10.5061/DRYAD.VDNCJSXRR
Wenger, Amelia
0000-0002-0433-6164
University of Queensland
Harris, Daniel
University of Queensland
Weber, Samuel
University of California, Irvine
Vaghi, Ferguson
Kolombangara Island Biodiversity Conservation Association
Nand, Yashika
Wildlife Conservation Society
Naisilisili, Waisea
Wildlife Conservation Society
Hughes, Alec
Wildlife Conservation Society
Delevaux, Jade
University of Hawaii at Manoa
Klein, Carissa
University of Queensland
Watson, James
University of Queensland
Mumby, Peter
University of Queensland
Jupiter, Stacy
Wildlife Conservation Society
Best-practice forestry management delivers diminishing returns for coral
reefs with increased land-clearing
Dryad
dataset
2020
coastal development
land use change
Marine conservation
ridge to reef
sediment runoff
sustainable development
Australian Research Council
https://ror.org/05mmh0f86
LP150100934
National Science Foundation
https://ror.org/021nxhr62
EF-1427453
2020-08-17T00:00:00Z
2020-08-17T00:00:00Z
en
13217648 bytes
3
CC0 1.0 Universal (CC0 1.0) Public Domain Dedication
Protection of coastal ecosystems from deforestation may be the best way to
protect coral reefs from sediment runoff. However, given the importance of
generating economic activities for coastal livelihoods, the prohibition of
development is often not feasible. In light of this, logging
codes-of-practice have been developed to mitigate the impacts of logging
on downstream ecosystems. However, no studies have assessed whether
managed land-clearing can occur in tandem with coral reef conservation
goals. This study quantifies the impacts of current land use and the risk
of potential logging activities on downstream coral reef condition and
fisheries using a novel suite of linked land-sea models, using
Kolombangara Island in the Solomon Islands as a case study. Further, we
examine the ability of erosion reduction strategies stipulated in logging
codes-of-practice to reduce these impacts as clearing extent increases. We
found that with present-day land use, reductions in live and branching
coral cover and increases in turf algae were associated with exposure to
sediment runoff from catchments and log ponds. Critically, reductions in
fish grazer abundance and biomass were associated with increasing sediment
runoff, a functional group that accounts for ~25% of subsistence fishing.
At low clearing extents, although best management practices minimises the
exposure of coral reefs to increased runoff, it would still result in 32%
of the reef experiencing an increase in sediment exposure. If clearing
extent increased, best management practices would have no impact, with a
staggering 89% of coral reef area at risk compared to logging with no
management. Synthesis and applications: Assessing trade-offs between
coastal development and protection of marine resources is a challenge for
decision makers globally. Although development activities requiring
clearing can be important for livelihoods, our results demonstrate that
new logging in intact forest risks downstream resources important for both
food and livelihood security. Importantly, our approach allows for
spatially-explicit recommendations for where terrestrial management might
best complement marine management. Finally, given the critical degradation
feedback loops that increased sediment runoff can reinforce on coral
reefs, minimising sediment runoff could play an important role in helping
coral reefs recover from climate-related disturbances.
Bathymetry Bathymetry is a key driver of coral reef communities and
sediment dispersal (Rude et al., 2015, Knudby et al., 2010), yet the
freely-available bathymetry data [The General Bathymetric Chart of the
Oceans (GEBCO)] was not suitable for our nearshore study region given how
coarse-scale it is. We therefore developed a tool on Google Earth Engine
(GEE) to calculate bathymetry from Sentinel-2 satellite imagery. While the
GEBCO data spatial resolution is 30 arc-seconds (roughly 930 x 930 m grid
cells at our study site) the spatial resolution of the GEE generated
bathymetry layer is 10 x 10 m grid cells. Since the region is frequently
cloud covered, a composite approach had to be used into order to create a
spatially complete output. All Sentinel-2 image from the study region with
less than 50% scene cloud cover were included. For each image, top of
atmosphere reflectance values were re-scaled from 0 to 1 then clouds and
land were masked using the image QA bands and a simple normalized
difference vegetation index (NDVI) threshold respectively. Areas of
sunglint were identified as water pixels where red values were greater
than 0.06 and red values were greater than 0.15. Deglinting was carried
out using a modified version of Hedley et al. (2005). We used a linear
regression to find the slope between the NIR band and the blue, green, and
red bands. The blue, green, red, and NIR bands were then deglinted on a
pixel by pixel basis by subtracting the product of the band slope and the
difference of the NIR value and NIR value of an open ocean pixel. The
in-situ depth points (see Table S1) were used to extract the
glint-corrected blue, green, red, and NIR pixels values for their
locations. Depth for each image was then calculated using the Lyzenga
method (Lyzenga et al., 2006). We ran a multiple linear regression between
the in-situ depth and the glint corrected values of the four bands and
calculated the first percentile values for the four glint-corrected bands
and a composite bathymetry layer was created by taking the mean depth
calculated for all available images. The accuracy was checked with a
linear regression with r2 = 0.59 and Root Mean Square Error = 1.8, with
the Google Earth Engine tool overestimating depth. Wave exposure and
sediment entrainment Wave exposure was determined by using the model
Simulating WAves Nearshore (SWAN) 41.20 (Booij et al., 1999). This model
was developed specifically for propagating deepwater waves to nearshore
environments and has been used extensively in a variety of coastal
settings, including coral reefs (e.g. Baldock et al. (2014)). The wave
boundary conditions for the model were determined from the average
long-term deep water wave height from 1979-2014 from the National Oceanic
and Atmospheric Administration (NOAA) Wave Watch 3. The bathymetry for the
SWAN model was a merged global bathymetry layer (GEBCO) reprojected to 100
x 100 m grid cells for deep water environments and the 10 x 10 m shallow
water bathymetry derived from GEE used for the reef zones. Significant
wave height (Hs) and wave period (T) were extracted from the SWAN model
for the reef slopes of Kolombangara. Significant wave height from SWAN was
used to produce a Rayleigh distributed histogram of wave heights that
represent the likely wave climate at the river discharge sites (Svendsen,
2006). The bin range of the histogram was every 0.01 m and between 0 to
twice the maximum Hs modelled for the reef slope (i.e. an approximation of
the largest individual waves, (Komar, 1998). This histogram represents the
synthetic wave climate for the nearshore zones of Kolombangara reef. The
near-bed wave orbital velocities from the synthetic wave climate was
determined at 10 metres depth using linear wave theory (Komar, 1998).
Sediment model Sediment transport routines were developed to determine the
likely location of settlement of suspended sediment from the sediment load
discharge. Annual sediment load discharges into the reef environment were
derived from running InVEST SDR models for each river outlet on
Kolombangara (Wenger et al. 2018). The transport, deposition, and
resuspension of suspended sediment depends on several complex physical and
bio-geochemical processes (Bainbridge et al., 2018, Soulsby, 1997). Given
the data limitations, we were not able to account for all of these
processes, including the role of freshwater in dispersing sediment
particles, complex transport of sediment of different sizes, or
flocculation of suspended particulate matter. However, we were able to
account for the likely distribution of suspended sediment, shallow water
bathymetry, ocean current speed and direction, and nearshore wave climate.
This approach produced sediment transport maps that err on the side of
longer transport distances of very fine sediment particles known to
adversely impact coral reef organisms (Bainbridge et al., 2018). The
percentage of sediment entrained on the coral reef slope was determined
from the proportion of near-bed wave orbital velocities in the synthetic
wave climate that were above the critical orbital velocity required to
entrain mud (grain size 63 µm), as defined by methods in Soulsby (1997).
That is, if 40% of the wave orbital velocities were above the critical
entrainment velocity, 40% of the sediment load was transported from the
location of river outflow. The transport direction of entrained sediment
in suspension was determined from the average current speed and direction
between 1993 – 2015 derived from the E.U. Copernicus Marine Service
Information (GLORYS12V1 product). The current speed and fall velocity of
fine muds was used to determine the transport distance of entrained
sediment by multiplying current speed by the time taken for sediment to
fall to 10 m depth (e.g. Rude et al. 2015, Soulsby 1997). If sediment
moved to a location where sediment entrainment was still likely it was
further transported until it settled in a lower energy zone. This process
was applied for each logging scenario and the respective sediment load
discharge into the reef environment. Since nearshore coral reefs have
evolved in variable water quality conditions and may be adapted to turbid
environments (Anthony and Fabricius, 2000), the difference in sediment
load between no-logging (i.e. prior to human impact) and logging scenarios
were also examined to understand the change in sediment load a reef can
tolerate before an ecosystems shift occurs. No-logging sediment run-off
was determined by artificially re-foresting the islands and examining what
the natural runoff regimes were.
File name descriptions: bathymetry.tif-the final bathymetry layer produced
by the desribed methods sediment_pres.tif- A map of patterns of sediment
runoff from present day landuse s1_10.tif-A map of patterns of sediment
runoff from logging activities with no management in place and 10% area
cleared s1_20.tif-A map of patterns of sediment runoff from logging
activities with no management in place and 20% area cleared s1_30.tif-A
map of patterns of sediment runoff from logging activities with no
management in place and 30% area cleared s1_40.tif-A map of patterns of
sediment runoff from logging activities with no management in place and
40% area cleared s6_10.tif-A map of patterns of sediment runoff from
logging activities with no management in place and 10% area cleared
s6_20.tif-A map of patterns of sediment runoff from logging activities
with no management in place and 20% area cleared s6_30.tif-A map of
patterns of sediment runoff from logging activities with no management in
place and 30% area cleared s6_40.tif-A map of patterns of sediment runoff
from logging activities with no management in place and 40% area cleared
waves.tif-the final waves layer produced by the desribed methods