Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning
Abstract Despite the success of species distribution models (SDMs) in rare species conservation planning, inherent uncertainty in predictions can facilitate misuse or misapplication. Field validation is important as model predictions can vary greatly due to low species prevalence or spatial bias; tr...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2023-08-01
|
| Series: | Conservation Science and Practice |
| Subjects: | |
| Online Access: | https://doi.org/10.1111/csp2.12978 |
| _version_ | 1827877290473684992 |
|---|---|
| author | Sara Johnson Brenda Molano‐Flores David Zaya |
| author_facet | Sara Johnson Brenda Molano‐Flores David Zaya |
| author_sort | Sara Johnson |
| collection | DOAJ |
| description | Abstract Despite the success of species distribution models (SDMs) in rare species conservation planning, inherent uncertainty in predictions can facilitate misuse or misapplication. Field validation is important as model predictions can vary greatly due to low species prevalence or spatial bias; traits frequently encountered in rare species data. This study outlines an effective SDM approach to map the potential distribution of a Florida endemic plant species, Macbridea alba, which is transferable to other rare species that: (1) utilize presence data to create species distribution maps, (2) conduct field validation to gather independent data, (3) and use independent data for model evaluation and improvement. Independent data was used to assess each model's ability to predict the occurrence of M. alba and identify habitat for conservation and reintroduction efforts. Our study documents that for low prevalence species, models can produce significant variation in optimized threshold and evaluation metrics, which impact outcomes when converting continuous predictions to binary predictions, a common practice in model application. Our species distribution models were useful in finding new populations of M. alba. Field validation is vital to improving model performance, threshold selection, and evaluating model design and should be carefully considered when suggesting model use and application. |
| first_indexed | 2024-03-12T17:35:21Z |
| format | Article |
| id | doaj.art-2e8ca7a81e4f47b394d5b8133364f98b |
| institution | Directory Open Access Journal |
| issn | 2578-4854 |
| language | English |
| last_indexed | 2024-03-12T17:35:21Z |
| publishDate | 2023-08-01 |
| publisher | Wiley |
| record_format | Article |
| series | Conservation Science and Practice |
| spelling | doaj.art-2e8ca7a81e4f47b394d5b8133364f98b2023-08-04T11:36:29ZengWileyConservation Science and Practice2578-48542023-08-0158n/an/a10.1111/csp2.12978Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planningSara Johnson0Brenda Molano‐Flores1David Zaya2Department of Natural Resources and Environmental Sciences University of Illinois: Urbana‐Champaign Champaign Illinois USAIllinois Natural History Survey Champaign Illinois USAIllinois Natural History Survey Champaign Illinois USAAbstract Despite the success of species distribution models (SDMs) in rare species conservation planning, inherent uncertainty in predictions can facilitate misuse or misapplication. Field validation is important as model predictions can vary greatly due to low species prevalence or spatial bias; traits frequently encountered in rare species data. This study outlines an effective SDM approach to map the potential distribution of a Florida endemic plant species, Macbridea alba, which is transferable to other rare species that: (1) utilize presence data to create species distribution maps, (2) conduct field validation to gather independent data, (3) and use independent data for model evaluation and improvement. Independent data was used to assess each model's ability to predict the occurrence of M. alba and identify habitat for conservation and reintroduction efforts. Our study documents that for low prevalence species, models can produce significant variation in optimized threshold and evaluation metrics, which impact outcomes when converting continuous predictions to binary predictions, a common practice in model application. Our species distribution models were useful in finding new populations of M. alba. Field validation is vital to improving model performance, threshold selection, and evaluating model design and should be carefully considered when suggesting model use and application.https://doi.org/10.1111/csp2.12978calibrationconservation planningdiscriminationfield validationMaxentmodel evaluation |
| spellingShingle | Sara Johnson Brenda Molano‐Flores David Zaya Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning Conservation Science and Practice calibration conservation planning discrimination field validation Maxent model evaluation |
| title | Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| title_full | Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| title_fullStr | Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| title_full_unstemmed | Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| title_short | Field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| title_sort | field validation as a tool for mitigating uncertainty in species distribution modeling for conservation planning |
| topic | calibration conservation planning discrimination field validation Maxent model evaluation |
| url | https://doi.org/10.1111/csp2.12978 |
| work_keys_str_mv | AT sarajohnson fieldvalidationasatoolformitigatinguncertaintyinspeciesdistributionmodelingforconservationplanning AT brendamolanoflores fieldvalidationasatoolformitigatinguncertaintyinspeciesdistributionmodelingforconservationplanning AT davidzaya fieldvalidationasatoolformitigatinguncertaintyinspeciesdistributionmodelingforconservationplanning |