BIOMOD – optimizing predictions of species distributions and projecting potential future shifts under global change

A new computation framework (BIOMOD: BIOdiversity MODelling) is presented, which aims to maximize the predictive accuracy of current species distributions and the reliability of future potential distributions using different types of statistical modelling methods. BIOMOD capitalizes on the different techniques used in static modelling to provide spatial predictions. It computes, for each species and in the same package, the four most widely used modelling techniques in species predictions, namely Generalized Linear Models (GLM), Generalized Additive Models (GAM), Classification and Regression Tree analysis (CART) and Artificial Neural Networks (ANN). BIOMOD was applied to 61 species of trees in Europe using climatic quantities as explanatory variables of current distributions. On average, all the different modelling methods yielded very good agreement between observed and predicted distributions. However, the relative performance of different techniques was idiosyncratic across species, suggesting that the most accurate model varies between species. The results of this evaluation also highlight that slight differences between current predictions from different modelling techniques are exacerbated in future projections. Therefore, it is difficult to assess the reliability of alternative projections without validation techniques or expert opinion. It is concluded that rather than using a single modelling technique to predict the distribution of several species, it would be more reliable to use a framework assessing different models for each species and selecting the most accurate one using both evaluation methods and expert knowledge.     

Ensemble forecasting of species distributions

Concern over implications of climate change for biodiversity has led to the use of bioclimatic models to forecast the range shifts of species under future climate-change scenarios. Recent studies have demonstrated that projections by alternative models can be so variable as to compromise their usefulness for guiding policy decisions. Here, we advocate the use of multiple models within an ensemble forecasting framework and describe alternative approaches to the analysis of bioclimatic ensembles, including bounding box, consensus and probabilistic techniques. We argue that, although improved accuracy can be delivered through the traditional tasks of trying to build better models with improved data, more robust forecasts can also be achieved if ensemble forecasts are produced and analysed appropriately.     

Uncertainty in ensemble forecasting of species distribution

Species distribution modelling has been widely applied in order to assess the potential impacts of climate change on biodiversity. Many methodological decisions, taken during the modelling process and forecasts, may, however, lead to a large variability in the assessment of future impacts. Using measures of species range change and turnover, the potential impacts of climate change on French stream fish species and assemblages were evaluated. Our main focus was to quantify the uncertainty in the projections of these impacts arising from four sources of uncertainty: initial datasets (Data), statistical methods [species distribution models (SDM)], general circulation models (GCM), and gas emission scenarios (GES). Several modalities of the aforementioned uncertainty sources were combined in an ensemble forecasting framework resulting in 8400 different projections. The variance explained by each source was then extracted from this whole ensemble of projections. Overall, SDM contributed to the largest variation in projections, followed by GCM, whose contribution increased over time equalling almost the proportion of variance explained by SDM in 2080. Data and GES had little influence on the variability in projections. Future projections of range change were more consistent for species with a large geographical extent (i.e., distribution along latitudinal or stream gradients) or with restricted environmental requirements (i.e., small thermal or elevation ranges). Variability in projections of turnover was spatially structured at the scale of France, indicating that certain particular geographical areas should be considered with care when projecting the potential impacts of climate change. The results of this study, therefore, emphasized that particular attention should be paid to the use of predictions ensembles resulting from the application of several statistical methods and climate models. Moreover, forecasted impacts of climate change should always be provided with an assessment of their uncertainty, so that management and conservation decisions can be taken in the full knowledge of their reliability.     

Ecosystem function and species loss – a microcosm study

There are too many kinds of organisms to be able to study and manage each, yet the loss of a single species can sometimes unravel an ecosystem. Such `fusewire species'– critical in the same sense that an electrical fuse can cut out a whole circuit – would be a rewarding focus for research and management effort. However, this approach can only be effective if these `fusewires' represent but a small proportion of the number of species in the system.  

Aim

To demonstrate methods for measuring what proportion of the species in a system are critical to ecosystem function.  

Methods

The prevalence of fusewire species was measured in manipulative experiments on an aquatic microcosm.  

Results

No single genus deletion caused changes in key characteristics of the system.  

Main conclusions

Comparison of these results with other published studies shows that the proportion of critical fusewire species varies amongst different ecosystems. The oxidation pond microcosms were shown to contain no single species indispensable to system function. They appear to be ill-suited to a management strategy which focuses on priority eukaryote species. However, a single study provides no evidence that this result is general or even typical of other kinds of ecosystems; it is presented here as an empirical model. Other methods of investigation are available; they are less experimentally rigorous but more practical. These could provide important guidance in planning an approach to management in a particular ecosystem.     

NeuralEnsembles: a neural network based ensemble forecasting program for habitat and bioclimatic suitability analysis

NeuralEnsembles is an integrated modeling and assessment tool for predicting areas of species habitat/bioclimatic suitability based on presence/absence data. This free, Windows based program, which comes with a friendly graphical user interface, generates predictions using ensembles of artificial neural networks. Models can quickly and easily be produced for multiple species and subsequently be extrapolated either to new regions or under different future climate scenarios. An array of options is provided for optimizing the construction and training of ensemble models. Main outputs of the program include text files of suitability predictions, maps and various statistical measures of model performance and accuracy.     

Determinants of species evenness in a neotropical bat ensemble

Evenness is an important property of communities. Species richness alone does not capture the fact that one or a few species may dominate total abundance and biomass of a community. This in turn has important consequences for ecosystem functioning and species interactions. Evenness has been observed to vary systematically along environmental and productivity gradients. However, a truly general theory about which factors control evenness in a community has yet to emerge. Prior research on evenness has suggested that high richness, biomass and abundance should lead to lower community evenness in our study system of bats in Panama. However, only few empirical studies examine the simultaneous effects of species richness, biomass or abundance on evenness. For the first time, we applied path analysis in the study of evenness to tease apart the relative importance and direction (positive or negative) of causality among these three factors. As predicted, we found that evenness decreases with increasing species richness, abundance and biomass. The negative effect of abundance was mediated by the positive joint effect of biomass and richness. The selected models varied in the strength of the correlation between the three variables with evenness but their direction was consistent. Overall, we argue that rarity, high mobility and differences in resource availability at sites with lower environmental stress can explain the negative effects of richness on evenness.     

Geomorphological disturbance is necessary for predicting fine‐scale species distributions

Disturbances related to geomorphological processes are frequent, widespread and often intense at high latitudes and altitudes, affecting the fine‐scale distribution of many plant species. While the inclusion of physical disturbances into models of species geographic ranges is widely recommended, no studies have yet tested the utility of field‐quantified geomorphological disturbances for terrestrial species distribution modelling. Here we apply generalized additive models and boosted regression trees to examine if the explicit inclusion of terrestrial and fluvial geomorphological variables alters species distribution models for 154 vascular plant, bryophyte and lichen species in north European mountain tundra. The inclusion of these disturbances significantly improved both the explanatory and predictive power of distribution models, with consistent results for all three species groups. Spatial distribution predictions changed considerably for some species after the inclusion of disturbance variables, with fluvial disturbances generating strongly linear features for species influenced by erosion or sediment deposition. As a consequence, models incorporating geomorphological variables produced markedly more refined distribution maps than simpler models. Predictions of species distributions will thus benefit strongly from the inclusion of fine‐scale geomorphological variables, particularly in areas of active earth surface processes, enabling more accurate forecasting of future species ranges under changing conditions.     

Chemical diversity – highlighting a species richness and ecosystem function disconnect

The lack of predictability in litter-mix studies may result from the low correlation between species number and the traits that drive the processes under observation. From the standpoint of litter-quality-dependent ecological processes, we propose that litter chemical qualities are functional traits and introduce a multivariate index of chemical diversity (CD_Q) based on Rao's quadratic entropy to describe the compositional heterogeneity of litter and foliar mixtures. Using published data from temperate and tropical forest systems to illustrate the relationship between species richness and chemical diversity, we show the variation of chemical diversity based on profiles of total nutrient concentrations (N, P, K, Ca and Mg) with species richness. We discuss how this behavior may explain the idiosyncratic responses exhibited in litter-mix experiments and how it may contribute to the observed dominance of species identity over species diversity. As a summary of resource heterogeneity relevant to detritivore and microbial processes, the chemical diversity index is potentially a better predictor of diversity effects on nutrient dynamics than species richness. Finally, we propose the use of infrared spectroscopy techniques for a rapid and more comprehensive determination of foliar and litter chemical composition to provide a more information-rich index.     

Time development of probability distributions for interacting species

A general solution to the dynamical equation for the probability distribution associated withn interacting species is obtained by employing the author's generic canonical expression for the rate functions. Interacting species models with limit-cycle dynamics and no stable equilibrium points feature probability distributions that are asymptotic for large values oft to Dirac δ-distributions concentrated on the limit-cycles, as illustrated here for an analytically solvable two-species model. For ann-species Volterra model, a stationary or temporally-averaged probability distribution should generally be much more complicated than the specialized Poisson form studied by Kerner and others.     

The great tit (Parus major) – a misclassified ring species

The great tit (Parus major) has been considered to be the most typical example of an avian ring species. The terminal taxa of the ring (major and minor sectors) are supposed to be reproductively isolated in a zone of secondary contact in the middle Amur valley, Siberia. Our study combines molecular markers (cytochrome‐b), bioacoustic analyses and morphological characters to judge the ring species status of the great tit complex. Despite a notable percentage of intermediately coloured birds in the mixed population of middle Amur, a lack of mitochondrial introgression between the major and minor sectors and a small number of true hybrids among voucher specimens from this area suggest at least a partial reproductive barrier between both sectors. In contrast, variation of morphological and especially acoustic characters along the ring‐shaped area and the phylogenetic structure of the P. major group do not match the ring species concept. Bioacoustic and molecular data (cytochrome‐b sequences) reveal two large and closely related subspecies blocks, the sectors major and bokharensis in the Western Palaearctic and central Asia, and the sectors minor and cinereus in the Eastern Palaearctic and South‐east Asia, respectively. The two western sectors diverged only recently (0.5 Mya) and they were separated from the eastern group by Pleistocene events about 1.5 Mya. Songs from allopatric regions of the two subspecies blocks differ distinctly in frequency parameters and element composition. In the area of secondary contact, males of all phenotypes share the same frequency range of song, close to the range of the typical minor song. Hybrids and major males sing mixed repertoires of typical major and minor strophe types as well as mixed strophes. In contrast, phenotypic minor males display only pure minor strophes. Differences in mate choice and mating success based on repertoire size are believed to uphold the reproductive barrier between major and minor birds in the area of sympatry. Taxonomic consequences suggest three separate species in the Parus major complex: Parus major s.s. (including the very closely related bokharensis sector), Parus minor and Parus cinereus. © 2005 The Linnean Society of London, Biological Journal of the Linnean Society, 2005, 86 , 153–174.     

Amphibian phylogeography: a model for understanding historical aspects of species distributions

Phylogeographic analysis has become a major tool for investigating historical aspects of biogeography and population genetic structure. Anuran amphibians are particularly informative subjects for phylogeographic research on account of their global distribution, high degree of population genetic structure and ease of sampling. Studies on all the world's inhabited continents have demonstrated the nature and locations of refugia, including the Gulf Coast in North America and the Mediterranean peninsulas in Europe during the Pleistocene glaciations; the importance of vicariance events such as the uplift of the Andes in shaping modern distributions; and colonization routes in temperate zones during postglacial climatic amelioration. Features identified as important to amphibian biogeography, notably mountain ranges, large rivers such as the Amazon and climatic fluctuations, are common to many other taxa. New analytical methods based on coalescent, Bayesian and likelihood approaches permit more rigorous hypothesis testing than has hitherto been possible and offer the prospect of even more detailed insights into species and population history in future work.     

Species–area relationships of butterflies in Europe and species richness forecasting

Species–area relationships (SARs) of European butterfly species (Rhopalocera) appear to follow power functions with Mediterranean butterflies having a much higher slope value (z=0.49) compared to the slope for the northern and eastern European countries (z=0.10). A simulated process of species extinction by a stepwise density dependent random elimination of species affected species–area patterns differently. For Mediterranean countries SAR slopes decreased, for other European countries slopes increased during the extinction process. Comparisons of species numbers before and after extinction with those predicted by a classical SAR approach differed widely and revealed that SARs are not able to predict future species numbers at local scales. For Mediterranean countries the classical SAR approach underestimated the number of species remaining after simulated extinction, for all other European countries SARs highly overestimated species numbers. These contrasting patterns indicate that changes in SAR patterns do not unequivocally point to changes in species diversity or community structure as assumed by current theory. On the other hand, the results strongly indicate that simplified applications of SARs for forecasting might give misimpressions about species loss and future biodiversity if the initial community structure, especially relative densities and numbers of species with restricted range size, are not taken into account.     

Near term climate projections for invasive species distributions

Climate change and invasive species pose important conservation issues separately, and should be examined together. We used existing long term climate datasets for the US to project potential climate change into the future at a finer spatial and temporal resolution than the climate change scenarios generally available. These fine scale projections, along with new species distribution modeling techniques to forecast the potential extent of invasive species, can provide useful information to aide conservation and invasive species management efforts. We created habitat suitability maps for Pueraria montana (kudzu) under current climatic conditions and potential average conditions up to 30 years in the future. We examined how the potential distribution of this species will be affected by changing climate, and the management implications associated with these changes. Our models indicated that P. montana may increase its distribution particularly in the Northeast with climate change and may decrease in other areas.     

The control of rank-abundance distributions by a competitive despotic species

Accounting for differences in abundances among species remains a high priority for community ecology. While there has been more than 80 years of work on trying to explain the characteristic S shape of rank-abundance distributions (RADs), there has been recent conjecture that the form may not depend on ecological processes per se but may be a general phenomenon arising in many unrelated disciplines. We show that the RAD shape can be influenced by an ecological process, namely, interference competition. The noisy miner (Manorina melanocephala) is a hyperaggressive, ‘despotic’ bird that occurs over much of eastern Australia (>10⁶ km²). We compiled data for bird communities from 350 locations within its range, which were collected using standard avian survey methods. We used hierarchical Bayesian models to show that the RAD shape was much altered when the abundance of the strong interactor exceeded a threshold density; RADs consistently were steeper when the density of the noisy miner ≥2.5 birds ha^?1. The structure of bird communities at sites where the noisy miner exceeded this density was very different from that at sites where the densities fell below the threshold: species richness and Shannon diversity were much reduced, but mean abundances and mean avian biomass per site did not differ substantially.     

Search and navigation in dynamic environments – from individual behaviors to population distributions

Animal movement receives widespread attention within ecology and behavior. However, much research is restricted within isolated sub-disciplines focusing on single phenomena such as navigation (e.g. homing behavior), search strategies (e.g. Levy flights) or theoretical considerations of optimal population dispersion (e.g. ideal free distribution). To help synthesize existing research, we outline a unifying conceptual framework that integrates individual-level behaviors and population-level spatial distributions with respect to spatio-temporal resource dynamics. We distinguish among (1) non-oriented movements based on diffusion and kinesis in response to proximate stimuli, (2) oriented movements utilizing perceptual cues of distant targets, and (3) memory mechanisms that assume prior knowledge of a target's location. Species' use of these mechanisms depends on life-history traits and resource dynamics, which together shape population-level patterns. Resources with little spatial variability should facilitate sedentary ranges, whereas resources with predictable seasonal variation in spatial distributions should generate migratory patterns. A third pattern, 'nomadism', should emerge when resource distributions are unpredictable in both space and time. We summarize recent advances in analyses of animal trajectories and outline three major components on which future studies should focus: (1) integration across alternative movement mechanisms involving links between state variables and specific mechanisms, (2) consideration of dynamics in resource landscapes or environments that include resource gradients in predictability, variability, scale, and abundance, and finally (3) quantitative methods to distinguish among population distributions. We suggest that combining techniques such as evolutionary programming and pattern oriented modeling will help to build strong links between underlying movement mechanisms and broad-scale population distributions.     

Spatial scaling of species abundance distributions

Species abundance distributions are an essential tool in describing the biodiversity of ecological communities. We now know that their shape changes as a function of the size of area sampled. Here we analyze the scaling properties of species abundance distributions by using the moments of the logarithmically transformed number of individuals. We find that the moments as a function of area size are well fitted by power laws and we use this pattern to estimate the species abundance distribution for areas larger than those sampled. To reconstruct the species abundance distribution from its moments, we use discrete Tchebichef polynomials. We exemplify the method with data on tree and shrub species from a 50 ha plot of tropical rain forest on Barro Colorado Island, Panama. We test the method within the 50 ha plot, and then we extrapolate the species abundance distribution for areas up to 5 km². Our results project that for areas above 50 ha the species abundance distributions have a bimodal shape with a local maximum occurring for the singleton classes and that this maximum increases with sampled area size.     

Spatiotemporal population distributions and their implications for species coexistence in a variable environment

A population experiences environmental variation both directly, through effects on life history parameters such as fecundity, and indirectly, through effects on the population distributions of competitors and thus on the distribution of competition. Which spatial and temporal scales of environmental variation most influence the coexistence of two species thus depends in part on the degree to which the resident population responds to different scales of variation. In this paper, I calculate an approximation for a spatiotemporal population distribution as the result of a filter function convolved with the environmental variation. I find that there is no straightforward connection between spatial or temporal scales inherent to an organism's life history, such as mean lifetime or dispersal distance, and the population's sensitivity to variation at different scales. Rather, life history traits interact sensitively with the way environmental variation affects the organism. I comment on the implications for variation-mediated coexistence.