10.5061/DRYAD.RS776P3
Sussman, Allison L.
Michigan State University
Gardner, Beth
University of Washington
Adams, Evan M.
University of Washington
Salas, Leo
Point Blue Conservation Science
Kenow, Kevin P.
United States Geological Survey
Luukkonen, David R.
Michigan State University
Monfils, Michael J.
Michigan State University
Mueller, William P.
Western Great Lakes Bird and Bat Observatory Port Washington Wisconsin
Williams, Kathryn A.
Biodiversity Research Institute
Leduc-Lapierre, Michele
Great Lakes Commission Ann Arbor Michigan
Zipkin, Elise F.
Michigan State University
Data from: A comparative analysis of common methods to identify waterbird
hotspots
Dryad
dataset
2019
gamma distribution
persistence
Getis-Ord Gi*
Kernel density estimation
Spatial models
lognormal distribution
Parametric and nonparametric models
Great Lakes
spatial statistics
2019-05-15T13:28:54Z
2019-05-15T13:28:54Z
en
https://doi.org/10.1111/2041-210x.13209
15459124 bytes
1
CC0 1.0 Universal (CC0 1.0) Public Domain Dedication
1. Hotspot analysis is a commonly used method in ecology and conservation
to identify areas of high biodiversity or conservation concern. However,
delineating and mapping hotspots is subjective and various approaches can
lead to different conclusions with regard to the classification of
particular areas as hotspots, complicating long-term conservation
planning. 2. We present a comparative analysis of recent approaches for
identifying waterbird hotspots, with the goal of developing insights about
the appropriate use of these methods. We selected four commonly used
measures to identify persistent areas of high use: kernel density
estimation, Getis-Ord Gi*, hotspot persistence, and hotspots conditional
on presence, which represent the range of quantitative hotspot estimation
approaches used in waterbird analyses. We applied each of the methods to
aerial survey waterbird count data collected in the Great Lakes from
2012-2014. For each approach, we identified areas of high use for seven
species/species groups and then compared the results across all methods
and with mean effort-corrected counts. 3. Our results indicate that formal
hotspot analysis frameworks do not always lead to the same conclusions.
The kernel density and Getis-Ord Gi* methods yielded the most similar
results across all species analyzed and were generally correlated with
mean effort-corrected count data. We found that these two models can
differ substantially from the hotspot persistence and hotspots conditional
on presence estimation approaches, which were not consistently similar to
one another. The hotspot persistence approach differed most significantly
from the other methods but is the only method to explicitly account for
temporal variation. 4. We recommend considering the ecological question
and scale of conservation or management activities prior to designing
survey methodologies. Deciding the appropriate definition and scale for
analysis is critical for interpretation of hotspot analysis results as is
inclusion of important covariates. Combining hotspot analysis methods
using an integrative approach, either within a single analysis or
post-hoc, could lead to greater consistency in the identification of
waterbird hotspots.
alldata_attributed_v04122017More information can be found on the Midwest
Avian Data Center, a regional node of the Avian Knowledge Network, hosted
by Point Blue Conservation Science
(http://data.pointblue.org/partners/mwadc/) or through the Zipkin
Quantitative Ecology Lab GitHub page (https://zipkinlab.github.io/).
Great Lakes