What's going on in Bill Kristan's lab
For a list of courses I teach, go here.
For a list of articles with downloadable reprints, go here.
My students and I work on a variety of topics in applied and theoretical ecology. My own past and ongoing projects have included both mathematical and field-oriented approaches, but I'm particularly interested in topics in basic or theoretical ecology that promise to have important applications to conservation.
Subsidized Predation by crows and ravens
Predators can affect the distribution and abundance of their prey. Under normal circumstances predators are not expected to drive their prey to extinction, because as prey densities decline predators should either switch to more abundant prey or starvation will reduce their numbers. If human-provided resources, such as food or water, prevent declines in predator populations in response to declines in prey densities then predators become a much greater threat to prey. These "subsidized" predators have been implicated in the decline of threatened and endangered species.
One of the species I study, the Common Raven (Corvus corax), is a good example of a subsidized predator in the Mojave Desert. Ravens are strongly associated with human developments in the Mojave, and are known to attack juvenile desert tortoise (Gopherus agassizii), a threatened species. Ravens make use of several types of resources obtained from people: they consume anthropogenic foods at landfills, road-killed carrion on roads, drink irrigation runoff, and nest in artificial platforms and ornamental trees. Ravens that breed in close proximity to human-provided resources experience high reproductive rates and enhanced juvenile survival.
Interestingly, breeding ravens and non-breeding ravens are ecologically quite different. Non-breeding ravens will aggregate in large groups at abundant food sources, such as landfills, but breeding ravens defend territories and maintain distance between nests. Consequently, breeding ravens are often found at greater distances from human developments than are non-breeders, but breeding birds do not reach the population densities that non-breeders can attain. Because of these ecological differences, we wished to study whether predatory activity was also different for breeding and non-breeding birds, and how the spatial pattern of distribution of ravens translated into a spatial pattern of predation risk to tortoises. For this work I used attacks by ravens on styrofoam models of juvenile tortoises to measure risk of predation. Risk of predation for desert tortoises is elevated in places where ravens are more abundant, and at successful raven nests. The large aggregations of ravens found at landfills are generally composed of non-breeding individuals, whereas breeding birds are more evenly dispersed out to greater distances from human developments. This means that both non-breeding and breeding ravens are responsible for elevated predation risk, both near developments and away from them.
More recently my students (Linnae DeCamp and Crystel Doyle) have started to study these processes at the San Diego Wild Animal Park and surrounding reserve lands owned and managed by the Zoological Society of San Diego. The park itself attracts large numbers of crows as well as ravens, which take advantage of animal feed and wastes. We will be studying whether crows and ravens attracted to the Park elevate predation risk for animals in surrounding undeveloped lands (the raven to the left is inspecting one of our artificial nests). Additionally, the Park supports breeding colonies of native wading birds, such as Black-crowned Night Herons. We will be assessing whether these birds respond to the spatial distribution of crows and ravens in their nest site choices within the Park.
Invasive Exotic Plants
Invasive exotic species are widely acknowledged to be second only to habitat destruction as threats to biodiversity. The most destructive invasive plants are capable of supplanting native species, along with animal species that depend on native vegetation.
One of my graduate students (Carolyn Martus) is working on invasive Mexican fan palms in riparian areas of northern San Diego County. Fan palms are common ornamental plantings in this region, and appear to be favored by runoff from impervious surfaces in urbanizing landscapes. Urbanizing areas may therefore function both as a seed source and as a source of water for seedlings.
I am also working on models that predict risk of invasion by cheatgrass and red brome in the Owyhee Uplands of southeastern Oregon. These grasses are favored by fire, and once established they prevent recovery of native sagebrush scrub.
Wetlands mapping
We are involved in mapping coastal wetlands in San Diego County, as part of a larger effort to map coastal wetlands throughout coastal southern California, in collaboration with CSU Northridge and the Southern California Coastal Water Research Program. We are mapping wetlands by hand, using high-resolution, seamless imagery from the National Agricultural Imagery Program, and other sources. For more information about the mapping project, see: http://www.socalwetlands.com
As part of this project, one of my graduate students (Chrystal Barry) will be attempting to predict the distributions of several invasive exotic plant species using multivariate statistical methods. If this is successful, it should prove substantially less time consuming than digitizing maps by hand.
Bird and carnivore communities in coastal shrublands
I have previously studied community composition of birds and small mammals in fragmented coastal sage-scrub (CSS) habitats. My colleagues and I found that edges between patches of CSS habitat and urban developments frequently have lower abundances of birds than "interior" sites that are distant from a developed edge. However, although the changes in abundance seen at edges are superficially similar among different species of birds, the underlying responses differ. Some species (such as cactus wrens) were found at reduced abundance because cactus was less common at edges, whereas other species (such as sage sparrows) were less abundant in spite of similar habitat suitabilities at edges and interior areas.
Mammalian predators also differ among distances to edge, but small mammals are not strongly edge responsive. Carnivore and small mammal communities tended to change in parallel among different study areas, but carnivores did not induce an edge effect in small mammals.
One of my students (Mike Tucker) is measuring responses by several species to drinking water sources (guzzlers) in coastal shrublands.
Theoretical and statistical ecology
Habitat models are grounded in the observation that animals are non-randomly distributed in the environment, and that by understanding the habitat associations of a species we should be able to predict its distribution. Unfortunately, in spite of this seemingly reasonable expectation, habitat modeling is a prime example of an area of ecology that frequently fails to be predictive. I suspect that part of the reason for this is that statistical habitat modeling produces over-simplified representations of the habitat selection selection process in most animals. I am working on a variety of related topics that are designed to help improve the predictiveness of habitat models.
My first work in this area was a descriptive model of consequences for population dynamics of relaxing the usual assumption that animals are able to identify and use the highest quality habitat available to them. For many animals, habitat selection more closely resembles forecasting than it resembles an optimal choice, because weather, food, and predators can all change substantially between the time that a habitat choice is made and the time that the consequences of the choice are realized. Animals will typically need to select habitat based on "cues" that determine the habitat's "attractiveness", which are only indirect indicators of habitat quality. My model treated habitat attractiveness and habitat quality as two different properties of the environment that could either be positively related (i.e. high-quality habitat is attractive) or negatively related (i.e. low-quality habitat is attractive). With this same simple model it was possible to generate two previously distinct population structures: source-sink dynamics (when attractiveness was positively related to quality) and "ecological traps" (when attractiveness was negatively related to quality).
Recently, I have elaborated on this simple model to more realistically represent the complexity of the habitat selection process. In real populations habitat attractiveness may be determined by many different cues rather than a single one, and habitat quality will be the integrated effects of habitat on multiple life history variables such as survival probability and fecundity. The simplest way to represent this complexity was to make both habitat attractiveness and habitat quality bivariate: habitat quality is now represented by two different demographic parameters, reproductive output and survival probability, that are each affected only by one of two habitat variables. An interesting feature of this approach is that changing the correlation between habitat gradients in the simulated "landscape" also changes the correlation between survival and fecundity within the animal population, and this alters population growth rate. If an animal is allowed to select the best habitat, distributions are still affected by changes in distributions of the variables, but ecological traps will not occur. 
If instead the animal is allowed to select habitat based on only one habitat variable (and thus only either survival probability or fecundity), it's possible to create ecological traps in habitats in which only high reproductive success or high survival probability can be achieved but not both.
In addition to these difficulties with representing habitat selection realistically in a habitat model, statistical habitat models are empirical fits to particular data sets that may generalize poorly to different times and places. As an example, habitat use is frequently density-dependent, in that high-quality habitats are used at a variety of population densities but poor-quality habitat is only used when population densities are high. A model fitted at high population densities and then used to predict distributions of animals at low population densities will generally over-predict. I am developing an approach to adjusting statistical models to account for changes in population density. A fortuitous side-effect of the method is that it's possible to test for density-dependent habitat use.
Ancient history
Long, long ago in a land far, far away I worked on bald eagles and ospreys. It was a lot of fun.
Want to join?
I am currently willing to consider accepting graduate students that are interested in landscape ecology, statistical modeling, and GIS. Highly-motivated students with general interests in field ecological studies are also always welcome to apply.
In addition to their regular coursework, members of my lab participate in weekly lab meetings in which we discuss each other's work, and read and discuss papers of mutual interest.
Undergraduates that would like to participate in my lab should familiarize themselves with the work we do, and should have an interest in one or more of the major aspects of our work.