Final Report Summary - SAVBIRD (Climate change and birds in the African savannah)
The SAVBIRD project set out with the aim of identifying and understanding the processes that are driving range changes in savannah birds in East Africa. Two parts to the work were envisaged, one primarily statistical involving the development and application of new methodologies that could be transferred to any region of the world and one field-based focussed on the specific mechanisms changing the distributions in savannas of Tanzania.
Statistical methods for the analysis of distributions and distribution change are of considerable interest to scientists studying the impacts of global change. A quick search of the scientific literature reveals that between 2005 and 2010 over 850 papers were published using the key words 'species distribution model' or 'niche model', whilst only 79 had been published in the previous five years. The explosive growth has continued in recent years. Such a new research field needs rapid development of ideas and methods, as the problems with early methods are identified by the research community. In the SAVBIRD project we have developed a new method for analysis of distributions that addresses some of the key issues identified by recent work:
1. it can accurately characterise non-random relationships with known environmental covariates even in the presence of influential but unknown covariates
2. it can enforce ecologically realistic functional forms to environmental relationships
3. it is appropriate to the data type and structure
4. it allows proper model comparison and the identification of important factors and
5. it can incorporate information on observer effort.
A paper describing these advances has been submitted to a high profile ecological journal (Ecology Letters). We have also studied the uncertainties associated with modelling and projecting species distributions and have discovered that uncertainty has often been underestimated in recent studies potentially leading to false alarms. We have identified (and developed) areas where development of new statistical tools will improve predictions from distribution models and have identified gaps in the research that will need filling in future work. These results have been presented at the Royal Society of London and published in the 'Philosophical transactions' of the Royal Society of London.
Additional methodological advances have focussed on methods for the analysis of data where no direct measure of observer effort is available. In such cases (which are remarkably common: museums across the world contain millions of specimens where the collection location is known, but the surveyed area is not) special care needs to be used when modelling distributions as absence of records may simply mean no-one has visited an area to look for species. This project has developed a new method for modelling distributions under such conditions that, in tests using simulated data, are a significant improvement on current methods and a paper describing the methods is currently being drafted.
In contrast to the general success of the statistical components of the project, the field-based components ran into considerable logistical difficulties early in the research programme when the researcher's permit to undertake much of the field research in Tanzania was cancelled. This was only re-instated in April 2012, leaving no time to implement the full field programme. As a consequence of this problem no experiments were run and the only field research implemented was focussed on observation work which has fed into the modelling component.
The main area where the fieldwork has been useful is in contributing data towards the application of the new methods to determine the recent changes in the distributions of Tanzanian birds. The extent to which climate change might diminish the efficacy of protected areas is one of the most pressing conservation questions that many researchers have addressed from either a theoretical or strictly model-based approach. Such results suggest that climate-driven species distribution shifts will leave protected areas impoverished and species inadequately protected while other evidence suggests that intact ecosystems within protected areas will be resilient to change. Using historical data from the Tanzanian bird atlas and contemporary data from that source and supplemented by the fellow's fieldwork, this project tackled the problem directly, investigating how recent changes in distribution of 139 Tanzanian savannah bird species are linked to climate change, protected area status and land degradation. We uncovered the first evidence of climate-driven range shifts for an African bird community and showed that distribution changes are also related to protected area status and the degree of land degradation. Importantly, we found that the continued maintenance of existing protected areas can be an appropriate conservation response to the challenge of climate and environmental change. These results are perhaps particularly surprising as the protected areas in Tanzania are managed without regard for birds. These results are particularly important to conservation managers and climate change policy makers in Tanzania and elsewhere and have been well received when presented to audiences of conservation managers and scientists alike. A paper describing the work has been submitted.
Despite the limited fieldwork possible to generate these results, the lack of a specific research permit for most of the fellowship period led to a major change in the original plan, with considerably more effort spent on the statistical components that originally planned. In order to develop some field aspects we diversified from the original aims, enabling the fellow to get involved in a number of small projects that would not otherwise have been possible. Some of these small projects have also been extremely productive.
A key product from these extra activities has been the identification of 10 key issues for the managed of savannas in eastern and southern Africa, based on discussions from a workshop in Serengeti involving researchers and conservationists from east and southern Africa. Following the workshop the ideas have been collated and processed by the fellow into a paper currently in review with the Journal of Applied Ecology. The management conclusions have been shared widely with land managers in Tanzania already and we hope the paper will become a landmark in savannah management research.
Other smaller projects include collaborations on analysis of Jackal home ranges in southern Serengeti that are currently been extended by a Master's student supervised by the fellow and the development of a consortium of conservation organisations seeking to understand how fires can be used to mitigate the impacts of changes in grazing patterns and carbon dioxide (CO2) in and around the Serengeti ecosystem. This partnership has recently led to the fellow submitting a grant application on behalf of the partners to the United Kingdom (UK) government for GBP 228 382, with an additional GBP 166 480 of matched funding from the partners (total value GBP 394 862 or EUR 470 972).
Statistical methods for the analysis of distributions and distribution change are of considerable interest to scientists studying the impacts of global change. A quick search of the scientific literature reveals that between 2005 and 2010 over 850 papers were published using the key words 'species distribution model' or 'niche model', whilst only 79 had been published in the previous five years. The explosive growth has continued in recent years. Such a new research field needs rapid development of ideas and methods, as the problems with early methods are identified by the research community. In the SAVBIRD project we have developed a new method for analysis of distributions that addresses some of the key issues identified by recent work:
1. it can accurately characterise non-random relationships with known environmental covariates even in the presence of influential but unknown covariates
2. it can enforce ecologically realistic functional forms to environmental relationships
3. it is appropriate to the data type and structure
4. it allows proper model comparison and the identification of important factors and
5. it can incorporate information on observer effort.
A paper describing these advances has been submitted to a high profile ecological journal (Ecology Letters). We have also studied the uncertainties associated with modelling and projecting species distributions and have discovered that uncertainty has often been underestimated in recent studies potentially leading to false alarms. We have identified (and developed) areas where development of new statistical tools will improve predictions from distribution models and have identified gaps in the research that will need filling in future work. These results have been presented at the Royal Society of London and published in the 'Philosophical transactions' of the Royal Society of London.
Additional methodological advances have focussed on methods for the analysis of data where no direct measure of observer effort is available. In such cases (which are remarkably common: museums across the world contain millions of specimens where the collection location is known, but the surveyed area is not) special care needs to be used when modelling distributions as absence of records may simply mean no-one has visited an area to look for species. This project has developed a new method for modelling distributions under such conditions that, in tests using simulated data, are a significant improvement on current methods and a paper describing the methods is currently being drafted.
In contrast to the general success of the statistical components of the project, the field-based components ran into considerable logistical difficulties early in the research programme when the researcher's permit to undertake much of the field research in Tanzania was cancelled. This was only re-instated in April 2012, leaving no time to implement the full field programme. As a consequence of this problem no experiments were run and the only field research implemented was focussed on observation work which has fed into the modelling component.
The main area where the fieldwork has been useful is in contributing data towards the application of the new methods to determine the recent changes in the distributions of Tanzanian birds. The extent to which climate change might diminish the efficacy of protected areas is one of the most pressing conservation questions that many researchers have addressed from either a theoretical or strictly model-based approach. Such results suggest that climate-driven species distribution shifts will leave protected areas impoverished and species inadequately protected while other evidence suggests that intact ecosystems within protected areas will be resilient to change. Using historical data from the Tanzanian bird atlas and contemporary data from that source and supplemented by the fellow's fieldwork, this project tackled the problem directly, investigating how recent changes in distribution of 139 Tanzanian savannah bird species are linked to climate change, protected area status and land degradation. We uncovered the first evidence of climate-driven range shifts for an African bird community and showed that distribution changes are also related to protected area status and the degree of land degradation. Importantly, we found that the continued maintenance of existing protected areas can be an appropriate conservation response to the challenge of climate and environmental change. These results are perhaps particularly surprising as the protected areas in Tanzania are managed without regard for birds. These results are particularly important to conservation managers and climate change policy makers in Tanzania and elsewhere and have been well received when presented to audiences of conservation managers and scientists alike. A paper describing the work has been submitted.
Despite the limited fieldwork possible to generate these results, the lack of a specific research permit for most of the fellowship period led to a major change in the original plan, with considerably more effort spent on the statistical components that originally planned. In order to develop some field aspects we diversified from the original aims, enabling the fellow to get involved in a number of small projects that would not otherwise have been possible. Some of these small projects have also been extremely productive.
A key product from these extra activities has been the identification of 10 key issues for the managed of savannas in eastern and southern Africa, based on discussions from a workshop in Serengeti involving researchers and conservationists from east and southern Africa. Following the workshop the ideas have been collated and processed by the fellow into a paper currently in review with the Journal of Applied Ecology. The management conclusions have been shared widely with land managers in Tanzania already and we hope the paper will become a landmark in savannah management research.
Other smaller projects include collaborations on analysis of Jackal home ranges in southern Serengeti that are currently been extended by a Master's student supervised by the fellow and the development of a consortium of conservation organisations seeking to understand how fires can be used to mitigate the impacts of changes in grazing patterns and carbon dioxide (CO2) in and around the Serengeti ecosystem. This partnership has recently led to the fellow submitting a grant application on behalf of the partners to the United Kingdom (UK) government for GBP 228 382, with an additional GBP 166 480 of matched funding from the partners (total value GBP 394 862 or EUR 470 972).