Actions for Predicting the abundance and distribution of freshwater fishes under global change

Email RIS file
Bookmark
Report an Issue

Predicting the abundance and distribution of freshwater fishes under global change

Author
Custer, Chris
Published
[University Park, Pennsylvania] : Pennsylvania State University, 2024.
Physical Description
1 electronic document
Additional Creators
Wagner, Tyler
Access Online

Availability

I Want It
I Want It

Finding items...

Graduate Program
Restrictions on Access
Open Access.
Summary
Freshwater ecosystems are important components of global biodiversity and are experiencing greater rates of biodiversity loss compared to terrestrial and marine ecosystems due to anthropogenic activities and climate change. Despite containing only a small portion of Earth's water, freshwater systems contain nearly 40% of global fish biodiversity. There is a critical need to better understand the ecological drivers and effects of climate change on the distributions and abundance of freshwater fishes in order to inform management, conservation, and climate adaptation strategies. Advances in the application and development of statistical models that predict species abundance and distributions are needed for a better understanding of the relationship between fishes and environmental conditions -- both current and predicted future conditions. This is particularly important under the context of global change since many organisms are experiencing novel habitat conditions due to disturbances, including invasive species, human development, and climate change. My dissertation quantified and developed species distribution and abundance modeling frameworks that accounted for species and spatial dependencies to inform fisheries management and conservation in a changing world. Although the focus of my analyses were on freshwater fish assemblages, the quantitative techniques used and developed throughout my research are applicable to a wide array of organisms. However, freshwater fishes were ideal case studies as they are ecologically and socioeconomically important organisms worldwide, and many of the advanced statistical techniques explored within my research had only been sparsely applied to the modeling efforts of their abundance and distributions. Species distribution models that are both species and spatially dependent are relatively novel advancements in the field of quantitative ecology, and accounting for such dependencies has been shown to improve predictions and reduce uncertainty. My dissertation can be summarized into two main groupings. First, I explored the potential effectiveness and relative importance of these dependencies using stream fish distribution and abundance data. Second, I developed a species and spatially dependent abundance model that incorporated thermal physiology to scale predictions under novel climate conditions, and apply this model to ecologically and socioeconomically important lake fishes in >10,000 north temperate lakes. My first chapter sought to quantify the relative importance of biotic and abiotic factors in landscape-based models of stream fish distributions. Fish distributions are frequently predicted using remotely sensed habitat variables that characterize the adjacent landscape and serve as proxies for instream habitat. Recent advancements in statistical methodology, however, allow for leveraging fish assemblage co-occurrence data when predicting species distributions. This is important because assemblage composition (co-occurrence) likely provides better information about instream habitat compared to landscape-derived metrics and therefore may improve predictions. Joint species distribution models are a popular technique for accounting for species dependencies but are limited in the inference potential of the estimated dependencies. Thus, to better understand the value of using fish assemblage data within stream fish modeling efforts, I fit a conditional random field (CRF) model to quantify the relative importance of fish assemblage co-occurrence, landscape-derived habitat variables, and interactions between these two predictor groups. Predictive performance was also compared against single-species models. The results suggested that species co-occurrence data both improved predictive performance and were more informative than abiotic data alone. These findings illustrate the value of fish assemblage data for landscape-scale species distribution modeling. My second chapter investigated the effectiveness of using non-dendritic spatial dependencies within dendritic stream networks. Species distribution models are vital for understanding the relationships between species and environmental conditions. Predictions from species distribution models are inherently a spatial problem, and research has suggested that accounting for spatial dependencies can improve predictions. For stream fishes, dendritic spatial models may not always be implementable by practitioners for various reasons. I compared non-spatial models with two types of non-dendritic spatial models for predicting stream fish abundance across large, heterogeneous landscapes through a simulation study and cross-validation analysis. The simulation study and cross-validation both suggested that results may vary across parameter estimated, performance metric used to measure predictive accuracy, and by species. Generally, accounting for spatial dependence improved predictive performance for all five species but this wasn't consistent across space. My findings highlight the need to consider multiple approaches for accounting for spatial dependencies, as the best option will vary by research goals and data available. My third chapter developed a novel modeling framework that incorporates a species thermal physiology along with species and spatial dependencies. Predicting the effects of warming temperatures on the abundance and distribution of organisms under future climate scenarios often requires extrapolating species-environment correlations to climatic conditions not currently experienced by a species, which can result in unrealistic predictions. For poikilotherms, incorporating species' thermal physiology to inform extrapolations under novel thermal conditions can result in more realistic predictions. Recently, a novel modeling framework called physiologically guided abundance (PGA) models was developed to provide better predictions of abundance under novel climate conditions. The PGA framework fuses a species' thermal response curve with a correlative niche model, allowing for predictions of abundance to account for both environmental predictors and thermal tolerances. I extend the PGA framework to incorporate both species and spatial dependencies, called the joint species, spatially dependent PGA (jsPGA) model, using the basis function approach. Models that incorporate species and spatial dependencies may improve predictions by capturing correlations present in ecological data that are not accounted for by predictor variables. I then applied the jsPGA to a case study of eight fish species across thousands Minnesota, USA lakes to predict their abundance at the end of the century under different climate change scenarios. Predictions reflected the species' varied thermal physiologies where warm-adapted species were predicted to experience increased abundances, but cold-adapted species were predicted to experience significant decreases in abundance and high probabilities of extirpation. The jsPGA model provides a new tool for predicting changes in abundance, distribution, and extirpation probability of poikilotherms under novel thermal conditions. My fourth chapter applied the jsPGA to a large, regional multi-state lake fish dataset across the Midwestern US. Freshwater ecosystems are experiencing rapid rates of biodiversity loss and habitat degradation due to climate change. However, the effects of climate change on the abundance and distribution of temperate lake fishes is difficult to predict due to heterogeneous thermal environments and species thermal tolerances. While accounting for species-specific thermal physiology, uncertainty in species thermal response curves, variability across global climate models, and lake--watershed characteristics, I predicted species response of eight ecologically and socioeconomically important lake fishes across seven states and more than 10,000 lakes in the Midwestern US. The three most warm tolerant species were predicted to increase in relative abundance at more than half the lakes within our study region. Conversely, the coldwater adapted species was predicted to decrease in abundance across the entire study region with high probabilities of extirpation at nearly 25% of these lakes. This study is one of the first large-scale, lake-specific predictions of the effects of climate change on lake fish assemblages across the Midwest US.
Furthermore, it has wide-ranging implications on the threats that poikilotherms face as thermal habitats continue to warm and become unsuitable within their current distributions. Species distribution models provide ecologists the ability to correlate observed abundance and ranges to potentially important environmental constraints and habitat requirements, which is important for management and conservation efforts under global change. I quantified the effectiveness of extending these models to account for species and spatial dependencies to better understand their effects on these modeling efforts. I also developed a novel modeling framework that incorporates a species' thermal physiology into an abundance model while also accounting for potential species and spatial dependencies unexplained by potentially important environmental factors.
Other Subject(s)
Genre(s)
Dissertation Note
Ph.D. Pennsylvania State University 2024.
Technical Details
The full text of the dissertation is available as an Adobe Acrobat .pdf file ; Adobe Acrobat Reader required to view the file.
View MARC record | catkey: 45625067