| dc.description.abstract |
<p>Over 50% of Western Hemisphere shorebird species are in decline due to ongoing
habitat loss and degradation. Many shorebird species require flooded habitat to rest
and feed during migratory movements spanning thousands of miles between breeding and
wintering grounds every spring and fall. In particular, shorebirds require shallowly
flooded habitat (water depth <15cm deep)—due to their morphology (i.e., bill and tarsus
length), many species are excluded from exploiting invertebrate prey resources in
deeper waters. While habitat-associations for shorebirds are relatively well understood
from observational studies, the distribution of suitable shorebird habitat over the
broad areas used by these species during migration is not well described. In some
regions of high wetland loss, shorebirds are heavily reliant on a core network of
remaining human-managed wetlands and flood-irrigated agricultural fields. Refuges
also provide substantial flooded habitat resources; however, these have typically
been designed and managed to match the habitat needs of waterfowl, which can use much
deeper water than shorebirds. Effective conservation strategies for migratory shorebirds
will require improved understanding of flooded habitat suitability patterns over large
migratory pathways, as well as knowledge of how species respond to habitat fluctuations
over time. </p><p>We analyzed water extent dynamics across the Sacramento Valley of
California, a globally important shorebird stopover site, for a 1983-2015 Landsat
time series, and evaluated the effect of climate on water extent. Satellite measurements
of surface water offer promise for understanding wetland habitat availability at broad
spatial and temporal scales. A range of methods can detect open water from imagery,
including supervised classification approaches and thresholds for spectral bands and
indices. Thresholds provide a time advantage; however, there is no universally superior
index, nor single best threshold for all instances. We used random forest to model
the presence or absence of water from >6,200 reference pixels, and derived an optimal
water probability threshold for our study area using receiver operating characteristic
curves. An optimized mid-infrared (1.5–1.7 µm) threshold identified open water in
the Sacramento Valley of California at 30-m resolution with an average of 90% producer’s
accuracy, comparable to approaches that require more intensive user input. SLC-off
Landsat 7 imagery was integrated by applying a customized interpolation that mapped
water in missing data gaps with 99% user’s accuracy. On average we detected open water
on ~26,000 ha (~3% of the study area) in early April at the peak of shorebird migration,
while water extent increased five-fold after the migration rush. Over the last three
decades, late March water extent declined by ~1,300 ha per year, primarily due to
changes in the extent and timing of agricultural flood-irrigation. Water within shorebird
habitats was significantly associated with an index of water availability at the peak
of migration. Our approach can be used to optimize thresholds for time series analysis
and near-real-time mapping in other regions, and requires only marginally more time
than generating a confusion matrix. </p><p>Two dimensional representations of flooded
habitat are insufficient to capture dynamic changes within the narrow water depth
range that is effectively accessible to migratory shorebirds. We developed a method
to quantify shallow water habitat distributions in inland non-tidal wetlands, and
assessed how water management practices have affected the amount of shorebird habitat
in Sacramento National Wildlife Refuge Complex (SNWRC), California. We produced water
depth distributions and modeled optimal habitat (<10 cm deep) within 23 managed wetlands
using high-resolution topography and fixed-point water depth records. We also demonstrated
that habitat availability, specifically suitable water depth ranges, can be tracked
from satellite imagery and high-resolution topography. We found that wetlands with
lower topographic roughness may have a higher potential to provide shorebird habitat
and that strategically reducing water levels could increase habitat extent. Over 50%
of the wetlands measured provided optimal habitat across <10% of their area at the
peak of migration in early April, and most provided a brief duration of shallow water
habitat. Reducing water volumes could increase the proportion of optimal habitat by
1– 1,678% (mean = 294 %) compared to actual volumes measured at peak spring migration
in 2016. For wetlands with a high habitat potential, beginning wetland drawdown earlier
and extending drawdown time could dramatically improve habitat conditions at the peak
of shorebird migration. Our approach can be adapted to track dynamic hydrologic changes
at broader spatial scales as additional high-resolution topographic (e.g., lidar,
drone imagery photogrammetry) and optical remote sensing data (e.g., Planet imagery,
drone photography) become available. </p><p>Attempting to model the response of a
community of shorebird species to flooded habitat dynamics from local to landscape
scale necessitates a rich dataset including field observations of shorebird habitat
use as well as information regarding regional habitat conditions over multiple time
periods. Bringing together these data sources results in several challenges for classical
statistical approaches, including overdispersion, fixed and random effects due to
repeated measures, irregular temporal intervals, and missing data. We investigated
how spring migration habitat use by 19 shorebird species at 327 wetland survey locations
across SNWRC responded to flooded habitat fluctuations at multiple spatial scales
from 1997-2015 using a generalized joint attribute modelling approach. In this analysis,
we integrated shorebird census records and habitat conditions documented in the field
with a suite of landscape-level habitat measurements derived from satellite imagery,
as well as water availability, water allocation and land use information. We found
that abundance by species peaked in late March and early April at SNWRC. Shorebird
abundance responded positively to the amount of flooded habitat at a given wetland
survey location. The total amount of water detected was the most important landscape
habitat measure; shorebirds were less likely to be observed at high abundance at SNWRC
wetlands when greater flooded habitat extent was present on the surrounding landscape.
We found that human land and water management were influential drivers of shorebird
habitat use. Water allocation information and reservoir storage resulted in better
model fit (i.e., lower DIC) than including measures of surface water availability
or drought conditions. Furthermore, the amount of landscape flooded habitat on agricultural
land produced a better fit than considering all flooded habitat, or flooded habitat
detected in wetlands. We found that the most relevant scale for measuring landscape
flooded habitat was within 2-10 km of wetland survey locations; this distance could
be a useful guideline for monitoring habitat conditions and targeting creation of
supplemental habitat to bolster the existing wetland network in the Sacramento Valley.</p>
|
|
