This content will become publicly available on January 1, 2027

Title: High‐frequency monitoring data reveal substantial variability in the intrinsic predictability of ecosystem dynamics
Abstract As ecology becomes a more predictive discipline, identifying the intrinsic predictability, or stochasticity, of ecosystem variables across space and time is needed to help guide the development of ecological models and forecasts. For example, if an ecological time series has high intrinsic predictability, then a high‐performing model should presumably be able to replicate its dynamics. Conversely, if an ecological variable has low intrinsic predictability, then no model—regardless of its performance—will be able to replicate its dynamics. However, despite the proliferation of ecological models and forecasts, the intrinsic predictability of ecological variables remains largely unknown. To bridge this gap, we analyzed a >4‐year time series of high‐frequency sensor data collected from replicate freshwater ecosystems to determine how intrinsic predictability (quantified as permutation entropy) differs among ecological variables, seasons, and ecosystems. We observed greater differences in predictability among ecological variables and days of year than between ecosystems. Although intrinsic predictability was generally low for all variables, it was still significantly higher than white noise, indicating complex yet predictable dynamics. We observed the highest predictability for physical ecosystem variables (e.g., water temperature) and the lowest predictability for biological variables (e.g., phytoplankton biomass), with chemical variables (e.g., dissolved oxygen) intermediate. We observed substantial seasonal differences in predictability among variables: surface water temperature and dissolved organic matter exhibited their highest levels of predictability in autumn, whereas surface chlorophyll and bottom‐layer dissolved oxygen and temperature exhibited highest predictability in summer. Periods of anoxia (low oxygen) were associated with the highest levels of predictability in dissolved oxygen over the time series. Altogether, our analysis highlights how intrinsic predictability data can both guide ecological model development and improve our understanding of how ecological predictability varies across space and time.  more » « less
Award ID(s):
2327030 2318861 1933016 2452117
PAR ID:
10693629
Author(s) / Creator(s):
 ;  ;  ;  
Publisher / Repository:
Wiley
Date Published:
Journal Name:
Ecology
Volume:
107
Issue:
1
ISSN:
0012-9658
Subject(s) / Keyword(s):
Sponsoring Org:
National Science Foundation
More Like this
  1. Near‐term, iterative ecological forecasts can be used to help understand and proactively manage ecosystems. To date, more forecasts have been developed for aquatic ecosystems than other ecosystems worldwide, likely motivated by the pressing need to conserve these essential and threatened ecosystems and increasing the availability of high‐frequency data. Forecasters have implemented many different modeling approaches to forecast freshwater variables, which have demonstrated promise at individual sites. However, a comprehensive analysis of the performance of varying forecast models across multiple sites is needed to understand broader controls on forecast performance. Forecasting challenges (i.e., community‐scale efforts to generate forecasts while also developing shared software, training materials, and best practices) present a useful platform for bridging this gap to evaluate how a range of modeling methods perform across axes of space, time, and ecological systems. Here, we analyzed forecasts from the aquatics theme of the National Ecological Observatory Network (NEON) Forecasting Challenge hosted by the Ecological Forecasting Initiative. Over 100,000 probabilistic forecasts of water temperature and dissolved oxygen concentration for 1–30 days ahead across seven NEON‐monitored lakes were submitted in 2023. We assessed how forecast performance varied among models with different structures, covariates, and sources of uncertainty relative to baseline null models. A similar proportion of forecast models were skillful across both variables (34%–40%), although more individual models outperformed the baseline models in forecasting water temperature (10 models out of 29) than dissolved oxygen (6 models out of 15). These top performing models came from a range of classes and structures. For water temperature, we found that forecast skill degraded with increases in forecast horizons, process‐based models, and models that included air temperature as a covariate generally exhibited the highest forecast performance, and that the most skillful forecasts often accounted for more sources of uncertainty than the lower performing models. The most skillful forecasts were for sites where observations were most divergent from historical conditions (resulting in poor baseline model performance). Overall, the NEON Forecasting Challenge provides an exciting opportunity for a model intercomparison to learn about the relative strengths of a diverse suite of models and advance our understanding of freshwater ecosystem predictability. 
    more » « less
  2. Abstract Dissolved organic matter (DOM) plays an important role in aquatic carbon cycling and is a valuable metric of ecosystem functioning and water quality in freshwater ecosystems. Despite its importance for biogeochemical cycling and water quality, no near‐term iterative forecasts have previously been developed for freshwater DOM concentrations. To advance both our understanding of freshwater DOM dynamics and management, we developed 1–34 days‐ahead forecasts of fluorescent DOM (fDOM) in three drinking water reservoirs. These temperate reservoirs are co‐located in Virginia, USA and experience variable DOM dynamics (range: 5–27 QSU (quinine sulfate units)). We developed six different forecasting models to predict fDOM in each reservoir. Three models were time series models based on forecasted drivers (water temperature and meteorology) that were updated daily from high‐frequency fDOM sensors. The other forecast models included a neural network machine learning model and two baseline reference models (day‐of‐year mean and persistence). Altogether, our forecasts were able to capture observed dynamics over a year in all three reservoirs, with one time series model outperforming the baseline models across the full 34‐day forecast horizon. Aggregated across reservoirs and models over a year, forecast RMSE increased from 0.7 to 4.1 QSU over the 1–34 days‐ahead forecast horizon. Forecast skill varied substantially across seasons, with greatest accuracy in the spring and winter compared to the summer and fall across reservoirs. These forecasts can help improve our understanding of the predictability of DOM and inform management in freshwater ecosystems as carbon dynamics become more variable due to global change. 
    more » « less
  3. Abstract Many ecological systems dominated by stochastic dynamics can produce complex time series that inherently limit forecast accuracy. The ‘intrinsic predictability’ of these systems can be approximated by a time series complexity metric called weighted permutation entropy (WPE). While WPE is a useful metric to gauge forecast performance prior to model building, it is sensitive to noise and may be biased depending on the length of the time series. Here, we introduce a simple randomized permutation test (rWPE) to assess whether a time series is intrinsically more predictable than white noise.We apply rWPE to both simulated and empirical data to assess its performance and usefulness. To do this, we simulate population dynamics under various scenarios, including a linear trend, chaotic, periodic and equilibrium dynamics. We further test this approach with observed abundance time series for 932 species across four orders of animals from the Global Population Dynamics Database. Finally, using Adélie (Pygoscelis adeliae) and emperor penguin (Aptenodytes forsteri) time series as case studies, we demonstrate the application of rWPE to multiple populations for a single species.We show that rWPE can determine whether a system is significantly more predictable than white noise, even with time series as short as 10 years that show an apparent trend under biologically realistic stochasticity levels. Additionally, rWPE has statistical power close to 100% when time series are at least 30 time steps long and show chaotic or periodic dynamics. Power decreases to ~10% under equilibrium dynamics, irrespective of time series length. Among four classes of animal taxa, mammals have the highest relative frequency (28%) of time series that are both longer than 30 time steps and indistinguishable from white noise in terms of complexity, followed by insects (16%), birds (16%) and bony fishes (11%).rWPE is a straightforward and useful method widely applicable to any time series, including short ones. By informing forecasters of the inherent limitations to a system's predictability, it can guide a modeller's expectations for forecast performance. 
    more » « less
  4. null (Ed.)
    Short-term forecasts of nonlinear dynamics are important for risk-assessment studies and to inform sustainable decision-making for physical, biological and financial problems, among others. Generally, the accuracy of short-term forecasts depends upon two main factors: the capacity of learning algorithms to generalize well on unseen data and the intrinsic predictability of the dynamics. While generalization skills of learning algorithms can be assessed with well-established methods, estimating the predictability of the underlying nonlinear generating process from empirical time series remains a big challenge. Here, we show that, in changing environments, the predictability of nonlinear dynamics can be associated with the time-varying stability of the system with respect to smooth changes in model parameters, i.e. its local structural stability. Using synthetic data, we demonstrate that forecasts from locally structurally unstable states in smoothly changing environments can produce significantly large prediction errors, and we provide a systematic methodology to identify these states from data. Finally, we illustrate the practical applicability of our results using an empirical dataset. Overall, this study provides a framework to associate an uncertainty level with short-term forecasts made in smoothly changing environments. 
    more » « less
  5. ABSTRACT With novel ocean conditions rapidly appearing as the result of climate change, basing decisions about fisheries and other ocean activities on historical conditions is no longer tenable. There is instead a widespread need for shifting ecological baselines to more effectively guide decisions into the future. What has not been as widely recognised is that the relevant timescales differ substantially across ocean‐related decisions, from lead times of hours to decades depending on the decision being made, and that this range necessitates a matching range of ecological forecast products across similar timescales. At the moment, a predictability gap exists at intermediate timescales, from multi‐annual to multi‐decadal forecasts. Because most fisheries and many other ocean activities rely on biological conditions like fish abundance or distribution, the ecological inertia of organismal growth, generational turnover, movement, and food web dynamics can help push ecological forecasts further across this gap. To realise this potential for more effective and usable ecological forecasts, coordinated research and implementation at the intersection of biology, climate science, social science, and decision‐making is needed. These efforts will be critical for forecasting shifting ecosystem baselines and sustaining fisheries, ocean ecosystems, and the ocean economy in the coming decades of rapid change. 
    more » « less