Progress 09/01/14 to 08/31/15
Outputs Target Audience:The primary achievements in this period were technical in nature, and we plan to focus on commercialization and outreach in the next period. Nonetheless, wehave submitted one manuscript and arein the process of preparinganother manuscript for publicationin the peer-reviewed literature, both of which will primarily reach a scientific and technical audience. Changes/Problems:The overall objectives have not undergone major revisions during the first year of this project.We have successfully completed many of our technical objectives (see 'Accomplishments' section). However, we encountered sometechnical challengesindeveloping a multi-speciesmodeling framework that allows for real-time, dynamic linkages among (single species) models.The first approach that weimplemented involved "inter-process communication"(whereby two or more separate software programs pass information among one another). Although we succeeded in achievingmulti-species models that dynamically share information, the processing speed for these interacting models was nearly four orders of magnitude slower in speed than the equivalent modelsrun withoutinter-species interactions. Therefore we deemed this method not commercially viable.For the same reason, we were also unable to meet our goal ofallowing dynamic feedbacksbetween population models and habitat suitability models (e.g.,implementation of habitat improvement efforts is contingent upon the status of a wildlife population). Afterconsultation with computer science experts, we have begun implementing an alternative approach for modelinginter-species interactions and population-habitat feedbacks. This approach involvesdeveloping a simplified population modeling framework (analogous to our existingRamas Metapop software, although a complete re-write of this software is outside the scope of this project) intheJava programming language. With this approach, we will be able toload multiple software components (e.g.,single-species population models) intoa computing environment with a shared address space, thereby enabling efficientpassage of information among multiple modeling components. In addition, this approach allows us to take advantage of the manypowerful softwarelibraries developed for the Java programming language.We are confident thatthis approachwill ultimately result in a software that isflexible and easily expandable,enabling this system to serve as apowerfulplatformformodeling wildlife population dynamics in complex, changing landscapes. What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?The accomplishments of this period were largely technical in nature. In the next reporting period, we plan to devote more time and effort to disseminating the practical tools resulting from our efforts. We have submitted onemanuscript for publication already, and other manuscript is in preparation.We submitted amanuscript entitled "Sensitivity of quasi-extinction risk", documenting a novel method for evaluating the expected change in extinction risk resulting from a small perturbation in anysimulation model parameter. This paper was submittedfor publication inthejournal, "Methods in Ecology and Evolution". An additional paper, currently in preparation, will address the topic of sensitivity analysis in wildlife populations more generally, outlining a set of best practices. The innovations and ideas from both papers will be implemented within the software framework we are developing. What do you plan to do during the next reporting period to accomplish the goals?As described in the "Changes/Problems" section of this progress report, the biggest technical challenge we encountered was enabling our existing population modeling software (Ramas GIS) to dynamically share data with other modeling components (e.g., other Ramas GIS instances, habitat suitability models). Our initialattempt to accomplish this task, while successful, led to unacceptably slow run-times, and was abandoned. Instead, we have begun implementing a simplified, Java-basedpopulation modeling framework thatcan run alongside all other software modules wehavedeveloped (e.g., habitat suitability,habitat management/protection), operating in acommon address-space and thereby facilitating efficient exchange of information acrossmodules. Alladditional goals andobjectives for the next reporting periodhave not been altered from the original proposal. These tasks includedeveloping new modules(e.g., accessing and processinghuman population density data,interfacing with existing frameworks for land-use change modeling) andperforming/documentinga rigorous case studywith this software.Finally, we willdevelop extensive documentation to accompanyour software release, including a user manual and a technical description of the equations and algorithms implemented therein.
Impacts What was accomplished under these goals?
Habitat loss isgenerally considered to be the biggestthreat to wildlife worldwide, and conversely, habitat restorationis widely acknowledged to be the most effective solution for wildlife conservation.Quantitative methods for mapping habitat and projecting how habitat will change under various global change scenarios (e.g., climate and land-use change)have advanced dramatically in recent years, and accessible software technologies are available for performing this task.Unfortunately,although several currently available software tools enableusers to model wildlife population viability in complex landscapes,no software tool currently exists toevaluate how wildlife populationswill respond to regional and global change. Nonetheless, as the processes of climate and land-use change intensify, questions about wildlife population response to global changeare being asked with increasing frequency by ecologists, state and federal agencies, and other stakeholder organizations. To this end, we aredeveloping a software tool for evaluating wildlife population viability in complex and dynamiclandscapes.We are confident that this tool will improve our clients ability to perform ecological impact assessments and develop regional development solutions in a complex and changing world. During the first year of this two-year SBIR Phase II project, we were able to make substantial progress towardmeeting thegoals stated in the original project narrative. In the following section, westate each of our originalproject goals,followed by a brief accounting ofourprogress to date in achieving these goals: (1) Expand the Phase I prototype module to integrate spatially referenced human population data (e.g., from census records), climate change models (e.g., IPCC forecasts), and habitat suitability models (e.g., ecological niche models). As described in the project narrative, we have developed an integrated software framework formodelingmultipledrivers of landscape changesimultaneously, and for modeling the impacts of these changes on habitat suitability for one or more wildlife species. Toaccomplish this task, we have leveragedthe Kepler/Ptolemy IIsoftware framework, which providesmethods for loading data from external sources and for sharing data amongsoftware modules.We successfully developed a software module fordescribing habitat suitability usingecological niche models (statistical descriptions of habitat suitability as a function of an arbitrary number of spatial layers). In addition, we developed a module that can be used to download all occurrence locations (data necessaryfor fitting ecological niche models)for any species in theonline Global Biodiversity Information Facility (GBIF) database.Finally, we developed a software module for generating a time series of habitat suitability maps, using this habitat suitability model in conjunction with climate change projections (derived from general circulation models, or GCMs) and land-use change projections.Currently,it remains the user's responsibility todownload the spatial predictor layersfor the habitat suitability model (e.g., satellite derived land-cover data, human population density). However, we are attempting to developsoftware modulesthat enable users to downloadadditional data layers from the world-wide web (analogous to the GBIF module described above). In particular, we hope to develop a module for accessing and processing data on human population density (a key driver of habitat suitability and land-use change). (2) Enable specification of feedbacks among the drivers of habitat change The software frameworkwe have developed is capable of modeling feedbacks and interactions among the drivers of habitat change.In thesecond year of this project, we intend toimplement such feedbacks and interactionsas part of a worked example/ case study. (3) Enable the specification of conservation strategies (e.g., habitat restoration) that are conditional upon the results from routine monitoring for multiple focal species The software framework we have developed is already capable of simulating/forecastingtheeffectsof conservation strategies on habitat suitability. For example,users can implement scenarios in whichcertain landparcels (e.g., wildlife refuges, protected areas) are protected from development. In addition, users canevaluate habitat suitability underalternative climate change scenarios --e.g.,representingreduced-emissions or "business asusual" scenarios. However, wehave encountered somechallenges inspecifyingconservation strategies that are conditional upon thecurrent status of the wildlife population model (e.g., habitat management actions are undertaken only if the population density falls below a certain threshold).As described in the"Changes/Problems" section of thisprogressreport, ourfirst attempt to dynamically share data from the widlife populationmodel with other modules (e.g., other wildlife population models, habitat change models) resulted in prohibitively long execution times. We have begun to implement an alternative approach for enabling feedbacks between species dynamics and habitat dynamics (see below). (4) Integrate the resulting habitat change module with multiple instances of the existing RAMAS GIS wildlife population modeling platform (one instance for each species). The software framework we have developed is capable of simultaneously specifying habitat suitability for multiple species and running wildlife population models using theRAMAS GISwildlife population modeling framework. In yeartwo of this project, we will demonstrate this capabilityas part of a worked example/ case study. (5) Develop and implement tools for model validation, uncertainty analysis, visualization of model structure and results, and scientific workflow management. One of the first tasks we undertook in the first year of this project was to identify scientific workflow management platforms that could be leveraged to achieve our goals.We decided to build ourmodeling frameworkas an extension of the existing Kepler/Ptolemy IIsoftware for scientific workflow management. This framework was particularlyappealing for many reasons: (1) by facilitating model sharing and execution, it is straightforward toshare complex analyses in a form that can be easily scrutinized (via an appealing visual interface) and executed by others, (2) it can automatically store a provenance trace, so results can always be linked to the correct generating model, (3) it provides a suite of existing features, from data extraction and preparation to visualization of results,that would be too labor-intensive for us to develop from scratch. Kepler/Ptolemy is licensed undera BSD license, which is compatible with commercial use. For commercialization, we are developing a simplified user interface for Kepler, which (for ease of use) will retain only those features necessary for the typical Ramas user. Finally, we are in the process of developing modules to facilitate global sensitivity analysisand model validation/ cross-validation (see below).We developed an innovation forsensitivity analysiswhich has led to our first manuscript submission under this grant, detailed below. (6) Demonstrate the application of the proposed approach with a rigorous case study. We plan to develop, implement, anddocumenta rigorous case study in year two of this project.
Publications
|