Presentation Review
Over the course of this class we were tasked with reviewing research articles that utilized how GIS is applied through the fields of Landscape Ecology, Health Geography and Crime Analysis. This process has allowed us to investigate some of the real-world applications that GIS has been used to further research while getting a better sense of how to improve the approach that some of these groups have taken. Working with my partner Graham, we identified and reviewed three papers across the three fields that we discussed in class: the habitat suitability of southern Florida for invasive Burmese pythons (Landscape Ecology), Environmental Influences on Allergy Development in Children (Health Geography) and a Spatial-Temporal analysis of violent trauma crime hot spots in Vancouver (Crime Analysis). You can find our review of each of these papers below. In critically thinking and analyzing these papers we are better able to identify weaknesses and strengths in our own work.
Landscape Ecology: A review of ‘Modeling relative habitat suitability of southern Florida for invasive Burmese pythons (Pyhton molurus bivittatus) by Mustacio et al.
Burmese pythons (Python molurus bivittatus) are a successful generalist invader in the southern Florida Everglades, and a threat to native species and conservation efforts. Recently, Burmese pythons have begun to occupy areas dissimilar to their native ranges and to the Everglades. Mutascio et al.’s goal is to utilize an Ecological Niche Model (ENM), and correct for geographic sampling bias, to build knowledge on the relative habitat suitability for pythons in southern Florida, and northward of their current range. The authors believe that land-cover classes that include wet areas, as well as proximity to water bodies, will be the strongest predictors of habitat suitability for pythons, based on the background knowledge surrounding the Burmese python’s life history strategy and native ranges.
The authors sourced their data from the Early Detection and Distribution Mapping System (EDDMapS), which pools from other databases, scientific studies, and volunteer observations. They utilized 3 types of landscape variables: fine-scale land cover, home range-level land cover, and distance to open freshwater or wetland. Because of the strong geographic sampling bias in the EDDMapS database, 10 bias correction scenarios were run to determine which best captured the sampling effort from the dataset. Maximum Entropy Modelling (MaxEnt) was then utilized for the ENM, and to assess the relative significance of each of the environmental variables in explaining the output of the model.
ENM’s are often commonly used to predict potential ranges of invasive species, based on climate-matching data and their native ranges, but the authors instead used ENM to provide a clearer understanding of relative habitat suitability, which does not have the potential pitfalls that predicting a species’ range does. Therefore, their procedures are appropriate for their research.
The authors found that a Binary Bias Sampling with Data (B-SWD) bias correction best fit their data and model. Their output showed that home range-level land cover had a 63.3% contribution to the model’s output, distance to open freshwater or wetland a 24.7% contribution, and fine-scale land cover a 12.1% contribution. At the home range scale, estuarine and freshwater wetlands were the most important cover types for suitable python habitat, and that proximity to fresh water was an important predictor in python home ranges, which were consistent with their initial predictions. At finer scale resolutions, the authors also found that agricultural land and canals also become important predictors for suitability, which may potentially facilitate a northward expansion of the pythons’ range. They also found a high concurrency between their habitat suitability map and the home range-level habitat layer (77.2% concurrency for estuaries, 91.3% for freshwater wetlands).
The authors made a great deal of effort to correct for the geographic sampling bias inherent to their database, which adds to their validity of their findings. Their results regarding the most suitable habitat types for Burmese pythons make sense in relation to the current, but limited, understanding about their life history strategies. Overall, we would rate this paper a 7 out of 10. While the corrections for sampling bias they utilized could be used in a wide variety of other ecological studies, and their habitat suitability map is solid, the results they provide aren’t particularly novel, and having not conducted the sampling themselves reduced the potential for analyses, and limited the conclusions that could be drawn from their results.
References
Mutascio, H.E., Pittman, S.E., Zollner, P.A. and D’Acunto, L.E. (2017). Modeling relative habitat suitability of southern Florida for invasive Burmese pythons (Pyhton molurus bivittatus). Landscape Ecology, 1-18. DOI: 10.1007/s10980-017-0597-5.
Health Geography: A review of “A GIS-Based Assessment of Environmental Influences on Allergy Development in Children” by Lam et al.
Experiencing respiratory diseases early as children often results in lifelong breathing issues that negatively impact the health of an individual. By studying what can cause allergens during childhood, researchers can better understand how to formulate preventative strategies in childhood healthcare. Lam et al’s goal of this paper is to investigate the relative importance of indoor and outdoor environmental variables on two common allergies: asthma (heavy wheezing) and rhino-conjunctivitis (frequent sneezing, runny nose or congestion). To do so, the authors combined results taken from a national questionnaire survey with GIS spatial analyses to determine how allergies developed in children in Hong Kong.
Survey data was collected from a parent answered questionnaire conducted through the International Study of Asthma and Allergies in Childhood (ISAAC). To minimize the geographic variation in the survey data, the researchers narrowed down their study to the Yuen Long District and focused on 508 preschool children between the ages of 4 and 6. This survey yielded information on the prevalence and incidence of the two respiratory issues as well as the indoor environmental conditions the children experienced growing up. Conditions included whether or not the household had pets, exposure to moisture or mold, exposure to second hand tobacco smoke and air condition installations. Chi-squared and multivariate tests were employed to confidently assess the statistics. The research team then manually geocoded each of the houses where children were identified as showing symptoms of asthma or rhino- conjunctivitis, that were mapped into the 29 District Council Constituency Areas (DCAA). GIS
was used to place them into four categories of exposure to outdoor pollutants by using the distance to the nearest major traffic roads (roads with an average of 30,000 daily cars) as a proxy.
The results of their statistical analysis indicated that second hand smoke and mold exposure had the greatest influence on allergy development in children. Mold exposure led to 2.5 times more cases of asthma and twice as many cases of rhino-conjunctivitis when compared to children who were not exposed to any mold. While their GIS-based results yielded a weak relationship between exposure to pollutants and allergy development, Lam et al. indicated that their findings may be limited due to the fact that only 0.79% of the participants were living within close proximity to the high traffic areas. They also cautioned that this metric may be flawed as means of transportation to and from school and high study wide air condition rates in households likely limited the effects of outdoor air pollutants.
Overall, we gave Lam et al.’s paper a score of 7 out of 10. We believe that while the researchers did an excellent job of combining two different approaches, acknowledging the limitations in their findings and structuring their paper in an easy to follow way; the paper can be improved upon to create a stronger case. Their findings would have benefited from highlighting the necessity of separating the allergens into two categories (prevalence and incidence) and they could have also taken their spatial analysis further by incorporating things like population density, income levels and proximity to major industrial facilities.
References
Lam, A., Wong, G. W. K., Poon, C. M. P., and Lee, S. S. (2014). A GIS-Based Assessment of Environmental Influences on Allergy Development in Children. Asia-Pacfic Journal of Public Health. 26:6, 575-587. DOI: 10.1177/1010539511428488
Crime Analysis: A Review of ‘A GIS-based spatiotemporal analysis of violent trauma hotspots in Vancouver, Canada: identification, contextualisation and intervention’ by Walker, Schuurman and Hameed.
Interpersonal violence has enormous economic, societal, and health costs for governments and individuals involved with its prevalence. The goal of this study was to examine the relationship between a variety of socioeconomic variables, the prevalence of violent injury, and how incidents of violent crime are associated, both spatially and temporally, in Vancouver BC. Prior research has how strong correlations with injuries within space-time; the authors attempted to build on this work by incorporating more variables associated with the crime itself, as well as a measure of localized social deprivation, to attempt to contribute to injury prevention and law enforcement in the city.
To conduct their analyses, the authors collected metro Vancouver injury data from 2001 to 2008, excluded self-inflicted, lethal, or less severe injuries for a total of 575 incidents, and classified these injuries under 7 different mechanisms. They first utilized a Pearson’s chi-squared test to determine the independence of the variables with day or night time occurrences, imported the dataset into GIS software, and conducted a kernel density estimation to produce hot spot maps, then contrasted these maps between male and female victims. The authors then used a space-time permutation model within a spatial scan statistic to identify spatial-temporal hotspots within their dataset. Finally, the authors integrated a variable called the Vancouver Area Neighbourhood Deprivation Index (VANDIX), created by Bell et al. in 2007 by incorporating census data and health expert consultations. An index score was calculated for each census tract, classified into quintiles, and incorporated into GIS to identify the proportion of violent injuries within each quintile. The methods they utilized for their analysis seemed appropriate for their research goals, though separating time of day into more components, such as dawn and dusk, may have added more breadth to their findings.
The researchers concluded that the incidence of violent injuries is spatially and temporally clustered. The results of the hot spot analysis indicated that these incidents were often located in low income, high density residences with the Downtown Eastside neighborhood possessing the greatest risk. Day and night cycles significantly impacted the occurrence of incidents and the mechanisms used. While a large proportion of victims were male, both genders experienced most of their reported violent injuries at night and a shift towards more severe mechanisms of injury was also noticed. The researchers believe that a peak in violence around 2am corresponds to the closing time of bars when there was a higher encounter rate between intoxicated individuals on the streets.. Gunshots, which accounted for a third of all cases, increased by 358% at night. Walker and his team also highlighted the Downtown Eastside, the Night Club district, Commercial and the Stadium Arena as major night time hot spots. With the VANDIX variable, 77% of all the incidents were found within neighborhoods in the two most deprived quantiles. The nightclub district and Metrotown were the only clusters that did not experience demographic deprivation. While the nightclub district can be explained by the aforementioned heavy concentration of alcohol consumption, Metrotown represents a unique case. Despite a low concentration of alcohol serving facilities and above average residents, built environment factors like being a busy shopping center and a transit node have driven up the rate of violent injuries.
The researchers go on to mention preventative and proactive techniques to dissuade further acts of violence. These discussions include using cautionary posters in night club washrooms, pricing liquor licenses along a range of closing times to spread out traffic throughout the night and an emphasis on police officers building community relationships. Overall, we gave this paper an 8 out of 10. We felt that the researchers performed a robust analysis, considered all of the possible factors impacting their findings, and presented their results clearly. The cartography and visualized graphs were also straightforward and visually appealing. One consideration would be to expand the length of their study beyond the seven-year period.