Iterative near-term ecological forecasting: Needs, opportunities, and challenges

The emergent interactions that govern biodiversity change

Observational studies have not yet shown that environmental variables can explain pervasive nonlinear patterns of species abundance, because those patterns could result from (indirect) interactions with other species (e.g., competition), and models only estimate direct responses. The experiments that could extract these indirect effects at regional to continental scales are not feasible. Here, a biophysical approach quantifies environment- species interactions (ESI) that govern community change from field data. Just as species interactions depend on population abundances, so too do the effects of environment, as when drought is amplified by competition. By embedding dynamic ESI within framework that admits data gathered on different scales, we quantify responses that are induced indirectly through other species, including probabilistic uncertainty in parameters, model specification, and data. Simulation demonstrates that ESI are needed for accurate interpretation. Analysis demonstrates how nonlinear responses arise even when their direct responses to environment are linear. Applications to experimental lakes and the Breeding Bird Survey (BBS) yield contrasting estimates of ESI. In closed lakes, interactions involving phytoplankton and their zooplankton grazers play a large role. By contrast, ESI are weak in BBS, as expected where year-to-year movement degrades the link between local population growth and species interactions. In both cases, nonlinear responses to environmental gradients are induced by interactions between species. Stability analysis indicates stability in the closed-system lakes and instability in BBS. The probabilistic framework has direct application to conservation planning that must weigh risk assessments for entire habitats and communities against competing interests.

Forecasting unprecedented ecological fluctuations

Forecasting 'Black Swan' events in ecosystems is an important but challenging task. Many ecosystems display aperiodic fluctuations in species abundance spanning orders of magnitude in scale, which have vast environmental and economic impact. Empirical evidence and theoretical analyses suggest that these dynamics are in a regime where system nonlinearities limit accurate forecasting of unprecedented events due to poor extrapolation of historical data to unsampled states. Leveraging increasingly available long-term high-frequency ecological tracking data, we analyze multiple natural and experimental ecosystems (marine plankton, intertidal mollusks, and deciduous forest), and recover hidden linearity embedded in universal 'scaling laws' of species dynamics. We then develop a method using these scaling laws to reduce data dependence in ecological forecasting and accurately predict extreme events beyond the span of historical observations in diverse ecosystems.

Sea ice predicts long-term trends in Adélie penguin population growth, but not annual fluctuations: Results from a range-wide multiscale analysis

Understanding the scales at which environmental variability affects populations is critical for projecting population dynamics and species distributions in rapidly changing environments. Here we used a multilevel Bayesian analysis of range-wide survey data for Adélie penguins to characterize multidecadal and annual effects of sea ice on population growth. We found that mean sea ice concentration at breeding colonies (i.e., "prevailing" environmental conditions) had robust nonlinear effects on multidecadal population trends and explained over 85% of the variance in mean population growth rates among sites. In contrast, despite considerable year-to-year fluctuations in abundance at most breeding colonies, annual sea ice fluctuations often explained less than 10% of the temporal variance in population growth rates. Our study provides an understanding of the spatially and temporally dynamic environmental factors that define the range limits of Adélie penguins, further establishing this iconic marine predator as a true sea ice obligate and providing a firm basis for projection under scenarios of future climate change. Yet, given the weak effects of annual sea ice relative to the large unexplained variance in year-to-year growth rates, the ability to generate useful short-term forecasts of Adélie penguin breeding abundance will be extremely limited. Our approach provides a powerful framework for linking short- and longer term population processes to environmental conditions that can be applied to any species, facilitating a richer understanding of ecological predictability and sensitivity to global change.

Earth and field observations underpin metapopulation dynamics in complex landscapes: Near-term study on carabids

Understanding risks to biodiversity requires predictions of the spatial distribution of species adapting to changing ecosystems and, to that end, Earth observations integrating field surveys prove essential as they provide key numbers for assessing landscape-wide biodiversity scenarios. Here, we develop, and apply to a relevant case study, a method suited to merge Earth/field observations with spatially explicit stochastic metapopulation models to study the near-term ecological dynamics of target species in complex terrains. Our framework incorporates the use of species distribution models for a reasoned estimation of the initial presence of the target species and accounts for imperfect and incomplete detection of the species presence in the study area. It also uses a metapopulation fitness function derived from Earth observation data subsuming the ecological niche of the target species. This framework is applied to contrast occupancy of two species of carabids (Pterostichus flavofemoratus, Carabus depressus) observed in the context of a large ecological monitoring program carried out within the Gran Paradiso National Park (GPNP, Italy). Results suggest that the proposed framework may indeed exploit the hallmarks of spatially explicit ecological approaches and of remote Earth observations. The model reproduces well the observed in situ data. Moreover, it projects in the near term the two species' presence both in space and in time, highlighting the features of the metapopulation dynamics of colonization and extinction, and their expected trends within verifiable timeframes.

Forecasting intensifying disturbance effects on coral reefs

Anticipating future changes of an ecosystem's dynamics requires knowledge of how its key communities respond to current environmental regimes. The Great Barrier Reef (GBR) is under threat, with rapid changes of its reef-building hard coral (HC) community structure already evident across broad spatial scales. While several underlying relationships between HC and multiple disturbances have been documented, responses of other benthic communities to disturbances are not well understood. Here we used statistical modelling to explore the effects of broad-scale climate-related disturbances on benthic communities to predict their structure under scenarios of increasing disturbance frequency. We parameterized a multivariate model using the composition of benthic communities estimated by 145,000 observations from the northern GBR between 2012 and 2017. During this time, surveyed reefs were variously impacted by two tropical cyclones and two heat stress events that resulted in extensive HC mortality. This unprecedented sequence of disturbances was used to estimate the effects of discrete versus interacting disturbances on the compositional structure of HC, soft corals (SC) and algae. Discrete disturbances increased the prevalence of algae relative to HC while the interaction between cyclones and heat stress was the main driver of the increase in SC relative to algae and HC. Predictions from disturbance scenarios included relative increases in algae versus SC that varied by the frequency and types of disturbance interactions. However, high uncertainty of compositional changes in the presence of several disturbances shows that responses of algae and SC to the decline in HC needs further research. Better understanding of the effects of multiple disturbances on benthic communities as a whole is essential for predicting the future status of coral reefs and managing them in the light of new environmental regimes. The approach we develop here opens new opportunities for reaching this goal.

Low-cost observations and experiments return a high value in plant phenology research

Plant ecologists in the Anthropocene are tasked with documenting, interpreting, and predicting how plants respond to environmental change. Phenology, the timing of seasonal biological events including leaf-out, flowering, fruiting, and leaf senescence, is among the most visible and oft-recorded facets of plant ecology. Climate-driven shifts in plant phenology can alter reproductive success, interspecific competition, and trophic interactions. Low-cost phenology research, including observational records and experimental manipulations, is fundamental to our understanding of both the mechanisms and effects of phenological change in plant populations, species, and communities. Traditions of local-scale botanical phenology observations and data leveraged from written records and natural history collections provide the historical context for recent observations of changing phenologies. New technology facilitates expanding the spatial, taxonomic, and human interest in this research by combining contemporary field observations by researchers and open access community science (e.g., USA National Phenology Network) and available climate data. Established experimental techniques, such as twig cutting and common garden experiments, are low-cost methods for studying the mechanisms and drivers of plant phenology, enabling researchers to observe phenological responses under novel environmental conditions. We discuss the strengths, limitations, potential hidden costs (i.e., volunteer and student labor), and promise of each of these methods for addressing emerging questions in plant phenology research. Applied thoughtfully, economically, and creatively, many low-cost approaches offer novel opportunities to fill gaps in our geographic, taxonomic, and mechanistic understanding of plant phenology worldwide.

Forecasting semi-arid biome shifts in the Anthropocene

Shrub encroachment, forest decline and wildfires have caused large-scale changes in semi-arid vegetation over the past 50 years. Climate is a primary determinant of plant growth in semi-arid ecosystems, yet it remains difficult to forecast large-scale vegetation shifts (i.e. biome shifts) in response to climate change. We highlight recent advances from four conceptual perspectives that are improving forecasts of semi-arid biome shifts. Moving from small to large scales, first, tree-level models that simulate the carbon costs of drought-induced plant hydraulic failure are improving predictions of delayed-mortality responses to drought. Second, tracer-informed water flow models are improving predictions of species coexistence as a function of climate. Third, new applications of ecohydrological models are beginning to simulate small-scale water movement processes at large scales. Fourth, remotely-sensed measurements of plant traits such as relative canopy moisture are providing early-warning signals that predict forest mortality more than a year in advance. We suggest that a community of researchers using modeling approaches (e.g. machine learning) that can integrate these perspectives will rapidly improve forecasts of semi-arid biome shifts. Better forecasts can be expected to help prevent catastrophic changes in vegetation states by identifying improved monitoring approaches and by prioritizing high-risk areas for management.

Predicting population change from models based on habitat availability and utilization

The need to understand the impacts of land management for conservation, agriculture and disease prevention are driving demand for new predictive ecology approaches that can reliably forecast future changes in population size. Currently, although the link between habitat composition and animal population dynamics is undisputed, its function has not been quantified in a way that enables accurate prediction of population change in nature. Here, using 12 house sparrow colonies as a proof-of-concept, we apply recent theoretical advances to predict population growth or decline from detailed data on habitat composition and habitat selection. We show, for the first time, that statistical population models using derived covariates constructed from parametric descriptions of habitat composition and habitat selection can explain an impressive 92% of observed population variation. More importantly, they provide excellent predictive power under cross-validation, anticipating 81% of variability in population change. These models may be embedded in readily available generalized linear modelling frameworks, allowing their rapid application to field systems. Furthermore, we use optimization on our sample of sparrow colonies to demonstrate how such models, linking populations to their habitats, permit the design of practical and environmentally sound habitat manipulations for managing populations.

Data-driven modelling and spatial complexity supports heterogeneity-based integrative management for eliminating Simulium neavei-transmitted river blindness

Concern is emerging regarding the challenges posed by spatial complexity for modelling and managing the area-wide elimination of parasitic infections. While this has led to calls for applying heterogeneity-based approaches for addressing this complexity, questions related to spatial scale, the discovery of locally-relevant models, and its interaction with options for interrupting parasite transmission remain to be resolved. We used a data-driven modelling framework applied to infection data gathered from different monitoring sites to investigate these questions in the context of understanding the transmission dynamics and efforts to eliminate Simulium neavei- transmitted onchocerciasis, a macroparasitic disease that causes river blindness in Western Uganda and other regions of Africa. We demonstrate that our Bayesian-based data-model assimilation technique is able to discover onchocerciasis models that reflect local transmission conditions reliably. Key management variables such as infection breakpoints and required durations of drug interventions for achieving elimination varied spatially due to site-specific parameter constraining; however, this spatial effect was found to operate at the larger focus level, although intriguingly including vector control overcame this variability. These results show that data-driven modelling based on spatial datasets and model-data fusing methodologies will be critical to identifying both the scale-dependent models and heterogeneity-based options required for supporting the successful elimination of S. neavei-borne onchocerciasis.

A statistical forecasting approach to metapopulation viability analysis

Conservation of at-risk species is aided by reliable forecasts of the consequences of environmental change and management actions on population viability. Forecasts from conventional population viability analysis (PVA) are made using a two-step procedure in which parameters are estimated, or elicited from expert opinion, and then plugged into a stochastic population model without accounting for parameter uncertainty. Recently developed statistical PVAs differ because forecasts are made conditional on models fitted to empirical data. The statistical forecasting approach allows for uncertainty about parameters, but it has rarely been applied in metapopulation contexts where spatially explicit inference is needed about colonization and extinction dynamics and other forms of stochasticity that influence metapopulation viability. We conducted a statistical metapopulation viability analysis (MPVA) using 11 yr of data on the federally threatened Chiricahua leopard frog (Lithobates chiricahuensis) to forecast responses to landscape heterogeneity, drought, environmental stochasticity, and management. We evaluated several future environmental scenarios and pond restoration options designed to reduce extinction risk. Forecasts over a 50-yr time horizon indicated that metapopulation extinction risk was <4% for all scenarios, but uncertainty was high. Without pond restoration, extinction risk is forecasted to be 3.9% (95% CI 0-37%) by year 2066. Restoring six ponds by increasing their hydroperiod reduced extinction risk to <1% and greatly reduced uncertainty (95% CI 0-2%). Our results suggest that managers can mitigate the impacts of drought and environmental stochasticity on metapopulation viability by maintaining ponds that hold water throughout the year and keeping them free of invasive predators. Our study illustrates the utility of the spatially explicit statistical forecasting approach to MPVA in conservation planning efforts.

Spatial and temporal turnover of soil microbial communities is not linked to function in a primary tropical forest

The spatial and temporal linkages between turnover of soil microbial communities and their associated functions remain largely unexplored in terrestrial ecosystems. Yet defining these relationships and how they vary across ecosystems and microbial lineages is key to incorporating microbial communities into ecological forecasts and ecosystem models. To define linkages between turnover of soil bacterial and fungal communities and their function we sampled fungal and bacterial composition, abundance, and enzyme activities across a 3-ha area of wet tropical primary forest over 2 yr. We show that fungal and bacterial communities both exhibited temporal turnover, but turnover of both groups was much lower than in temperate ecosystems. Turnover over time was driven by gain and loss of microbial taxa and not changes in abundance of individual species present in multiple samples. Only fungi varied over space with idiosyncratic variation that did not increase linearly with distance among sampling locations. Only phosphorus-acquiring enzyme activities were linked to shifts in septate, decomposer fungal abundance; no enzymes were affected by composition or diversity of fungi or bacteria. Although temporal and spatial variation in composition was appreciable, because turnover of microbial communities did not alter the functional repertoire of decomposing enzymes, functional redundancy among taxa may be high in this ecosystem. Slow temporal turnover of tropical soil microbial communities and large functional redundancy suggests that shifts in abundance of particular functional groups may capture ecosystem function more accurately than composition in these heterogeneous ecosystems.

Are laboratory growth rate experiments relevant to explaining bloom-forming cyanobacteria distributions at global scale?

Predicting algal population dynamics using models informed by experimental data has been used as a strategy to inform the management and control of harmful cyanobacterial blooms. We selected toxic bloom-forming species Microcystis spp. and Raphidiopsis raciborskii (basionym Cylindrospermopsis raciborskii) for further examination as they dominate in 78 % and 17 %, respectively, of freshwater cyanobacterial blooms (cyanoHABs) reported globally over the past 30 years. Field measurements of cyanoHABs are typically based on biomass accumulation, but laboratory experiments typically measure growth rates, which are an important variable in cyanoHAB models. Our objective was to determine the usefulness of laboratory studies of these cyanoHAB growth rates for simulating the species dominance at a global scale. We synthesized growth responses of M. aeruginosa and R. raciborskii from 20 and 16 culture studies, respectively, to predict growth rates as a function of two environmental variables, light and temperature. Predicted growth rates of R. raciborskii exceeded those of M. aeruginosa at temperatures ≳ 25 °C and light intensities ≳ 150 μmol photons m^-2 s^-1. Field observations of biomass accumulation, however, show that M. aeruginosa dominates over R. raciborskii, irrespective of climatic zones. The mismatch between biomass accumulation measured in the field, and what is predicted from growth rate measured in the laboratory, hinders effective use of culture studies to predict formation of cyanoHABs in the natural environment. The usefulness of growth rates measured may therefore be limited, and field experiments should instead be designed to examine key physiological attributes such as colony formation, buoyancy regulation and photoadaptation. Improving prediction of cyanoHABs in a changing climate requires a more effective integration of field and laboratory approaches, and an explicit consideration of strain-level variability.

Informing management decisions for ecological networks, using dynamic models calibrated to noisy time-series data

Well-intentioned environmental management can backfire, causing unforeseen damage. To avoid this, managers and ecologists seek accurate predictions of the ecosystem-wide impacts of interventions, given small and imprecise datasets, which is an incredibly difficult task. We generated and analysed thousands of ecosystem population time series to investigate whether fitted models can aid decision-makers to select interventions. Using these time-series data (sparse and noisy datasets drawn from deterministic Lotka-Volterra systems with two to nine species, of known network structure), dynamic model forecasts of whether a species' future population will be positively or negatively affected by rapid eradication of another species were correct > 70% of the time. Although 70% correct classifications is only slightly better than an uninformative prediction (50%), this classification accuracy can be feasibly improved by increasing monitoring accuracy and frequency. Our findings suggest that models may not need to produce well-constrained predictions before they can inform decisions that improve environmental outcomes.

Automated data-intensive forecasting of plant phenology throughout the United States

Phenology, the timing of cyclical and seasonal natural phenomena such as flowering and leaf out, is an integral part of ecological systems with impacts on human activities like environmental management, tourism, and agriculture. As a result, there are numerous potential applications for actionable predictions of when phenological events will occur. However, despite the availability of phenological data with large spatial, temporal, and taxonomic extents, and numerous phenology models, there have been no automated species-level forecasts of plant phenology. This is due in part to the challenges of building a system that integrates large volumes of climate observations and forecasts, uses that data to fit models and make predictions for large numbers of species, and consistently disseminates the results of these forecasts in interpretable ways. Here, we describe a new near-term phenology-forecasting system that makes predictions for the timing of budburst, flowers, ripe fruit, and fall colors for 78 species across the United States up to 6 months in advance and is updated every four days. We use the lessons learned in developing this system to provide guidance developing large-scale near-term ecological forecast systems more generally, to help advance the use of automated forecasting in ecology.

Constraining estimates of terrestrial carbon uptake: new opportunities using long-term satellite observations and data assimilation

The response of terrestrial carbon uptake to increasing atmospheric [CO₂ ], that is the CO₂ fertilization effect (CFE), remains a key area of uncertainty in carbon cycle science. Here we provide a perspective on how satellite observations could be better used to understand and constrain CFE. We then highlight data assimilation (DA) as an effective way to reconcile different satellite datasets and systematically constrain carbon uptake trends in Earth System Models. As a proof-of-concept, we show that joint DA of multiple independent satellite datasets reduced model ensemble error by better constraining unobservable processes and variables, including those directly impacted by CFE. DA of multiple satellite datasets offers a powerful technique that could improve understanding of CFE and enable more accurate forecasts of terrestrial carbon uptake.

Conservation palaeobiology and the shape of things to come

Conservation decision-making is a forward-looking process that involves choices among alternative images of how the future will unfold. Scenarios, easily understood as stories about plausible futures, are emerging as a powerful approach used by the conservation community to define a range of socio-ecological futures when standard, predictive modelling approaches to decision-making are inappropriate, providing a framework for making robust decisions under uncertainties. Conservation palaeobiologists can help the conservation community imagine the future. The utility of the past centres on orienting us to the present-grounding the future in the realm of what is plausible-by providing context against which to think about future scenarios, which may help stakeholders and decision-makers to develop a new mental map of a conservation problem, inspiring our intentions and moving us purposefully toward a desirable tomorrow. This article is part of a discussion meeting issue 'The past is a foreign country: how much can the fossil record actually inform conservation?'

Remote sensing and modeling fusion for investigating the ecosystem water-carbon coupling processes

The water and carbon cycles are tightly linked and play a key role in the material and energy flows between terrestrial ecosystems and the atmosphere, but the interactions of water and carbon cycles are not quite clear. The global climate change and intensive human activities could also complicate the water and carbon coupling processes. Better understanding the coupled water-carbon cycles and their spatiotemporal evolution can inform management and decision-making efforts regarding carbon uptake, food production, water resources, and climate change. The integration of remote sensing and numeric modeling is an attractive approach to address the challenge. Remote sensing can provide extensive data for a number of variables at regional scale and support models, whereas process-based modeling can facilitate investigating the processes that remote sensing cannot well handle (e.g., below-ground and lateral material movement) and backcast/forecast the impacts of environmental change. Over the past twenty years, an increasing number of studies using a variety of remote sensing products together with numeric models have examined the water-carbon interactions. This article reviewed the methodologies for integrating remote sensing data into these models and the modeling of water-carbon coupling processes. We first summarized the major remote sensing datasets and models used for studying the coupled water-carbon cycles. We then provided an overview of the methods for integrating remote sensing data into water-carbon models, and discussed their strengths and challenges. We also prospected the development of potential new remote sensing datasets, modeling methods, and their potential applications in the field of eco-hydrology.

Accumulating evidence in ecology: Once is not enough

Many published studies in ecological science are viewed as stand-alone investigations that purport to provide new insights into how ecological systems behave based on single analyses. But it is rare for results of single studies to provide definitive results, as evidenced in current discussions of the "reproducibility crisis" in science. The key step in science is the comparison of hypothesis-based predictions with observations, where the predictions are typically generated by hypothesis-specific models. Repeating this step allows us to gain confidence in the predictive ability of a model, and its corresponding hypothesis, and thus to accumulate evidence and eventually knowledge. This accumulation may occur via an ad hoc approach, via meta-analyses, or via a more systematic approach based on the anticipated evolution of an information state. We argue the merits of this latter approach, provide an example, and discuss implications for designing sequences of studies focused on a particular question. We conclude by discussing current data collection programs that are preadapted to use this approach and argue that expanded use would increase the rate of learning in ecology, as well as our confidence in what is learned.

What Have Long-Term Field Studies Taught Us About Population Dynamics?

Long-term studies have been crucial to the advancement of population biology, especially our understanding of population dynamics. We argue that this progress arises from three key characteristics of long-term research. First, long-term data are necessary to observe the heterogeneity that drives most population processes. Second, long-term studies often inherently lead to novel insights. Finally, long-term field studies can serve as model systems for population biology, allowing for theory and methods to be tested under well-characterized conditions. We illustrate these ideas in three long-term field systems that have made outsized contributions to our understanding of population ecology, evolution, and conservation biology. We then highlight three emerging areas to which long-term field studies are well positioned to contribute in the future: ecological forecasting, genomics, and macrosystems ecology. Overcoming the obstacles associated with maintaining long-term studies requires continued emphasis on recognizing the benefits of such studies to ensure that long-term research continues to have a substantial impact on elucidating population biology.

Accelerating river blindness elimination by supplementing MDA with a vegetation "slash and clear" vector control strategy: a data-driven modeling analysis

Attention is increasingly focusing on how best to accelerate progress toward meeting the WHO's 2030 goals for neglected tropical diseases (NTDs). For river blindness, a major NTD targeted for elimination, there is a long history of using vector control to suppress transmission, but traditional larvicide-based approaches are limited in their utility. One innovative and sustainable approach, "slash and clear", involves clearing vegetation from breeding areas, and recent field trials indicate that this technique very effectively reduces the biting density of Simulium damnosum s.s. In this study, we use a Bayesian data-driven mathematical modeling approach to investigate the potential impact of this intervention on human onchocerciasis infection. We developed a novel "slash and clear" model describing the effect of the intervention on seasonal black fly biting rates and coupled this with our population dynamics model of Onchocerca volvulus transmission. Our results indicate that supplementing annual drug treatments with "slash and clear" can significantly accelerate the achievement of onchocerciasis elimination. The efficacy of the intervention is not very sensitive to the timing of implementation, and the impact is meaningful even if vegetation is cleared only once per year. As such, this community-driven technique will represent an important option for achieving and sustaining O. volvulus elimination.

Projected urban growth in the southeastern USA puts small streams at risk

Future land-use development has the potential to profoundly affect the health of aquatic ecosystems in the coming decades. We developed regression models predicting the loss of sensitive fish (R² = 0.39) and macroinvertebrate (R² = 0.64) taxa as a function of urban and agricultural land uses and applied them to projected urbanization of the rapidly urbanizing Piedmont ecoregion of the southeastern USA for 2030 and 2060. The regression models are based on a 2014 investigation of water quality and ecology of 75 wadeable streams across the region. Based on these projections, stream kilometers experiencing >50% loss of sensitive fish and invertebrate taxa will nearly quadruple to 19,500 and 38,950 km by 2060 (16 and 32% of small stream kilometers in the region), respectively. Uncertainty was assessed using the 20 and 80% probability of urbanization for the land-use projection model and using the 95% confidence intervals for the regression models. Adverse effects on stream health were linked to elevated concentrations of contaminants and nutrients, low dissolved oxygen, and streamflow alteration, all associated with urbanization. The results of this analysis provide a warning of potential risks from future urbanization and perhaps some guidance on how those risks might be mitigated.

Quantifying trends and uncertainty in prehistoric forest composition in the upper Midwestern United States

Forest ecosystems in eastern North America have been in flux for the last several thousand years, well before Euro‐American land clearance and the 20th‐century onset of anthropogenic climate change. However, the magnitude and uncertainty of prehistoric vegetation change have been difficult to quantify because of the multiple ecological, dispersal, and sedimentary processes that govern the relationship between forest composition and fossil pollen assemblages. Here we extend STEPPS, a Bayesian hierarchical spatiotemporal pollen–vegetation model, to estimate changes in forest composition in the upper Midwestern United States from about 2,100 to 300 yr ago. Using this approach, we find evidence for large changes in the relative abundance of some species, and significant changes in community composition. However, these changes took place against a regional background of changes that were small in magnitude or not statistically significant, suggesting complexity in the spatiotemporal patterns of forest dynamics. The single largest change is the infilling of Tsuga canadensis in northern Wisconsin over the past 2,000 yr. Despite range infilling, the range limit of T. canadensis was largely stable, with modest expansion westward. The regional ecotone between temperate hardwood forests and northern mixed hardwood/conifer forests shifted southwestward by 15–20 km in Minnesota and northwestern Wisconsin. Fraxinus, Ulmus, and other mesic hardwoods expanded in the Big Woods region of southern Minnesota. The increasing density of paleoecological data networks and advances in statistical modeling approaches now enables the confident detection of subtle but significant changes in forest composition over the last 2,000 yr.

Linking demographic and food-web models to understand management trade-offs

Alternatives in ecosystem-based management often differ with respect to trade-offs between ecosystem values. Ecosystem or food-web models and demographic models are typically employed to evaluate alternatives, but the approaches are rarely integrated to uncover conflicts between values. We applied multistate models to a capture-recapture dataset on common guillemots Uria aalge breeding in the Baltic Sea to identify factors influencing survival. The estimated relationships were employed together with Ecopath-with-Ecosim food-web model simulations to project guillemot survival under six future scenarios incorporating climate change. The scenarios were based on management alternatives for eutrophication and cod fisheries, issues considered top priority for regional management, but without known direct effects on the guillemot population. Our demographic models identified prey quantity (abundance and biomass of sprat Sprattus sprattus) as the main factor influencing guillemot survival. Most scenarios resulted in projections of increased survival, in the near (2016-2040) and distant (2060-2085) future. However, in the scenario of reduced nutrient input and precautionary cod fishing, guillemot survival was projected to be lower in both future periods due to lower sprat stocks. Matrix population models suggested a substantial decline of the guillemot population in the near future, 24% per 10 years, and a smaller reduction, 1.1% per 10 years, in the distant future. To date, many stakeholders and Baltic Sea governments have supported reduced nutrient input and precautionary cod fishing and implementation is underway. Negative effects on nonfocal species have previously not been uncovered, but our results show that the scenario is likely to negatively impact the guillemot population. Linking model results allowed identifying trade-offs associated with management alternatives. This information is critical to thorough evaluation by decision-makers, but not easily obtained by food-web models or demographic models in isolation. Appropriate datasets are often available, making it feasible to apply a linked approach for better-informed decisions in ecosystem-based management.

Variation across river channels in demographic dynamics of a riparian herb with threatened status: management and conservation implications

PREMISE

Gesneria pauciflora is a rare, threatened plant in riparian forests. Periodic disturbances, expected in this habitat, could influence demographic dynamics on plant populations, yet their impact may not be the same across the watershed. We hypothesized that differences in disturbances between the main channel and tributaries may lead to spatial dissimilarities in population growth rate (λ), structure, and fecundity.

METHODS

In the Maricao River Watershed in Puerto Rico, 1277 plants were tagged and monitored for 1.5 years. Every 6 months, we measured plant size and recorded survival, fecundity, and appearance of seedlings. These variables were used in integral projection models to assess the population status of G. pauciflora.

RESULTS

Plants in the main channel were smaller but more likely to flower and fruit than those in the tributaries. Overall mortality was greater in the main channel and greater during the rainy season. At both sites, λ ranged from 0.9114 to 0.9865, and survival/growth of larger plants had a greater effect on λ (>0.90) regardless of site.

CONCLUSIONS

Values for population growth rates suggest that G. pauciflora is declining across the watershed. Higher mortality rates in the main channel (more-perturbed sites) might drive G. pauciflora to reproduce at smaller sizes, while tributaries (less-perturbed sites) might be better for growth and lead to larger plant sizes. Extreme climatic events are expected to increase in the Caribbean and might decrease the population if the population is left unmanaged. Management strategies that reduce the time plants require to reach larger sizes might be necessary to increase λ, and reintroduction using cuttings might be a possible solution.

A scalable model of vegetation transitions using deep neural networks

In times of rapid global change, anticipating vegetation changes and assessing their impacts is of key relevance to managers and policy makers. Yet, predicting vegetation dynamics often suffers from an inherent scale mismatch, with abundant data and process understanding being available at a fine spatial grain, but the relevance for decision-making is increasing with spatial extent.We present a novel approach for scaling vegetation dynamics (SVD), using deep learning to predict vegetation transitions. Vegetation is discretized into a large number (103-106) of potential states based on its structure, composition and functioning. Transition probabilities between states are estimated via a deep neural network (DNN) trained on observed or simulated vegetation transitions in combination with environmental variables. The impact of vegetation transitions on important ecological indicators is quantified by probabilistically linking attributes such as carbon storage and biodiversity to vegetation states.Here, we describe the SVD approach and present results of applying the framework in a meta-modelling context. We trained a DNN using simulations of a process-based forest landscape model for a complex mountain forest landscape under different climate scenarios. Subsequently, we evaluated the ability of SVD to project long-term vegetation dynamics and the resulting changes in forest carbon storage and biodiversity. SVD captured spatial (e.g. elevational gradients) and temporal (e.g. species succession) patterns of vegetation dynamics well, and responded realistically to changing environmental conditions. In addition, we tested the computational efficiency of the approach, highlighting the utility of SVD for country- to continental scale applications. SVD is the-to our knowledge-first vegetation model harnessing deep neural networks. The approach has high predictive accuracy and is able to generalize well beyond training data. SVD was designed to run on widely available input data (e.g. vegetation states defined from remote sensing, gridded global climate datasets) and exceeds the computational performance of currently available highly optimized landscape models by three to four orders of magnitude. We conclude that SVD is a promising approach for combining detailed process knowledge on fine-grained ecosystem processes with the increasingly available big ecological datasets for improved large-scale projections of vegetation dynamics.

BOARD INVITED REVIEW: Prospects for improving management of animal disease introductions using disease-dynamic models

Management and policy decisions are continually made to mitigate disease introductions in animal populations despite often limited surveillance data or knowledge of disease transmission processes. Science-based management is broadly recognized as leading to more effective decisions yet application of models to actively guide disease surveillance and mitigate risks remains limited. Disease-dynamic models are an efficient method of providing information for management decisions because of their ability to integrate and evaluate multiple, complex processes simultaneously while accounting for uncertainty common in animal diseases. Here we review disease introduction pathways and transmission processes crucial for informing disease management and models at the interface of domestic animals and wildlife. We describe how disease transmission models can improve disease management and present a conceptual framework for integrating disease models into the decision process using adaptive management principles. We apply our framework to a case study of African swine fever virus in wild and domestic swine to demonstrate how disease-dynamic models can improve mitigation of introduction risk. We also identify opportunities to improve the application of disease models to support decision-making to manage disease at the interface of domestic and wild animals. First, scientists must focus on objective-driven models providing practical predictions that are useful to those managing disease. In order for practical model predictions to be incorporated into disease management a recognition that modeling is a means to improve management and outcomes is important. This will be most successful when done in a cross-disciplinary environment that includes scientists and decision-makers representing wildlife and domestic animal health. Lastly, including economic principles of value-of-information and cost-benefit analysis in disease-dynamic models can facilitate more efficient management decisions and improve communication of model forecasts. Integration of disease-dynamic models into management and decision-making processes is expected to improve surveillance systems, risk mitigations, outbreak preparedness, and outbreak response activities.

A concise guide to developing and using quantitative models in conservation management

Quantitative models are powerful tools for informing conservation management and decision-making. As applied modeling is increasingly used to address conservation problems, guidelines are required to clarify the scope of modeling applications and to facilitate the impact and acceptance of models by practitioners. We identify three key roles for quantitative models in conservation management: (a) to assess the extent of a conservation problem; (b) to provide insights into the dynamics of complex social and ecological systems; and, (c) to evaluate the efficacy of proposed conservation interventions. We describe 10 recommendations to facilitate the acceptance of quantitative models in conservation management, providing a basis for good practice to guide their development and evaluation in conservation applications. We structure these recommendations within four established phases of model construction, enabling their integration within existing workflows: (a) design (two recommendations); (b) specification (two); (c) evaluation (one); and (d) inference (five). Quantitative modeling can support effective conservation management provided that both managers and modelers understand and agree on the place for models in conservation. Our concise review and recommendations will assist conservation managers and modelers to collaborate in the development of quantitative models that are fit-for-purpose, and to trust and use these models appropriately while understanding key drivers of uncertainty.

Water provisioning improvement through payment for ecosystem services

We assess whether a Payments for Ecosystem Services (PES) programme met its objectives of reducing soil erosion and yielding water in an environmental protected area, the Guariroba River Basin, Midwestern Brazil. We measured rainfall and water discharge throughout 2012 and 2016. During the same period, soil and water conservation practices were performed in the basin, such as: building level terraces and riparian vegetation recovery. We separated streamflow into baseflow and direct runoff, then we evaluted the baseflow index that indicated that groundwater significantly contributes to total flow. Therefore, to investigate the effects on streamflow, we performed a trend analysis in the baseflow time series using the Mann-Kendall test. In addition, we analysed the efficiency of soil erosion regulation practices over time, considering the total payment and the trends found in the baseflow. Whereas precipitation records present a decreasing trend (1 mm month^-1), baseflow tends to increase by 0.018 m³ s^-1 in the same period. Our findings show that soil conservation practices performed in the basin increase baseflow and also provide a better resilience to endure extreme events such as drought based on an increase in forest areas and soil conservation practices such as level terrace.

Evaluating the effects of landscape structure on the recovery of an invasive vertebrate after population control

CONTEXT

Effective landscape control of invasive species is context-dependent due to the interplay between the landscape structure, local population dynamics, and metapopulation processes. We use a modelling approach incorporating these three elements to explore the drivers of recovery of populations of invasive species after control.

OBJECTIVES

We aim to improve our understanding of the factors influencing the landscape-level control of invasive species.

METHODS

We focus on the case study of invasive brushtail possum (Trichosurus vulpecula) control in New Zealand. We assess how 13 covariates describing the landscape, patch, and population features influence the time of population recovery to a management density threshold of two possums/ha. We demonstrate the effects of those covariates on population recovery under three scenarios of population growth: logistic growth, strong Allee effects, and weak Allee effects.

RESULTS

Recovery times were rapid regardless of the simulated population dynamics (average recovery time < 2 years), although populations experiencing Allee effects took longer to recover than those growing logistically. Our results indicate that habitat availability and patch area play a key role in reducing times to recovery after control, and this relationship is consistent across the three simulated scenarios.

CONCLUSIONS

The control of invasive possum populations in patchy landscapes would benefit from a patch-level management approach (considering each patch as an independent management unit), whereas simple landscapes would be better controlled by taking a landscape-level view (the landscape as the management unit). Future research should test the predictions of our models with empirical data to refine control operations.

Landscape heterogeneity is key to forecasting outcomes of plant reintroduction

Conservation and restoration projects often involve starting new populations by introducing individuals into portions of their native or projected range. Such efforts can help meet many related goals, including habitat creation, ecosystem service provisioning, assisted migration, and the reintroduction of imperiled species following local extirpation. The outcomes of reintroduction efforts, however, are highly variable, with results ranging from local extinction to dramatic population growth; reasons for this variation remain unclear. Here, we ask whether population growth following plant reintroductions is governed by variation at two scales: the scale of individual habitat patches to which individuals are reintroduced, and larger among-landscape scales in which similar patches may be situated in landscapes that differ in matrix type, soil conditions, and other factors. Quantifying demographic variation at these two scales will help prioritize locations for introduction and, once introductions take place, forecast population growth. This work took place within a large-scale habitat fragmentation experiment, where individuals of two perennial forb species were reintroduced into eight replicate ~50-ha landscapes, each containing a set of five ~1-ha patches that varied in their degree of isolation (connected by habitat corridors or unconnected) and edge-to-area ratio. Using data on individual growth, survival, reproductive output, and recruitment collected one to two years after reintroduction, we developed models to forecast population growth, then compared forecasts to observed population sizes, three and six years later. Both the type of patch (connected and unconnected) and identity of the landscape to which individuals were reintroduced had effects on forecasted population growth rates, but only variation associated with landscape identity was an accurate predictor of subsequently observed population growth rates. Models that did not include landscape identity had minimal forecasting ability, revealing the key importance of variation at this scale for accurate prediction. Of the five demographic rates used to model population dynamics, seed production was the most important source of forecast error in population growth rates. Our results point to the importance of accounting for landscape-scale variation in demographic models and demonstrate how such models might assist with prioritizing particular landscapes for species reintroduction projects.

Modelling future safe and just operating spaces in regional social-ecological systems

Shaping social-ecological systems towards sustainable, desirable and equitable futures is often hampered by complex human-natural feedbacks, emergence and nonlinearities. Consequently, the future of systems vulnerable to collapse is uncertain under plausible trajectories of environmental change, socioeconomic development and decision-making. We develop a modelling approach that incorporates driver interactions and feedbacks to operationalise future "safe and just operating spaces" for sustainable development. Monte Carlo simulations of fish catch from India's Chilika lagoon are compared to conditions that are ecologically and socioeconomically desirable as per today's norms. Akin to a satellite-navigation system, the model identifies multidimensional pathways giving at least a 75% chance of achieving the desirable future, whilst simultaneously diverting the system away from undesirable pathways. Critically for regional governance, the driver limits and trade-offs associated with regulating the resource are realised. More widely, this approach represents an adaptable framework that explores the resilience of social-ecological interactions and feedbacks underpinning regional sustainable development.

Pollinator size and its consequences: Robust estimates of body size in pollinating insects

Body size is an integral functional trait that underlies pollination-related ecological processes, yet it is often impractical to measure directly. Allometric scaling laws have been used to overcome this problem. However, most existing models rely upon small sample sizes, geographically restricted sampling and have limited applicability for non-bee taxa. Allometric models that consider biogeography, phylogenetic relatedness, and intraspecific variation are urgently required to ensure greater accuracy. We measured body size as dry weight and intertegular distance (ITD) of 391 bee species (4,035 specimens) and 103 hoverfly species (399 specimens) across four biogeographic regions: Australia, Europe, North America, and South America. We updated existing models within a Bayesian mixed-model framework to test the power of ITD to predict interspecific variation in pollinator dry weight in interaction with different co-variates: phylogeny or taxonomy, sexual dimorphism, and biogeographic region. In addition, we used ordinary least squares regression to assess intraspecific dry weight ~ ITD relationships for ten bees and five hoverfly species. Including co-variates led to more robust interspecific body size predictions for both bees and hoverflies relative to models with the ITD alone. In contrast, at the intraspecific level, our results demonstrate that the ITD is an inconsistent predictor of body size for bees and hoverflies. The use of allometric scaling laws to estimate body size is more suitable for interspecific comparative analyses than assessing intraspecific variation. Collectively, these models form the basis of the dynamic R package, "pollimetry," which provides a comprehensive resource for allometric pollination research worldwide.

Harnessing Deep Learning in Ecology: An Example Predicting Bark Beetle Outbreaks

Addressing current global challenges such as biodiversity loss, global change, and increasing demands for ecosystem services requires improved ecological prediction. Recent increases in data availability, process understanding, and computing power are fostering quantitative approaches in ecology. However, flexible methodological frameworks are needed to utilize these developments towards improved ecological prediction. Deep learning is a rapidly evolving branch of machine learning, yet has received only little attention in ecology to date. It refers to the training of deep neural networks (DNNs), i.e. artificial neural networks consisting of many layers and a large number of neurons. We here provide a reproducible example (including code and data) of designing, training, and applying DNNs for ecological prediction. Using bark beetle outbreaks in conifer-dominated forests as an example, we show that DNNs are well able to predict both short-term infestation risk at the local scale and long-term outbreak dynamics at the landscape level. We furthermore highlight that DNNs have better overall performance than more conventional approaches to predicting bark beetle outbreak dynamics. We conclude that DNNs have high potential to form the backbone of a comprehensive disturbance forecasting system. More broadly, we argue for an increased utilization of the predictive power of DNNs for a wide range of ecological problems.

Developing a modern data workflow for regularly updated data

Over the past decade, biology has undergone a data revolution in how researchers collect data and the amount of data being collected. An emerging challenge that has received limited attention in biology is managing, working with, and providing access to data under continual active collection. Regularly updated data present unique challenges in quality assurance and control, data publication, archiving, and reproducibility. We developed a workflow for a long-term ecological study that addresses many of the challenges associated with managing this type of data. We do this by leveraging existing tools to 1) perform quality assurance and control; 2) import, restructure, version, and archive data; 3) rapidly publish new data in ways that ensure appropriate credit to all contributors; and 4) automate most steps in the data pipeline to reduce the time and effort required by researchers. The workflow leverages tools from software development, including version control and continuous integration, to create a modern data management system that automates the pipeline.

Seasonal difference in temporal transferability of an ecological model: near-term predictions of lemming outbreak abundances

Ecological models have been criticized for a lack of validation of their temporal transferability. Here we answer this call by investigating the temporal transferability of a dynamic state-space model developed to estimate season-dependent biotic and climatic predictors of spatial variability in outbreak abundance of the Norwegian lemming. Modelled summer and winter dynamics parametrized by spatial trapping data from one cyclic outbreak were validated with data from a subsequent outbreak. There was a distinct difference in model transferability between seasons. Summer dynamics had good temporal transferability, displaying ecological models' potential to be temporally transferable. However, the winter dynamics transferred poorly. This discrepancy is likely due to a temporal inconsistency in the ability of the climate predictor (i.e. elevation) to reflect the winter conditions affecting lemmings both directly and indirectly. We conclude that there is an urgent need for data and models that yield better predictions of winter processes, in particular in face of the expected rapid climate change in the Arctic.

Transferability of biotic interactions: Temporal consistency of arctic plant-rodent relationships is poor

Variability in biotic interaction strength is an integral part of food web functioning. However, the consequences of the spatial and temporal variability of biotic interactions are poorly known, in particular for predicting species abundance and distribution. The amplitude of rodent population cycles (i.e., peak-phase abundances) has been hypothesized to be determined by vegetation properties in tundra ecosystems. We assessed the spatial and temporal predictability of food and shelter plants effects on peak-phase small rodent abundance during two consecutive rodent population peaks. Rodent abundance was related to both food and shelter biomass during the first peak, and spatial transferability was mostly good. Yet, the temporal transferability of our models to the next population peak was poorer. Plant-rodent interactions are thus temporally variable and likely more complex than simple one-directional (bottom-up) relationships or variably overruled by other biotic interactions and abiotic factors. We propose that parametrizing a more complete set of functional links within food webs across abiotic and biotic contexts would improve transferability of biotic interaction models. Such attempts are currently constrained by the lack of data with replicated estimates of key players in food webs. Enhanced collaboration between researchers whose main research interests lay in different parts of the food web could ameliorate this.