I am a post-doc based in Andy Purvis’ lab. I work on the PREDICTS (Projecting Responses of Ecological Diversity In Changing Terrestrial Systems) project: a NERC-funded collaboration to investigate how biodiversity in terrestrial communities responds to human pressures.
Before joining PREDICTS I completed a PhD in and an MSc both at Imperial College London. Between 1995 to 2008 I worked as a software engineer.
2009 – 2012 PhD Computational Ecology, Imperial College London, supervised by Dan Reuman and by Rich Williams and Lucas Joppa at Microsoft Research, Cambridge.
MSc in Ecology, Evolution & Conservation, Imperial College London.
I have taught modules on several of Imperial College's ecology MSc courses, covering version control, programming in R and in Python and creating pipelines for reproducible science.
Within the PREDICTS projects we use meta-analytical techniques to model biodiversity responses at a global scale using data collected at local levels. Our results will provide indicators of responses both at a global scale and at smaller scales, such as within a nation or for a particular biome. In addition to increasing understanding of the patterns of decline seen so far, PREDICTS will produce quantitative projections under different IPCC storylines with the aim of providing inputs to conservation policy such as the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services.
An essential part of the project is a database of biodiversity measurements and my first year on the project has been spent designing and implementing processes and software to manage all aspects of our data - collection, validation, curation, visualisation and extraction. To carry out these tasks we use a range of in-house tools written in Python, with many based around the wonderful Django web-development framework together with GeoDjango and a PostgreSQLdatabase with the PostGIS spatial and geographical extensions, running on Linux. We are aiming for good coverage of major taxonomic groups, biomes, countries, habitats, use intensities and latitudes - we visualise and assess these using tools and datasets including R, Catalogue of Life, Google maps, d3.js and the Terrestrial Ecoregions of the World and Biodiversity Hotspots GIS datasets. We collect published or in-press data on terrestrial biodiversity or community composition, and currently have more than 1.5 million measurements of over 27,000 named taxa taken at over 13,000 locations.
I will spend the next year further developing and applying our modelling framework, based around internal R packages, that relates spatial and temporal variation in biodiversity to human pressures. I will expand our database to include species' traits such as body size, functional group and trophic level, and information about habitat changes, derived from remote-sensed data products such as land-cover maps and MODIS vegetation indices.
During my PhD I examined the relationship between the dynamics and structure of complex, multi-trophic ecological communities. This has historically been a difficult area to investigate because mathematical models of the dynamics of systems of realistic complexity have a large number of unmeasured parameters, and whole-community data are limited and typically comprise only a snapshot or time-averaged picture. The resulting 'plague of parameters' means most studies of multi-species population dynamics have been very theoretical.
In ground-breaking work, Yodzis and Innes (1992) Am. Nat. (doi - 10.1086/285380) proposed a class of dynamical models parameterised using physiological allometries. These models are a synthesis of allometric scaling and Lotka-Volterra style dynamical models: model parameters are computed from empirically observed inter-specific power-law relationships between physiological rates and body masses. This approach avoids the need to derive species- or population-specific parameters, sacrificing some accuracy for generality, and making it possible to investigate the dynamics of complex communities. Allometrically parameterised models therefore offer a potential cure for the plague of parameters. These models have been used in a large number of theoretical studies that have drawn conclusions on a wide range of topics. Despite their increasing use, this class of dynamical models are rarely tested against empirical data.
I examined this modelling approach and some of its assumptions. I confronted the model with empirical data from a complex, multi-trophic lake community and used Imperial College's High Performance Computing Linux services to search the space of those parameters that cannot be determined using physiological allometries. Many important empirical patterns were reproducible as outcomes of dynamics, and were not reproducible when parameters did not follow physiological allometries. Results validate the usefulness, when parameters follow physiological allometries, of classic differential-equation models for understanding whole-community dynamics and the structure-dynamics relationship (Hudson and Reuman, 2013, Proc R. Soc B. doi - 10.1098/rspb.2013.1901).
Other outcomes from my work were the publication of a new dataset of field metabolic rate data of individual birds and mammals together with an analysis of this data using linear mixed-effects models, leading to a better understanding of one of the model's principal assumptions (Hudson et al, 2013, J. Anim. Ecol. doi - 10.1111/1365-2656.12086). In collaboration with Guy Woodward's group, Eoin O'Gorman, Mark Ledger and others, I produced an open-source R package - Cheddar - for analysing and visualising empirical food-web data (Hudson et al, 2013, Methods Ecol. Evol. doi - 10.1111/2041-210X.12005).