Representing species distributions in spatially-explicit ecosystem models from presence-only data

Arnaud Grüss, Michael D. Drexler, Emily Chancellor, Cameron H. Ainsworth, Jeffrey S. Gleason, John M. Tirpak, Matthew S. Love, Elizabeth A Babcock

Research output: Contribution to journalArticle

Abstract

Spatially-explicit ecosystem models are increasingly considered for informing fisheries management. The inputs of these models that determine species distributions in the modeled system are critical. There is a need for methods to estimate species distributions in spatially-explicit ecosystem models from presence-only data. To address this need, we used a method relying on binomial generalized additive models integrating environmental covariates. This method allows for the production of distribution maps for ecosystem models such as Atlantis models, and of preference functions for Ecospace models; preference functions define the preferences of species groups for certain environmental parameter values and are employed by Ecospace to allocate species group biomasses spatially. The key step of the method we used is the objective generation of pseudo-absences for each month, by sampling with replacement the centroids of the cells of fine-scale spatial grids defined for each month. To demonstrate the method, we applied it to the diving bird and surface-feeding bird groups represented in the Atlantis model for the Gulf of Mexico (GOM), and to the seabird group represented in the Ecospace model for the West Florida Shelf. We also employed the distribution maps we constructed to provide a basis for a hypothetical marine protected area (MPA) planning scenario aiming to secure food for seabirds. Specifically, we produced a hotspot map for seabirds for the U.S. GOM from the distribution map of the species group, and then combined the hotspot map for seabirds with a hotspot map for their main prey, forage fish (Clupeidae and Exocoetidae), to determine where the hotspots of seabirds and forage fish overlap. This analysis suggested that, to secure forage fish for seabirds in the U.S. GOM, hypothetical MPAs should be implemented primarily within the coastal regions of the Louisiana-Texas shelf, within the coastal region of West Florida located between Sarasota and Naples and/or within Apalachee Bay, Florida.

LanguageEnglish (US)
Pages89-105
Number of pages17
JournalFisheries Research
Volume210
DOIs
StatePublished - Feb 1 2019

Fingerprint

biogeography
seabirds
seabird
ecosystems
ecosystem
Gulf of Mexico
forage
hot spot
fish
Exocoetidae
Clupeidae
environmental models
methodology
bird
distribution
birds
fisheries management
diving
fishery management
conservation areas

Keywords

  • Presence-only data
  • Pseudo-absences
  • Spatially-explicit ecosystem models
  • Species distribution models
  • U.S. Gulf of Mexico

ASJC Scopus subject areas

  • Aquatic Science

Cite this

Representing species distributions in spatially-explicit ecosystem models from presence-only data. / Grüss, Arnaud; Drexler, Michael D.; Chancellor, Emily; Ainsworth, Cameron H.; Gleason, Jeffrey S.; Tirpak, John M.; Love, Matthew S.; Babcock, Elizabeth A.

In: Fisheries Research, Vol. 210, 01.02.2019, p. 89-105.

Research output: Contribution to journalArticle

Grüss, A, Drexler, MD, Chancellor, E, Ainsworth, CH, Gleason, JS, Tirpak, JM, Love, MS & Babcock, EA 2019, 'Representing species distributions in spatially-explicit ecosystem models from presence-only data' Fisheries Research, vol. 210, pp. 89-105. https://doi.org/10.1016/j.fishres.2018.10.011
Grüss, Arnaud ; Drexler, Michael D. ; Chancellor, Emily ; Ainsworth, Cameron H. ; Gleason, Jeffrey S. ; Tirpak, John M. ; Love, Matthew S. ; Babcock, Elizabeth A. / Representing species distributions in spatially-explicit ecosystem models from presence-only data. In: Fisheries Research. 2019 ; Vol. 210. pp. 89-105.
@article{27dfa2c914024a31abafa38b0b9c3dae,
title = "Representing species distributions in spatially-explicit ecosystem models from presence-only data",
abstract = "Spatially-explicit ecosystem models are increasingly considered for informing fisheries management. The inputs of these models that determine species distributions in the modeled system are critical. There is a need for methods to estimate species distributions in spatially-explicit ecosystem models from presence-only data. To address this need, we used a method relying on binomial generalized additive models integrating environmental covariates. This method allows for the production of distribution maps for ecosystem models such as Atlantis models, and of preference functions for Ecospace models; preference functions define the preferences of species groups for certain environmental parameter values and are employed by Ecospace to allocate species group biomasses spatially. The key step of the method we used is the objective generation of pseudo-absences for each month, by sampling with replacement the centroids of the cells of fine-scale spatial grids defined for each month. To demonstrate the method, we applied it to the diving bird and surface-feeding bird groups represented in the Atlantis model for the Gulf of Mexico (GOM), and to the seabird group represented in the Ecospace model for the West Florida Shelf. We also employed the distribution maps we constructed to provide a basis for a hypothetical marine protected area (MPA) planning scenario aiming to secure food for seabirds. Specifically, we produced a hotspot map for seabirds for the U.S. GOM from the distribution map of the species group, and then combined the hotspot map for seabirds with a hotspot map for their main prey, forage fish (Clupeidae and Exocoetidae), to determine where the hotspots of seabirds and forage fish overlap. This analysis suggested that, to secure forage fish for seabirds in the U.S. GOM, hypothetical MPAs should be implemented primarily within the coastal regions of the Louisiana-Texas shelf, within the coastal region of West Florida located between Sarasota and Naples and/or within Apalachee Bay, Florida.",
keywords = "Presence-only data, Pseudo-absences, Spatially-explicit ecosystem models, Species distribution models, U.S. Gulf of Mexico",
author = "Arnaud Gr{\"u}ss and Drexler, {Michael D.} and Emily Chancellor and Ainsworth, {Cameron H.} and Gleason, {Jeffrey S.} and Tirpak, {John M.} and Love, {Matthew S.} and Babcock, {Elizabeth A}",
year = "2019",
month = "2",
day = "1",
doi = "10.1016/j.fishres.2018.10.011",
language = "English (US)",
volume = "210",
pages = "89--105",
journal = "Fisheries Research",
issn = "0165-7836",
publisher = "Elsevier",

}

TY - JOUR

T1 - Representing species distributions in spatially-explicit ecosystem models from presence-only data

AU - Grüss, Arnaud

AU - Drexler, Michael D.

AU - Chancellor, Emily

AU - Ainsworth, Cameron H.

AU - Gleason, Jeffrey S.

AU - Tirpak, John M.

AU - Love, Matthew S.

AU - Babcock, Elizabeth A

PY - 2019/2/1

Y1 - 2019/2/1

N2 - Spatially-explicit ecosystem models are increasingly considered for informing fisheries management. The inputs of these models that determine species distributions in the modeled system are critical. There is a need for methods to estimate species distributions in spatially-explicit ecosystem models from presence-only data. To address this need, we used a method relying on binomial generalized additive models integrating environmental covariates. This method allows for the production of distribution maps for ecosystem models such as Atlantis models, and of preference functions for Ecospace models; preference functions define the preferences of species groups for certain environmental parameter values and are employed by Ecospace to allocate species group biomasses spatially. The key step of the method we used is the objective generation of pseudo-absences for each month, by sampling with replacement the centroids of the cells of fine-scale spatial grids defined for each month. To demonstrate the method, we applied it to the diving bird and surface-feeding bird groups represented in the Atlantis model for the Gulf of Mexico (GOM), and to the seabird group represented in the Ecospace model for the West Florida Shelf. We also employed the distribution maps we constructed to provide a basis for a hypothetical marine protected area (MPA) planning scenario aiming to secure food for seabirds. Specifically, we produced a hotspot map for seabirds for the U.S. GOM from the distribution map of the species group, and then combined the hotspot map for seabirds with a hotspot map for their main prey, forage fish (Clupeidae and Exocoetidae), to determine where the hotspots of seabirds and forage fish overlap. This analysis suggested that, to secure forage fish for seabirds in the U.S. GOM, hypothetical MPAs should be implemented primarily within the coastal regions of the Louisiana-Texas shelf, within the coastal region of West Florida located between Sarasota and Naples and/or within Apalachee Bay, Florida.

AB - Spatially-explicit ecosystem models are increasingly considered for informing fisheries management. The inputs of these models that determine species distributions in the modeled system are critical. There is a need for methods to estimate species distributions in spatially-explicit ecosystem models from presence-only data. To address this need, we used a method relying on binomial generalized additive models integrating environmental covariates. This method allows for the production of distribution maps for ecosystem models such as Atlantis models, and of preference functions for Ecospace models; preference functions define the preferences of species groups for certain environmental parameter values and are employed by Ecospace to allocate species group biomasses spatially. The key step of the method we used is the objective generation of pseudo-absences for each month, by sampling with replacement the centroids of the cells of fine-scale spatial grids defined for each month. To demonstrate the method, we applied it to the diving bird and surface-feeding bird groups represented in the Atlantis model for the Gulf of Mexico (GOM), and to the seabird group represented in the Ecospace model for the West Florida Shelf. We also employed the distribution maps we constructed to provide a basis for a hypothetical marine protected area (MPA) planning scenario aiming to secure food for seabirds. Specifically, we produced a hotspot map for seabirds for the U.S. GOM from the distribution map of the species group, and then combined the hotspot map for seabirds with a hotspot map for their main prey, forage fish (Clupeidae and Exocoetidae), to determine where the hotspots of seabirds and forage fish overlap. This analysis suggested that, to secure forage fish for seabirds in the U.S. GOM, hypothetical MPAs should be implemented primarily within the coastal regions of the Louisiana-Texas shelf, within the coastal region of West Florida located between Sarasota and Naples and/or within Apalachee Bay, Florida.

KW - Presence-only data

KW - Pseudo-absences

KW - Spatially-explicit ecosystem models

KW - Species distribution models

KW - U.S. Gulf of Mexico

UR - http://www.scopus.com/inward/record.url?scp=85055248363&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055248363&partnerID=8YFLogxK

U2 - 10.1016/j.fishres.2018.10.011

DO - 10.1016/j.fishres.2018.10.011

M3 - Article

VL - 210

SP - 89

EP - 105

JO - Fisheries Research

T2 - Fisheries Research

JF - Fisheries Research

SN - 0165-7836

ER -