An official website of the United States government
Abstract Society increasingly demands accurate predictions of complex ecosystem processes under novel conditions to address environmental challenges. However, obtaining the process‐level knowledge required to do so does not necessarily align with the burgeoning use in ecology of correlative model selection criteria, such as Akaike information criterion. These criteria select models based on their ability to reproduce outcomes, not on their ability to accurately represent causal effects. Causal understanding does not require matching outcomes, but rather involves identifying model forms and parameter values that accurately describe processes. We contend that researchers can arrive at incorrect conclusions about cause‐and‐effect relationships by relying on information criteria. We illustrate via a specific example that inference extending beyond prediction into causality can be seriously misled by information‐theoretic evidence. Finally, we identify a solution space to bridge the gap between the correlative inference provided by model selection criteria and a process‐based understanding of ecological systems.
more »
« less
Denitrification and the challenge of scaling microsite knowledge to the globe
Abstract Our knowledge of microbial processes—who is responsible for what, the rates at which they occur, and the substrates consumed and products produced—is imperfect for many if not most taxa, but even less is known about how microsite processes scale to the ecosystem and thence the globe. In both natural and managed environments, scaling links fundamental knowledge to application and also allows for global assessments of the importance of microbial processes. But rarely is scaling straightforward: More often than not, process rates in situ are distributed in a highly skewed fashion, under the influence of multiple interacting controls, and thus often difficult to sample, quantify, and predict. To date, quantitative models of many important processes fail to capture daily, seasonal, and annual fluxes with the precision needed to effect meaningful management outcomes. Nitrogen cycle processes are a case in point, and denitrification is a prime example. Statistical models based on machine learning can improve predictability and identify the best environmental predictors but are—by themselves—insufficient for revealing process‐level knowledge gaps or predicting outcomes under novel environmental conditions. Hybrid models that incorporate well‐calibrated process models as predictors for machine learning algorithms can provide both improved understanding and more reliable forecasts under environmental conditions not yet experienced. Incorporating trait‐based models into such efforts promises to improve predictions and understanding still further, but much more development is needed.
more »
« less
- Award ID(s):
- 2224712
- PAR ID:
- 10470587
- Publisher / Repository:
- Wiley
- Date Published:
- Journal Name:
- mLife
- Volume:
- 2
- Issue:
- 3
- ISSN:
- 2770-100X
- Page Range / eLocation ID:
- 229 to 238
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Plants have a surprising capacity to alter their environmental conditions to create adequate niches for survival and stress tolerance. This process of environmental transformation, commonly referred to as “extended phenotypes” or “niche construction”, has historically been studied in the domain of ecology, but this is a process that is pervasive across the plant kingdom. Furthermore, research is beginning to show that plants’ extended phenotypes shape the assembly and function of closely associated microbial communities. Incorporation and understanding the role that plant-extended phenotypes play in agriculture may offer novel, bioinspired methods to manage our arable soil microbiomes. Here, we review the challenges agriculture faces, the plant extended phenotypes we know to shape the microbiome, and the potential utilization of this knowledge to improve the environmental impact of agriculture. Understanding how plant extended phenotypes shape microbial communities could be a key to creating a sustainable future with both plants and microbiomes in consideration.more » « less
-
Projecting climate change is a generalization problem: We extrapolate the recent past using physical models across past, present, and future climates. Current climate models require representations of processes that occur at scales smaller than model grid size, which have been the main source of model projection uncertainty. Recent machine learning (ML) algorithms hold promise to improve such process representations but tend to extrapolate poorly to climate regimes that they were not trained on. To get the best of the physical and statistical worlds, we propose a framework, termed “climate-invariant” ML, incorporating knowledge of climate processes into ML algorithms, and show that it can maintain high offline accuracy across a wide range of climate conditions and configurations in three distinct atmospheric models. Our results suggest that explicitly incorporating physical knowledge into data-driven models of Earth system processes can improve their consistency, data efficiency, and generalizability across climate regimes.more » « less
-
Abstract Ecosystem functions and services are under threat from anthropogenic global change at a planetary scale. Microorganisms are the dominant drivers of nearly all ecosystem functions and therefore ecosystem-scale responses are dependent on responses of resident microbial communities. However, the specific characteristics of microbial communities that contribute to ecosystem stability under anthropogenic stress are unknown. We evaluated bacterial drivers of ecosystem stability by generating wide experimental gradients of bacterial diversity in soils, applying stress to the soils, and measuring responses of several microbial-mediated ecosystem processes, including C and N cycling rates and soil enzyme activities. Some processes (e.g., C mineralization) exhibited positive correlations with bacterial diversity and losses of diversity resulted in reduced stability of nearly all processes. However, comprehensive evaluation of all potential bacterial drivers of the processes revealed that bacterial α diversity per se was never among the most important predictors of ecosystem functions. Instead, key predictors included total microbial biomass, 16S gene abundance, bacterial ASV membership, and abundances of specific prokaryotic taxa and functional groups (e.g., nitrifying taxa). These results suggest that bacterial α diversity may be a useful indicator of soil ecosystem function and stability, but that other characteristics of bacterial communities are stronger statistical predictors of ecosystem function and better reflect the biological mechanisms by which microbial communities influence ecosystems. Overall, our results provide insight into the role of microorganisms in supporting ecosystem function and stability by identifying specific characteristics of bacterial communities that are critical for understanding and predicting ecosystem responses to global change.more » « less
-
null (Ed.)Abstract Understanding the physical drivers of seasonal hydroclimatic variability and improving predictive skill remains a challenge with important socioeconomic and environmental implications for many regions around the world. Physics-based deterministic models show limited ability to predict precipitation as the lead time increases, due to imperfect representation of physical processes and incomplete knowledge of initial conditions. Similarly, statistical methods drawing upon established climate teleconnections have low prediction skill due to the complex nature of the climate system. Recently, promising data-driven approaches have been proposed, but they often suffer from overparameterization and overfitting due to the short observational record, and they often do not account for spatiotemporal dependencies among covariates (i.e., predictors such as sea surface temperatures). This study addresses these challenges via a predictive model based on a graph-guided regularizer that simultaneously promotes similarity of predictive weights for highly correlated covariates and enforces sparsity in the covariate domain. This approach both decreases the effective dimensionality of the problem and identifies the most predictive features without specifying them a priori. We use large ensemble simulations from a climate model to construct this regularizer, reducing the structural uncertainty in the estimation. We apply the learned model to predict winter precipitation in the southwestern United States using sea surface temperatures over the entire Pacific basin, and demonstrate its superiority compared to other regularization approaches and statistical models informed by known teleconnections. Our results highlight the potential to combine optimally the space–time structure of predictor variables learned from climate models with new graph-based regularizers to improve seasonal prediction.more » « less
