Ways to test stochastic dynamic programming models empirically

Stochastic dynamic programming (SDP) models are widely used to predict optimal behavioural and life history strategies. We discuss a diversity of ways to test SDP models empirically, taking as our main illustration a model of the daily singing routine of birds. One approach to verification is to quantify model parameters, but most SDP models are schematic. Because predictions are therefore qualitative, testing several predictions is desirable. How state determines behaviour (the policy) is a central prediction that should be examined directly if both state and behaviour are measurable. Complementary predictions concern how behaviour and state change through time, but information is discarded by considering behaviour rather than state, by looking only at average state rather than its distribution, and by not following individuals. We identify the various circumstances in which an individual's state/behaviour at one time is correlated with its state/behaviour at a later time. When there are several state variables the relationships between them may be informative. Often model parameters represent environmental conditions that can also be viewed as state variables. Experimental manipulation of the environment has several advantages as a test, but a problem is uncertainty over how much the organism's policy will adjust. As an example we allow birds to use different assumptions about how well past weather predicts future weather. We advocate mirroring planned empirical investigations on the computer to investigate which manipulations and predictions will best test a model. Copyright 2000 The Association for the Study of Animal Behaviour.     

How mutualisms arise in phytoplankton communities: building eco‐evolutionary principles for aquatic microbes

Extensive sampling and metagenomics analyses of plankton communities across all aquatic environments are beginning to provide insights into the ecology of microbial communities. In particular, the importance of metabolic exchanges that provide a foundation for ecological interactions between microorganisms has emerged as a key factor in forging such communities. Here we show how both studies of environmental samples and physiological experimentation in the laboratory with defined microbial co‐cultures are being used to decipher the metabolic and molecular underpinnings of such exchanges. In addition, we explain how metabolic modelling may be used to conduct investigations in reverse, deducing novel molecular exchanges from analysis of large‐scale data sets, which can identify persistently co‐occurring species. Finally, we consider how knowledge of microbial community ecology can be built into evolutionary theories tailored to these species’ unique lifestyles. We propose a novel model for the evolution of metabolic auxotrophy in microorganisms that arises as a result of symbiosis, termed the Foraging‐to‐Farming hypothesis. The model has testable predictions, fits several known examples of mutualism in the aquatic world, and sheds light on how interactions, which cement dependencies within communities of microorganisms, might be initiated.     

How host plant variability influences the advantages to learning: a theoretical model for oviposition behaviour in Lepidoptera

Learning can allow individuals to increase their fitness in particular environments. The advantage to learning depends on the predictability of the environment and the extent to which animals can adjust their behaviour. Earlier general models have investigated when environmental predictability might favour the evolution of learning in foraging animals. Here, we construct a theoretical model that predicts the advantages to learning using a specific biological example: oviposition in the Lepidoptera. Our model includes environmental and behavioural complexities relevant to host selection in these insects and tests whether the predictions of the general models still hold. Our results demonstrate how the advantage of learning is maximised when within-generation variability is minimised (the local environment consists mainly of a single host plant species) and between-generation variability is maximised (different host plant species are the most common in different generations). We discuss how our results: (a) can be applied to recent empirical work in different lepidopteran species and (b) predict an important role of learning in lepidopteran agricultural pests.     

Forecasting species range dynamics with process‐explicit models: matching methods to applications

Knowing where species occur is fundamental to many ecological and environmental applications. Species distribution models (SDMs) are typically based on correlations between species occurrence data and environmental predictors, with ecological processes captured only implicitly. However, there is a growing interest in approaches that explicitly model processes such as physiology, dispersal, demography and biotic interactions. These models are believed to offer more robust predictions, particularly when extrapolating to novel conditions. Many process–explicit approaches are now available, but it is not clear how we can best draw on this expanded modelling toolbox to address ecological problems and inform management decisions. Here, we review a range of process–explicit models to determine their strengths and limitations, as well as their current use. Focusing on four common applications of SDMs – regulatory planning, extinction risk, climate refugia and invasive species – we then explore which models best meet management needs. We identify barriers to more widespread and effective use of process‐explicit models and outline how these might be overcome. As well as technical and data challenges, there is a pressing need for more thorough evaluation of model predictions to guide investment in method development and ensure the promise of these new approaches is fully realised.     

Assessing and improving the transferability of current global spatial prediction models

Aim

Global-scale maps of the environment are an important source of information for researchers and decision makers. Often, these maps are created by training machine learning algorithms on field-sampled reference data using remote sensing information as predictors. Since field samples are often sparse and clustered in geographic space, model prediction requires a transfer of the trained model to regions where no reference data are available. However, recent studies question the feasibility of predictions far beyond the location of training data.

Innovation

We propose a novel workflow for spatial predictive mapping that leverages recent developments in this field and combines them in innovative ways with the aim of improved model transferability and performance assessment. We demonstrate, evaluate and discuss the workflow with data from recently published global environmental maps.

Main conclusions

Reducing predictors to those relevant for spatial prediction leads to an increase of model transferability and map accuracy without a decrease of prediction quality in areas with high sampling density. Still, reliable gap-free global predictions were not possible, highlighting that global maps and their evaluation are hampered by limited availability of reference data.     

Climate change,species distribution models,and physiological performance metrics: predicting when biogeographic models are likely to fail

Modeling the biogeographic consequences of climate change requires confidence in model predictions under novel conditions. However, models often fail when extended to new locales, and such instances have been used as evidence of a change in physiological tolerance, that is, a fundamental niche shift. We explore an alternative explanation and propose a method for predicting the likelihood of failure based on physiological performance curves and environmental variance in the original and new environments. We define the transient event margin (TEM) as the gap between energetic performance failure, defined as CT_max, and the upper lethal limit, defined as LT_max. If TEM is large relative to environmental fluctuations, models will likely fail in new locales. If TEM is small relative to environmental fluctuations, models are likely to be robust for new locales, even when mechanism is unknown. Using temperature, we predict when biogeographic models are likely to fail and illustrate this with a case study. We suggest that failure is predictable from an understanding of how climate drives nonlethal physiological responses, but for many species such data have not been collected. Successful biogeographic forecasting thus depends on understanding when the mechanisms limiting distribution of a species will differ among geographic regions, or at different times, resulting in realized niche shifts. TEM allows prediction of the likelihood of such model failure.     

Universally sloppy parameter sensitivities in systems biology models

Quantitative computational models play an increasingly important role in modern biology. Such models typically involve many free parameters, and assigning their values is often a substantial obstacle to model development. Directly measuring in vivo biochemical parameters is difficult, and collectively fitting them to other experimental data often yields large parameter uncertainties. Nevertheless, in earlier work we showed in a growth-factor-signaling model that collective fitting could yield well-constrained predictions, even when it left individual parameters very poorly constrained. We also showed that the model had a “sloppy” spectrum of parameter sensitivities, with eigenvalues roughly evenly distributed over many decades. Here we use a collection of models from the literature to test whether such sloppy spectra are common in systems biology. Strikingly, we find that every model we examine has a sloppy spectrum of sensitivities. We also test several consequences of this sloppiness for building predictive models. In particular, sloppiness suggests that collective fits to even large amounts of ideal time-series data will often leave many parameters poorly constrained. Tests over our model collection are consistent with this suggestion. This difficulty with collective fits may seem to argue for direct parameter measurements, but sloppiness also implies that such measurements must be formidably precise and complete to usefully constrain many model predictions. We confirm this implication in our growth-factor-signaling model. Our results suggest that sloppy sensitivity spectra are universal in systems biology models. The prevalence of sloppiness highlights the power of collective fits and suggests that modelers should focus on predictions rather than on parameters.     

From topography to hydrology—The modifiable area unit problem impacts freshwater species distribution models

Species distribution models (SDMs) are statistical tools to identify potentially suitable habitats for species. For SDMs in river ecosystems, species occurrences and predictor data are often aggregated across subcatchments that serve as modeling units. The level of aggregation (i.e., model resolution) influences the statistical relationships between species occurrences and environmental predictors—a phenomenon known as the modifiable area unit problem (MAUP), making model outputs directly contingent on the model resolution. Here, we test how model performance, predictor importance, and the spatial congruence of species predictions depend on the model resolution (i.e., average subcatchment size) of SDMs. We modeled the potential habitat suitability of 50 native fish species in the upper Danube catchment at 10 different model resolutions. Model resolutions were derived using a 90‐m digital‐elevation model by using the GRASS‐GIS module r.watershed. Here, we decreased the average subcatchment size gradually from 632 to 2 km². We then ran ensemble SDMs based on five algorithms using topographical, climatic, hydrological, and land‐use predictors for each species and resolution. Model evaluation scores were consistently high, as sensitivity and True Skill Statistic values ranged from 86.1–93.2 and 0.61–0.73, respectively. The most contributing predictor changed from topography at coarse, to hydrology at fine resolutions. Climate predictors played an intermediate role for all resolutions, while land use was of little importance. Regarding the predicted habitat suitability, we identified a spatial filtering from coarse to intermediate resolutions. The predicted habitat suitability within a coarse resolution was not ported to all smaller, nested subcatchments, but only to a fraction that held the suitable environmental conditions. Across finer resolutions, the mapped predictions were spatially congruent without such filter effect. We show that freshwater SDM predictions can have consistently high evaluation scores while mapped predictions differ significantly and are highly contingent on the underlying subcatchment size. We encourage building freshwater SDMs across multiple catchment sizes, to assess model variability and uncertainties in model outcomes emerging from the MAUP.     

Unpacking the mechanisms captured by a correlative species distribution model to improve predictions of climate refugia

Climate refugia are regions that animals can retreat to, persist in and potentially then expand from under changing environmental conditions. Most forecasts of climate change refugia for species are based on correlative species distribution models (SDMs) using long‐term climate averages, projected to future climate scenarios. Limitations of such methods include the need to extrapolate into novel environments and uncertainty regarding the extent to which proximate variables included in the model capture processes driving distribution limits (and thus can be assumed to provide reliable predictions under new conditions). These limitations are well documented; however, their impact on the quality of climate refugia predictions is difficult to quantify. Here, we develop a detailed bioenergetics model for the koala. It indicates that range limits are driven by heat‐induced water stress, with the timing of rainfall and heat waves limiting the koala in the warmer parts of its range. We compare refugia predictions from the bioenergetics model with predictions from a suite of competing correlative SDMs under a range of future climate scenarios. SDMs were fitted using combinations of long‐term climate and weather extremes variables, to test how well each set of predictions captures the knowledge embedded in the bioenergetics model. Correlative models produced broadly similar predictions to the bioenergetics model across much of the species' current range – with SDMs that included weather extremes showing highest congruence. However, predictions in some regions diverged significantly when projecting to future climates due to the breakdown in correlation between climate variables. We provide unique insight into the mechanisms driving koala distribution and illustrate the importance of subtle relationships between the timing of weather events, particularly rain relative to hot‐spells, in driving species–climate relationships and distributions. By unpacking the mechanisms captured by correlative SDMs, we can increase our certainty in forecasts of climate change impacts on species.     

Predicting Ecological Roles in the Rhizosphere Using Metabolome and Transportome Modeling

The ability to obtain complete genome sequences from bacteria in environmental samples, such as soil samples from the rhizosphere, has highlighted the microbial diversity and complexity of environmental communities. However, new algorithms to analyze genome sequence information in the context of community structure are needed to enhance our understanding of the specific ecological roles of these organisms in soil environments. We present a machine learning approach using sequenced Pseudomonad genomes coupled with outputs of metabolic and transportomic computational models for identifying the most predictive molecular mechanisms indicative of a Pseudomonad’s ecological role in the rhizosphere: a biofilm, biocontrol agent, promoter of plant growth, or plant pathogen. Computational predictions of ecological niche were highly accurate overall with models trained on transportomic model output being the most accurate (Leave One Out Validation F-scores between 0.82 and 0.89). The strongest predictive molecular mechanism features for rhizosphere ecological niche overlap with many previously reported analyses of Pseudomonad interactions in the rhizosphere, suggesting that this approach successfully informs a system-scale level understanding of how Pseudomonads sense and interact with their environments. The observation that an organism’s transportome is highly predictive of its ecological niche is a novel discovery and may have implications in our understanding microbial ecology. The framework developed here can be generalized to the analysis of any bacteria across a wide range of environments and ecological niches making this approach a powerful tool for providing insights into functional predictions from bacterial genomic data.     

Step selection techniques uncover the environmental predictors of space use patterns in flocks of Amazonian birds

Understanding the behavioral decisions behind animal movement and space use patterns is a key challenge for behavioral ecology. Tools to quantify these patterns from movement and animal–habitat interactions are vital for transforming ecology into a predictive science. This is particularly important in environments undergoing rapid anthropogenic changes, such as the Amazon rainforest, where animals face novel landscapes. Insectivorous bird flocks are key elements of avian biodiversity in the Amazonian ecosystem. Therefore, disentangling and quantifying the drivers behind their movement and space use patterns is of great importance for Amazonian conservation. We use a step selection function (SSF) approach to uncover environmental drivers behind movement choices. This is used to construct a mechanistic model, from which we derive predicted utilization distributions (home ranges) of flocks. We show that movement decisions are significantly influenced by canopy height and topography, but depletion and renewal of resources do not appear to affect movement significantly. We quantify the magnitude of these effects and demonstrate that they are helpful for understanding various heterogeneous aspects of space use. We compare our results to recent analytic derivations of space use, demonstrating that the analytic approximation is only accurate when assuming that there is no persistence in the animals' movement. Our model can be translated into other environments or hypothetical scenarios, such as those given by proposed future anthropogenic actions, to make predictions of spatial patterns in bird flocks. Furthermore, our approach is quite general, so could potentially be used to understand the drivers of movement and spatial patterns for a wide variety of animal communities.     

The role of model selection in describing stochastic ecological processes

A great deal of variation in ecological processes can be generated through variation in environmental conditions. Models describing the underlying processes should be able to account for this variation. However, models may not be flexible enough, so that different realizations of a process may be better described by different models. This may lead to uncertainty in model selection.
Here we examine the question of whether two empirical models can provide consistent fits to different realizations of a process affected by environmental variation. We further examine the sensitivity of the model predictions to the amount of data available and the selection of the model. To study this, we simulated pollen dispersal patterns under varying wind conditions and then investigated whether the datasets consistently supported the same model. The role of the model selection and the impact of the spatial range over which the dispersal distances were observed were assessed by comparing model predictions at long dispersal distances.
There was no consistent pattern of one model providing a better fit than the other across simulations. The model providing better fit varied depending on the range of distances over which the dispersal patterns were observed, and on the amount of long-distance dispersal. The model predictions were found to be very sensitive to the selection of the model.
The variation between datasets produced with the same underlying mechanisms cannot be easily described using one model, which also limits our ability to reliably predict the underlying process. Therefore, the amount of information about a model choice provided by an individual field study may be rather limited. If we are to understand processes that are affected by environmental variation then we have to observe the range of possible outcomes of the processes under varying spatio-temporal conditions.     

Modelling large herbivore movement decisions: Beyond food availability as a predictor of ranging patterns

The ability of animals to adapt to their changing environment will depend in part on shifts in their ranging patterns, but when and why individuals choose to move requires detailed understanding of their decision‐making processes. We develop a simple decision‐making model accounting for resource availability in habitually used ranges. We suggest that disparities between model predictions and animal tracking data indicate additional factors influencing movement decisions, which may be identified given detailed system‐specific knowledge. The model was evaluated using movement data from satellite‐tracked elephants (Loxodonta africana) inhabiting the Amboseli basin in Kenya, moving from savannah areas with low quality but constant resource availability, to areas with temporally constrained higher nutrient availability. Overall, the model fits the data well: There was a good correlation between predicted and observed locations for the combined data from all elephants, but variation between individuals in how well the model fits. For those elephants where model predictions were less successful, additional factors likely to affect movement decisions, including reproduction, anthropogenic threats, memory and perception are suggested. This protocol for building and testing decision‐making models should contribute to success in attempts to preserve sufficient space for large herbivores in their increasingly human‐dominated ecosystems.     

When mother knows best: A population genetic model of transgenerational versus intragenerational plasticity

Many organisms exhibit phenotypic plasticity; producing alternate phenotypes depending on the environment. Individuals can be plastic (intragenerational or direct plasticity), wherein individuals of the same genotype produce different phenotypes in response to the environments they experience. Alternatively, an individual's phenotype may be under the control of its parents, usually the mother (transgenerational or indirect plasticity), so that mother's genotype determines the phenotype produced by a given genotype of her offspring. Under what conditions does plasticity evolve to have intragenerational as opposed to transgenerational genetic control? To explore this question, we present a population genetic model for the evolution of transgenerational and intragenerational plasticity. We hypothesize that the capacity for plasticity incurs a fitness cost, which is borne either by the individual developing the plastic phenotype or by its mother. We also hypothesize that individuals are imperfect predictors of future environments and their capacity for plasticity can lead them occasionally to make a low‐fitness phenotype for a particular environment. When the cost, benefit and error parameters are equal, we show that there is no evolutionary advantage to intragenerational over transgenerational plasticity, although the rate of evolution of transgenerational plasticity is half the rate for intragenerational plasticity, as predicted by theory on indirect genetic effects. We find that transgenerational plasticity evolves when mothers are better predictors of future environments than offspring or when the fitness cost of the capacity for plasticity is more readily borne by a mother than by her developing offspring. We discuss different natural systems with either direct intragenerational plasticity or indirect transgenerational plasticity and find a pattern qualitatively in accord with the predictions of our model.     

Climate change and nesting behaviour in vertebrates: a review of the ecological threats and potential for adaptive responses

Nest building is a taxonomically widespread and diverse trait that allows animals to alter local environments to create optimal conditions for offspring development. However, there is growing evidence that climate change is adversely affecting nest‐building in animals directly, for example via sea‐level rises that flood nests, reduced availability of building materials, and suboptimal sex allocation in species exhibiting temperature‐dependent sex determination. Climate change is also affecting nesting species indirectly, via range shifts into suboptimal nesting areas, reduced quality of nest‐building environments, and changes in interactions with nest predators and parasites. The ability of animals to adapt to sustained and rapid environmental change is crucial for the long‐term persistence of many species. Many animals are known to be capable of adjusting nesting behaviour adaptively across environmental gradients and in line with seasonal changes, and this existing plasticity potentially facilitates adaptation to anthropogenic climate change. However, whilst alterations in nesting phenology, site selection and design may facilitate short‐term adaptations, the ability of nest‐building animals to adapt over longer timescales is likely to be influenced by the heritable basis of such behaviour. We urgently need to understand how the behaviour and ecology of nest‐building in animals is affected by climate change, and particularly how altered patterns of nesting behaviour affect individual fitness and population persistence. We begin our review by summarising how predictable variation in environmental conditions influences nest‐building animals, before highlighting the ecological threats facing nest‐building animals experiencing anthropogenic climate change and examining the potential for changes in nest location and/or design to provide adaptive short‐ and long‐term responses to changing environmental conditions. We end by identifying areas that we believe warrant the most urgent attention for further research.     

Hunt or hide: How insularity and urbanization affect foraging decisions in lizards

Foraging decisions should reflect a balance between costs and benefits of alternative strategies. Predation risk and resource availability in the environment may be crucial in deciding how cautious individuals should behave during foraging. These costs and benefits will vary in time and context, meaning that animals should be able to adjust their foraging behaviour to new or altered environments. Studying how animals do this is essential to understand their survival in these environments. In this study, we investigated the effect of both insularity and urbanization on risk‐taking and neophobia during foraging in the Dalmatian wall lizard (Podarcis melisellensis). Small islets tend to have both a lower number of predators and less resources. Therefore, islet populations were expected to show more risk‐taking behaviour and less neophobia in a foraging context. Previous studies on behaviour of urban lizards have yielded inconsistent results, but due to a lack of both predators and arthropod prey in urban habitats, we expected urban lizards to also take more risks and behave less neophobic. We sampled several inhabited and uninhabited locations on Vis (Croatia) and surrounding islets. Risk‐taking behaviour was tested by measuring the latency of lizards to feed in the presence of a predator model, and neophobia by measuring the latency to feed in the presence of a novel object. We found that islet lizards do indeed take more risks and were less vigilant, but not less neophobic. Urban and rural lizards did not differ in any of these behaviours, which is in sharp contrast with previous work on mammals and birds. The behavioural differences between islet and island lizards were novel, but not unexpected findings and are in line with the theory of “island tameness”. The effect of urbanization on the behaviour of animals seems to be more complex and might vary among taxa.     

Fear responses in domestic chicks as a function of the social environment

The open field or novel environment has been used to assess fear in many species but its validity for the domestic fowl has recently been questioned. Based primarily on experiments which involved manipulation of the social environment Gallup and Suarez proposed that, contrary to an emotionality or fear interpretation, “open–field behaviour in chickens represents a compromise between opposing tendencies to reinstate contact with conspecifics and to minimize detection in the face of possible predation”. Predictions which can be made from the Gallup and Suarez model and from the fear hypothesis were tested by examining the effects of manipulating the social environment, during rearing and testing, on the open–field, hole–in–the–wall box and tonic immobility responses of domestic chicks. The results were inconsistent with predictions made from the Gallup and Suarez model but they conformed to the fear hypothesis. Furthermore, they were consistent with the majority of findings reported in the literature. Thus, while the opposing tendencies of reinstatement and predator evasion are, almost undoubtedly, important in many situations there remains considerable evidence for the role of fear in regulating the responses of domestic chicks to novel environments such as the open field. The two interpretations should not be considered mutually incompatible.     

BIO-LGCA: A cellular automaton modelling class for analysing collective cell migration

Collective dynamics in multicellular systems such as biological organs and tissues plays a key role in biological development, regeneration, and pathological conditions. Collective tissue dynamics—understood as population behaviour arising from the interplay of the constituting discrete cells—can be studied with on- and off-lattice agent-based models. However, classical on-lattice agent-based models, also known as cellular automata, fail to replicate key aspects of collective migration, which is a central instance of collective behaviour in multicellular systems. To overcome drawbacks of classical on-lattice models, we introduce an on-lattice, agent-based modelling class for collective cell migration, which we call biological lattice-gas cellular automaton (BIO-LGCA). The BIO-LGCA is characterised by synchronous time updates, and the explicit consideration of individual cell velocities. While rules in classical cellular automata are typically chosen ad hoc, rules for cell-cell and cell-environment interactions in the BIO-LGCA can also be derived from experimental cell migration data or biophysical laws for individual cell migration. We introduce elementary BIO-LGCA models of fundamental cell interactions, which may be combined in a modular fashion to model complex multicellular phenomena. We exemplify the mathematical mean-field analysis of specific BIO-LGCA models, which allows to explain collective behaviour. The first example predicts the formation of clusters in adhesively interacting cells. The second example is based on a novel BIO-LGCA combining adhesive interactions and alignment. For this model, our analysis clarifies the nature of the recently discovered invasion plasticity of breast cancer cells in heterogeneous environments.     

Anticipated effects of abiotic environmental change on intraspecific social interactions

Social interactions are ubiquitous across the animal kingdom. A variety of ecological and evolutionary processes are dependent on social interactions, such as movement, disease spread, information transmission, and density-dependent reproduction and survival. Social interactions, like any behaviour, are context dependent, varying with environmental conditions. Currently, environments are changing rapidly across multiple dimensions, becoming warmer and more variable, while habitats are increasingly fragmented and contaminated with pollutants. Social interactions are expected to change in response to these stressors and to continue to change into the future. However, a comprehensive understanding of the form and magnitude of the effects of these environmental changes on social interactions is currently lacking. Focusing on four major forms of rapid environmental change currently occurring, we review how these changing environmental gradients are expected to have immediate effects on social interactions such as communication, agonistic behaviours, and group formation, which will thereby induce changes in social organisation including mating systems, dominance hierarchies, and collective behaviour. Our review covers intraspecific variation in social interactions across environments, including studies in both the wild and in laboratory settings, and across a range of taxa. The expected responses of social behaviour to environmental change are diverse, but we identify several general themes. First, very dry, variable, fragmented, or polluted environments are likely to destabilise existing social systems. This occurs as these conditions limit the energy available for complex social interactions and affect dissimilar phenotypes differently. Second, a given environmental change can lead to opposite responses in social behaviour, and the direction of the response often hinges on the natural history of the organism in question. Third, our review highlights the fact that changes in environmental factors are not occurring in isolation: multiple factors are changing simultaneously, which may have antagonistic or synergistic effects, and more work should be done to understand these combined effects. We close by identifying methodological and analytical techniques that might help to study the response of social interactions to changing environments, highlight consistent patterns among taxa, and predict subsequent evolutionary change. We expect that the changes in social interactions that we document here will have consequences for individuals, groups, and for the ecology and evolution of populations, and therefore warrant a central place in the study of animal populations, particularly in an era of rapid environmental change.