Abstract
Environmental variability often has substantial impacts on natural populations and communities through its effects on the performance of individuals. Because organisms' responses to environmental conditions are often nonlinear (e.g., decreasing performance on both sides of an optimal temperature), the mean response is often different from the response in the mean environment. Ye et. al. 2020, proposed testing for the presence of such variance effects on individual or population growth rates by estimating the "Jensen Effect", the difference in average growth rates under varying versus fixed environments, in functional single index models for environmental effects on growth. In this paper, we extend this analysis to effect of environmental variance on reproduction and survival, which have count and binary outcomes. In the standard generalized linear models used to analyze such data the direction of the Jensen Effect is tacitly assumed a priori by the model's link function. Here we extend the methods of Ye et. al. 2020 using a generalized single index model to test whether this assumed direction is contradicted by the data. We show that our test has reasonable power under mild alternatives, but requires sample sizes that are larger than are often available. We demonstrate our methods on a long-term time series of plant ground cover on the Idaho steppe.
Introduction
In ecology, the "Jensen Effect"—derived from Jensen’s Inequality—describes a fundamental principle: if an organism’s response to its environment is nonlinear, the average response in a fluctuating environment will not equal the response to a constant, average environment. For instance, if a plant thrives at a specific temperature but suffers at both higher and lower extremes (a concave relationship), environmental variability will inherently reduce its average performance compared to a stable climate.
While this concept is well-understood, the statistical tools commonly used to measure it often introduce a bias known as "mathematical determinism." Researchers typically employ Generalized Linear Models (GLMs) to handle non-normal data like offspring counts or survival rates. These models rely on fixed "link functions" (such as log or logit) to ensure predictions remain biologically plausible. However, these functions have inherent curvatures; a log link is always concave, meaning the model assumes variability is harmful before the data is even processed. This paper presents a framework using Generalized Single Index Models (GSIMs) to let the data, rather than the model's structure, determine the impact of environmental variance.
Methodology
To address the limitations of standard GLMs, the authors extended Functional Single Index Models into a generalized framework. This approach involves three key components:
- The Single Index: Rather than treating dozens of weather variables (temperature, precipitation, etc.) as independent predictors, the model collapses environmental history into a single scalar value—the "index"—using a weighted sum.
- Non-parametric Link Functions: Instead of a fixed link function, the model uses smoothing splines to estimate the shape of the relationship ($g$) between the environmental index and the biological response. This allows the relationship to be convex, concave, or sigmoidal based on empirical evidence.
- Hypothesis Testing: The authors adapted the "SiZer" (Significant Zero crossing of derivatives) method to create a one-sided test. This determines if the estimated Jensen Effect significantly differs from, or even opposes, the direction predicted by a standard fixed-link model. The Delta method was employed to calculate the variance of the Jensen Effect for rigorous inference.
Empirical Application: The Idaho Steppe
The model was applied to a high-quality, long-term dataset from the U.S. Sheep Experiment Station (USSES) in the Idaho steppe, covering the years 1926 to 1957. The study focused on four key species: the sagebrush (Artemisia tripartita) and three perennial bunchgrasses (Pseudoroegneria spicata, Hesperostipa comata, and Poa secunda). The dataset provided a robust foundation for analysis, including over 1,000 growth transitions and up to 5,410 survival observations for certain species.
Key Results and Insights
The GSIM analysis revealed that the impact of environmental variability is highly species-specific and varies across life stages:
- Growth: For the sagebrush Artemisia, environmental variability is clearly detrimental. The data supported a concave relationship, confirming that stable conditions favor larger plant size.
- Reproduction: The models identified two distinct strategies. The grasses P. spicata and H. comata follow a "Goldilocks" strategy, reproducing best in average conditions; consequently, variability hurts them. In contrast, Artemisia and the grass Poa secunda reproduce most successfully at environmental extremes. For these species, a stable, average climate would be a reproductive disadvantage.
- Survival: Across all species, variability was generally found to be harmful to survival, though the effects were often statistically similar to those predicted by standard logistic models.
These findings suggest a mechanism for niche partitioning. By responding differently to environmental fluctuations, these species can coexist in the same landscape. While one species dominates during "average" years, others take advantage of rare extremes, preventing any single species from outcompeting the others indefinitely.
Limitations and Conclusion
While GSIMs provide a powerful alternative to traditional GLMs, they are "data-hungry." Simulations indicated that the test requires large sample sizes (n ≈ 1000) to maintain high power, especially as environmental noise increases. Furthermore, the smoothing splines used in the model tend to bias results toward linearity, which limited the authors to one-sided tests (checking if the data contradicts the default link function).
In conclusion, this research highlights the danger of assuming the direction of environmental effects through model selection. By utilizing GSIMs, ecologists can more accurately determine whether a changing, more variable climate will act as a threat or a surprising boon to specific species and life stages.
Summary updated