Biomedical subjects

Greg Dwyer

Publications and source records attributed to Greg Dwyer.

7 recordsLinked to original sources

PubMed2006

Uncertainty in predictions of disease spread and public health responses to bioterrorism and emerging diseases.

Concerns over bioterrorism and emerging diseases have led to the widespread use of epidemic models for evaluating public health strategies. Partly because epidemic models often capture the dynamics of prior epidemics remarkably well, little attention has been paid to how uncertainty in parameter estimates might affect model predictions. To understand such effects, we used Bayesian statistics to rigorously estimate the uncertainty in the parameters of an epidemic model, focusing on smallpox bioterrorism. We then used a vaccination model to translate the uncertainty in the model parameters into uncertainty in which of two vaccination strategies would provide a better response to bioterrorism, mass vaccination, or vaccination of social contacts, so-called "trace vaccination." Our results show that the uncertainty in the model parameters is remarkably high and that this uncertainty has important implications for vaccination strategies. For example, under one plausible scenario, the most likely outcome is that mass vaccination would save approximately 100,000 more lives than trace vaccination. Because of the high uncertainty in the parameters, however, there is also a substantial probability that mass vaccination would save 200,000 or more lives than trace vaccination. In addition to providing the best response to the most likely outcome, mass vaccination thus has the advantage of preventing outcomes that are only slightly less likely but that are substantially more horrific. Rigorous estimates of uncertainty thus can reveal hidden advantages of public health strategies, suggesting that formal uncertainty estimation should play a key role in planning for epidemics.

Bayes Theorem↗

PubMed2006

Combining population-dynamic and ecophysiological models to predict climate-induced insect range shifts.

Hundreds of species are shifting their ranges in response to recent climate warming. To predict how continued climate warming will affect the potential, or “bioclimatic range,” of a skipper butterfly, we present a population‐dynamic model of range shift in which population growth is a function of temperature. We estimate the parameters of this model using previously published data for Atalopedes campestris. Summer and winter temperatures affect population growth rate independently in this species and therefore interact as potential range‐limiting factors. Our model predicts a two‐phase response to climate change; one range‐limiting factor gradually becomes dominant, even if warming occurs steadily along a thermally linear landscape. Whether the range shift accelerates or decelerates and whether the number of generations per year at the range edge increases or decreases depend on whether summer or winter warms faster. To estimate the uncertainty in our predictions of range shift, we use a parametric bootstrap of biological parameter values. Our results show that even modest amounts of data yield predictions with reasonably small confidence intervals, indicating that ecophysiological models can be useful in predicting range changes. Nevertheless, the confidence intervals are sensitive to regional differences in the underlying thermal landscape and the warming scenario.

Animals↗

PubMed2005

Resource-dependent dispersal and the speed of biological invasions.

Many mobile organisms exhibit resource-dependent movement in which movement rates adjust to changes in local resource densities through changes in either the probability of moving or the distance moved. Such changes may have important consequences for invasions because reductions in resources behind an invasion front may cause higher dispersal while simultaneously reducing population growth behind the front and thus lowering the number of dispersers. Intuiting how the interplay between population growth and dispersal affects invasions is difficult without mathematical models, yet most models assume dispersal rates are constant. Here we present spatial-spread models that allow for consumer-resource interactions and resource-dependent dispersal. Our results show that when resources affect the probability of dispersal, then the invasion dynamics are no different than if resources did not affect dispersal. When resources instead affect the distance dispersed, however, the invasion dynamics are strongly affected by the strength of the consumer-resource interaction, and population cycles behind the wave front lead to fluctuating rates of spread. Our results suggest that for actively dispersing invaders, invasion dynamics can be determined by species interactions. More practically, our work suggests that reducing invader densities behind the front may be a useful method of slowing an invader's rate of spread.

Animals↗

PubMed2005

Combining stochastic models with experiments to understand the dynamics of monarch butterfly colonization.

Stochastic models are of increasing importance in ecology but are usually only applied to observational data. Here we use a stochastic population model to combine experimental and observational data to understand the colonization of old fields by monarch butterflies Danaus plexippus. We experimentally tested for density dependence in oviposition rates when predators were excluded, and we measured predation rates under natural conditions. Significance tests on the resulting data showed that both oviposition and predation were density dependent but could not show how oviposition and mortality combine to determine egg densities in nature. We therefore used our data to calculate the Akaike Information Criterion to choose between a nested suite of stochastic models that differed in their oviposition and mortality terms. When we simply fit the models to the observational data, the best model assumed density independence in both oviposition and predation. When we instead first estimated the oviposition rate at low density from experimental data, however, the best model included density dependence in oviposition, and a model that included density dependence in both oviposition and predation performed nearly as well. This result is consistent with our experiments and suggests that experiments can enhance the usefulness of stochastic models in ecology.

Animal Feed↗

PubMed2004

Should models of disease dynamics in herbivorous insects include the effects of variability in host-plant foliage quality?

Interactions between insects and their baculovirus pathogens are often described using simple disease models. Baculoviruses, however, are transmitted when insects consume virus-contaminated foliage, and foliage variability, whether within or between host-plant species, can affect viral infectiousness. Insect-baculovirus interactions may thus be embedded in a tritrophic interaction with the insect's host plant, but disease models include only the host and the pathogen. We tested these models by measuring the transmission of a baculovirus of gypsy moths (Lymantria dispar) on red oak (Quercus rubra) and white oak (Quercus alba) in the field in six experiments over four years. In all experiments, there were only weak effects of host-tree species, and in only one did the best-fitting model include tree species effects. These weak effects of foliage variability on transmission were not due to a lack of foliage variability on viral infectiousness, because when larvae were force-fed virus-contaminated foliage, infection rates were higher on white oak. Our results suggest that feeding behavior plays an important role in baculovirus transmission and that models can usefully describe baculovirus dynamics even without including foliage variability. Our work provides a clear example of how two-species models are sometimes sufficient to describe what appear to be tritrophic interactions.

Animals↗

PubMed2004

The combined effects of pathogens and predators on insect outbreaks.

The economic damage caused by episodic outbreaks of forest-defoliating insects has spurred much research, yet why such outbreaks occur remains unclear. Theoretical biologists argue that outbreaks are driven by specialist pathogens or parasitoids, because host-pathogen and host-parasitoid models show large-amplitude, long-period cycles resembling time series of outbreaks. Field biologists counter that outbreaks occur when generalist predators fail, because predation in low-density defoliator populations is usually high enough to prevent outbreaks. Neither explanation is sufficient, however, because the time between outbreaks in the data is far more variable than in host-pathogen and host-parasitoid models, and far shorter than in generalist-predator models. Here we show that insect outbreaks can be explained by a model that includes both a generalist predator and a specialist pathogen. In this host-pathogen-predator model, stochasticity causes defoliator densities to fluctuate erratically between an equilibrium maintained by the predator, and cycles driven by the pathogen. Outbreaks in this model occur at long but irregular intervals, matching the data. Our results suggest that explanations of insect outbreaks must go beyond classical models to consider interactions among multiple species.

Animals↗

PubMed2000

Pathogen-Driven Outbreaks in Forest Defoliators Revisited: Building Models from Experimental Data.

Models of outbreaks in forest-defoliating insects are typically built from a priori considerations and tested only with long time series of abundances. We instead present a model built from experimental data on the gypsy moth and its nuclear polyhedrosis virus, which has been extensively tested with epidemic data. These data have identified key details of the gypsy moth-virus interaction that are missing from earlier models, including seasonality in host reproduction, delays between host infection and death, and heterogeneity among hosts in their susceptibility to the virus. Allowing for these details produces models in which annual epidemics are followed by bouts of reproduction among surviving hosts and leads to quite different conclusions than earlier models. First, these models suggest that pathogen-driven outbreaks in forest defoliators occur partly because newly hatched insect larvae have higher average susceptibility than do older larvae. Second, the models show that a combination of seasonality and delays between infection and death can lead to unstable cycles in the absence of a stabilizing mechanism; these cycles, however, are stabilized by the levels of heterogeneity in susceptibility that we have observed in our experimental data. Moreover, our experimental estimates of virus transmission rates and levels of heterogeneity in susceptibility in gypsy moth populations give model dynamics that closely approximate the dynamics of real gypsy moth populations. Although we built our models from data for gypsy moth, our models are, nevertheless, quite general. Our conclusions are therefore likely to be true, not just for other defoliator-pathogen interactions, but for many host-pathogen interactions in which seasonality plays an important role. Our models thus give qualitative insight into the dynamics of host-pathogen interactions, while providing a quantitative interpretation of our gypsy moth-virus data.

Lymantria dispar↗