Landscape Ecology
Health Geography
Crime Analysis
Landscape Ecology (link to presentation slides)
The bobcat is a wide-ranging carnivore in North America, occupies relatively large home ranges, and exhibits low reproductive rates, territoriality, and male-biased dispersal. They move long distances within their home ranges and travel mostly by night, with males traveling farther and faster than females. Use a variety of land cover types for travel, including stream valleys and associated ridge lines, thick under-stories and rocky ledges
Building on recent advances in field movement ecology, the goal of the authors was to determine the characteristics of bobcat movement, habitat selection, and the environmental covariants influencing these characteristics. In determining the characteristics that influence the bobcat’s movement and habitat selection, movement corridors for this species can be identified and then used for conservation and planning purposes.
This study was done in Vermont. To collect bobcat data, 41 bobcat individuals were captured and fitted with GPS collars between 2005-2007. They were captured either by padded leg trap or in cages depending on season and probability of human interaction. The GPS collars captured temporal and spatial data in x,y location points. Data was screened to only include trajectories with consecutive location points less than 20-25 minutes apart which resulted in 13 of the original 41 bobcat individuals being suitable for movement analysis. The short time interval reduced the risk that trajectories recreated from discrete point data differed from the true path of the individual. The x,y location data was used to created a point feature class for each bobcat in ArcGIS and Hawth’s Tools analysis package was used to create continuous movement paths from the location data. Movement data was also used to calculate step length and speed of travel.
Land cover types were categorized based on the National Land Cover Database (NLCD) then the 16 land cover types present in the study area were reclassified into five classes: forest, wetland, developed, agriculture, and scrub-rock. To test habitat preference between the five land cover types, compositional analysis was used. Compositional analysis was also used to evaluate the effect of topography on bobcat movement. To do this, the composition of land cover types in a given space, be it used or available, summed to 1.0. Using methods by Aebischer et al. (1993) log-ratio trans-formations were used, which ensures linear independence of the proportions and allows the use of multivariate compositional analysis to determine preference for one landcover type over another. This was then used to create a “difference matrix”. The authors tested for selection in the form of slope, forest edge, forest core, and wetland edge vegetation (variables that are not proportions). Edge and core habitats were determined using 30 m pixel landcover raster while road density was calculated based on orthophotos using ArcGIS. To compare potential corridor designs between two state management areas within the bobcat study area, least cost path analysis and circuit theory analysis were used by the authors.
I believe the methodological approach and procedures used were appropriate for the research question. While the sample size used in this study was small, the authors found that the results were consistent with other studies of bobcats. Due to the nature of GPS collars and the limitations caused by collar bias (i.e., collars cannot receive a GPS signal in some landscape conditions so those locations are not included), the approach of this paper was appropriate and I do not think that there would be another feasible way to obtain the same data without disturbing the bobcats or landscape.
Bobcats showed strong preferences for forest and scrub and rock land cover types while favouring forest edge, wetland edge, and areas with high stream density during movement. Bobcat movement through least-preferred habitat types was faster while movement in preferred habitat types was slower which indicated that facilitating movement and landscape connectivity within disturbed landscapes is important. Based on this data, the authors call for a shift in corridor design from designs based on expert opinions to designs based on empirical data and the use of pixel resistance. By designing corridors in this manner, the resulting corridor will be more reflective of the focus species true behaviour and this will also work to make the process of corridor design more object and transparent to all stakeholders involved.
Paper rating: 8/10
Abouelezz, H. G., Donovan, T. M., Mickey, R. M., Murdoch, J. D., Freemand, M., & Royar, K. (2018). Landscape composition mediates movement and habitat selection in bobcats (Lynx rufus): implications for conservation planning. Landscape Ecology, 33, 1301-1318. https://doi.org/10.1007/s10980-018-0654-8
Health Geography (link to presentation slides)
In this paper, authors Ngui and Vanasse assess the accessibility of mental health services in southwest Montreal. The paper focuses on the potential (i.e., the probable utilization of services) spatial accessibility of mental health services and on identifying areas that are underserved or at risk of being underserved (Ngui & Vanasse, 2012). Two research questions are asked by the authors: 1) “[w]hat is the current spatial distribution of mental health care facilities in the southwest of Montreal,” and 2) “[w]hich health care territories of the southwest that do not possess enough mental health care facilities are appropriate candidates (sic) location for new mental health services?” (Ngui & Vanasse, 2012, p. 196).
To conduct their assessment, the authors use a two-step floating catchment area (2SFCA) method. The 2SFCA method was chosen because other methods tend to inflate accessibility scores in poor-access areas and as these areas are usually of the most interest to health administrators and policy makers, calculating accurate scores is important (Ngui & Vanasse, 2012). Additionally, using only a single-step floating catchment area method only accounts for the supply available (i.e., the number of mental health services) while ignoring demand (Ngui & Vanasse, 2012). To conduct the 2SFCA method, three parameters are needed: demand, supply, and the computation of accessibility measures (Ngui & Vanasse, 2012). The 2SFCA calculates a ratio of suppliers to residents within a service area centred on a supplier’s location then sums the ratios for residents who live in areas where services overlap (Ngui & Vanasse, 2012).
The results of the assessment indicate that mental health care services are globally spatially clustered with services concentrated in a southwest to northeast direction with highest densities present in three of the five health territories (Ngui & Vanasse, 2012). Mental health facilities are accessible by 90.5% of the dissemination areas but as distances increase from 1 km to 3 km, the number of services available per 10,000 persons decreases from 10.9 to 1.4 (Ngui & Vanasse, 2012). The authors highlight that this decrease occurs because the 2SFCA method accounts for potential users (population) and the number of services so that as the catchment area increases (from 1 km to 3 km), the number of potential users also increases which causes the number of available services per 10,000 persons to decrease (Ngui & Vanasse, 2012). The results of this assessment allow health administrators to identify specific places with mental health care facility shortages where accessibility improvements are needed (Ngui & Vanasse, 2012). The results also indicate a correlation between high accessibility scores of mental health services and proximity to Montreal’s central business district which follows trends in other North American cities where doctors and health services tend to cluster in the downtown and surrounding areas (Ngui & Vanasse, 2012).
The limitations of the article addressed by the authors include aggregation-error, the reliability of geocoding, and edge effects. As this paper only considers potential spatial accessibility, the authors acknowledge that residents living in areas with high scores may face aspatial barriers to accessing mental health care (Ngui & Vanasse, 2012). Areas needing further study are identified as how the variation in accessibility corresponds to socioeconomic characteristics of the population, implications of access to a personal vehicle versus public transit, and comparisons of service accessibility over time (Ngui & Vanasse, 2012).
Overall, this article accomplishes its goal and does so in a clear, concise, and easily understood manner. The limitations of the data and results are clearly explained and they do not detract from the results of the study. Provided similar data was available in other provinces or cities, this analysis could be carried out for other Canadian cities to better understand accessibility of mental health care services.
Paper rating: 8.5/10
Ngui, A. N., & Vanasse, A. (2012). Assessing spatial accessibility to mental health facilities in an urban environment. Spatial and Spatio-temporal Epidemiology, 3, 195-203. doi:10.1016/j.sste.2011.11.001
Crime Analysis (link to presentation slides)
In this paper, “Proximity to Alcohol-Serving Establishments and Crime Probabilities in Savannah, Georgia: A Statistical and GIS Analysis,” authors Kumar and Waylor explore a link between probability of crime and locational proximity to alcohol-service locations. This paper presents in-depth analyses of the spatial distribution of crime and crime densities and probabilities in relation to their proximity to alcohol-serving establishments. The authors specifically hypothesize that crime probability and density is an inverse function of distance from alcohol-service sites (Kumar & Waylor, 2003, p. 134).
To familiarize readers with this topic, the authors provide background research and a literature review. A link between alcohol consumption and violence is made clear by the authors through this and the existence of “crime hot spots” around alcohol-serving and selling sites (Kumar & Waylor, 2003). A longitudinal study over 38 years found that assault rates are “significantly related” to alcohol consumption in bars and restaurants while another study found that five of its top ten crime “hot spots” were clustered around alcohol-serving establishments (Kumar & Waylor, 2003).
Savannah, Georgia was selected for this study because of the availability of geocoded crime data (Kumar & Waylor, 2003). Crime data was obtained from the Savannah Police Department while road network data was obtained from the Georgia GIS Clearing House (Kumar & Waylor, 2003). The authors used five methods in their analysis: random point generation, logistic regression, crime classification, crime density, and crime probability determination although the first two were the main methods of analysis (Kumar & Waylor, 2003). The random point generation was done using Grid Module in ArcGIS 8.1 through several steps. First an image of Savannah was generated with random numbers ranging from 0 to 1 distributed randomly across the built-up area (as all crimes occurred in built-up areas) (Kumar & Waylor, 2003). Next, the total number of pixels for the built-up area was counted and the random number cut-off value (RNCOV) was calculated by dividing the actual number of crimes by the total number of pixels to produce a number of pixels approximately equal to the total number of crimes (Kumar & Waylor, 2003). Finally, the centroid coordinates of these random pixels were used to create the random points vector layer that was used to calculate the probabilities of crime (Kumar & Waylor, 2003). Logistic regression was used to assess crime probabilities in relation to the distance from the crime instances and random points to the nearest alcohol-serving establishment (Kumar & Waylor, 2003). The use of logistic regression allowed the authors to investigate the relationship between a dependent, binary variable (crime and random location together) and an independent variable (distance from random locations and crime incidences to the closest alcohol-serving establishment) (Kumar & Waylor, 2003). The results of these analyses show a strong correlation between distance from alcohol-serving establishments and incidences of crime as the number of crimes per square mile decreases with distance from alcohol-service sites as does the probability of crime (Kumar & Waylor, 2003). The regression coefficients for all crimes are negative and statistically significant at 0.001 which further reinforces the authors’ initial hypothesis of crime probability and density being an inverse function of distance from alcohol services (Kumar & Waylor, 2003).
Overall, this paper clearly set out its goals and hypothesis. The background research and literature review provided context for why this analysis was needed. The limitations are clearly reported and do not detract from the results while potential areas for future study and application of the author’s methods are practical and feasible. The paper’s use of tables and maps allows for quick interpretation of results without overwhelming the reader.
Rating: 9/10
Kumar, N. & Taylor, C. R. M. (2003). Proximity to Alcohol-Serving Establishments and Crime Probabilities in Savannah, Georgia: A Statistical and GIS Analysis. Southeastern Geographer, 43(1), 125-141. https://doi.org/10.1353/sgo.2003.0015