Joshua J. Lawler

Random Forests for Classification in Ecology

Classification procedures are some of the most widely used statistical methods in ecology. Random forests (RF) is a new and powerful statistical classifier that is well established in other disciplines but is relatively unknown in ecology. Advantages of RF compared to other statistical classifiers include (1) very high classification accuracy; (2) a novel method of determining variable importance; (3) ability to model complex interactions among predictor variables; (4) flexibility to perform several types of statistical data analysis, including regression, classification, survival analysis, and unsupervised learning; and (5) an algorithm for imputing missing values. We compared the accuracies of RF and four other commonly used statistical classifiers using data on invasive plant species presence in Lava Beds National Monument, California, USA, rare lichen species presence in the Pacific Northwest, USA, and nest sites for cavity nesting birds in the Uinta Mountains, Utah, USA. We observed high classification accuracy in all applications as measured by cross-validation and, in the case of the lichen data, by independent test data, when comparing RF to other common classification methods. We also observed that the variables that RF identified as most important for classifying invasive plant species coincided with expectations based on the literature.

Machine Learning Methods Without Tears: A Primer for Ecologists

Machine learning methods, a family of statistical techniques with origins in the field of artificial intelligence, are recognized as holding great promise for the advancement of understanding and prediction about ecological phenomena. These modeling techniques are flexible enough to handle complex problems with multiple interacting elements and typically outcompete traditional approaches (e.g., generalized linear models), making them ideal for modeling ecological systems. Despite their inherent advantages, a review of the literature reveals only a modest use of these approaches in ecology as compared to other disciplines. One potential explanation for this lack of interest is that machine learning techniques do not fall neatly into the class of statistical modeling approaches with which most ecologists are familiar. In this paper, we provide an introduction to three machine learning approaches that can be broadly used by ecologists: classification and regression trees, artificial neural networks, and evolutionary computation. For each approach, we provide a brief background to the methodology, give examples of its application in ecology, describe model development and implementation, discuss strengths and weaknesses, explore the availability of statistical software, and provide an illustrative example. Although the ecological application of machine learning approaches has increased, there remains considerable skepticism with respect to the role of these techniques in ecology. Our review encourages a greater understanding of machine learning approaches and promotes their future application and utilization, while also providing a basis from which ecologists can make informed decisions about whether to select or avoid these approaches in their future modeling endeavors.

Projected climate-induced faunal change in the Western Hemisphere

Climate change is predicted to be one of the greatest drivers of ecological change in the coming century. Increases in temperature over the last century have clearly been linked to shifts in species distributions. Given the magnitude of projected future climatic changes, we can expect even larger range shifts in the coming century. These changes will, in turn, alter ecological communities and the functioning of ecosystems. Despite the seriousness of predicted climate change, the uncertainty in climate-change projections makes it difficult for conservation managers and planners to proactively respond to climate stresses. To address one aspect of this uncertainty, we identified predictions of faunal change for which a high level of consensus was exhibited by different climate models. Specifically, we assessed the potential effects of 30 coupled atmosphere-ocean general circulation model (AOGCM) future-climate simulations on the geographic ranges of 2954 species of birds, mammals, and amphibians in the Western Hemisphere. Eighty percent of the climate projections based on a relatively low greenhouse-gas emissions scenario result in the local loss of at least 10% of the vertebrate fauna over much of North and South America. The largest changes in fauna are predicted for the tundra, Central America, and the Andes Mountains where, assuming no dispersal constraints, specific areas are likely to experience over 90% turnover, so that faunal distributions in the future will bear little resemblance to those of today.

Efficiency of incentives to jointly increase carbon sequestration and species conservation on a landscape

We develop an integrated model to predict private land-use decisions in response to policy incentives designed to increase the provision of carbon sequestration and species conservation across heterogeneous landscapes. Using data from the Willamette Basin, Oregon, we compare the provision of carbon sequestration and species conservation under five simple policies that offer payments for conservation. We evaluate policy performance compared with the maximum feasible combinations of carbon sequestration and species conservation on the landscape for various conservation budgets. None of the conservation payment policies produce increases in carbon sequestration and species conservation that approach the maximum potential gains on the landscape. Our results show that policies aimed at increasing the provision of carbon sequestration do not necessarily increase species conservation and that highly targeted policies do not necessarily do as well as more general policies.

Dispersal will limit ability of mammals to track climate change in the Western Hemisphere

As they have in response to past climatic changes, many species will shift their distributions in response to modern climate change. However, due to the unprecedented rapidity of projected climatic changes, some species may not be able to move their ranges fast enough to track shifts in suitable climates and associated habitats. Here, we investigate the ability of 493 mammals to keep pace with projected climatic changes in the Western Hemisphere. We modeled the velocities at which species will likely need to move to keep pace with projected changes in suitable climates. We compared these velocities with the velocities at which species are able to move as a function of dispersal distances and dispersal frequencies. Across the Western Hemisphere, on average, 9.2% of mammals at a given location will likely be unable to keep pace with climate change. In some places, up to 39% of mammals may be unable to track shifts in suitable climates. Eighty-seven percent of mammalian species are expected to experience reductions in range size and 20% of these range reductions will likely be due to limited dispersal abilities as opposed to reductions in the area of suitable climate. Because climate change will likely outpace the response capacity of many mammals, mammalian vulnerability to climate change may be more extensive than previously anticipated.

Climate change adaptation strategies for resource management and conservation planning.

Recent rapid changes in the Earths climate have altered ecological systems around the globe. Global warming has been linked to changes in physiology, phenology, species distributions, interspecific interactions, and disturbance regimes. Projected future climate change will undoubtedly result in even more dramatic shifts in the states of many ecosystems. These shifts will provide one of the largest challenges to natural resource managers and conservation planners. Managing natural resources and ecosystems in the face of uncertain climate requires new approaches. Here, the many adaptation strategies that have been proposed for managing natural systems in a changing climate are reviewed. Most of the recommended approaches are general principles and many are tools that managers are already using. What is new is a turning toward a more agile management perspective. To address climate change, managers will need to act over different spatial and temporal scales. The focus of restoration will need to shift from historic species assemblages to potential future ecosystem services. Active adaptive management based on potential future climate impact scenarios will need to be a part of everyday operations. And triage will likely become a critical option. Although many concepts and tools for addressing climate change have been proposed, key pieces of information are still missing. To successfully manage for climate change, a better understanding will be needed of which species and systems will likely be most affected by climate change, how to preserve and enhance the evolutionary capacity of species, how to implement effective adaptive management in new systems, and perhaps most importantly, in which situations and systems will the general adaptation strategies that have been proposed work and how can they be effectively applied.

Projected land-use change impacts on ecosystem services in the United States

Significance Land-use change affects the provision of ecosystem services and wildlife habitat. We project land-use change from 2001 to 2051 for the contiguous United States under two scenarios reflecting continuation of 1990s trends and high crop demand more reflective of the recent past. These scenarios result in large differences in land-use trajectories that generate increases in carbon storage, timber production, food production from increased yields (even with declines in cropland area), and >10% decreases in habitat for one-quarter of modeled species. We analyzed three policy alternatives that provide incentives to maintain and expand forest cover, conserve natural habitats, and limit urban sprawl. Policy interventions need to be aggressive to significantly alter underlying land-use trends and shift the trajectory of ecosystem service provision. Providing food, timber, energy, housing, and other goods and services, while maintaining ecosystem functions and biodiversity that underpin their sustainable supply, is one of the great challenges of our time. Understanding the drivers of land-use change and how policies can alter land-use change will be critical to meeting this challenge. Here we project land-use change in the contiguous United States to 2051 under two plausible baseline trajectories of economic conditions to illustrate how differences in underlying market forces can have large impacts on land-use with cascading effects on ecosystem services and wildlife habitat. We project a large increase in croplands (28.2 million ha) under a scenario with high crop demand mirroring conditions starting in 2007, compared with a loss of cropland (11.2 million ha) mirroring conditions in the 1990s. Projected land-use changes result in increases in carbon storage, timber production, food production from increased yields, and >10% decreases in habitat for 25% of modeled species. We also analyze policy alternatives designed to encourage forest cover and natural landscapes and reduce urban expansion. Although these policy scenarios modify baseline land-use patterns, they do not reverse powerful underlying trends. Policy interventions need to be aggressive to significantly alter underlying land-use change trends and shift the trajectory of ecosystem service provision.

Will extreme climatic events facilitate biological invasions

Extreme climatic events (ECEs) – such as unusual heat waves, hurricanes, floods, and droughts – can dramatically affect ecological and evolutionary processes, and these events are projected to become more frequent and more intense with ongoing climate change. However, the implications of ECEs for biological invasions remain poorly understood. Using concepts and empirical evidence from invasion ecology, we identify mechanisms by which ECEs may influence the invasion process, from initial introduction through establishment and spread. We summarize how ECEs can enhance invasions by promoting the transport of propagules into new regions, by decreasing the resistance of native communities to establishment, and also sometimes by putting existing non-native species at a competitive disadvantage. Finally, we outline priority research areas and management approaches for anticipating future risks of unwanted invasions following ECEs. Given predicted increases in both ECE occurrence and rates of species introduction...

Resource management in a changing and uncertain climate

Climate change is altering ecological systems throughout the world. Managing these systems in a way that ignores climate change will likely fail to meet management objectives. The uncertainty in projected climate-change impacts is one of the greatest challenges facing managers attempting to address global change. In order to select successful management strategies, managers need to understand the uncertainty inherent in projected climate impacts and how these uncertainties affect the outcomes of management activities. Perhaps the most important tool for managing ecological systems in the face of climate change is active adaptive management, in which systems are closely monitored and management strategies are altered to address expected and ongoing changes. Here, we discuss the uncertainty inherent in different types of data on potential climate impacts and explore climate projections and potential management responses at three sites in North America. The Central Valley of California, the headwaters of the...

Applied Climate-Change Analysis: The Climate Wizard Tool

Background Although the message of “global climate change” is catalyzing international action, it is local and regional changes that directly affect people and ecosystems and are of immediate concern to scientists, managers, and policy makers. A major barrier preventing informed climate-change adaptation planning is the difficulty accessing, analyzing, and interpreting climate-change information. To address this problem, we developed a powerful, yet easy to use, web-based tool called Climate Wizard (http://ClimateWizard.org) that provides non-climate specialists with simple analyses and innovative graphical depictions for conveying how climate has and is projected to change within specific geographic areas throughout the world. Methodology/Principal Findings To demonstrate the Climate Wizard, we explored historic trends and future departures (anomalies) in temperature and precipitation globally, and within specific latitudinal zones and countries. We found the greatest temperature increases during 1951–2002 occurred in northern hemisphere countries (especially during January–April), but the latitude of greatest temperature change varied throughout the year, sinusoidally ranging from approximately 50°N during February-March to 10°N during August-September. Precipitation decreases occurred most commonly in countries between 0–20°N, and increases mostly occurred outside of this latitudinal region. Similarly, a quantile ensemble analysis based on projections from 16 General Circulation Models (GCMs) for 2070–2099 identified the median projected change within countries, which showed both latitudinal and regional patterns in projected temperature and precipitation change. Conclusions/Significance The results of these analyses are consistent with those reported by the Intergovernmental Panel on Climate Change, but at the same time, they provide examples of how Climate Wizard can be used to explore regionally- and temporally-specific analyses of climate change. Moreover, Climate Wizard is not a static product, but rather a data analysis framework designed to be used for climate change impact and adaption planning, which can be expanded to include other information, such as downscaled future projections of hydrology, soil moisture, wildfire, vegetation, marine conditions, disease, and agricultural productivity.