Preventable fine sediment export from the Burdekin River catchment reduces coastal seagrass abundance and increases dugong mortality within the Townsville region of the Great Barrier Reef, Australia

Abstract

The coastal seagrass meadows in the Townsville region of the Great Barrier Reef are crucial seagrass foraging habitat for endangered dugong populations. Deteriorating coastal water quality and in situ light levels reduce the extent of these meadows, particularly in years with significant terrestrial runoff from the nearby Burdekin River catchment. However, uncertainty surrounds the impact of variable seagrass abundance on dugong carrying capacity. Here, I demonstrate that a power-law relationship with exponent value of − 1 (R2 ~ 0.87) links mortality data with predicted changes in annual above ground seagrass biomass. This relationship indicates that the dugong carrying capacity of the region is tightly coupled to the biomass of seagrass available for metabolism. Thus, mortality rates increase precipitously following large flood events with a response lag of < 12-months. The management implications of this result are discussed in terms of climate scenarios that indicate an increased future likelihood of extreme flood events.

Full text

UNCORRECTED PROOF
ARTICLE INFO
Article history:
Received 18 March 2016
Received in revised form 11 October
2016
Accepted 20 October 2016
Available online xxx
Keywords:
Seagrass
Dugong
Turbidity
Runoff
Resilience
Bayesian network
ABSTRACT
The coastal seagrass meadows in the Townsville region of the Great Barrier Reef are crucial seagrass foraging habitat
for endangered dugong populations. Deteriorating coastal water quality and in situ light levels reduce the extent of these
meadows, particularly in years with significant terrestrial runoff from the nearby Burdekin River catchment. However,
uncertainty surrounds the impact of variable seagrass abundance on dugong carrying capacity. Here, I demonstrate that
a power-law relationship with exponent value of − 1 (R2~ 0.87) links mortality data with predicted changes in annual
above ground seagrass biomass. This relationship indicates that the dugong carrying capacity of the region is tightly cou-
pled to the biomass of seagrass available for metabolism. Thus, mortality rates increase precipitously following large
flood events with a response lag of < 12-months. The management implications of this result are discussed in terms of
climate scenarios that indicate an increased future likelihood of extreme flood events.
© 2016 Published by Elsevier Ltd.
Marine Pollution Bulletin xxx (2016) xxx-xxx
Contents lists available at ScienceDirect
Marine Pollution Bulletin
journal homepage: www.elsevier.com
Preventable fine sediment export from the Burdekin River catchment reduces coastal
seagrass abundance and increases dugong mortality within the Townsville region of
the Great Barrier Reef, Australia
Scott A Wooldridge ⁎
Catchment to Reef Management Solutions Ltd, Newcastle, NSW, Australia
1. Introduction
Coastal seagrass meadows are an essential component of healthy
marine ecosystems. Seagrasses provide crucial nursery habitat for nu-
merous marine organisms as well as other ecologically important and
economically valuable ecosystem services such as coastal protection,
nutrient cycling and particle trapping (Coles et al., 1993; Costanza et
al., 1997; Cullen-Unsworth et al., 2014). The extensive coastal sea-
grass meadows found in the Townsville region of the Great Barrier
Reef (GBR, Australia; Fig. 1a) are of particular importance since they
are globally significant habitat for critically-endangered dugong pop-
ulations. This significance was recognised by the declaration of a
Dugong Protected Area that encompasses Cleveland Bay and adjoin-
ing Bowling Green Bay (Great Barrier Reef Marine Park Authority,
2003).
Despite their known importance, experts consider the coastal sea-
grass meadows within the Townsville region to be at ‘high risk’
of periodic declines and/or loss (Rasheed et al., 2007; Grech et al.,
2008, 2011; Coles et al., 2015). This inherent risk profile is con-
sistent with the fact that the ‘observed’ extent of coastal seagrass
meadows in the Townsville region, is often considerably less than
the predicted ‘potential’ (Grech and Coles, 2010; Davies et al., 2013;
Coles et al., 2015). Deteriorating water clarity in the coastal zone is
a key contributor to the loss of potential seagrass habitat (Waycott
et al., 2005; Petus et al., 2014; Coles et al., 2015). Reduction in
benthic light, measured as photosynthetic active radiation (PAR) is
a primary factor controlling the health and abundance of seagrasses
(Duarte, 1991; Longstaff and
⁎Corresponding author.
Email address: swooldri23@gmail.com (S.A. Wooldridge)
Dennison, 1999; Ralph et al., 2007; Collier et al., 2012). Even rela-
tively minor changes in PAR availability can lead to large-scale loss
of seagrasses over relatively short time scales (~ weeks to months)
(Collier et al., 2012).
Poor coastal water clarity may thus be acting to limit the dugong
population carrying capacity of the Townsville region. Dugongs feed
almost exclusively on shallow water (< 10 m) seagrasses, most no-
tably from the families Potamogetonaceae and Hydrocharitaceae,
particularly pioneer species such as Halophila and Halodule genera
(Preen, 1995a, 1995b). This dependency provides a strong connec-
tion between coastal seagrass availability and dugong population dy-
namics, with several studies from other regions of the GBR attribut-
ing local declines in dugong density to major seagrass loss (Preen and
Marsh, 1995; Marsh and Kwan, 2008; Meager and Limpus, 2014).
Moreover, the preferred Halophila and Halodule seagrass species
have very low tolerance to prolonged low-light conditions (Longstaff
and Dennison, 1999; Collier et al., 2012). This means that any re-
duction in coastal water quality can quickly (~ weeks) act to shrink
dugong feeding areas, and ultimately lower the dugong carrying ca-
pacity of a region.
A range of natural and anthropogenic sources (e.g. localised dredg-
ing, coastal and urban development) contribute to the threat of dete-
riorating coastal water clarity in the Townsville region (reviewed by
Haynes et al., 2001). However, it is the impact of sediment and nutri-
ent pollutants exported from the 133,400 km2Burdekin River catch-
ment (Fig. 1b) that is now understood to be the consistent (long-term)
dominating feature (Logan et al., 2013; Fabricius et al., 2014). The
Burdekin River catchment is the single greatest source of suspended
sediments into the GBR lagoon (Kroon et al., 2012). During an av-
erage year, ~ 4 million tonnes are exported through the agricultural-
http://dx.doi.org/10.1016/j.marpolbul.2016.10.053
0025-326/© 2016 Published by Elsevier Ltd.

UNCORRECTED PROOF
2 Marine Pollution Bulletin xxx (2016) xxx-xxx
Fig. 1. (a) The coastal waters of the Townsville region (shaded area) contain extensive seagrass meadows, particularly in the sheltered, north facing bays (e.g. Cleveland Bay, Bowl-
ing Green Bay, Bushland Beach). (b) The water clarity in these bays is strongly impacted by annual sediment export from the nearby Burdekin River. The Burdekin River catchment
and adjoining Fitzroy River catchment make up two thirds of the total watershed area draining into the Great Barrier Reef lagoon.
dominated river drainage network, which represents ~ 25% of total
loads entering into GBR (Kroon et al., 2012). The erosion processes
supplying suspended sediment material to the Burdekin River have
been magnified by modern agricultural management practices. Most
notably, this is due to the impact of rangeland beef grazing and the as-
sociated loss of pasture cover, leading to hillslope and gully erosion
(Thorburn et al., 2013; Bartley et al., 2014). In total, it is estimated that
annual sediment export from Burdekin River is now ~ 8 times higher
than existed before European settlement ~ 160 years ago (Kroon et al.,
2012).
Controversy had surrounded the significance of anthropogenic al-
terations to Burdekin River sediment loads in terms of the resultant
impact on coastal water clarity. Originally, it was believed that GBR
water clarity was not limited by modern sediment supply (Larcombe
and Woolfe, 1999) since thick deposits of terrigenous sediments al-
ready existed along the coastal wedge – having accumulated over ge-
ological timescales (Belperio, 1983). However, well calibrated satel-
lite observations clearly demonstrate that new materials do signifi-
cantly contribute to reducing water clarity in the GBR lagoon, even
within the more turbid coastal zone (Weeks et al., 2012; Logan et
al., 2013; Fabricius et al., 2014). In summary, geological deposits to-
gether with newly imported materials, most specifically the fine sedi-
ment fraction, collectively determine coastal water clarity (further de-
tails are provided as Supplementary material, ‘Sediment dynamics in
the Townsville Region’).
The importance of this new conceptual understanding should not
be understated, since it implies that successful land management prac-
tices leading to load reductions in fine sediment and nutrient export
from the Burdekin River could measurably improve coastal water clar-
ity in the Townsville region – with potentially significant ecosys-
tem health benefits for seagrass abundance and dependent foragers
(e.g., dugongs, green turtles). However, fundamental pieces of this
scientific puzzle are currently missing in order to assess these po-
tential benefits. Firstly, a quantitative platform is not available to
map the impact of variable water clarity on the health and extent
of coastal (< 10 m) seagrass
meadows. Secondly, uncertainty surrounds the impact of variable sea-
grass abundance on dugong carrying capacity. Finally, only cursory
estimates are available for the potential improvement in coastal water
clarity due to targeted reductions in end-of-river fine sediment loads
from the Burdekin catchment (Brodie et al., 2016). In this paper, I
utilise existing data sets to provide scientific solutions for the first two
issues. The development of a predictive decision support tool that can
resolve the final issue is currently underway, and will be reported else-
where.
2. Methods
I chose Cleveland Bay as a representative location with which to
calibrate quantitative water clarity relationships for the Townsville
region, particularly in relation to seagrass health. The decision was
guided by the extensive database that exists for Cleveland Bay, which
includes historical time-series for water quality, climate and seagrass
abundance for the period of interest (2002 − 2012). The key model-
ling challenge was to link observed changes in the long-term trend of
daily water clarity with changes in seagrass abundance within Cleve-
land Bay. This task is complicated by the fact that simulating changes
in seagrass abundance requires insight into how both meadow area
(km2) and within-meadow cover (%) vary under different water clarity
regimes (Fig. 2).
2.1. Burdekin River sediment loads and coastal water clarity
Remotely-sensed (MODIS-Aqua) estimates of mean daily ‘Photic
depth’ (Z%; unit: m) have previously been developed to test the in-
fluence of daily Burdekin River loads on coastal water clarity across
the period 2002–2012 (c.f. Logan et al., 2013; Fabricius et al., 2014).
Photic depth is a measure to quantify light availability (as photosyn-
thetically active radiation, PAR) relative to the light at the water sur-
face. For example, Z10% is the photic depth where 10% of surface
PAR is still available (Weeks et al., 2012). The GBR-validated algo

UNCORRECTED PROOF
Marine Pollution Bulletin xxx (2016) xxx-xxx 3
Fig. 2. Conceptual representation of how water clarity impacts upon the availability of seagrass for dugong feeding within Cleveland Bay (Townsville). (a) High turbidity (= low
water clarity) limits the transmission of photosynthetic active radiation (PAR) needed to maintain seagrass health, thereby reducing both seagrass cover (%) and the maximum col-
onization depth. (b) Management actions that improve water clarity can reverse this trend, leading to higher percent seagrass cover and deeper colonization depths. The resultant
increase in seagrass biomass benefits the health and sustainable population size of dependent foragers (e.g., dugongs). The relatively steep depth profile within Cleveland Bay means
that even relatively small changes in colonization depth can lead to large area changes in viable seagrass habitat.
rithm permits Z10% to be directly equated with traditional in situ sec-
chi depth measurements of water clarity (Weeks et al., 2012).
After statistically removing the near instantaneous effects of lo-
calised (e.g. waves, tides and bathymetry) and seasonal (wet/dry) dri-
vers, the long-term trend cycle in daily mean photic depth (Z10%)
dataset clearly demonstrates that Burdekin River discharge is a strong
determinate of intra- and inter-annual variation in coastal water clar-
ity for the Townsville region. This is made particularly obvious by the
fact that there were strong differences in the freshwater discharge vol-
umes of the Burdekin River between water years, with four dry water
years (2003–2006) being followed by six wet years with on average
64.4% greater discharge volumes (2007–2012) (Fig. S1a). In this case,
annual mean photic depth was 20–30% reduced in the six wet com-
pared to four dry years (Fig. 3a).
2.2. Light attenuation model
Several researchers have investigated the minimum light require-
ments for coastal Queensland seagrass species (Collier et al., 2012;
Chartrand et al., 2012; Petrou et al., 2013). For the present study, I
was particularly interested in minimum light level ‘thresholds’ that
align with the response dynamic recorded by the longer-term trend in
daily mean photic depth (e.g., needed to survive ~ 3 month periods of
depressed light levels resulting from large flood events). Of particu-
lar relevance are the in situ observations of Collier et al. (2012), who
found that for the longer-term survival of Queensland coastal seagrass
species the percent (%) of days below 3 mol m− 2 d− 1 was the best pre-
dictor of in situ seagrass losses. For example, at Magnetic Island the
developed relationship suggests that 30 days below the threshold in
any given 3-month period leads to a 50% decline in seagrass cover,
and ~ 60 days a 100% decline (Collier et al., 2012).
To help identify the location (and temporal duration) of excur-
sions below the 3 mol m− 2 d− 1 minimum light threshold, I applied
the Beer-Lambert equation to spatially extrapolate benthic irradiance
across the entire modelling domain:
where Ezis the benthic irradiance, E0is irradiance beneath the sur-
face corrected for 5% surface reflectance (Kirk, 1991), KDis a light
attenuation coefficient, and Zis water depth. Surface irradiance (E0,
μmol s− 1 m−) observations were available from a local weather station
at Magnetic Island, whilst water depths (m below MSL) were avail-
able for every 100 m grid cell (Beaman, 2012). A calibration of the
light attenuation coefficient (KD, units) for the local turbidity regime
has previously been developed based on the optical properties of the
Townsville coastal zone (Cooper et al., 2008).
2.3. Seagrass
The Reef Rescue Marine Monitoring Program (MMP) employs
‘Seagrass-Watch’ protocols (McKenzie et al., 2007; www.
seagrasswatch.org) to conduct intensive small-scale (50 m × 50 m
area) seagrass monitoring at mainland coastal sites (Bushland Beach
and Cape Pallarenda) and on Magnetic Island (Picnic Bay). Sites are
monitored for seagrass cover and species composition. The main shal-
low water species are Halophila ovalis,Halodule uninervis,Zostera
capricorni, and Cymodocea serrulata (Fig. S2; Davies et al., 2013).
Photographic standards have been developed that help visualise
changes in percent seagrass cover (%) for a typical coastal commu-
nity type in the Townsville region (Fig. S3; www.seagrasswatch.org).
Empirical calibrations also exist that link percent cover (%) and above
ground dry weight biomass estimates (gDW m− 2) for the dominant
Halophila,Halodule and Zostera species (Fig. S4; Thomas, 2008).
Across the study period (2002–2012), the three MMP sites vari-
ously recorded very similar temporal trends in percent cover changes,
with maximum values (30–40%) around 2007 and minimum values
(0–5%) in 2011 (Fig. 3b). Linked changes in daily mean photic depth
(1)

UNCORRECTED PROOF
4 Marine Pollution Bulletin xxx (2016) xxx-xxx
Fig. 3. Time-series changes (~ 3 month scale) for (a) coastal photic depth, and (b) %
seagrass cover. (c) The strong temporal coherence (R2= 0.67) identifies photic depth as
a good predictor of changes in % seagrass cover, with a response lag < 6-months. It is
important to note that during the study period (2002–2012) a seed bank was always pre-
sent at each study site (Davies et al., 2013). In the absence of a seed bank or remnant
seagrass population, recovery of disturbed seagrass meadows can become disconnected
from temporal changes in water clarity (Rasheed et al., 2014).
(Fig. 3a) map as a strong driver of this temporal variability (R2= 0.67,
Fig. 3c).
2.4. Dugong mortality data
The Queensland Marine Wildlife Strandings and Mortality Pro-
gram maintain an online database (StrandNet) that records the death
of all dugongs in Queensland, Australia. Records are obtained from
government departments, community groups, environmental organisa-
tions, and the general public. All records are verified by experts. The
probable cause of death is established through necropsies by veteri-
narians, examination of carcasses by trained staff or, in some cases,
through photos and/or case histories. Only dugong mortalities that
were either attributed to natural or unidentified causes are consid-
ered in this paper (i.e., the analysis does not include boat strikes or
drownings). Across the analysis period (2002–2012) no mass mortal-
ity events (e.g. land strandings) occurred in the dataset, though clear
annual changes in mortality numbers are evident (Fig. S5). In most
years, the Townsville region dominates the response profile (up to
50%), especially in the larger freshwater discharge years (2007–2012).
There are obvious limitations with the StrandNet data that need to
be considered in the present analysis. For example, it is understood
that the data only captures a proportion of the total number of dead
dugongs. The number of carcasses that reach the shoreline depends
on factors such as currents, winds, carcass buoyancy, and losses to
scavengers (Peltier et al., 2012). There is also a potential bias for a
higher reporting of deaths in regions that are closer to urban centres.
However, as long as these reporting errors remain relatively constant
through time (e.g., the ratio of the number of dugong washed ashore
versus total number of dead) then ecological impacts can still be attrib-
uted, even if absolute mortality numbers remain uncertain (see Results
for details).
2.5. Decision support framework: EcoNet
The Netica (http://www.norsys.com) network editor was used to
construct a spatial Bayesian Network (BN) model (Wooldridge and
Done, 2004, 2009) from the above outlined relationships that link
changes in water clarity and seagrass abundance within Cleveland
Bay. A BN is a decision support tool which, when applied in an eco-
logical setting, depicts and quantifies the strength (i.e. certainty/uncer-
tainty) of dependency relations among environmental and ecological
factors that might influence outcomes of interest – in this case, the
likelihood that the benthic light regime at any given 100 m grid cell
location within Cleveland Bay can support a viable seagrass commu-
nity for the chosen depth, water clarity and surface irradiance combi-
nation (c.f., Eq. (1)). That is, by integrating the known spatial values
for E0,Z, and coincident photic depth (used to calculate KD; Cooper et
al., 2008) it was possible to apply Eq. (1) and calculate benthic irradi-
ance (Ez) at every 100 m grid cell in the model domain; including the
long-run (seasonal) likelihood that Ezwas above the 3 mol m− 2 d− 1
threshold level needed to maintain a viable seagrass community. In
this case, ‘likelihood’ was calculated based on the % of days (in
any given 3-month period/season) below the minimum 3 mol m− 2 d− 1
light threshold (i.e. 33% = 30/90 days; 67% = 60/90 days). Consistent
with the local seagrass response data (c.f. Collier et al., 2012), any
location that had > 60 days below the minimum light threshold was
deemed unlikely to contain a viable seagrass community.
For the present application, the spatial BN model (herein referred
to as EcoNet) was hard coded to reproduce the outlined (data-driven)
parent → child regression relationships. However, to account for un-
certainty in the calibrations, a standard 10% error (assumed normally
distributed) was applied around the central tendency of each determin-
istic relationship. In this way, the implication of ‘local’ uncertainty in
each parent → child relationship is faithfully propagated via the laws
of Bayesian probability (Lauritzen and Spiegelhalter, 1988), through
to the final ‘system-level’ prediction node (viz., the likelihood that the
location is viable seagrass habitat for the chosen input scenario).
3. Results
3.1. Viable seagrass habitat
EcoNet was applied with both summer and winter irradiance data.
Mean summer surface irradiance (PAR ~ 550 μmol s− 1 m− 2) is ap-
proximately 60% higher than winter levels (PAR
~ 350 μmol s− 1 m− 2). It has previously been discussed how this large
seasonal difference (alone) is a strong driver of change in the maxi-
mum depth of light penetration, and by inference viable seagrass habi-
tat within Cleveland Bay (Anthony et al., 2004). Fig. 4 helps to visu-
alise the comparative difference in spatial extent of suitable seagrass
habitat for a low flow (‘dry’) year (e.g. 2007) and a high flow (‘wet’)
year (e.g. 2011). In this case Z10% differs by ~ 1 m. It is important to
remember that EcoNet only accounts for benthic ‘light threshold’ con

UNCORRECTED PROOF
Marine Pollution Bulletin xxx (2016) xxx-xxx 5
Fig. 4. Map of suitable areas within Cleveland Bay (Townsville) that can support viable seagrass communities based on benthic light constraints. The meadow area shrinks by
~ 15–20% in a wet (2011) versus dry (2007) year. However, the reduction in total seagrass biomass will be > 15–20% since the percent seagrass cover within the suitable areas also
falls disproportionately (as per Figs. 2, 3c). The constrained meadow area in winter (versus summer) is driven by large seasonal differences (~ 60%) in surface irradiance levels.
straints when assessing viable seagrass habitat. It is known that other
factors, such as sediment deposition and scouring, tidal fluctuations,
desiccation, fluctuating temperature and variable salinity can also
limit the capacity for coastal benthos to support a viable seagrass com-
munity (summarised by Carruthers et al., 2002). As such, it is impor-
tant to note that EcoNet represents a first order prediction of ‘poten-
tial’ (versus realised) viable seagrass habitat.
3.2. Seagrass biomass
Changes in viable seagrass habitat area were converted into equiv-
alent biomass estimates by combining the expanded meadow area
(km2) with the concurrent within-meadow percent cover (%) for the
predicted water clarity regime (viz. Fig. 3c); and it subsequent con-
version into an above ground dry weight biomass equivalent (viz.
Fig. S4). To help simplify this process, a ‘typical’ seagrass com-
munity in Cleveland Bay was assumed to be well represented by a
mixture of 70% Halophila/Halodule and 30% Zostera (see Davies et
al., 2013), and in the absence of better information, within-meadow
cover (%) was assumed to be uniform across the viable seagrass
meadow area. The annual total biomass was represented by the av-
erage of summer and win
ter values. For the Cleveland Bay study domian, it is evident that large
changes in annual seagrass biomass (up to 10-fold) can occur between
comparatively wet and dry years.
3.3. Dugong mortality
Comparison of dugong mortality count data for the Townsville re-
gion (Fig. S5) with predicted changes in annual above ground sea-
grass biomass (Fig. 5a) shows strong temporal trends in the form of
a power-law relationship with exponent value of − 1 (R2~ 0.87; Fig.
5b). Power-laws are a common attribute arising from metabolic analy-
ses of ecological systems (reviewed by Brown et al., 2004; Marquet et
al. 2005). This makes sense because the maximum number of individ-
uals that a species can achieve in a given area, will be proportional to
the ratio between the rate of resource supply per unit area and the av-
erage per individual rate of resource use (i.e. needed to sustain metab-
olism). An exponent value of − 1 is consistent with ‘the linear biomass
hypothesis’ (c.f. Sheldon et al., 1986), which has commonly been ob-
served in aquatic ecosystems. Moreover, the exponent value of − 1 can
still arise even if total dugong mortality counts are underrepresented
(e.g. due to carcasses that don't wash ashore), provided that the missed

UNCORRECTED PROOF
6 Marine Pollution Bulletin xxx (2016) xxx-xxx
Fig. 5. (a) Predicted annual change in above ground dry weight of seagrass biomass
within the Cleveland Bay model domain. (b) Power-law relationship between the avail-
able biomass of seagrass and the number of dead dugong in the Townsville region.
proportion remain relatively constant through time – such that re-
ported counts act as a proxy record for total mortality.
A nice feature of power-law relationships is that they are scale in-
variant, that is, they possess the same statistical properties at any ob-
servational scale. Thus, a change in the scale of the independent vari-
able (above ground seagrass biomass) preserves the functional form of
the original relationship. In practical terms, this means that the same
metabolic processes are at work no matter what the scale of analysis.
For the present application, this provides confidence that the represen-
tative study domain (Cleveland Bay) captures the correct functional
response of dugong populations to limiting seagrass supply across
their entire feeding range. Namely, that the dugong carrying capacity
of the Townsville region is tightly coupled to the biomass of seagrass
available for metabolism.
4. Discussion and future prospects
The coastal areas of the Great Barrier Reef (GBR) support a signif-
icant proportion of the worlds remaining dugong populations, which
is one of the reasons for its World Heritage listing (Great Barrier
Reef Marine Park Authority, 1981). Yet despite this fact, a recent sur-
vey (Nov 2011) in the Southern GBR, which spans over 1000 km of
coastline from Rockhampton → Townsville → Cairns, indicates that
the population size fell to its lowest size (~ 500 dugong; Stobtzick
et al., 2012) since surveys began in the mid-1980s. Strong evidence
also exists to suggest that this decline has been occurring for many
decades before the aerial surveys began (Marsh et al., 2005). Quanti-
tative frameworks (e.g. EcoNet) that can help to establish the proximal
drivers of this decline, including methods to halt and reverse the trend,
are thus critical for the management of the World Heritage Area.
As confirmed here, there can be little doubt that seagrass abun-
dance along the coastal areas of the southern GBR is significantly
reduced by
poor water clarity arising from both the immediate (flood) and longer
term impact of terrestrial fine sediment export (Figs. 3, 4, 5). Shrink-
ing seagrass meadow areas challenge the dugong carrying capacity
of a region (Sheppard et al., 2007). When their food supply fails, in-
dividual dugongs variously exhibit one of two functional responses:
(1) move from the affected area in search of seagrass with unknown
consequences, or (2) stay and consume any remaining seagrass and
low quality food such as algae, postpone breeding and ultimately risk
mortality (reviewed by Marsh et al., 2011). Whilst it is known that
dugongs have the capacity to move hundreds of kilometres in only
a few days (Sheppard et al., 2006), it is presently unknown which
combination of factors (e.g. age, sex, physical condition, matrilineally
transmitted learned behaviour) underpin apparently highly individual-
istic movement patterns (reviewed by Marsh et al., 2011). For exam-
ple, in one study of 10 dugongs, only four made substantial journeys
(Preen reported in Marsh et al., 1999). It is important to note that in
most cases, the decision to move or stay may be equally perilous in
terms of mortality risk since there is no a priori guarantee that distant
seagrass meadows will be in better condition.
The potential for dugongs to migrate away from impacted areas
challenges the development of regional relationships that endeavour to
link seagrass abundance and dugong carrying capacity, and this sce-
nario may confound the historical low numbers of dugongs observed
in the southern GBR surveys in 2011. Yet, there are reasons why
this may not be the case for this particular region. The southern GBR
dugong population includes key seagrass feeding areas (e.g. Shoalwa-
ter Bay, Ince Bay, Newry Region, Upstart Bay, Bowling Green Bay,
Cleveland Bay, Hinchinbrook; Lawler et al., 2002) that are impacted
by river discharge from just two major river basins, namely the Bur-
dekin and Fitzroy catchments. Taken together these two river sys-
tems make up two thirds of the total watershed area draining into the
GBR, accounting for > 70% of the annual sediment load, > 75% of
the anthropogenic total nitrogen load, and ~ 55% of the anthropogenic
total phosphorus load exported to the GBR lagoon (Kroon et al.,
2012). The combined annual impact from these river systems thus has
the capacity to disrupt coastal water clarity and seagrass abundance
for well over a 1000 km from Rockhampton → Townsvill → Cairns
(Fabricius et al., 2016). This scenario may limit the capacity of dugong
to escape the ‘reach’ of significant seagrass losses, particularly in
large flood years. Indeed, this may be a contributing factor as to why
StrandNet consistently identifies the Townsville region as a mortality
‘hotspot’ (Fig. S5). For example in 2011, Townsville accounted for
~ 50% of the total recorded deaths along the entire Queensland coast-
line.
The fact that no calves were seen in the Southern GBR during the
2011 survey (Stobtzick et al., 2012) lends support to the belief that
low seagrass abundance was a key factor responsible for the high mor-
tality rates observed in the Townsville region (see also Fuentes et al.,
2016). The impact of food shortages on dugong calf counts is expected
to be lagged by two years (Marsh et al., 2011), indicating that seagrass
was already of limited availability in the southern GBR before the se-
vere floods and cyclone (Yasi) of the summer of 2010–11. Seagrass
monitoring data from across the region records the details of the situ-
ation: (i) 75% of the monitored sites had declined in abundance over
the previous 12 months, (ii) 80% showed a declining long-term trend
(5–10 years); (iii) 55% of sites exhibited a shrinking meadow area,
(iv) 90% of sites were suffering from light limitation, and (v) the vast
majority of sites had limited to no seed production, essential for rapid
recovery (see McKenzie et al., 2012).
The regional decline in seagrass abundance across this period was
most obviously exacerbated by six consecutive very wet years
(2007–2012), wherein annual discharge volumes were on average
64.4% greater than normal (Fig. S1a). The impact of a prolonged
(multi-year) wet period can be understood to escalate regional losses
in seagrass abundance. Firstly, the concatenated load of resuspendable

UNCORRECTED PROOF
Marine Pollution Bulletin xxx (2016) xxx-xxx 7
material within the coastal zone magnifies the proportion of seagrass
habitat affected by non-sustainable light limitation, thereby shrink-
ing meadow areas. Secondly, the prolonged reduction in water clar-
ity diminishes the capacity of seagrass to build energy reserves even
within the remaining viable areas, leading to lower reproductive effort
and seed production. With a healthy seed bank, and adequate in situ
light levels the estimated recovery time for a pioneer community of
25% of previous mean cover (Halophila dominated) is expected to be
1–2 years (McKenzie et al., 2012). However, recovery times are much
less predictable (if at all) once adult plants and seed banks are lost.
Under this scenario, the recovery process is reliant on chancy vege-
tative fragment dispersal from outside the impacted area, the likeli-
hood of which is further diminished as seagrass meadows shrink and
fragment (Rasheed et al., 2014). Slow recovery of heavily impacted
seagrass meadows represent a severe impediment to the maintenance
of healthy and viable dugong populations, i.e., high initial starvation
rates (= reduced number of breeding age females) are compounded by
delays (> 3–5 years) in breeding rates until seagrass abundance returns
to levels that can sustain dugongs in good condition (Marsh and Kwan,
2008).
It could be argued that the coastal ecology of the GBR was simply
‘unlucky’ to experience six consecutive very wet years (2007–2012).
However, this would be to ignore historical data for the region, which
demonstrates that variability in rainfall and river flow has increased
during the twentieth century - with more very wet and very dry ex-
tremes than in earlier centuries (Lough et al., 2015). This trend is also
consistent with predictions for the region as a consequence of global
warming (reviewed by Lough et al., 2006). This new (emerging) cli-
mate regime highlights the need for a strong management focus on
strategies that increase the resilience of coastal seagrass communities,
i.e., reduce seagrass losses in the first instance, and increase recovery
rates between disturbances.
The EcoNet model makes clear that improving long-term ambi-
ent water clarity, as achieved by reducing fine sediment flood export
from the nearby Burdekin River, is fundamental for enhancing sea-
grass resilience in the Townsville region. Encouragingly, an ambition
of the recently-announced ‘Reef 2050 Long-Term Sustainability Plan’
(Australian Government Policy Document, 2015) includes sediment
load reductions of 50% in the Burdekin and Fitzroy rivers by 2025.
However, it remains to be quantitatively tested whether a 50% reduc-
tion in sediment loads is sufficient to avoid mass dugong starvation/
mortality events in the Southern GBR, as occurred in 2011. An exten-
sion of the EcoNet model is currently underway that will facilitate this
computation, and will be reported elsewhere. The simplest inference
from the present analysis is that the required level of improvement
in long-term average daily photic depth will need to be ~ 0.6–1.0 m.
Importantly, the current analysis also highlights that the water clarity
benefits arising from effective sediment reduction strategies can occur
over a relatively short timeframe. For example, photic depth improved
by ~ 1.0 m during the 3-year ‘dry’ period (= surrogate for low sedi-
ment loads) that occurred from 2004 to 2007. The challenge for reef
managers and policy makers is thus to: (i) confirm that the 50% tar-
get level reduction in sediment loads is sufficient from an ecological
point of view, and (ii) maximise the speed of catchment remediation
and land-use practice change needed to bring about end-of-river im-
provements.
Acknowledgements
The ideas outlined in this manuscript benefited from discussions
with staff from the Centre for Tropical Water and Aquatic Ecosystem
Research (James Cook University), including Jon Brodie, Dr. Colette
Thomas, Dr. Catherine Collier, Dr. Steve Lewis, and Dr. Zoe Bain-
bridge. The final manuscript also benefited from suggestions provided
by Prof. Helene Marsh (James Cook University). I am extremely
grateful to Olivia Unicomb (Newcastle University) for creating Fig. 2.
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.
1016/j.marpolbul.2016.10.053.
References
Anthony, K., Ridd, P.V., Orpin, A.R., Lough, J., 2004. Temporal variation of light
availability in coastal benthic habitats: effects of clouds, turbidity, and tides. Lim-
nol. Oceanogr. 49, 2201–2211.
Australian Government Policy Document, 2015. The Reef 2050 Plan. (< http://
www.environment.gov.au/marine/gbr/long-term-sustainability-plan >).
Bartley, R., Bainbridge, Z.T., Lewis, S.E., Kroon, F.J., Brodie, J.E., Wilkinson, S.N.,
Silburn, M.D., 2014. Linking sediment impacts on coral reefs to watershed
sources, processes and management. Sci. Total Environ. 468–469, 1138–1153.
Beaman, R.J., 2012. Great Barrier Reef and Coral Sea Bathymetry. James Cook Uni-
versity, Australia (< http://deepreef.org >).
Belperio, A.P., 1983. Terrigenous sedimentation in the central Great Barrier Reef la-
goon: a model from the Burdekin region. BMR J. Aust. Geol. Geo-
phys. 8, 179–190.
Brodie, J.E., Lewis, S.E., Collier, C.J., Wooldridge, S.A., Bainbridge, Z.T., Water-
house, J., Rasheed, M.A., Honchin, C., Holmes, G., Fabricius, K.E., 2016. Setting
ecologically relevant targets for river pollutant loads to meet marine water quality
requirements for the Great Barrier Reef, Australia: a preliminary methodology and
analysis. Ocean Coast. Manag. (in press).
Brown, J.H., Gillooly, J.F., Allen, A.P., Savage, V.M., West, G.B., 2004. Toward a
metabolic theory of ecology. Ecology 85, 1771–1789.
Carruthers, T.J.B., Dennison, W.C., Longstaff, B.J., Waycott, M., Abal, E.G., McKen-
zie, L.J., W.J., L.L., 2002. Seagrass habitats of northeast Australia: models of key
processes and controls. Bull. Mar. Sci. 71, 1153–1169.
Chartrand, K.M., Ralph, P.J., Petrou, K., Rasheed, M.A., 2012. Development of a
Light-Based Seagrass Management Approach for the Gladstone Western Basin
Dredging Program. DEEDI Publication, Fisheries Queensland, Cairns, 92.
Coles, R.G., Lee Long, W.J., Watson, R.A., Derbyshire, K.J., 1993. Distribution of
seagrasses, and their fish and penaeid prawn communities, in Cairns harbour, a
tropical estuary, Northern Queensland, Australia. Mar. Freshw. Res. 44, 193–210.
Coles, R.G., Rasheed, M.A., McKenzie, L.J., Grech, A., York, P.H., Sheaves, M.,
McKenna, S., Bryant, C., 2015. The Great Barrier Reef World Heritage Area sea-
grasses: managing this iconic Australian ecosystem resource for the future. Estuar.
Coast. Shelf Sci. 153, A1–A12.
Collier, C.J., Waycott, M., McKenzie, L.J., 2012. Light thresholds derived from sea-
grass loss in the coastal zone of the northern Great Barrier Reef, Australia. Ecol.
Indic. 23, 211–219.
Cooper, T.F., Ridd, P.V., Ulstrup, K.E., Humphrey, C., Slivkoff, M., Fabricius, K.E.,
2008. Temporal dynamics in coral bioindicators for water quality on coastal coral
reefs of the Great Barrier Reef. Mar. Freshw. Res. 59, 703–716.
Costanza, R., D'Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg,
K., Naeem, S., O'Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., van den Belt,
M., 1997. The value of the world's ecosystem services and natural capital. Na-
ture 387, 253–260.
Cullen-Unsworth, L.C., Nordlund, L.M., Paddock, J., Baker, S., McKenzie, L.J.,
Unsworth, R.K.F., 2014. Seagrass meadows globally as a coupled social–ecologi-
cal system: implications for human wellbeing. Mar. Pollut. Bull. 83, 387–397.
Davies, J.N., McKenna, S.A., Rasheed, M.A., 2013. Port of Townsville Long-Term
Seagrass Monitoring: September 2012. Centre for Tropical Water & Aquatic
Ecosystem Research (TropWATER) Publication, James Cook University,
Cairns (32 p).
Duarte, C.M., 1991. Seagrass depth limits. Aquat. Bot. 40, 363–377.
Fabricius, K.E., Logan, M., Weeks, S., Brodie, J.E., 2014. The effects of river run-off
on water clarity across the central Great Barrier Reef. Mar. Pollut.
Bull. 84, 191–200.
Fabricius, K.E., Logan, M., Weeks, S.J., Lewis, S.E., Brodie, J.E., 2016. Changes in
water clarity in response to river discharges on the Great Barrier Reef continental
shelf: 2002–2013. Estuar. Coast. Shelf Sci. 173, A1–A15.
Fuentes, M.M.P.B., Delean, S., Grayson, J., Lavender, S., Logan, M., Marsh, H., 2016.
Spatial and temporal variation in the effects of climate variables on dugong calf
production. PLoS ONE 11 (6), e0155675http://dx.doi.org/10.1371/journal.pone.
0155675.
Great Barrier Reef Marine Park Authority, 1981. Nomination of the Great Barrier Reef
by the Commonwealth of Australia for Inclusion on the World Heritage List. UN-
ESCO, 37p.
Great Barrier Reef Marine Park Authority, 2003. Great Barrier Reef Marine Park Zon-
ing Plan 2003.
Grech, A., Coles, R., 2010. An ecosystem-scale predictive model of coastal seagrass
distribution. Aquat. Conserv. 20, 437–444.

UNCORRECTED PROOF
8 Marine Pollution Bulletin xxx (2016) xxx-xxx
Grech, A., Coles, R., McKenzie, L., Rasheed, M., 2008. Spatial risk assessment for
coastal seagrass habitats in the Great Barrier Reef World Heritage Area: a case
study of the dry and wet tropics. In: Report to the Marine and Tropical Sciences
Research Facility. Reef and Rainforest Research Centre Limited, Cairns (24 p).
Grech, A., Coles, R., Marsh, H., 2011. A broad-scale assessment of the risk to coastal
seagrasses from cumulative threats. Mar. Policy 35, 560–567.
Haynes, D., Brodie, J., Christie, C., Devlin, M., Michalek-Wagner, K., Morris, S.,
Ramsay, M., Storrie, J., Waterhouse, J., Yorkston, H., 2001. Great Barrier Reef
Water Quality: Current Issues. Great Barrier Reef Marine Park Authority,
Townsville.
Kirk, J., 1991. Volume scattering function, average cosines, and the underwater light
field. Limnol. Oceanogr. 36, 455–467.
Kroon, F.J., Kuhnert, P.M., Henderson, B.L., Wilikinson, S.N., Kinsey-Henderson, A.,
Brodie, J.E., Turner, R.D.R., 2012. River loads of suspended solids, nitrogen,
phosphorus and herbicides delivered to the Great Barrier Reef lagoon. Mar. Pollut.
Bull. 65, 167–181.
Larcombe, P., Woolfe, K., 1999. Increased sediment supply to the Great Barrier Reef
will not increase sediment accumulation at most coral reefs. Coral
Reefs 18, 163–169.
Lauritzen, S.L., Spiegelhalter, D.J., 1988. Local computations with probabilities on
graphical structures and their application to expert systems. J. Royal Stat.
Soc. B50, 157–194.
Lawler, I., Marsh, H., McDonald, B., Stokes, T., 2002. Current State of Knowledge:
Dugongs in the Great Barrier Reef. CRC Reef Research Centre, Townsville.
Logan, M., Fabricius, K.E., Weeks, S., Canto, M., Noonan, S., Wolanski, E., Brodie,
J.E., 2013. The Relationship between Burdekin River Discharges and Photic Depth
in the Central Great Barrier Reef. Report to the National Environmental Research
Program. Reef and Rainforest Research Centre Limited, Cairns (29 p).
Longstaff, B.J., Dennison, W.C., 1999. Seagrass survival during pulsed turbidity
events: the effects of light deprivation on the seagrasses Halodule pinifolia and
Halophila ovalis. Aquat. Bot. 65, 105–121.
Lough, J.M., Berkelmans, R., van Oppen, M.J.H., Wooldridge, S.A., Steinberg, C.R.,
2006. The Great Barrier Reef and climate change. Bull. Aust. Meteorol. Oceanogr.
Soc. 19, 53–58 (Pages).
Lough, J.M., Lewis, S.E., Cantin, N.E., 2015. Freshwater impacts in the central Great
Barrier Reef: 1648–2011. Coral Reefs 34, 739–751.
Marsh, H., Kwan, D., 2008. Temporal variability in the life history and reproductive
biology of female dugongs in Torres Strait: the likely role of sea grass dieback.
Cont. Shelf Res. 28, 2152–2159.
Marsh, H., De'ath, G., Gribble, N., Lane, B., 2005. Historical marine population esti-
mates: triggers or targets for conservation? The dugong case study. Ecol.
Appl. 15, 481–492.
Marsh, H., Eros, C., Corkeron, P., Breen, B., 1999. A conservation strategy for
dugongs: implications for Australian research. Mar. Freshw. Res. 50, 979–990.
Marsh, H., O'Shea, T.J., Reynolds, J.E., 2011. Ecology and Conservation of the Sire-
nia: Dugongs and Manatees. Cambridge University Press, Cambridge.
McKenzie, L.J., Campbell, S.J., Vidler, K.E., Mellors, J.E., 2007. Seagrass-Watch:
Manual for Mapping and Monitoring Seagrass Resources. Seagrass-Watch HQ,
Cairns (114p).
McKenzie, L.J., Collier, C., Waycott, M., 2012. Reef Rescue Marine Monitoring Pro-
gram: Inshore Seagrass, Annual Report for the Sampling Period 1st July 2010–31st
May 2011. Fisheries Queensland, Cairns (177p).
Meager, J.J., Limpus, C., 2014. Mortality of inshore marine mammals in eastern Aus-
tralia is predicted by freshwater discharge and air temperature. PLoS ONE 9 (4),
e94849http://dx.doi.org/10.1371/journal.pone.0094849.
Peltier, H., Dabin, W., Daniel, P., Van Canneyt, O., Dorémus, G., Huon, M., Ridoux,
V., 2012. The significance of stranding data as indicators of cetacean populations
at sea: modelling the drift of cetacean carcasses. Ecol. Indic. 18, 278–290.
Petrou, K., Jimenez-Denness, I., Chartrand, K.M., McCormack, C., Rasheed, M.A.,
Ralph, P.J., 2013. Seasonal heterogeneity in the impact of air-exposure on the pho-
tophysiology of two tropical intertidal seagrass species (Zostera muelleri and
Halophila ovalis). Mar. Ecol. Prog. Ser. 482, 93–106.
Petus, C., Collier, C.J., Devlin, M., Rasheed, M., McKenna, S., 2014. Using MODIS
data for understanding changes in seagrass meadow health: a case study in the
Great Barrier Reef (Australia). Mar. Environ. Res. 98, 68–85.
Preen, A., 1995. Impacts of dugong foraging on seagrass habitats: observational and
experimental evidence for cultivation grazing. Mar. Ecol. Progr. Ser. 124,
201–213.
Preen, A., 1995. Diet of dugongs: are they omnivores?. J. Mammalogy 76, 163–171.
Preen, A., Marsh, H., 1995. Response of dugongs to large-scale loss of seagrass from
Hervey Bay, Queensland, Australia. Wildl. Res. 22, 507–519.
Ralph, P.J., Durako, M.J., Enriquez, S., Collier, C.J., Doblin, M.A., 2007. Impact of
light limitation on seagrasses. J. Exp. Mar. Biol. Ecol. 350, 176–193.
Rasheed, M.A., Taylor, H.A., Coles, R.G., McKenzie, L.J., 2007. Coastal seagrass
habitats at risk from human activity in the Great Barrier Reef World Heritage
Area: review of areas to be targeted for monitoring. In: Report to the Marine and
Tropical Sciences Research Facility. Reef and Rainforest Research Centre Lim-
ited, Cairns (122 p).
Rasheed, M.A., McKenna, S.A., Cater, A.B., Coles, R.G., 2014. Contrasting recovery
of shallow and deep water seagrass communities following climate associated
loses in tropical north Queensland, Australia. Mar. Pollut. Bull. 83, 491–499.
Sheppard, J.K., Preen, A.R., Marsh, H., Lawler, I.R., Whiting, S.D., Jones, R.E., 2006.
Movement heterogeneity of dugongs, Dugong dugon, over large spatial scales. J.
Exp. Mar. Biol. Ecol. 334, 64–83.
Sheppard, J.K., Lawler, I.R., Marsh, H., 2007. Seagrass as pasture for seacows: land-
scape-level dugong habitat evaluation. Estuar. Coast. Shelf Sci. 71, 117–132.
Stobtzick, S., Hagihara, R., Grech, A., Marsh, H., 2012. Aerial survey of the urban
coast of Queensland to evaluate the response of the dugong population to wide-
spread effects of the extreme weather events of 2010–11. In: Final Report to the
Australian Marine Mammal Centre and the National Environmental Research Pro-
gram. (Published online by the Reef and Rainforest Research Centre Ltd., Cairns,
62p).
Thomas, C.R., 2008. Bayesian Belief Network Development for Ecological Decision
Support in Data-sparse Domains: A Case Study on Tropical Seagrass in the Great
Barrier Reef, Australia. PhD Thesis. Monash University, Victoria, Australia (412
p).
Thorburn, P.J., Wilkinson, S.N., Silburn, D.M., 2013. Water quality in agricultural
lands draining to the Great Barrier Reef: a review of causes, management and pri-
orities. Agric. Ecosyst. Environ. 180, 4–20.
Waycott, M., Longstaff, B.J., Mellors, J., 2005. Seagrass population dynamics and wa-
ter quality in the Great Barrier Reef region: a review and future research direc-
tions. Mar. Pollut. Bull. 51, 343–350.
Weeks, S., Werdell, P.J., Schaffelke, B., Canto, M., Lee, Z., Wilding, J.G., Feldman,
G.C., 2012. Satellite-derived photic depth on the Great Barrier Reef: spatio-tempo-
ral patterns of water clarity. Remote Sens. 4, 3781–3795.
Wooldridge, S.A., Done, T.J., 2004. Learning to predict large-scale coral bleaching
from past events: a Bayesian approach using remotely sensed data, in-situ data,
and environmental proxies. Coral Reefs 23, 96–108.
Wooldridge, S.A., Done, T.J., 2009. Improved water quality can ameliorate effects of
climate change on corals. Ecol. Appl. 19, 1492–1499.