Using stochastic epidemiological models to evaluate conservation strategies for endangered amphibians

Brian Drawert, Marc Griesemer, Linda Petzold, Cheryl J. BriggsJournal of The Royal Society Interface 2017 · DOI

TL;DR

This paper uses stochastic epidemiological models to evaluate the effectiveness of different one-time conservation strategies against the deadly chytrid fungus (Bd) in amphibians. The study reveals that while no single strategy guarantees success, treating infected frogs with antifungal agents offers the best chance of preventing extinction with the lowest risk to the host population, outperforming environmental disinfection or host culling.

What Problem Does This Paper Address?

Amphibian populations worldwide are facing severe declines, with the fungal pathogen Batrachochytrium dendrobatidis (Bd) being a major culprit behind widespread chytridiomycosis outbreaks. These outbreaks can lead to rapid population crashes and local extinctions, threatening amphibian biodiversity. While various conservation strategies have been proposed and implemented, their effectiveness in the wild remains uncertain, often proving labor-intensive, expensive, and sometimes unsuccessful. This paper aims to address this gap by using mathematical modeling to rigorously assess the potential of different one-time conservation interventions to prevent Bd-induced extinction in naive amphibian populations.

Methods

The researchers developed a mathematical model that simulates the dynamics of Bd outbreaks in a single amphibian host species, specifically inspired by the mountain yellow-legged frogs (Rana muscosa and Rana sierrae). The model incorporates both demographic stochasticity (randomness in population events) and within-host fungal load dynamics. They explored three versions of the model, each making different assumptions about how the fungal load on an individual frog changes over time.

They then implemented and evaluated three types of one-time conservation strategies:

  1. Clearing individuals of infection: This involves treating frogs with antifungal agents to reduce their fungal load.
  2. Reducing host density: This refers to culling or removing a portion of the frog population.
  3. Disinfecting the environment: This involves using antifungals to reduce the pool of infectious Bd zoospores in the water.

These strategies were tested by varying the timing of intervention (14, 28, or 56 days post-infection) and the fraction of the population or environment treated (25%, 50%, or 75%). The study employed both deterministic (ordinary differential equation) and stochastic (Gillespie algorithm) simulation approaches to capture the full range of potential outcomes. A global sensitivity analysis was conducted across a wide range of parameter values to understand how different biological parameters influence the effectiveness of these strategies.

Datasets

Not discussed in the paper. The study relies on a simulated system based on biological parameters and ranges derived from existing literature.

Key Insights and Ideas

  • Stochasticity is Crucial: The paper highlights that random chance (stochasticity) plays a significant role in determining the outcome of Bd outbreaks. Even with parameters that would lead to guaranteed extinction in a deterministic model, stochasticity can sometimes lead to fungal extinction or coexistence with the host.
  • "Partial Die-offs" and "Stuttering Chains": Two key stochastic phenomena were identified: "partial die-offs" where a fraction of the population dies, reducing the pathogen's transmission enough for the rest to survive, and "stuttering chains" where the infection spreads slowly and intermittently, failing to cause widespread mortality.
  • No Silver Bullet: For a large portion of the parameter space, none of the tested one-time conservation strategies were effective at preventing extinction. This underscores the difficulty of combating this devastating disease.
  • Treating Frogs is Best: Among the evaluated strategies, directly treating infected frogs with antifungal agents to reduce their fungal load emerged as the most promising. It offered the highest likelihood of a beneficial outcome and the lowest risk of negatively impacting the frog population.
  • Environmental Disinfection is Largely Ineffective: Cleaning the environment to reduce zoospores showed minimal impact, with positive and negative outcomes being equally likely.
  • Culling is Risky: While culling can sometimes be effective by reducing host density, it carries a significant risk of being detrimental to the population, as it directly removes individuals.
  • Zoospores' Background Death Rate is Less Important Than Expected: Surprisingly, the model showed low sensitivity to parameters governing the background death rate of zoospores, suggesting that strategies solely focused on reducing zoospore survival in the environment might be less effective than anticipated for preventing initial outbreaks.

Key Results

  • In deterministic models without intervention, frog extinction occurred in 100% of tested parameter combinations.
  • In stochastic models without intervention, frog extinction occurred in approximately 65% of cases, with fungal extinction in 24% and coexistence in a small fraction.
  • Cleaning frogs (reducing individual fungal load) had a positive effect in 20-30% of the parameter space and rarely had a negative effect.
  • Culling frogs had a higher chance of a positive effect than cleaning but also an equally large chance of being detrimental.
  • Cleaning the environment was largely ineffective, with only a small chance of any effect and equally likely positive and negative outcomes.
  • The effectiveness of strategies was most pronounced in groups with high stochasticity and high mortality (Groups B and C), where cleaning frogs showed a high probability of positive effect.
  • The timing of intervention mattered: earlier interventions (14 days) were generally more effective than later ones (56 days).

Limitations and Future Work

The authors acknowledge that their model simplifies complex ecological systems. For instance, they focused on a single host species and one-time interventions, whereas real-world scenarios often involve multiple species, continuous environmental pressures, and combinations of strategies. The model also doesn't fully explore the long-term effects of interventions or the role of probiotics, which are mentioned as a promising area for future research. The paper suggests that strategies that permanently alter host resistance or tolerance, such as stimulating the immune response or augmenting beneficial skin bacteria, warrant further investigation. Additionally, the model's findings on the limited effectiveness of environmental disinfection and the risk of culling highlight the need for careful consideration and prioritization of conservation efforts. The authors also note that their findings could be relevant for understanding and combating other emerging fungal diseases, such as Bsal in salamanders.

Summary updated

Abstract

Recent outbreaks of chytridiomycosis, the disease of amphibians caused by the fungal pathogen <i>Batrachochytrium dendrobatidis</i> (Bd), have contributed to population declines of numerous amphibian species worldwide. The devastating impacts of this disease have led researchers to attempt drastic conservation measures to prevent further extinctions and loss of biodiversity. The conservation measures can be labour-intensive or expensive, and in many cases have been unsuccessful. We developed a mathematical model of Bd outbreaks that includes the effects of demographic stochasticity and within-host fungal load dynamics. We investigated the impacts of one-time treatment conservation strategies during the disease outbreak that occurs following the initial arrival of Bd into a previously uninfected frog population. We found that for all versions of the model, for a large fraction of parameter space, none of the one-time treatment strategies are effective at preventing disease-induced extinction of the amphibian population. Of the strategies considered, treating frogs with antifungal agents to reduce their fungal load had the greatest likelihood of a beneficial outcome and the lowest risk of decreasing the persistence of the frog population, suggesting that this disease mitigation strategy should be prioritized over disinfecting the environment or reducing host density.
BibTeX
@article{drawert2017using,
  author = {Brian Drawert and Marc Griesemer and Linda Petzold and Cheryl J. Briggs},
  title = {{Using stochastic epidemiological models to evaluate conservation strategies for endangered amphibians}},
  journal = {Journal of The Royal Society Interface},
  year = {2017},
  doi = {10.1098/rsif.2017.0480},
}
Figure