entre 1992 e 2011, com o fim da pecuária no local. Investigações multinível e multiescala de relações espécie-paisagem ainda são incipientes, embora importantes aplicações destas relações já venham sendo feitas há décadas. / The relationship between the landscape and the distribution of species’ occurrence and abundance is one of the main questions in ecology, with important applications for the current period of intense land use and climate change. However, to infer about these relationships, we have to describe the landscape as closely as possible to how species actually realize it. We can describe landscapes at different hierarchical levels of the environment (e.g. mineral amount in soil, number of plants in a plot, forest cover area…), and each level can be described at different scales (resolution and extent). The best levels and scales to predict species’ occurrence/abundance are known as level of effect and scale (extent) of effect. In this PhD thesis, I use camera traps and hierarchical models to assess relationships between the landscape and mammals or birds. My goals are: (1) to evaluate possible relationships between the level and extent of effect to predict species’ abundance and species’ home range or mass; (2) to investigate soil features important to the distribution of mammals and birds engaged on geophagy (soil consumption); and (3) from species-landscape relations, to estimate the density of a species, the Gray Brocket in space for different dates. There is no support for species’ home range or mass as relevant traits related to the level and scale that a landscape should be described in order to predict species’ abundance. This highlight the importance of evaluating different levels and scales of a landscape when searching for species-landscape relationships. Local factors such as clay or minerals may be important for some species. I found that the Red Brocket and Collared Peccary select soils for consumption based on clay amount and type. The Collared Peccary also selects soil based on the concentration of trace minerals, as well as Blue Ground-dove, Hyacinth Macaw, Bare-faced Curassow, Chaco Chachalaca, and Pale-vented Pigeon. Describing the landscape at soil composition level may be important to evaluate the distribution of these and another species. Species-landscape relationships can be used to predict the abundance of species in space at different dates. I estimated the density of the Gray Brocket in 1992 (2.07 ind/km2) and 2011 (7.31 ind/km2), in a livestock region in northeastern Brazilian Pantanal where a reserve was established in 1997. Gray Brocket density increased 3.5 times between 1992 and 2011, after livestock ban. Multi-level and multi-scale approaches to investigate species-landscape relationships are still emerging, though important applications of such relationships have been done for decades.

In Peru, research and practices that contribute to Andean natural resources’ sustainable management are needed. The Nor Yauyos Cochas Landscape Reserve is home to a wild vicugna population, which can be viewed as a key resource for conservation and sustainable development. However, some activities of Tanta Peasant Community impact negatively on vicugna’s habitat. The research aimed to determine spatial relations and impacts of Tanta’s activities on vicugna’s habitat and distribution over communal territory.A participatory mapping workshop was applied to determine vicugna’s actual distribution and local activities that could influence vicugna’s habitat. The species’ potential habitat was estimated with a species distribution model named Maxent. Spatial relations between vicugna’s actual distribution, its potential habitat and communal activities were analysed. Results indicate that potential habitat is determined by environmental conditions, while human presence and domestic livestock determine vicugna’s actual distribution. Based on the research process, recommendations relating vicugna’s sustainable management in the study area are given.The results are valuable to local community and conservation agents. Main contributions consist in generating a space for exchanging knowledge during the workshop, as well as the integration of analysis methods in physical and human geography. / En el Perú son necesarias investigación y prácticas que contribuyan al manejo sostenible de los recursos alto-andinos. La Reserva Paisajística Nor Yauyos Cochas alberga una población silvestre de vicuñas, que pueden ser clave para la conservación y desarrollo sostenible. Sin embargo, en la Comunidad Campesina de Tanta algunas actividades impactan negativamente en el hábitat de la vicuña. La investigación buscó determinar las relaciones espaciales e impactos de las actividades de la Comunidad Campesina de Tanta sobre el hábitat y la distribución de la vicuña en el territorio comunal. Mediante un taller de mapeo participativo se determinaron la distribución actual de las vicuñas y las actividades comunales que pueden influir sobre su hábitat. El hábitat potencial de la especie se estimó con el modelo de distribución de especies Maxent. Se analizaron las relaciones espaciales entre la distribución actual de la vicuña, su hábitat potencial y las actividades comunales. Los resultados indican que el hábitat potencial está determinado por condiciones ambientales, mientras que la distribución actual está determinada por la presencia humana y del ganado doméstico. En base al proceso de investigación se incluyen recomendaciones en relación al manejo sostenible de la vicuña en el área de estudio. Los resultados son de interés para la comunidad local y agentes de conservación. Aportes principales consisten en la generación de un espacio de intercambio de conocimientos en el taller, así como la integración de métodos de análisis en geografía física y humana.

abstract: This dissertation investigates spatial and temporal changes in land cover and plant species distributions on Cyprus in the past, present and future (1973-2070). Landsat image analysis supports inference of land cover changes following the political division of the island of Cyprus in 1974. Urban growth in Nicosia, Larnaka and Limasol, as well as increased development along the southern coastline, is clearly evident between 1973 and 2011. Forests of the Troodos and Kyrenia Ranges remain relatively stable, with transitions occurring most frequently between agricultural land covers and shrub/herbaceous land covers. Vegetation models were constructed for twenty-two plant species of Cyprus using Maxent to predict potentially suitable areas of occurrence. Modern vegetation models were constructed from presence-only data collected by field surveys conducted between 2008 and 2011. These models provide a baseline for the assessment of potential species distributions under two climate change scenarios (A1b and A2) for the years 2030, 2050, and 2070. Climate change in Cyprus is likely to influence habitat availability, particularly for high elevation species as the relatively low elevation mountain ranges and small latitudinal range prevent species from shifting to areas of suitable environmental conditions. The loss of suitable habitat for some species may allow the introduction of non-native plant species or the expansion of generalists currently excluded from these areas. Results from future projections indicate the loss of suitable areas for most species by the year 2030 under both climate regimes and all four endemic species (Cedrus brevifolia, Helianthemum obtusifolium, Pterocephalus multiflorus, and Quercus alnifolia) are predicted to lose all suitable environments as soon as 2030. As striking exceptions Prunus dulcis (almond), Ficus carica (fig), Punica granatum (pomegranate) and Olea europaea (olive), which occur as both wild varieties and orchard cultigens, will expand under both scenarios. Land cover and species distribution maps are evaluated in concert to create a more detailed interpretation of the Cypriot landscape and to discuss the potential implications of climate change for land cover and plant species distributions. / Dissertation/Thesis / Ph.D. Geography 2013

No description available.

Golden eagles (Aquila chrysaetos canadensis Linnaeus 1758) are declining in some areas throughout their Nearctic range (Sauer et al. 2011). This reduction is linked to changes in their habitat caused by human activities. Golden eagles inhabit an extensive range of environments (Watson 1997, Kochert et al. 2002). In the American Continent, the golden eagle's range encompasses Alaska, Canada, the United States and the Northern and Central portions of Mexico. Northern golden eagle populations migrate during winter to southern grounds, crossing international boundaries of Canada, the US and Mexico and therefore, their conservation is of trilateral concern. Golden eagles are protected by domestic laws in the three North American countries where they occur (FWCA 1997, BGEPA 1940, MBTA 1918, Lacey Act 1900, DOF 2002) and although the IUCN list the species as Least Concern, the A. c. canadensis subspecies has been protected by CITES since 1975 (Birdlife International 2012). While intensively studied in the United States, very little is known in Mexico about golden eagle ecology and their populations. As the national bird of Mexico, its conservation has been a priority for the Mexican government since its inclusion in the Endangered Species List in 1994 (SEDESOL 1994). Several threats jeopardize golden eagle populations throughout their range in North America: habitat alteration and fragmentation, electrocution, collisions with vehicles, collision with windmills and wires, poisoning from lead ingestion, drowning, shooting and trapping, and poaching for illegal wildlife trade. Mexican experts believe that a dramatic decline occurred over recent decades and that the remaining pairs have been restricted to remnant suitable habitat patches (SEMARNAP-INE 1999). Long-term survival of golden eagles largely depends on the effectiveness of current conservation efforts of habitat at a landscape level. Successful conservation and management requires accurate information on ecology of the species upon which decisions can be based. This study investigated habitat characteristics of the areas occupied by golden eagles and developed strategies for habitat management and protection to improve golden eagle viability in Chihuahua State. I surveyed a portion of Chihuahuan Desert Ecoregion in Mexico to locate golden eagle territories during 2009 and 2010. I located 30 golden eagle nesting territories and found similar composition of cover type, vegetation structure and prey indices between the territory cores and their buffer zones. Distance to most anthropogenic disturbance sources was similar between golden eagle sites and random areas (n=60). Grassland was the most common cover type, occurring in 100% of the nesting territories, and comprising 58% of the territories' area, suggesting a disproportionate use of this cover type compared to its overall availability (25% of the state area). I used landscape attributes such as topographic characteristics and human disturbances to model the probability of occurrence of golden eagles across the landscape. I used logistic regression to model the occurrence of golden eagles at two different landscape scales and selected the best model at a home range scale based on AIC values to develop a predictive map of golden eagle distribution in Chihuahua, Mexico. I found that at a home range scale, golden eagles' occurrence was positively related to open areas and terrain ruggedness and negatively to human settlements, while at a larger scale it was positively related to open areas and negatively related to forested areas. The results confirm that golden eagles are dependent on grasslands and rugged terrain. I developed predictive maps of golden eagle occurrence using a logistic regression and a Mahalanobis distance approach using the variables from the model chosen to compare the performance and output with logistic regression modeling. I analyzed the Mexican National Plan for Golden Eagle Recovery (PACE - Ã guila Real) and proposed a conservation strategy oriented to protect golden eagle populations and their habitat in Chihuahua, Mexico. This strategy integrates ecologic knowledge developed in the first two chapters and incorporates social participation of all stakeholders. The strategy recognizes the potential limitations of conservation implementation programs in Mexico and explores the potential opportunities to protect golden eagles populations and their habitat. / Ph. D.