Computational and Mathematical Research in Conservation Biology
This page describes my work on Computer Science applications in Conservation Biology, including links to papers
and software on the Maximum Entropy Method for
estimating a probability distribution from a limited number of
samples, and its application to modeling species distributions.
Please send mail to "phillips at-sign research.att.com"
with comments or suggestions.
Maxent Software for Modeling Species Distributions
Maxent is a self-contained Java
application for modeling species geographic distributions using the
Maximum Entropy Method. It has been developed in collaboration with
Rob Schapire and Miro Dudik, and is available for free download.
Species distribution modelling papers
- Arctic Vegetation Distribution Shifts and associated Feedbacks under Future Climate Change,
R. G. Pearson, S. J. Phillips, M. M. Loranty, P. S. A. Beck, T. Damoulas, S. J. Knight and S. J. Goetz
, Nature Climate Change, in press. [pdf] [supplement]
- On Estimating Probability of Presence from Use-Availability or Presence-Background Data,
S. J. Phillips and J. Elith, Ecology, in press.
- Weekly predictions of North Atlantic right whale Eubalaena glacialis
habitat reveal influence of prey abundance and seasonality of habitat preferences,
D. E. Pendleton, P. J. Sullivan, M. W. Brown, T. V. N. Cole, C. P. Good,
C. A. Mayo, B. C. Monger, S. J. Phillips, N. R. Record and A. J. Pershing,
Endangered Species Research 18, 2012, pp 147-161. [pdf]
- Inferring prevalence from presence-only data: a response to 'Can we model the probability of presence of species without absence data?', S. J. Phillips, Ecography 35, 2012, pp 385-387. [pdf]
- Logistic Methods for Resource Selection Functions
and Presence-Only Species Distribution Models, S. J. Phillips and J. Elith, Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence, 2011, pp 1384-1389. [pdf]
- A statistical explanation of MaxEnt for ecologists, J. Elith, S. J. Phillips, T. Hastie, M. Dudík, Y. E. Chee and C. Yates, Diversity and Distributions 17, 2011, pp 43-57. [pdf]
- The art of modelling range-shifting species, J. Elith, M. Kearney and S. J. Phillips, Methods in Ecology & Evolution 1, 2010, pp 330-342. [pdf] [appendix]
- POC-plots: calibrating species distribution models with presence-only data, S. J. Phillips and J. Elith, Ecology 91, 2010, pp 2476-2484. [pdf]
- Generative and Discriminative Learning with
Unknown Labeling Bias, M. Dudík and S. J. Phillips, Neural Information Processing Systems 2008, pp 401-408. [pdf]
- Sample Selection Bias and
Presence-Only Species Distribution Models: Implications for Background
and Pseudo-absence data, S. J. Phillips, M. Dudík, J. Elith, C.H. Graham, A. Lehmann, J. Leathwich and S. Ferrier, Ecological Applications 19(1), 2009, pp. 181-197. [pdf]
- Aligning Conservation Priorities Across Taxa in Madagascar with
High-Resolution Planning Tools, with C. Kremen, A. Cameron, A. Moilanen, S. J. Phillips, C. D. Thomas, H. Beentje, J. Dransfield, B. L. Fisher, F. Glaw, T. C. Good, G. J. Harper, R. J. Hijmans, D. C.
Lees, E. Louis Jr, R. A. Nussbaum, C. J. Raxworthy, A. Razafimpahanana, G. E.
Schatz, M. Vences, D. R. Vieites, P. C. Wright and M. L. Zjhra,
Science 320(5873), 2008, pp 222-226. [Full text available from Claire Kremen's website]
- Modeling of species distributions with Maxent: new extensions and
a comprehensive evaluation, S. J. Phillips and M. Dudík,
Ecography 31, 2008, pp
- Transferability, sample selection bias and background data in
presence-only modelling: a response to Peterson et al. (2007), S. J. Phillips,
Ecography 31, 2008, pp
- Maximum entropy density estimation with generalized regularization and an
application to species distrubtion modeling, M. Dudík, S. J. Phillips and R. E. Schapire,
Journal of Machine Learning Research 8, 2007, pp
- Novel methods improve prediction of species' distributions from occurrence data, J. Elith, C.H. Graham, R.P. Anderson, M. Dudík, S. Ferrier, A. Guisan, R.J. Hijmans, F. Huettmann, J.R. Leathwick, A. Lehmann, J. Li, L.G. Lohmann, B.A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J. McC. Overton, A.T. Peterson, S.J. Phillips, K. Richardson, R. Scachetti-Pereira, R.E. Schapire, J. Soberón, S. Williams, M.S. Wisz and N.E. Zimmermann, Ecography 29, 2006, pp 129-151. [pdf]
- Maximum entropy modeling of species geographic
distributions, S. J. Phillips, R. P. Anderson and R. E. Schapire,
Ecological Modelling 190(3-4), 2006, pp
- Correcting sample selection bias in maximum entropy density estimation,
M. Dudík, S. J. Phillips and R. E. Schapire,
in Proceedings of Neural Information Processing Systems, 2005.
- Performance guarantees for regularized maximum entropy density
estimation, M. Dudík, S. J. Phillips and R. E. Schapire, in Proceedings of the
Seventeenth Annual Conference on Computational Learning Theory, 2004, pp 472-486.
- A maximum entropy approach to species distribution modeling,
S. J. Phillips, R. E. Schapire and M. Dudík, in Proceedings of the Twenty-First
International Conference on Machine Learning, 2004, pp 655-662.
Other Conservation Topics
Graph-theoretic methods are useful for modeling spatial and temporal habitat connectivity. This paper uses the Viterbi algorithm to model dynamic refugia, by determining the most likely way for rainforest to have shifted through historical climates in a way that provides stable habitat for rainforest-dependent organisms.
Site prioritization indices are useful for conservation planning across broad scales, especially in a decision support framework. This paper investigates close connections between site prioritization and the mathematical theory of power in voting systems, and introduces a simple and effective new site prioritization index.
- Dynamic refugia and species persistence: tracking spatial shifts in habitat through time, C. H. Graham, J. VanDerWal, S. J. Phillips, C. Moritz and S. E. Williams, Ecography 33, 2010, pp. 1062-1069. [pdf]
The following paper describes and method for selecting an optimal set of sites that allows a group of species to migrate from their currently occupied areas to areas predicted to become suitable under climate change, with migration speed constained by the individual species' dispersal abilities. The method is based on "network flow", an optimization paradigm that is heavily used in modeling and optimizing telecom networks, among other applications.
- Voting power and target-based site prioritization, S. J. Phillips, A. Archer, R. L. Pressey, D. Torkornoo, D. Applegate, D. S. Johnson and M. E. Watts, Biological Conservation 143, 2010, pp. 1989-1997. [pdf]
Once sites are protected, optimization methods can be useful for informing managament and trading off the needs of different species. This paper uses integer programming methods to optimize the configuration of managed ponds and restored tidal marsh in San Francisco Bay, formerly used as evaporative salt ponds, to simultaneously maximize predicted abundance of a set of marsh- and pond-associated bird species.
- Optimizing Dispersal Corridors for the Cape Proteaceae Using Network Flow, S. J. Phillips, P. Williams, G. Midgley and A. Archer,
Ecological Applications 18(5), 2008, pp. 1200-1211. [pdf]
Species distribution models predict where suitable conditions for a species will be, but they do not directly say which suitable conditions are likely to be occupied. This paper integrates species distribution models with stochastic population models to model changes in a species' population size as suitable areas move under climate change:
- Optimizing wetland restoration and management for avian
communities using a mixed integer programming approach, D. Stralberg, D.L. Applegate, S. J. Phillips, M.P. Herzog, N. Nur and N. Warnock, Biological
Conservation 142, 2009, pp 94-109. [pdf]
- Predicting extinction risks under climate change: a new mechanistic
approach linking stochastic population models with dynamic bioclimatic
D. A. Keith, H. R. Akçakaya, W. Thuiller, G. F. Midgley, R. G. Pearson,
S. J. Phillips, H. M. Regan, M. B. Araújo and T. G. Rebelo,
Biology Letters 4, 2008, pp. 560-563. [pdf]
Communications technology has great potential for helping other industries
reduce their greenhouse gas emissions. Nevertheless, telecom providers need
to transition to more energy-efficient technology in order to
limit their own environmental footprint. This paper uses statistical methods
to partition electricity consumption by technology and network type in central
offices. We find that class-5 telephone switches should be a primary focus
for reducing central office energy consumption.
AT&T made a TV commercial highlighting my conservation work in Madagascar, as described in the Science article above. The commercial aired in Washington, DC in June, 2008. AT&T entered a related video of my Madagascar work in a coolest-corporate-science-job competition. Mongabay.com featured my Madagascar work too.
- A regression approach to infer electricity consumption of
legacy telecom equipment, S. J. Phillips, S. L. Woodward, M. D. Feuer, P. D. Magill, ACM SIGMETRICS Performance Evaluation Review, to appear. [pdf]
Together with Richard Pearson from the American Museum of Natural History, I teach a biannual week-long course on Species Distribution Modeling for Conservation Biologists at the AMNH's Southwestern Research Station in the Chiricahua Mountains of Arizona.
I am a member of a cross-organizational "Green team" inside AT&T, investigating ways to reduce the company's electricity consumption and carbon footprint. I'm managing a wiki (internal AT&T link here) with information about wind and solar power technology, news and meteorological information on the distribution of clean energy resources.
In my spare time, I am a volunteer for Transportation Alternatives' campaign to close the loop road of Central Park to car traffic and return it to its original recreational use. As Edward Abbey put it, "we have agreed not to drive our automobiles into cathedrals, concert halls, art museums, private bedrooms and other sanctums of our culture; we should treat our parks with the same deference."
Last modified September 6, 2011.