Steven Le Comber
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Steven Le Comber has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
- Sally C. Faulkner
School of Biological and Chemical Sciences Queen Mary University of London London E1 4NS UK
7 shared publications D. Kim Rossmo
6 shared publicationsMark D. Stevenson
5 shared publications- Robert Verity
MRC Centre for Global Infectious Disease Analysis, Imperial College London, London, UK
5 shared publications - Michael C. A. Stevens
School of Biological and Chemical Sciences Queen Mary University of London London E1 4NS UK
4 shared publications - Richard Alan Nichols
Queen Mary University of London
3 shared publications
Latest Publications
- Open Access
- 22 March 2021
- Published by Wiley in Ecological Applications
- Vol. 31 (5) , e2329
Spatially clustered count data provide more efficient search strategies in invasion biology and disease control
Geographic profiling, a mathematical model originally developed in criminology, is increasingly being used in ecology and epidemiology. Geographic profiling boasts a wide range of applications, such as finding source populations of invasive species or breeding sites of vectors of infectious disease. The model provides a cost‐effective approach for prioritizing search strategies for source locations and does so via simple data in the form of the positions of each observation, such as individual sightings of invasive species or cases of a disease. In doing so, however, classic geographic profiling approaches fail to make the distinction between those areas containing observed absences and those areas where no data were recorded. Absence data are generated via spatial sampling protocols but are often discarded during the inference process. Here we construct a geographic profiling model that resolves these issues by making inferences via count data, analyzing a set of discrete sentinel locations at which the number of encounters has been recorded. Crucially, in our model this number can be zero. We verify the ability of this new model to estimate source locations and other parameters of practical interest via a Bayesian power analysis. We also measure model performance via real‐world data in which the model infers breeding locations of mosquitoes in bromeliads in Miami‐Dade County, Florida, USA. In both cases, our novel model produces more efficient search strategies by shifting focus from those areas containing observed absences to those with no data, an improvement over existing models that treat these areas equally. Our model makes important improvements upon classic geographic profiling methods, which will significantly enhance real‐world efforts to develop conservation management plans and targeted interventions.
- 22 March 2021
- Published by Wiley in Ecological Applications
- Vol. 31 (5) , e2329
Open AccessSpatially clustered count data provide more efficient search strategies in invasion biology and disease control
Geographic profiling, a mathematical model originally developed in criminology, is increasingly being used in ecology and epidemiology. Geographic profiling boasts a wide range of applications, such as finding source populations of invasive species or breeding sites of vectors of infectious disease. The model provides a cost‐effective approach for prioritizing search strategies for source locations and does so via simple data in the form of the positions of each observation, such as individual sightings of invasive species or cases of a disease. In doing so, however, classic geographic profiling approaches fail to make the distinction between those areas containing observed absences and those areas where no data were recorded. Absence data are generated via spatial sampling protocols but are often discarded during the inference process. Here we construct a geographic profiling model that resolves these issues by making inferences via count data, analyzing a set of discrete sentinel locations at which the number of encounters has been recorded. Crucially, in our model this number can be zero. We verify the ability of this new model to estimate source locations and other parameters of practical interest via a Bayesian power analysis. We also measure model performance via real‐world data in which the model infers breeding locations of mosquitoes in bromeliads in Miami‐Dade County, Florida, USA. In both cases, our novel model produces more efficient search strategies by shifting focus from those areas containing observed absences to those with no data, an improvement over existing models that treat these areas equally. Our model makes important improvements upon classic geographic profiling methods, which will significantly enhance real‐world efforts to develop conservation management plans and targeted interventions.
- Open Access
Assessing the impact of taxon resolution on network structure
Constructing ecological networks has become an indispensable approach in understanding how different taxa interact. However, the methods used to generate data in network research vary widely among studies, potentially limiting our ability to compare results meaningfully. In particular, methods of classifying nodes vary in their precision, likely altering the architecture of the network studied. For example, rather than being classified as Linnaean species, taxa are regularly assigned to morphospecies in observational studies, or to molecular operational taxonomic units (MOTUs) in molecular studies, with the latter defined based on an arbitrary threshold of sequence similarity. Although the use of MOTUs in ecological networks holds great potential, especially for allowing rapid construction of large data sets of interactions, it is unclear how the choice of clustering threshold can influence the conclusions obtained. To test the impact of taxonomic precision on network architecture, we obtained and analyzed 16 data sets of ecological interactions, inferred from metabarcoding and observations. Our comparisons of networks constructed under a range of sequence thresholds for assigning taxa demonstrate that even small changes in node resolution can cause wide variation in almost all key metric values. Moreover, relative values of commonly used metrics such as robustness were seen to fluctuate continuously with node resolution, thereby potentially causing error in conclusions drawn when comparing multiple networks. In observational networks, we found that changing node resolution could, in some cases, lead to substantial changes to measurements of network topology. Overall, our findings highlight the importance of classifying nodes to the greatest precision possible, and demonstrate the need for caution when comparing networks that differ with respect to node resolution, even where taxonomic groups and interaction types are similar. In such cases, we recommend that comparisons of networks should focus on relative differences rather than absolute values between the networks studied.
Open AccessAssessing the impact of taxon resolution on network structure
Constructing ecological networks has become an indispensable approach in understanding how different taxa interact. However, the methods used to generate data in network research vary widely among studies, potentially limiting our ability to compare results meaningfully. In particular, methods of classifying nodes vary in their precision, likely altering the architecture of the network studied. For example, rather than being classified as Linnaean species, taxa are regularly assigned to morphospecies in observational studies, or to molecular operational taxonomic units (MOTUs) in molecular studies, with the latter defined based on an arbitrary threshold of sequence similarity. Although the use of MOTUs in ecological networks holds great potential, especially for allowing rapid construction of large data sets of interactions, it is unclear how the choice of clustering threshold can influence the conclusions obtained. To test the impact of taxonomic precision on network architecture, we obtained and analyzed 16 data sets of ecological interactions, inferred from metabarcoding and observations. Our comparisons of networks constructed under a range of sequence thresholds for assigning taxa demonstrate that even small changes in node resolution can cause wide variation in almost all key metric values. Moreover, relative values of commonly used metrics such as robustness were seen to fluctuate continuously with node resolution, thereby potentially causing error in conclusions drawn when comparing multiple networks. In observational networks, we found that changing node resolution could, in some cases, lead to substantial changes to measurements of network topology. Overall, our findings highlight the importance of classifying nodes to the greatest precision possible, and demonstrate the need for caution when comparing networks that differ with respect to node resolution, even where taxonomic groups and interaction types are similar. In such cases, we recommend that comparisons of networks should focus on relative differences rather than absolute values between the networks studied.
- Open Access
- 12 June 2020
- Published by Taylor & Francis in The Professional Geographer
- Vol. 72 (4) , 566-574
Investigating Sherlock Holmes: Using Geographic Profiling to Analyze the Novels of Arthur Conan Doyle
Central to spatial analysis of crime is the assumption that the offender has visited the locations analyzed. We show that even if this is incorrect, meaningful patterns can be identified. Using 390 locations from the Sherlock Holmes stories, we show that geographic profiling (GP) ranks Conan Doyle in the top 13 percent of 2,678 historical figures in London, despite the fact at the time of writing he had not visited all of the sites. Restricting the analysis to thirty authors contemporary with the Sherlock Holmes stories, Conan Doyle ranks first, with a hit score of 2.8 percent (above Holmes’s address at 221b Baker Street). Finally, we show that GP prioritizes sites strongly associated with Conan Doyle (example.g., his home) compared to those more tangentially associated with him, even when sites are close together. Our analysis, although mostly for amusement, underlines the ability of GP to extract useful information from complex data.
- 12 June 2020
- Published by Taylor & Francis in The Professional Geographer
- Vol. 72 (4) , 566-574
Open AccessInvestigating Sherlock Holmes: Using Geographic Profiling to Analyze the Novels of Arthur Conan Doyle
Central to spatial analysis of crime is the assumption that the offender has visited the locations analyzed. We show that even if this is incorrect, meaningful patterns can be identified. Using 390 locations from the Sherlock Holmes stories, we show that geographic profiling (GP) ranks Conan Doyle in the top 13 percent of 2,678 historical figures in London, despite the fact at the time of writing he had not visited all of the sites. Restricting the analysis to thirty authors contemporary with the Sherlock Holmes stories, Conan Doyle ranks first, with a hit score of 2.8 percent (above Holmes’s address at 221b Baker Street). Finally, we show that GP prioritizes sites strongly associated with Conan Doyle (example.g., his home) compared to those more tangentially associated with him, even when sites are close together. Our analysis, although mostly for amusement, underlines the ability of GP to extract useful information from complex data.
- Open Access
- 11 December 2019
- Published by Wiley in Journal of Zoology
- Vol. 312 (1) , 1-11
Understanding the origins of the ring‐necked parakeet in the UK
- 11 December 2019
- Published by Wiley in Journal of Zoology
- Vol. 312 (1) , 1-11
Open AccessUnderstanding the origins of the ring‐necked parakeet in the UK
- 22 July 2019
- Published by Taylor & Francis in Journal of Spatial Science
- Vol. 66 (2) , 183-194
Double cross: geographic profiling of V-2 impact sites
Between September 1944 and March 1945, more than a thousand V-2 rockets struck Great Britain, causing 2,754 deaths. M.I.5 responded with an ingenious scheme to mislead the Germans so they would shorten the range, missing Westminster and the docks. This is widely regarded as a success, with claims that the M.P.I. moved eastwards by two miles a week, saving 1,300 lives. Here, we use geographic profiling to show that this deception had little or no effect. Our results suggest rockets aimed at a single target form clusters of 30 km radius, swamping the claimed movement.
- 22 July 2019
- Published by Taylor & Francis in Journal of Spatial Science
- Vol. 66 (2) , 183-194
Double cross: geographic profiling of V-2 impact sites
Between September 1944 and March 1945, more than a thousand V-2 rockets struck Great Britain, causing 2,754 deaths. M.I.5 responded with an ingenious scheme to mislead the Germans so they would shorten the range, missing Westminster and the docks. This is widely regarded as a success, with claims that the M.P.I. moved eastwards by two miles a week, saving 1,300 lives. Here, we use geographic profiling to show that this deception had little or no effect. Our results suggest rockets aimed at a single target form clusters of 30 km radius, swamping the claimed movement.
- Open Access
- 27 April 2019
- Published by Springer Nature in BMC Plant Biology
- Vol. 19 (1) , 1-19
Early consequences of allopolyploidy alter floral evolution in Nicotiana (Solanaceae)
Polyploidy has played a major role in angiosperm evolution. Previous studies have examined polyploid phenotypes in comparison to their extant progenitors, but not in context of predicted progenitor phenotypes at allopolyploid origin. In addition, differences in the trends of polyploid versus diploid evolution have not been investigated. We use ancestral character-state reconstructions to estimate progenitor phenotype at allopolyploid origin to determine patterns of polyploid evolution leading to morphology of the extant species. We also compare trends in diploid versus allopolyploid evolution to determine if polyploidy modifies floral evolutionary patterns. Predicting the ancestral phenotype of a nascent allopolyploid from reconstructions of diploid phenotypes at the time of polyploid formation generates different phenotype predictions than when extant diploid phenotypes are used, the outcome of which can alter conclusions about polyploid evolution; however, most analyses yield the same results. Using ancestral reconstructions of diploid floral phenotypes indicate that young polyploids evolve shorter, wider corolla tubes, but older polyploids and diploids do not show any detectable evolutionary trends. Lability of the traits examined (floral shape, corolla tube length, and corolla tube width) differs across young and older polyploids and diploids. Corolla length is more evolutionarily labile in older polyploids and diploids. Polyploids do not display unique suites of floral characters based on both morphological and color traits, but some suites of characters may be evolving together and seem to have arisen multiple times within Nicotiana, perhaps due to the influence of pollinators. Young polyploids display different trends in floral evolution (shorter, wider corolla tubes, which may result in more generalist pollination) than older polyploids and diploids, suggesting that patterns of divergence are impacted by the early consequences of allopolyploidy, perhaps arising from genomic shock and/or subsequent genome stabilization associated with diploidization. Convergent evolution in floral morphology and color in Nicotiana can be consistent with pollinator preferences, suggesting that pollinators may have shaped floral evolution in Nicotiana.
- 27 April 2019
- Published by Springer Nature in BMC Plant Biology
- Vol. 19 (1) , 1-19
Open AccessEarly consequences of allopolyploidy alter floral evolution in Nicotiana (Solanaceae)
Polyploidy has played a major role in angiosperm evolution. Previous studies have examined polyploid phenotypes in comparison to their extant progenitors, but not in context of predicted progenitor phenotypes at allopolyploid origin. In addition, differences in the trends of polyploid versus diploid evolution have not been investigated. We use ancestral character-state reconstructions to estimate progenitor phenotype at allopolyploid origin to determine patterns of polyploid evolution leading to morphology of the extant species. We also compare trends in diploid versus allopolyploid evolution to determine if polyploidy modifies floral evolutionary patterns. Predicting the ancestral phenotype of a nascent allopolyploid from reconstructions of diploid phenotypes at the time of polyploid formation generates different phenotype predictions than when extant diploid phenotypes are used, the outcome of which can alter conclusions about polyploid evolution; however, most analyses yield the same results. Using ancestral reconstructions of diploid floral phenotypes indicate that young polyploids evolve shorter, wider corolla tubes, but older polyploids and diploids do not show any detectable evolutionary trends. Lability of the traits examined (floral shape, corolla tube length, and corolla tube width) differs across young and older polyploids and diploids. Corolla length is more evolutionarily labile in older polyploids and diploids. Polyploids do not display unique suites of floral characters based on both morphological and color traits, but some suites of characters may be evolving together and seem to have arisen multiple times within Nicotiana, perhaps due to the influence of pollinators. Young polyploids display different trends in floral evolution (shorter, wider corolla tubes, which may result in more generalist pollination) than older polyploids and diploids, suggesting that patterns of divergence are impacted by the early consequences of allopolyploidy, perhaps arising from genomic shock and/or subsequent genome stabilization associated with diploidization. Convergent evolution in floral morphology and color in Nicotiana can be consistent with pollinator preferences, suggesting that pollinators may have shaped floral evolution in Nicotiana.
- Open Access
Assessing the impact of taxon resolution on network structure, with implication for comparative ecology
Constructing networks has become an indispensable approach in understanding how different taxa interact. However, methodologies vary widely among studies, potentially limiting our ability to meaningfully compare results. In particular, how network architecture is influenced by the extent to which nodes are resolved to either taxa or taxonomic units is poorly understood. To address this, here we collate nine datasets of ecological interactions, from both observations and DNA metabarcoding, and construct networks under a range of commonly-used node resolutions. We demonstrate that small changes in node resolution can cause wide variation in almost all key metric values, including robustness and nestedness. Moreover, relative values of metrics such as robustness were seen to fluctuate continuously with node resolution, thereby potentially confounding comparisons of networks, as well as interpretations concerning their constituent ecological interactions. These findings highlight the need for care when comparing networks, especially where these differ with respect to node resolution. Statement of authorship: DRHB, SJR and ELC conceived of the project, DRHB facilitated fieldwork in Malaysia, DRHB, HFMO and ELC undertook field collections, DRHB, SCLC and HFMO analysed the data, and DRHB wrote the manuscript with input from all authors. Data Accessibility Statement: All data used in this analysis will be archived in Dryad and made available by DOI. Specific analysis scripts are available on GitHub with links given in the manuscript.
Open AccessAssessing the impact of taxon resolution on network structure, with implication for comparative ecology
Constructing networks has become an indispensable approach in understanding how different taxa interact. However, methodologies vary widely among studies, potentially limiting our ability to meaningfully compare results. In particular, how network architecture is influenced by the extent to which nodes are resolved to either taxa or taxonomic units is poorly understood. To address this, here we collate nine datasets of ecological interactions, from both observations and DNA metabarcoding, and construct networks under a range of commonly-used node resolutions. We demonstrate that small changes in node resolution can cause wide variation in almost all key metric values, including robustness and nestedness. Moreover, relative values of metrics such as robustness were seen to fluctuate continuously with node resolution, thereby potentially confounding comparisons of networks, as well as interpretations concerning their constituent ecological interactions. These findings highlight the need for care when comparing networks, especially where these differ with respect to node resolution. Statement of authorship: DRHB, SJR and ELC conceived of the project, DRHB facilitated fieldwork in Malaysia, DRHB, HFMO and ELC undertook field collections, DRHB, SCLC and HFMO analysed the data, and DRHB wrote the manuscript with input from all authors. Data Accessibility Statement: All data used in this analysis will be archived in Dryad and made available by DOI. Specific analysis scripts are available on GitHub with links given in the manuscript.
- 6 March 2018
- Published by Wiley in Conservation Biology
- Vol. 32 (3) , 685-693
A spatial approach to combatting wildlife crime
Poaching can have devastating impacts on animal and plant numbers, and in many countries has reached crisis levels, with illegal hunters employing increasingly sophisticated techniques. We used data from an 8‐year study in Savé Valley Conservancy, Zimbabwe, to show how geographic profiling—a mathematical technique originally developed in criminology and recently applied to animal foraging and epidemiology—can be adapted for use in investigations of wildlife crime. The data set contained information on over 10,000 incidents of illegal hunting and the deaths of 6,454 wild animals. We used a subset of data for which the illegal hunters’ identities were known. Our model identified the illegal hunters’ home villages based on the spatial locations of the hunting incidences (e.g., snares). Identification of the villages was improved by manipulating the probability surface inside the conservancy to reflect the fact that although the illegal hunters mostly live outside the conservancy, the majority of hunting occurs inside the conservancy (in criminology terms, commuter crime ). These results combined with rigorous simulations showed for the first time how geographic profiling can be combined with GIS data and applied to situations with more complex spatial patterns, for example, where landscape heterogeneity means some parts of the study area are less likely to be used (e.g., aquatic areas for terrestrial animals) or where landscape permeability differs (e.g., forest bats tend not to fly over open areas). More broadly, these results show how geographic profiling can be used to target antipoaching interventions more effectively and more efficiently and to develop management strategies and conservation plans in a range of conservation scenarios.
- 6 March 2018
- Published by Wiley in Conservation Biology
- Vol. 32 (3) , 685-693
A spatial approach to combatting wildlife crime
Poaching can have devastating impacts on animal and plant numbers, and in many countries has reached crisis levels, with illegal hunters employing increasingly sophisticated techniques. We used data from an 8‐year study in Savé Valley Conservancy, Zimbabwe, to show how geographic profiling—a mathematical technique originally developed in criminology and recently applied to animal foraging and epidemiology—can be adapted for use in investigations of wildlife crime. The data set contained information on over 10,000 incidents of illegal hunting and the deaths of 6,454 wild animals. We used a subset of data for which the illegal hunters’ identities were known. Our model identified the illegal hunters’ home villages based on the spatial locations of the hunting incidences (e.g., snares). Identification of the villages was improved by manipulating the probability surface inside the conservancy to reflect the fact that although the illegal hunters mostly live outside the conservancy, the majority of hunting occurs inside the conservancy (in criminology terms, commuter crime ). These results combined with rigorous simulations showed for the first time how geographic profiling can be combined with GIS data and applied to situations with more complex spatial patterns, for example, where landscape heterogeneity means some parts of the study area are less likely to be used (e.g., aquatic areas for terrestrial animals) or where landscape permeability differs (e.g., forest bats tend not to fly over open areas). More broadly, these results show how geographic profiling can be used to target antipoaching interventions more effectively and more efficiently and to develop management strategies and conservation plans in a range of conservation scenarios.
- 1 March 2017
- Published by Elsevier in Ecological Informatics
- Vol. 38, 76-81
The use of jackknifing for the evaluation of geographic profiling reliability
- 1 March 2017
- Published by Elsevier in Ecological Informatics
- Vol. 38, 76-81
The use of jackknifing for the evaluation of geographic profiling reliability
- 28 October 2016
- Published by Wiley in Diversity and Distributions
- Vol. 23 (1) , 104-112
Using geographic profiling to compare the value of sightings vs trap data in a biological invasion
Aim: The development of conservation plans, including those dealing with invasive species, is underpinned by the need to obtain reliable and accurate data. However, in many cases responding rapidly is equally critical.Location: The data were obtained from the Hebridean Mink Project, which was set up with the objective of removing mink from North Uist, Benbecula and South Uist.Methods: Here, we introduce an extension of the Dirichlet process mixture (DPM) model of geographic profiling that can be used to estimate source locations of invasions directly from spatial point pattern data without the need to specify dispersal parameters. We use this model to analyse a biological invasion of American mink (Neovison vison) in the Hebrides.Results: Our results suggest that sightings data – which are relatively easy and quick to acquire – can be used to capture much of the information about sources of invasive species that is obtained from the harder to acquire and more intensive trap data.Main conclusion: These results have important implications for the development of conservation plans and, in this case, in the early stages of biological invasions, when interventions are most likely to be successful.
- 28 October 2016
- Published by Wiley in Diversity and Distributions
- Vol. 23 (1) , 104-112
Using geographic profiling to compare the value of sightings vs trap data in a biological invasion
Aim: The development of conservation plans, including those dealing with invasive species, is underpinned by the need to obtain reliable and accurate data. However, in many cases responding rapidly is equally critical.Location: The data were obtained from the Hebridean Mink Project, which was set up with the objective of removing mink from North Uist, Benbecula and South Uist.Methods: Here, we introduce an extension of the Dirichlet process mixture (DPM) model of geographic profiling that can be used to estimate source locations of invasions directly from spatial point pattern data without the need to specify dispersal parameters. We use this model to analyse a biological invasion of American mink (Neovison vison) in the Hebrides.Results: Our results suggest that sightings data – which are relatively easy and quick to acquire – can be used to capture much of the information about sources of invasive species that is obtained from the harder to acquire and more intensive trap data.Main conclusion: These results have important implications for the development of conservation plans and, in this case, in the early stages of biological invasions, when interventions are most likely to be successful.