Final Report Summary - RANGESHIFT (Modelling the response of biodiversity to environmental change)
The research objectives of RANGESHIFT were to: develop an integrated landscape permeability modelling framework; determine how the spatial arrangement of landscape, including habitat fragmentation, affects the rate of species range-shifting, the degree of genetic diversity that persists following range shifting and population viability following range shift; and to derive robust decision rules for inclusion of habitats to existing networks.
To achieve these objectives, we utilised the software ‘Rangeshifter’ developed at the University of Aberdeen, which provides an excellent modelling platform for simulation of spatial eco-evolutionary dynamics and species’ responses to environmental changes. The programming effort in the project centred on extending ‘Rangeshifter’’s productivity via efficient batch execution. This was achieved through the development of the package ‘RUnShifter’ that accompanies the main program and allows full control of simulation settings from within the popular ‘R’ statistical environment.
Using these packages, in cooperation with researchers from Forest Research and University of Southampton, we have investigated population response of several artificial species (representing long and short range dispersers and a range of life-history strategies) to various woodland management scenarios (restoration, creation or improvement of forest habitats) on UK-based landscapes with varied habitat compositions. Preliminary results highlight the differences in the velocity of range-expansion of each of the examined “species”, depending on the management strategy and landscape composition. The rate at which the landscape changes can also play significant role for the range expansion process. Importantly, the simulations allow us to quantify the prediction uncertainty of the landscape management strategies.
We also developed an individual-based metapopulation model that incorporates genetics and range of mating systems. With this model, we investigated mutation accumulation during the formation of species range limits. Results of this study illustrate the important role deleterious mutations may play in range formation, and highlight this as an important area for further work, particularly focussing on the potentially conflicting effects that dispersal may have in reducing mutation while simultaneously increasing migration load in marginal populations.
Dispersal is a key factor influencing the speed of range shifting. Theory on the evolution of dispersal has been largely constructed on the assumption that individuals have no knowledge of the possible destinations. In reality, many organisms do not make such blind dispersal decisions. If information is available, individuals can use it in making better decisions, thereby increasing their fitness. Developments incorporating greater realism in modelling dispersal as three stage process (of departure, transient and settlement) are readily extendable to explore the evolution of prospecting behaviour in the context of dispersal. Therefore, we also investigate novel theoretical aspects of dispersal as an informed process. * Knowledge how individuals acquire and use information in order to make more adaptive dispersal decisions is essential for achieving progress in a variety of fields ranging from gene flow and speciation to the global persistence of species in the face of local extinction. Theory suggests that animals may use a combination of genetically encoded and acquired information to guide their behaviour. We looked at how the acquisition and use of information evolves in relation to their costs, as well as at the advantages and effects of utilising Bayesian information updating by the dispersers. We found that informed dispersal strategies may evolve when the costs of information acquisition during prospecting are low but only if there are mortality costs associated with dispersal movements. That is, selection favours informed dispersal strategies when the acquisition and use processes themselves were not expensive. * Under some environmental conditions it is expected that the information stored in the genes may be more useful than under other conditions, where the information an individual acquires itself is more useful. By formulating a model of ‘Bayesian dispersal’ we make an attempt at understanding how much weight should an organism put on the sampling information as opposed to innate information to make optimal settlement decisions under different environmental characteristics, and how well the evolution of these parameters works in terms of individual fitness or population abundance.
A valuable effect of the developments in the project is the programming library to facilitate rapid development of individual-based models. A large body of individual-based models (such as the ones mentioned above) share the same or very similar structure and basic elements. They consist of populations of entities carrying some individual properties, these entities are acting and interacting on a spatial grid, reproducing, possibly relocating. Majority of these models, despite having so much in common, have been written from scratch as there is no suitable programming framework available. Responding to this need, a programming library has been put together (“libibmod: a C++ class library for individual-based modelling”: available at). This library provides a suite of commonly used elements. These include base object classes for spatial grids (landscape), individuals (agents), as well as useful mathematical and statistical functions (e.g. random number distributions), and means for simulation parameter control.
To achieve these objectives, we utilised the software ‘Rangeshifter’ developed at the University of Aberdeen, which provides an excellent modelling platform for simulation of spatial eco-evolutionary dynamics and species’ responses to environmental changes. The programming effort in the project centred on extending ‘Rangeshifter’’s productivity via efficient batch execution. This was achieved through the development of the package ‘RUnShifter’ that accompanies the main program and allows full control of simulation settings from within the popular ‘R’ statistical environment.
Using these packages, in cooperation with researchers from Forest Research and University of Southampton, we have investigated population response of several artificial species (representing long and short range dispersers and a range of life-history strategies) to various woodland management scenarios (restoration, creation or improvement of forest habitats) on UK-based landscapes with varied habitat compositions. Preliminary results highlight the differences in the velocity of range-expansion of each of the examined “species”, depending on the management strategy and landscape composition. The rate at which the landscape changes can also play significant role for the range expansion process. Importantly, the simulations allow us to quantify the prediction uncertainty of the landscape management strategies.
We also developed an individual-based metapopulation model that incorporates genetics and range of mating systems. With this model, we investigated mutation accumulation during the formation of species range limits. Results of this study illustrate the important role deleterious mutations may play in range formation, and highlight this as an important area for further work, particularly focussing on the potentially conflicting effects that dispersal may have in reducing mutation while simultaneously increasing migration load in marginal populations.
Dispersal is a key factor influencing the speed of range shifting. Theory on the evolution of dispersal has been largely constructed on the assumption that individuals have no knowledge of the possible destinations. In reality, many organisms do not make such blind dispersal decisions. If information is available, individuals can use it in making better decisions, thereby increasing their fitness. Developments incorporating greater realism in modelling dispersal as three stage process (of departure, transient and settlement) are readily extendable to explore the evolution of prospecting behaviour in the context of dispersal. Therefore, we also investigate novel theoretical aspects of dispersal as an informed process. * Knowledge how individuals acquire and use information in order to make more adaptive dispersal decisions is essential for achieving progress in a variety of fields ranging from gene flow and speciation to the global persistence of species in the face of local extinction. Theory suggests that animals may use a combination of genetically encoded and acquired information to guide their behaviour. We looked at how the acquisition and use of information evolves in relation to their costs, as well as at the advantages and effects of utilising Bayesian information updating by the dispersers. We found that informed dispersal strategies may evolve when the costs of information acquisition during prospecting are low but only if there are mortality costs associated with dispersal movements. That is, selection favours informed dispersal strategies when the acquisition and use processes themselves were not expensive. * Under some environmental conditions it is expected that the information stored in the genes may be more useful than under other conditions, where the information an individual acquires itself is more useful. By formulating a model of ‘Bayesian dispersal’ we make an attempt at understanding how much weight should an organism put on the sampling information as opposed to innate information to make optimal settlement decisions under different environmental characteristics, and how well the evolution of these parameters works in terms of individual fitness or population abundance.
A valuable effect of the developments in the project is the programming library to facilitate rapid development of individual-based models. A large body of individual-based models (such as the ones mentioned above) share the same or very similar structure and basic elements. They consist of populations of entities carrying some individual properties, these entities are acting and interacting on a spatial grid, reproducing, possibly relocating. Majority of these models, despite having so much in common, have been written from scratch as there is no suitable programming framework available. Responding to this need, a programming library has been put together (“libibmod: a C++ class library for individual-based modelling”: available at