Forgot your password?
Document Actions
Long et al 97
SEAGRASS DIEBACK
IN NORTH WESTERN TORRES STRAIT
Brian Long
Timothy Skewes
Thomas Taranto
Mervyn Thomas
Peter Isdale
Roland Pitcher
Ian Poiner
June 1997
REPORT MR-GIS 97/6
T O R R E S S T R A I T S E A G R A S S D I E B A C K 2
Executive Summary
To assess the magnitude and extent of a reported seagrass decline in north-western
Torres Strait the distribution and abundance of seagrass from a survey of 332 sites in
November 1993 (post-impact) in a study area of 4,388 km2 was compared with historical
data of the distribution and abundance of seagrass collected at 498 sites in the study area
between 1986 and 1989 (pre-impact).
There was an estimated loss of 1,199 km2 of seagrass in the north-eastern region of the
study area, in the main impact area, which represented a 60% loss of seagrass there. In
contrast there was little change in the area of seagrass which was mapped for the
southern region of the study area.
On the foreshore areas at three locations where quantitative samples of seagrass biomass
were taken at 201 sites with a 0.07 m2 core a the biomass of a mixed assemblage of
seagrasses on the foreshore and shallow subtidal areas along the southern margin of
Boigu Island at the northern end of the study area was significantly lower in the post-
than pre-impact surveys. Before the November 1993 survey the seagrass beds were
numerically dominated by Cymodocea serrulata, Thalassia hemprichii and Enhalus
acoroides with high biomass, 66 g.m-2, whereas in the November 1993 survey only
T. hemprichii and to a lesser extent C. serrulata was common and biomass, 18 g.m-2, and
diversity was significantly lower. In contrast there were no significant differences
between estimates of seagrass biomass for the historical 1986–1989 and the November
1993 survey at two ‘control’ locations just south of the study area at Badu Island. The
species composition was also similar for the historical data and the November 1993
survey at these locations.
In addition to the changes in seagrass in the northern section of the study area there were
also large changes in the distribution and abundance of epifauna in this region over a
three year period from 1989 to 1993. Whereas there was dense or sparse epifauna present
throughout much of the study area north of Buru Island in May and June 1989, epifauna
was mostly sparse or absent in the November 1993 survey.
Analysis of a coral head of Porites sp. from Boigu Island indicated that freshwater runoff
from the Mai River in the 1990/91 wet season was very unusual with a short but intense
run-off at the end of the season, which correlated well with the changes in seagrass,
epifauna and areas with high abundance of sea urchins in the northern and north eastern
region of the study area.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 3
Introduction
The decline and loss of seagrass is being reported at an increasing rate around the world
(den Hartog, 1970; Preen et al., 1995). Changes in seagrass occur at a range of spatial
and temporal scales due to anthropogenic and natural causes, and the complex
interaction of the two (den Hartog 1970). Anthropogenic causes related to decrease in
water quality have been linked to seagrass decline in coastal waters (Walker and
McComb, 1992). Natural effects include cyclones and storm events which physically
removes seagrass by scouring, and indirectly through lowered salinity and increased
turbidity (Preen et al., 1995).
In the early 1990’s there were reports by rock lobster fishermen and Torres Strait
Islanders that seagrass was disappearing in north western Torres Strait. The anecdotal
evidence was corroborated by CSIRO scientists doing ongoing field work in north
western Torres Strait on the rock lobster Panulirus ornatus (Pitcher et al., 1994). In
addition to the seagrass decline they also noted that epifauna had declined and that there
was an abrupt increase in the abundance of sea urchins in the area. Concerns over the
possible repercussions of a seagrass dieback on the dugong, prawn and fish populations
prompted a survey of the seagrass, epifauna and sea urchins of north western Torres
Strait in November 1993 (hereafter called the dieback survey). There was extensive
qualitative historical data (1984–1989) of seagrass presence or absence in the inter-reefal
areas (Long and Poiner 1993). There was also detailed historical quantitative data of
seagrass biomass collected along the southern margin of Boigu Island and two bays at the
southern and eastern end of Badu Island (ibid.). These two historical data sets provided
the necessary information on the distribution and abundance of seagrass in the area
before the reported dieback to compare with data collected after the impact for a
Before/After/Control/Impact experimental design which is a prerequisite for detecting
an environmental impact (Underwood, 1993). To identify possible environmental causes
of the reported dieback freshwater run-off events from nearby rivers which discharge
into the area were correlated with changes in seagrass. The patterns of growth of coral
heads, which accurately record freshwater run-off events (Isdale, 1996), were analysed as
there was no data available on freshwater discharge by rivers into the study area.
Materials and Methods
Description of the study area
Torres Strait lies between the NW coast of Cape York Peninsula and the S coast of
Papua New Guinea, and connects the Coral and Arafura Sea (Fig. 1). The physical
oceanography and sedimentary geology has been described by Wolanski et al. (1988),
Harris (1988) and Bode and Mason (1994). The Straits are shallow (< 15 m) with strong
tidal currents due to large pressure gradients between the Arafura and Coral Sea (Bode
and Mason, 1994). Water speeds exceeding 2.5 m.s-1 occur in the narrow channels
between some islands and reefs (Admiralty, 1973).
T O R R E S S T R A I T S E A G R A S S D I E B A C K 4
PNG 142° 143° E
N
Mai River
PNG
Boigu Is.
Buru Is.
Dauan Is.
Aldai Reef
Mabuiag Is. Orman Reefs
10° S
Badu Is. Moa Is.
Torres Strait
Cape York
AUSTRALIA
0 50 100 km
Figure 1. Map of Torres Strait, showing the boundaries of the seagrass dieback survey done in November
1993.
The strong tidal currents have created sand waves in many areas of Torres Strait (Harris,
1988) including north western Torres Strait. There are two distinct seasons in Torres
Strait: a dry season which runs for seven months from May to November with an
average rainfall of 21.4 mm month-1, and a wet monsoon season which lasts for five
months from December to April with an average monthly rainfall of 311 mm at
Thursday Island (Admiralty, 1973). The prevailing winds for the two seasons are also
distinct with south-east trade winds blowing from E and SE 90% of the time during the
dry season whereas winds are more variable during the wet monsoon and blow from the
NE, N and NW for 30% of the time. The average wind speed is lower in the wet
monsoon, 5 knots.h-1, than dry season, 7.9 knots.h-1, and the number of calm days is also
lower in the dry season, < 1 day.month-1 than wet monsoon, 2.1 days.month-1. There are
more gales during the monsoon than dry season (6 and < 1 days.month-1 respectively).
There is little net flow of water through Torres Strait although there are seasonal
differences in the direction of net flow with a net westwardly flow over the dry season
with the south-east trade winds and a net eastwardly flow over the wet monsoon season
when westerlies and north westerlies prevail (Wolanski et al., 1988). The winds and
currents stir up the bottom sediments in shallow water areas of central Torres Strait
which results in a turbidity maximum zone in central Torres Strait (Harris, 1988).
Seagrass was reported to be declining in north western Torres Strait, north of Buru
Island, however, the extent of the dieback was not known so a study area large enough to
T O R R E S S T R A I T S E A G R A S S D I E B A C K 5
encompass much of north western Torres Strait was defined. The southern limits of the
study area were Badu and Moa Island, situated mid-way across the Straits; Boigu and
Dauan Island near the S coast of Papua New Guinea formed the northern boundary; the
eastern limit of the study area were formed by a line NE from Moa Island through the
Orman reefs and N to Dauan Island; and the 142nd meridian of longitude formed the
western boundary (Fig. 1).
Field sampling: Inter-reefal areas
Historical data
The percentage cover of seagrass in 0.25 m2 quadrats along 20 m transects were recorded
by divers at 253 subtidal sites in the study area in north western Torres Strait during
seven cruises from 1986 to 1989 as part of the seagrass study (Long and Poiner, 1993)
(Fig. 2a). At a further 112 sites the relative cover of seagrass was recorded by divers for
500 m x 2 m transects sampled during a lobster survey in 1989 (Pitcher et al., 1992).
Because the sampling methods differed the data from the seagrass study and the lobster
survey were converted to seagrass presence or absence data which gave a total of 365
sites sampled on the seabed in the study area from 1986 to 1989.
In addition to the historical data on the distribution and abundance of seagrass there was
extensive data on the distribution and abundance of epifauna in the region between
Boigu and Badu Island which was collected as part of the lobster survey in 1989 (see
Pitcher et al. 1992 for full details). Epifauna was recorded as dense, sparse, very sparse or
absent for the 500 m x 2 m transects by divers at the 112 sites sampled in the study area.
Sediment samples were taken at 9 sites within the study area during a seagrass survey in
1989.
November 1993 survey
The seagrass at 251 subtidal sites in the study area were sampled in November 1993 to
assess the magnitude and extent of the reported seagrass dieback (Fig. 2b). The study
area was first divided into primary sampling units which were each 4.5 km east-west and
4.2 km north-south. The primary sampling unit area, 18.9 km2, was chosen based on
estimates of the time it would take to sample a site (15 min), the time to travel between
sites and the total time (three weeks) available for field sampling. To give complete
sampling coverage of the study area all primary sampling units were sampled. Because it
was impractical to sample the whole primary sampling unit (18.9 km2) the area sampled
in the field were 100 m2 sites. The position of each site within each primary sampling unit
was chosen randomly. Global Positioning System (GPS) satellite navigation was used to
locate the sites in the field. At all sites divers searched an area of approximately 100 m2
and recorded the presence or absence of seagrass along with estimates of the abundance
of sea urchins (per 25 m2) as well as descriptions of the substratum and epifauna. The
epifauna cover was scored into four categories: dense, sparse, very sparse and absent to
conform with the categories devised by Pitcher et al. (1992). Water visibility (m) and
water depth (m) were also recorded at each site and a sediment sample was taken for
grain-size analysis.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 6
a). b).
10 km 10 km
Seagrass Seagrass
present
present
absent
absent
.
Figure 2. Seagrass presence or absence data from the a). historical (1986–1989) seagrass data; and b). November 1993 survey.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 7
Field sampling: Foreshore areas and shallow embayments
Historical data
Quantitative samples of seagrass biomass were collected with a 0.07 m2 shovel by divers
from a dinghy at 133 sites sampled on foreshore and shallow subtidal areas at three
locations in the study area (Fig. 3a & b). At each site 3 to 5 shovel samples were taken at
fixed intervals down a 20 m transect (for full details of the sampling see Long and Poiner
1993). The seagrass was placed in a divers mesh bag and returned to the surface where it
was sieved over the side of the dinghy in a 10 mm lug basket to remove the sediment
from the rhizomes, labelled and stored on ice for further processing at the CSIRO
marine laboratories, Cleveland, Qld. The location sampled on the foreshore area on the
southern side of Boigu Island was selected as the impact location in the
Before/After/Control/Impact analysis. Forty-three sites were sampled there in an area
of 3.413 km2 from 1986 to 1989 before the reported dieback (Fig. 3a). Two embayments
just south of the study area at the south and east end of Badu Island were selected as
control locations with 38 and 54 sites sampled from 1986 to 1989 in an area of 2.02 and
1.911 km2 respectively (Fig. 3b).
November 1993 survey
Quantitative sampling of seagrass biomass was done at Boigu and Badu Islands to match
up with the historical data sampled at these three locations in the
Before/After/Control/Impact (BACI) experimental design to test for significant changes
in seagrass biomass at these locations (Green 1993). At the location on the foreshore area
of the southern margin of Boigu Island (impact location), 24 core samples were taken
with a 0.07 m2 seagrass grab (Long et al. 1994) along two transects separated
approximately 1 km east-west and 170 m north-south to match up with the samples
taken there from 1986 to 1989 (Fig. 3a). At the east and south end of Badu Island, just
south of the study area (control locations) in sheltered embayments of the island, 21
samples were taken from each location respectively in a grid arrangement with 310 m
separation in both the east-west and north-south directions to match up with the samples
taken there from 1986 to 1989 (Fig. 3b).
Back at CSIRO Marine Laboratories the seagrass samples were placed in a 10% solution
of orthophosphoric acid for 30 min to clean the epiphytes from the shoots (refer to Long
and Poiner, 1993 for full details). The samples were rinsed in freshwater and separated
into species and the above-ground shoots and stems were separated from the below
ground rhizomes. The seagrass was dried at 60°C until a constant weight and weighed to
the nearest 0.1 g.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 8
Data Analysis
Mapping — seagrass presence or absence
Point-to-area spatial data transformations with a Geographic Information System (GIS)
was used to create Voroni maps of seagrass presence or absence for the study area for
the historical (1986–1989) and the November 1993 survey data. The two maps were
overlaid to produce a third which mapped areas where seagrass was present over the
entire eight year study period; seagrass was absent over the study period; seagrass was
present during the survey but was absent before the survey and; seagrass was absent
during the survey but was present before the survey. The last category mapped areas
where seagrass had disappeared.
a). b).
N
Mai River 0 5 10 km
Papua New Guinea
Badu Island
Boigu Island
N
0 2.5 5 km
Boigu Island - south
Badu Island - east
0 500 1000 m
0 500 m 1 km
Badu Island - south
0 1 km
Figure 3. Sites sampled on the foreshore in the impact area at a). the south side of Boigu Island; and
control areas at b). the east and south end of Badu Island. : Historical sites; circle, : November 1993
seagrass dieback survey; dotted line - convex hull around sample points.
BACI analysis
The seagrass biomass sampled at 135 sites on the foreshore and shallow water
embayments of Boigu and Badu Island from 1986 to 1989 were not suitable for a simple
BACI ANOVA because sampling was not random as the sites were spatially clumped
mainly along transects. Therefore large scale spatial trends at each location and each time
on the foreshore areas of Boigu and Badu Island were investigated by fitting a loess
response surface to latitude and longitude. The residuals from the trend surface were
then examined, first to assess the appropriateness of the assumptions that the residuals
were normally distributed with constant variance, and secondly to assess the extent of
T O R R E S S T R A I T S E A G R A S S D I E B A C K 9
any smaller scale spatial correlation. For each site and sample time, a robust estimate of
the semi-variogram was generated (Cressie,1993) and plotted against inter-sample
distance.
Because the location of samples differed between time points, any spatial trends or
dependencies would result in biased estimates of change over time. To accommodate
these effects, the different time points were compared by estimating the mean response
(from the fitted surface and/ or semi variogram) in a fixed polygon (i.e. the same
polygon was used for both times). For this analysis the spatial response was modelled as
a polynomial in latitude and longitude. Whilst less flexible than a loess surface, this yields
simple expressions for the estimated mean response, and for the standard error of the
estimated mean.
In the presence of spatial correlation, the standard error is based on equations given in
Cressie (1993). In the absence of such spatial correlation the standard error is simpler.
Since the estimated mean is simply a linear combination of the regression coefficients for
the fitted surface, it can be obtained directly from the estimated co-variance matrix of the
parameters.
The polygon for each site (Fig. 3) was based on a convex hull around the set of sample
points (aggregated over both sample times). A regular grid of points was generated within
the polygon, and the response integrated by a process of simple averaging. A fifty point
grid was used in each dimension.
Sediment grain size analysis
Sediment samples collected during the historical and November 1993 survey were
analysed for gran size fractions by the method given in Folk 1968 to give percentage
grain size for mud (< 62 µm), sand (62–1,000 µm) and gravel (> 1.0 mm). There were
too few sediment samples (9) taken in the study area for an accurate ANOVA of grain
size fractions so historical and November 1993 samples were matched by geographic
location and a paired sample t-test was used to compare the sediment grain sizes between
the historical and November 1993 surveys. To do this 17 samples taken in 1993 which
were within a 4 km radius of the 9 sites sampled in 1989 were paired up. The three grain
size fractions, gravel, sand and mud, of the November 1993 sediment samples were
averaged for each of the 9 sites to give a paired comparison.
Coral heads were collected from Boigu, Aldai and Mabuiag reef to provide proxt
information about the recent history of seasonal freshwater inputs to the area. The
timing and intensity of fluorescing bands in coral skeletons have been found to correlate
quite well with instrument records of river runoff from adjacent tropical coasts (Isdale,
1996). We used the bands in coral from Boigu to qualitatively reconstruct the freshwater
input history to the study area for the period of interest.
CSIRO scientists researching the tropical rock lobster have been monitoring three sites
in the study area north of Buru Island on a yearly basis from 1989 to 1993. Descriptions
of the epifauna and substrate type were recorded. This data provided ancillary
information on changes of seagrass and epifauna at these three sites over a six year
period.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 10
Results
Seagrass mapping: Inter reefal areas
Seagrass was distributed widely throughout the entire study area from 1986 to 1989
(Fig. 2a). The area of seagrass mapped with GIS, 3,078 km2, was larger than seagrass
absent, 1,188 km2 (Table 1). In contrast, the area of seagrass, 1,871 km2, was lower than
seagrass absent, 2,394 km2 (Fig. 2b). An overlay of the two maps of seagrass presence or
absence of the historical and November 1993 survey data indicated that the largest loss
of seagrass was in the north and north eastern region of the study area between Boigu
and Buru Island, and south east between Buru Island and the top of the Orman Reefs in
region A (Fig. 4). In region A, 1,993 km2, or two thirds (60.2%) of the area had seagrass
before the November 1993 survey and none during the survey. Very little of region A,
3.6%, had no seagrass before the survey and seagrass during the November 1993 survey.
There was a large area, 195 km2, in the north west of the study area which had no
seagrass over the seven year study period from 1986 to 1993. In contrast there was little
evidence to suggest that seagrass had declined in the southern half of the study area
between Buru and Badu Island in region B (Fig. 4). In region B, 2,201 km2, about half the
area (40%) had seagrass before and during the November 1993 survey; 23% had seagrass
before the November survey and no seagrass during; 19% had no seagrass before the
survey and seagrass during the November 1993 survey; and seagrass was absent over the
seven year study period in the remaining 17% of region B.
Table 1. Area analysis of seagrass change (km2) based on the overlay of maps of seagrass presence or
absence from historical 1986-1989 data and the November 1993 seagrass dieback survey data.
Seagrass Region A Region B Total
Absent before, absent now 289 365 653
Absent before, present now 72 412 485
Present before, absent now 1,199 515 1,714
Present before, present now 433 908 1,342
Total 1,993 2,200 4,194
Seagrass change: foreshore and shallow water embayments
The seagrass beds on the shallow water foreshore areas on the southern side of Boigu
Island were lush, mixed species assemblages from 1986 to 1989 numerically dominated
by Cymodocea serrulata, Thalassia hemprichii, Cymodocea rotundata and Enhalus acoroides (Fig. 5).
In contrast the seagrass beds there during the November 1993 survey were mainly
stunted Thalassia hemprichii with scattered Cymodocea serrulata. The detailed survey of east
and south Badu Island indicated that the species composition of seagrass had not
T O R R E S S T R A I T S E A G R A S S D I E B A C K 11
changed and were numerically dominated by Cymodocea rotundata and Thalassia hemprichii at
south Badu and Enhalus acoroides, C. serrulata and T. hemprichii at east Badu Island (Fig. 5).
An analysis of variance for the Loess surface indicated that there were statistically
significant medium-scale (100’s of metres) spatial trends at East Badu before the impact,
and South Boigu after the impact (Table 2).
Table 2. Fit of Loess trend surface by location and sampling occasion. ***: P < 0.0001.
Before After
Location F p Value F p Value
South Badu 0.30 0.966 2.50 0.074
East Badu 3.68 0.000*** 1.26 0.340
South Boigu 1.43 0.217 5.26 0.000***
Small Scale Spatial Dependencies
The Loess response surface was removed from the data before calculation of the semi-
variogram for all three locations, before and after impact (Fig 6a–c). All of the
variograms were essentially flat, which provided little evidence of any spatial dependency,
once the larger scale spatial trends were removed. Thus at the scale of spatial separation
in our study, different samples may be construed as essentially independent.
Fitted Response
There were no significant differences among locations before the reported seagrass
dieback but there were significant differences after the impact (Table 3). The control
versus impact contrast was also statistically significant for the change over time. That is,
the change over time for South Boigu was statistically significantly different from the
change over time for South and East Badu. Furthermore, this difference arose from a
greater decrease in seagrass biomass over time for South Boigu than the control
locations.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 12
Table 3. Fitted response for each location, Before and After Impact and the estimated change in mean
response over time for each location, and a comparison of the control locations (mean of East Badu and
South Badu versus South Boigu).
Pre Impact Post Impact Change
Location Mean s2 Mean s2 Mean s2
East Badu 4.032 0.064 3.503 0.204 0.529 0.268
South Badu 4.298 0.150 2.717 0.152 1.581 0.302
South Boigu 4.646 0.025 1.260 0.057 3.385 0.081
Control vs Impact -0.481 0.078 1.850 0.146 -2.330 0.224
Z-score -1.718 4.849 -4.926
T O R R E S S T R A I T S E A G R A S S D I E B A C K 13
Region A
N
land
reef
Region B AA
AP
PA
PP
0 50 100 km
Figure 4. Torres Strait: Area analysis of change of seagrass produced by overlaying a seagrass presence or
absence map based on data collected from 1986–1989 and a seagrass presence or absence map based on
data collected during the November 1993 seagrass dieback survey. PA: seagrass present before but absent
during the November 1993 survey; AA: seagrass absent before and during the November 1993 survey; AP:
seagrass absent before but present after the November 1993 survey; PP: seagrass present before and after
the November 1993 survey. For the main impact region A, and remaining study areas region B.
2
S e a g r a s s b io m a s s ( g . 0 .0 7 m )
4 .6 5 1 .2 6 4 .0 3 3 .5 0 4 .3 0 2 .7 2
100%
90
80
70
60
50
40
30
20
10
0 PR E PR E PR E
PO ST PO ST PO ST
S BO I E BA D S B AD
Im p a c t C o n tr o ls
Figure 5. Comparison of above-ground biomass broken down by species for east, south and west Badu,
and south Boigu. Sampling occasions — PRE: historical 1984 to 1989 data; DIE: November 1993 seagrass
dieback survey. Locations — EBAD: east Badu; SBAD: south Badu; SBOI: south Boigu and WBAD: west
Badu Island for : Cymodocea rotundata; : Cymodocea serrulata; : Enhalus acoroides; :
Halophila ovalis; : Syringodium isoetifolium; : Thalassia hemprichii; and : Halodule uninervis..
Vertical axis: proportion of total.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 14
a). Control Site: South Badu Control Site: South Badu b). Control Site: East Badu Control Site: East Badu
Pre Impact Post Impact Pre Impact Post Impact
2.5 2.5
3 3
2.0 2.0
1.5 2 1.5
2
gamma gamma gamma gamma
1.0 1.0
1 1
0.5 0.5
0 0.0 0 0.0
0.0 0.002 0.004 0.006 0.008 0.0 0.002 0.006 0.010 0.0 0.002 0.004 0.006 0.008 0.010 0.0 0.002 0.004 0.006 0.008
Distance (m) Distance (m) Distance (m) Distance (m)
T O R R E S S T R A I T S E A G R A S S D I E B A C K 15
c). Impact Site: South Boigu Control Site: South Boigu
Pre Impact Post Impact
2.0
0.6
1.5
0.4
1.0
Gamma
0.2
0.5
0.0 0.0
0.0 0.002 0.006 0.010 0.0 0.004 0.008 0.012
Distance (km) Distance (km)
Figure 6. Torres Strait Before and After Impact (BACI) empirical variogram for Control areas: a). South
Badu; and b). East Badu; and Impact area: c). South Boigu Island.
Environmental data
The sediments in the study area were mainly gravelly sands with mud (Fig. 7). The
average percentage fraction of gravel in region A of the study area in the November 1993
survey, 18.75%, was significantly lower than 1989, (33.37%) based on the paired t-test
(Table 4). The sand fraction in the November 1993 survey, 73.23%, was significantly
higher than 1989, 57.58%. In contrast, there was no significant difference in the
percentage mud between the two years with an pooled average of 8.5% mud. At a couple
of sites, we recorded on our data sheets that there was a thin layer (cm’s) of clean sand
overlying a muddy sand base.
Table 4. T-test for paired comparison of percentage grain size
_
fraction of gravel, sand and mud for sediment samples taken in
1989 and the November 1993 survey. D : mean difference of _
1989 samples - 1993 samples; sD: std. of the difference; sD
stderr. of the difference; ts : t-statistic. *: P < 0.05.
%Gravel %Sand %Mud
1989 33.37 57.58 9.04
1993 18.75 73.23 8.03
_
D 14.62 -15.65 1.02
sD 16.42 16.43 4.54
_
sD 5.47 5.48 1.51
St 2.67* -2.86* 0.67
T O R R E S S T R A I T S E A G R A S S D I E B A C K 16
Ancillary data
Epifauna was dense or sparse throughout much of the study area north of Buru Island in
1989 (Fig. 8a) whereas epifauna was sparse or absent in during the November 1993
survey (Fig. 8b). A visual comparison of the two maps indicated that epifauna had been
affected over a larger area than seagrass. At a couple of sites north of Buru Island we
recorded the presence of white dead corals with little algal growth on them which were
partly covered by clean sand.
% gravel
80
50
25
0 25 50 km
Figure 7. Bubble plots of percentage gravel fraction (> 1.0 mm) at 9 sites sampled in the study area during
1989 (clear bubbles) and 16 sites sampled within a 4 km radius of the 1989 samples during November 1993
(solid bubbles).
The sea urchin, Prionocidaris sp. was sampled only in region A of the study area and where
present, their abundance ranged from 2 to 20 animals per 25m2 (Fig. 9). To test for
significant differences between areas where seagrass was lost and the remaining areas an
ANOVA was done on square-root transformed abundance of urchins. The results
indicated that the abundance of Prionocidaris sp. was significantly higher (P < 0.005) in
areas where seagrass was lost (2.326 animals.25m2; se = 0.503) than remaining areas in
region A (0.365 animals.25m2; se = 0.293).
Examination of the coral heads from Boigu, Aldai and Mabuiag reefs indicated that only
the Porites spp.coral head from Boigu was suitable for providing a seasonal proxy record
of freshwater inputs to the area. The corals taken from Aldai and Mabuiag reef were
unsuitable because the corallite dimensions were coarse, a feature of most of the Favid
species . Analysis of the fluorescence in the coral head from Boigu indicated that there
T O R R E S S T R A I T S E A G R A S S D I E B A C K 17
were four abnormal seasons: 1982/83, 1983/84, 1984/85 and 1990/91 (Fig. 10). The
1982/83 and 1984/85 seasons were wetter than averga and run-off was in a single long
seasonal pulse. In contrast the 1983/84 season was apparently very dry. The 1990/91
season was very unusual. It started normally and run-off occurred as a consistent pulse
until near the end of the season, where reduced somewhat, but then showed a short but
intense run-off period shortly after December 1990.
Discussion
More than 1,400 km2 of seagrass north and north east of Buru Island in region A of the
study area was lost between 1989 and 1993 which represented a 60% reduction in
seagrass in the study area and a 10% reduction of the 13,425 km2 of seagrass estimated
for the whole of Torres Strait (Long and Poiner, 1997). The magnitude of the loss of
seagrass in Torres Strait is the same as a recently reported dieback of > 1000 km2 of
seagrass reported in 1992 for Hervey Bay (Preen et al., 1995).
The results of the BACI analysis indicated that there was a significant reduction of
seagrass on the foreshore areas along the southern margin of Boigu Island in the impact
area than the control locations at South and East Badu Island. Moreover, because there
was clear spatial structure in seagrass biomass for East Badu and South Boigu the analysis
of change of seagrass over time must take into account any differences in the position of
sample points. There were, however, no small scale spatial correlation among samples at
the distances sampled (10’s to 100’s m apart) in this study. Thus the different samples
may be construed as essentially independent. This finding greatly simplified the
estimation of the standard error since the estimated mean is simply a linear combination
of the regression coefficients for the fitted surface and can be obtained directly from the
estimated co-variance matrix of the parameters.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 18
a). b).
Boigu Is
Boigu Is
Buru Is
Epibenthos
Epibenthos dense
dense sparse
sparse
absent
absent
Badu Badu Moa Is
Figure 8. Distribution of epifauna in the study area in (a) 1989 lobster survey and (b) November 1993 survey.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 19
#
##
A #
#
#
##
#
##
### #
# #
#
#
# #
# ##
#
# # ## #
#
Prionocidaris sp.
per 100 m2
B
#
# # 80
40
10
N
0 50 100 km
Figure 9. Bubble plot of the sea urchin, Prionocidaris sp. per m2, estimated from the visual spot dives during
the November 1993 seagrass dieback survey. red areas: Seagrass absent after the impact and present before.
Region A: main area of impact; Region B: no impact.
1 9 9 3 /4 s u m m e r
250
F in e f lu o r e s c e n t b a n d s o o n a f te r D e c e m b e r 1 9 9 0
200
150
100
liv in g tis s u e
50
0 40 80 120 160 200
s te p s ( 0 .5 m m )
Figure 10. Plot of cross-sectional fluorescence of Porites sp. coral head from the reef flat at the east tip of
Boigu Island. Vertical axis: Fluorescence (relative units)
T O R R E S S T R A I T S E A G R A S S D I E B A C K 20
There have been a number of reported seagrass diebacks in Torres Strait in the early
1970’s which are in themselves anecdotal, but when taken together suggests that diebacks
are a natural event in Torres Strait. The two main rivers that empty into Torres Strait are
the Mai and the Fly river. There is no available data on river runoff by the Mai river.
There has been no marked changes in land clearing practises in the mainland catchment
of Papua New Guinea which empties into Torres Strait land and consequently land
degradation with concomitant increased runoff and suspended sediment can not explain
the loss of seagrass in this study. Preen et al. (1995) attributed the large loss of seagrass in
Hervey Bay to a complex array of factors associated with a cyclone and flooding which
was exacerbated by soil erosion due to land clearing in the adjoining catchment area.
Seagrass is critical habitat for many commercial important species of prawns and fish and
is an important food source for dugongs and turtles. The distribution of dugongs are
reliable indicators of seagrass distribution and disproportionately high concentrations of
dugongs are associated with large seagrass beds in Torres Strait and the Starcke region of
north Queensland (Preen et al. 1995). Torres Strait supports the largest reported
population of dugongs in Australia and possibly the world (Marsh 1995) and the main
area where the seagrass had disappeared coincided with the largest concentration of
dugongs in Torres Strait in 1987 (Marsh 1995). A recent survey of dugongs in 1989 in
Torres Strait has indicated that the largest concentrations of dugongs were now south of
Buru Island (H. Marsh pers. comm.) which indicates that the dieback has affected the
distribution of dugongs in northern Torres Strait. The seagrass dieback Hervey Bay was
followed by a reduction in the population of dugongs from 1937 to less than 200
dugongs (Preen et al. 1995).
The maps of seagrass presence or absence indicated a change of seagrass distribution in
inter reefal areas of north western Torres Strait. The biomass and community structure
of seagrass was also affected shown by the decrease in biomass and diversity and change
in species composition on the foreshore and shallow subtidal areas along the south side
of Boigu Island at the northern end of the study area. There was a significantly lower
biomass of seagrass on the foreshore areas and shallow subtidal areas along the south
side of Boigu Island during the November 1993 survey than before and diversity was also
significantly lower. There was a shift in the species composition of seagrass from a
community dominated by C. serrulata, T. hemprichii and E. acoroides to an assemblage of
low diversity dominated by a single species, T. hemprichii. In contrast, there were no
significant differences in biomass at the south and east end of Badu Island at the
southern end of the study area. The results of the BACI test and the seagrass mapping
suggests that the dieback was largely restricted to the northern part of the study area.
Anecdotal evidence provided by CSIRO fisheries doing research in the area on Panulirus
ornatus (tropical rock lobsters) also suggested that seagrass, lobsters and epifauna had
declined in the area (Pitcher et al. 1994). At three sites north of Buru Island in the study
area, descriptions of the epifauna and seagrass were recorded on a yearly basis from 1989
to 1993.
Lobsters were abundant at the three sites from 1989 to 1991 and were absent in 1992
and 1993. The Orman reef area, which is adjacent to the area of interest north of Buru
Island, normally has abundant lobster but has shown a decline in lobsters in 1993 based
on survey data and from changes in fishing effort by the rock lobster freezer boat fleet.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 21
There have been dramatic changes in seagrass abundance at the three sample sites since
1989. Seagrass abundance declined from high in 1990 (higher even than in 1989) to low
in 1991, to zero or near zero abundances in 1992. The decline was greatest for the ovoid
species, H. spinulosa mainly and H. ovalis to a lesser degree, in part because they were the
dominant of the two forms previously (CSIRO, unpublished data).
Sparse epibenthic 'garden' (hard corals, whips, sponges) bottom and Sargassum spp. were
a common feature of consolidated sand bottoms in the area with occasional reef
epibenthic assemblages on rock outcrops. There was a general decline in the amount of
epifauna at these three sites with little or no epifauna recorded there in 1992.
The disappearance of seagrass in this area was also accompanied by a decline in epifauna,
shown by the results of this study. The loss of seagrass could explain the reduction in
epifauna in the northern region of the study area as the removal of seagrass can
destabilise the sediments and often leads to blowout holes and a shift in the sediment
grain size distribution to sandier sediments which could affect epifauna. There were
significant changes in sediment grain size distribution in the northern region of the study
area with sandier sediments during the November 1993 survey than before the survey
which is consistent with this hypothesis. Alternately, the decline could have been caused
by the movement of sand and suspended sediment in the area which smothered epifauna
and reduced seagrass. There was evidence of this at a few sites where we recorded on the
data sheets that there was a layer of clean sand over a muddier sand base and at some
other sites where we recorded that the corals were partially buried by sand. There are
mobile sand waves in many regions of Torres Strait including the west end of Buru
Island and parts of the region are in a tidal bedload parting zone which is characterised
by strong tidal currents and shifting sand waves (Harris 1991).
There was evidence of an unusually large but short-lived run-off event from the Mai
River on the Papuan mainland north of Boigu at the end of the 1990/91 monsoon
season. Reduced salinity was probably not the cause of the seagrass decline because
seagrass was present on the foreshore areas along the south margin of Boigu Island and
also at the east end of Boigu Island which is only a few km from the mouth of the Mai
River. Thus the cause of the decline of seagrass can not be determined with any degree
of certainty, however, the information suggests that a complex interaction of
hydrological and sedimentary factors and river runoff event was responsible for the
dieback These factors resulted in the loss of large areas of seagrass in the subtidal inter-
reefal areas and a significant reduction in the biomass, but not complete loss of seagrass
in the intertidal and shallow subtidal foreshore areas of south Boigu Island.
The abundance of the sea urchin, Prionocidaris sp. was high (up to 1 per m2) in areas
where seagrass had disappeared in the north eastern part of the study area and was
significantly higher than other areas. Although they may not cause the dieback the large
numbers in some areas may interfere with the recovery of seagrass. This aspect of the
dieback requires further research.
The results of this study suggest that run-off pulses may well prove to be an important
factor in structuring the seagrass and epifauna communities near the major rivers on the
Papuan New Guinea coast which empty into Torres Strait.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 22
Acknowledgments
This study has been funded by the Australian Fisheries Management Authority as part of
its Torres Strait Protected Zone fisheries research program. Some results of this project
were based on data collected by CSIRO for their tropical rock lobster research.
References
Admiralty (1973). Australia Pilot. Vol. III. 6th ed. Oxford University Press, 320 pp.
Anon. (1996). Australian National Tide Tables 1996, Australian Hydrographic
Publication 11, Australian Government Publishing Service, Canberra, 353 pp.
Birch W.R. and Birch M. (1984). Succession and pattern of tropical intertidal seagrasses
in Cockle Bay, Queensland, Australia: a decade of observations. Aquatic Botany, 19:
343–367.
Bode L. and Mason L.B. (1994). Numerical modelling of tidal currents in Torres Strait
and the Gulf of Papua. Report to Victorian Institute of Marine Science, Melbourne,
Australia, 65 pp.
Bridges K.W., Phillips R.C. and Young P.C. (1982). Patterns of some seagrass
distributions in the Torres Strait, Queensland. Australian Journal of Marine and Freshwater
Research, 33: 273–283.
Brouns J.J.W.M. (1985). A comparison of the annual production and biomass in three
monospecific stands of the seagrass Thalassia hemprichii (Ehrenb.) Aschers. Aquatic
Botany, 23: 149–175.
Brouns J.J.W.M. (1987). Aspects of production and biomass of four seagrass species
(Cymodoceoideae) from Papua New Guinea. Aquatic Botany, 27: 333–362.
Coles R.G., Lee Long W.J. and Squire L.C. (1985). Seagrass beds and prawn nursery
grounds between Cape York and Cairns. Queensland Department of Primary
Industries Information Series, QI85017, 31 pp.
Cressie N.A. (1993). Statistics for Spatial Analysis. John Wiley and Sons, NY, 900 pp.
den Hartog C. (1970). The seagrasses of the world. North-Holland Publishing;
Amsterdam, 275 pp.
Dennison W.C. (1987). Effects of light on seagrass photosynthesis, growth and depth
distribution. Aquatic Botany, 27: 15–26
Folk R.L. (1968). ‘Petrology of Sedimentary Rocks.’ (University of Texas Press: Austin.)
Green R.H. (1993). Application of repeated measures designs in environmental impact
and monitoring studies. Australian Journal of Ecology, 18: 81–98
T O R R E S S T R A I T S E A G R A S S D I E B A C K 23
Harris P.T. (1991). Reversal of subtidal dune assymetries caused by seasonal reversing
wind-driven currents in Torres Strait, north eastern Australia. Cont. Shelf Res. 1991.
11/7: 655-662
Harris, P.T. (1988). Sediments, bedforms and bedload transport pathways on the
continental shelf adjacent to Torres Strait, Australia-Papua New Guinea, Continental
Shelf Research, 8: 979-1003.
Isdale P.J. (1996). Coral rain gauges: the fluorescence proxy record in reef corals.
Proceedings of the IGBP-PAGES/PEP-II Symposium on Paleoclimate and
Environmental variability in the Austral-Asia Transect during the past 2000 years.
IAHS Nagoya University, Japan. pp51-59.
Lee Long W.J., Coles R.G. and McKenzie L. J. (1996). Deepwater seagrasses in
northeastern Australia — how deep, how meaningful. In: Seagrass Biology [J. Kuo,
R.C. Phillips, D.I. Walker and H. Kirkman eds.], Proceedings of an International
Workshop, Rottnest Island, Western Australia 25–29 January 1996
Lee Long W.J., Mellors J.E. and Coles R.G. (1993). Seagrasses between Cape York and
Hervey Bay, Queensland, Australia. In: Tropical Seagrass Ecosystems; Structure and
Dynamics in the Indo West Pacific. Australian Journal of Marine and Freshwater Research,
44: 19–31
Long B.G. and Poiner I.R. (1993). Distribution and community structure of the
seagrasses of Torres Strait. Final report to the Torres Strait Fisheries Scientific
Committee, April 1993, 110 pp.
Long B.G. and Poiner I.R. (1997). Seagrass communities of Torres Strait. Report to
TSFSAC #27, 1997, 49 pp.
Long B.G. and Skewes T.D. (1994). Use of GIS for research and management of the
marine resources of Torres Strait. (Abstract) 6th Pacific Congress on Marine Science
and Technology, Townsville, Australia 4–8 July, p. 178
Long B.G., Skewes T.D. and Poiner I.R. (1994b). An efficient method for estimating
seagrass biomass. Aquatic Botany, 47: 277– 291
Long B.G., Skewes T.D. and Poiner I.R. (1995). Torres Strait marine geographic
information system. In: Recent advances in marine science and technology ‘94 [O.
Bellwood, H. Choat & N. Saxena (eds.)], 4–8 July, 1994, Townsville, Australia, pp.
231–239
Marsh H. (1995). Torres Strait Dugong, 1994. Stock Assessment Report, edited by the
Torres Strait Fisheries Assessment Group. AFMA, Canberra, pp.200
Pitcher C.R., Dennis D. and Skewes T. (1995). Report to TSFSAC.
Pitcher C.R., T.D. Skewes D.M. Dennis and J.H. Prescott (1992). Distribution of
seagrasses, substratum types and epibenthic macrobiota in Torres Strait, with notes on
pearl oyster abundance. Australian Journal of Marine and Freshwater Research, 43: 409–419
T O R R E S S T R A I T S E A G R A S S D I E B A C K 24
Pitcher, C.R., T.D. Skewes, D.M. Dennis (1994) Research for management of the ornate
rock lobster, Panulirus ornatus, fishery in Torres Strait: final report on CSIRO
research from 1990-1993. CSIRO Division of Fisheries Final Report, 47pp.
Poiner I.R., Staples D.J. and Kenyon R. (1987). Seagrass communities of the Gulf of
Carpentaria. Australian Journal of Marine and Freshwater Research, 38: 121–131
Preen, A.R., Lee Long, W.J. and Coles, R.G. (1995). Flood and cyclone related loss, and
partial recovery, of more than 1000 km2 of seagrass in Hervey Bay, Queensland,
Australia. Aquatic Botany, 52(1–2): 3 – 19.
Underwood A.J. (1993). The mechanics of spatially replicated sampling programmes to
detect environmental impacts in a variable world. Australian Journal of Ecology, 18: 99-
116
Walker, D.I. and McComb, A.J. (1992). Seagrass degradation in Australian coastal waters.
Marine Pollution Bulletin, 25(5-8): 191-195.
Wolanski E., Pickard G.L. and Jupp D.L.B. (1984). River plumes, coral reefs and mixing
in the Gulf of Papua and Northern Great Barrier Reef. Estuarine, Coastal and Shelf
Science, 18: 291–314
Wolanski, E., Ridd, P. and Inoue, M. (1988). Currents through Torres Strait. Journal of
Physical Oceanography, 18: 1535–1545.
IN NORTH WESTERN TORRES STRAIT
Brian Long
Timothy Skewes
Thomas Taranto
Mervyn Thomas
Peter Isdale
Roland Pitcher
Ian Poiner
June 1997
REPORT MR-GIS 97/6
T O R R E S S T R A I T S E A G R A S S D I E B A C K 2
Executive Summary
To assess the magnitude and extent of a reported seagrass decline in north-western
Torres Strait the distribution and abundance of seagrass from a survey of 332 sites in
November 1993 (post-impact) in a study area of 4,388 km2 was compared with historical
data of the distribution and abundance of seagrass collected at 498 sites in the study area
between 1986 and 1989 (pre-impact).
There was an estimated loss of 1,199 km2 of seagrass in the north-eastern region of the
study area, in the main impact area, which represented a 60% loss of seagrass there. In
contrast there was little change in the area of seagrass which was mapped for the
southern region of the study area.
On the foreshore areas at three locations where quantitative samples of seagrass biomass
were taken at 201 sites with a 0.07 m2 core a the biomass of a mixed assemblage of
seagrasses on the foreshore and shallow subtidal areas along the southern margin of
Boigu Island at the northern end of the study area was significantly lower in the post-
than pre-impact surveys. Before the November 1993 survey the seagrass beds were
numerically dominated by Cymodocea serrulata, Thalassia hemprichii and Enhalus
acoroides with high biomass, 66 g.m-2, whereas in the November 1993 survey only
T. hemprichii and to a lesser extent C. serrulata was common and biomass, 18 g.m-2, and
diversity was significantly lower. In contrast there were no significant differences
between estimates of seagrass biomass for the historical 1986–1989 and the November
1993 survey at two ‘control’ locations just south of the study area at Badu Island. The
species composition was also similar for the historical data and the November 1993
survey at these locations.
In addition to the changes in seagrass in the northern section of the study area there were
also large changes in the distribution and abundance of epifauna in this region over a
three year period from 1989 to 1993. Whereas there was dense or sparse epifauna present
throughout much of the study area north of Buru Island in May and June 1989, epifauna
was mostly sparse or absent in the November 1993 survey.
Analysis of a coral head of Porites sp. from Boigu Island indicated that freshwater runoff
from the Mai River in the 1990/91 wet season was very unusual with a short but intense
run-off at the end of the season, which correlated well with the changes in seagrass,
epifauna and areas with high abundance of sea urchins in the northern and north eastern
region of the study area.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 3
Introduction
The decline and loss of seagrass is being reported at an increasing rate around the world
(den Hartog, 1970; Preen et al., 1995). Changes in seagrass occur at a range of spatial
and temporal scales due to anthropogenic and natural causes, and the complex
interaction of the two (den Hartog 1970). Anthropogenic causes related to decrease in
water quality have been linked to seagrass decline in coastal waters (Walker and
McComb, 1992). Natural effects include cyclones and storm events which physically
removes seagrass by scouring, and indirectly through lowered salinity and increased
turbidity (Preen et al., 1995).
In the early 1990’s there were reports by rock lobster fishermen and Torres Strait
Islanders that seagrass was disappearing in north western Torres Strait. The anecdotal
evidence was corroborated by CSIRO scientists doing ongoing field work in north
western Torres Strait on the rock lobster Panulirus ornatus (Pitcher et al., 1994). In
addition to the seagrass decline they also noted that epifauna had declined and that there
was an abrupt increase in the abundance of sea urchins in the area. Concerns over the
possible repercussions of a seagrass dieback on the dugong, prawn and fish populations
prompted a survey of the seagrass, epifauna and sea urchins of north western Torres
Strait in November 1993 (hereafter called the dieback survey). There was extensive
qualitative historical data (1984–1989) of seagrass presence or absence in the inter-reefal
areas (Long and Poiner 1993). There was also detailed historical quantitative data of
seagrass biomass collected along the southern margin of Boigu Island and two bays at the
southern and eastern end of Badu Island (ibid.). These two historical data sets provided
the necessary information on the distribution and abundance of seagrass in the area
before the reported dieback to compare with data collected after the impact for a
Before/After/Control/Impact experimental design which is a prerequisite for detecting
an environmental impact (Underwood, 1993). To identify possible environmental causes
of the reported dieback freshwater run-off events from nearby rivers which discharge
into the area were correlated with changes in seagrass. The patterns of growth of coral
heads, which accurately record freshwater run-off events (Isdale, 1996), were analysed as
there was no data available on freshwater discharge by rivers into the study area.
Materials and Methods
Description of the study area
Torres Strait lies between the NW coast of Cape York Peninsula and the S coast of
Papua New Guinea, and connects the Coral and Arafura Sea (Fig. 1). The physical
oceanography and sedimentary geology has been described by Wolanski et al. (1988),
Harris (1988) and Bode and Mason (1994). The Straits are shallow (< 15 m) with strong
tidal currents due to large pressure gradients between the Arafura and Coral Sea (Bode
and Mason, 1994). Water speeds exceeding 2.5 m.s-1 occur in the narrow channels
between some islands and reefs (Admiralty, 1973).
T O R R E S S T R A I T S E A G R A S S D I E B A C K 4
PNG 142° 143° E
N
Mai River
PNG
Boigu Is.
Buru Is.
Dauan Is.
Aldai Reef
Mabuiag Is. Orman Reefs
10° S
Badu Is. Moa Is.
Torres Strait
Cape York
AUSTRALIA
0 50 100 km
Figure 1. Map of Torres Strait, showing the boundaries of the seagrass dieback survey done in November
1993.
The strong tidal currents have created sand waves in many areas of Torres Strait (Harris,
1988) including north western Torres Strait. There are two distinct seasons in Torres
Strait: a dry season which runs for seven months from May to November with an
average rainfall of 21.4 mm month-1, and a wet monsoon season which lasts for five
months from December to April with an average monthly rainfall of 311 mm at
Thursday Island (Admiralty, 1973). The prevailing winds for the two seasons are also
distinct with south-east trade winds blowing from E and SE 90% of the time during the
dry season whereas winds are more variable during the wet monsoon and blow from the
NE, N and NW for 30% of the time. The average wind speed is lower in the wet
monsoon, 5 knots.h-1, than dry season, 7.9 knots.h-1, and the number of calm days is also
lower in the dry season, < 1 day.month-1 than wet monsoon, 2.1 days.month-1. There are
more gales during the monsoon than dry season (6 and < 1 days.month-1 respectively).
There is little net flow of water through Torres Strait although there are seasonal
differences in the direction of net flow with a net westwardly flow over the dry season
with the south-east trade winds and a net eastwardly flow over the wet monsoon season
when westerlies and north westerlies prevail (Wolanski et al., 1988). The winds and
currents stir up the bottom sediments in shallow water areas of central Torres Strait
which results in a turbidity maximum zone in central Torres Strait (Harris, 1988).
Seagrass was reported to be declining in north western Torres Strait, north of Buru
Island, however, the extent of the dieback was not known so a study area large enough to
T O R R E S S T R A I T S E A G R A S S D I E B A C K 5
encompass much of north western Torres Strait was defined. The southern limits of the
study area were Badu and Moa Island, situated mid-way across the Straits; Boigu and
Dauan Island near the S coast of Papua New Guinea formed the northern boundary; the
eastern limit of the study area were formed by a line NE from Moa Island through the
Orman reefs and N to Dauan Island; and the 142nd meridian of longitude formed the
western boundary (Fig. 1).
Field sampling: Inter-reefal areas
Historical data
The percentage cover of seagrass in 0.25 m2 quadrats along 20 m transects were recorded
by divers at 253 subtidal sites in the study area in north western Torres Strait during
seven cruises from 1986 to 1989 as part of the seagrass study (Long and Poiner, 1993)
(Fig. 2a). At a further 112 sites the relative cover of seagrass was recorded by divers for
500 m x 2 m transects sampled during a lobster survey in 1989 (Pitcher et al., 1992).
Because the sampling methods differed the data from the seagrass study and the lobster
survey were converted to seagrass presence or absence data which gave a total of 365
sites sampled on the seabed in the study area from 1986 to 1989.
In addition to the historical data on the distribution and abundance of seagrass there was
extensive data on the distribution and abundance of epifauna in the region between
Boigu and Badu Island which was collected as part of the lobster survey in 1989 (see
Pitcher et al. 1992 for full details). Epifauna was recorded as dense, sparse, very sparse or
absent for the 500 m x 2 m transects by divers at the 112 sites sampled in the study area.
Sediment samples were taken at 9 sites within the study area during a seagrass survey in
1989.
November 1993 survey
The seagrass at 251 subtidal sites in the study area were sampled in November 1993 to
assess the magnitude and extent of the reported seagrass dieback (Fig. 2b). The study
area was first divided into primary sampling units which were each 4.5 km east-west and
4.2 km north-south. The primary sampling unit area, 18.9 km2, was chosen based on
estimates of the time it would take to sample a site (15 min), the time to travel between
sites and the total time (three weeks) available for field sampling. To give complete
sampling coverage of the study area all primary sampling units were sampled. Because it
was impractical to sample the whole primary sampling unit (18.9 km2) the area sampled
in the field were 100 m2 sites. The position of each site within each primary sampling unit
was chosen randomly. Global Positioning System (GPS) satellite navigation was used to
locate the sites in the field. At all sites divers searched an area of approximately 100 m2
and recorded the presence or absence of seagrass along with estimates of the abundance
of sea urchins (per 25 m2) as well as descriptions of the substratum and epifauna. The
epifauna cover was scored into four categories: dense, sparse, very sparse and absent to
conform with the categories devised by Pitcher et al. (1992). Water visibility (m) and
water depth (m) were also recorded at each site and a sediment sample was taken for
grain-size analysis.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 6
a). b).
10 km 10 km
Seagrass Seagrass
present
present
absent
absent
.
Figure 2. Seagrass presence or absence data from the a). historical (1986–1989) seagrass data; and b). November 1993 survey.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 7
Field sampling: Foreshore areas and shallow embayments
Historical data
Quantitative samples of seagrass biomass were collected with a 0.07 m2 shovel by divers
from a dinghy at 133 sites sampled on foreshore and shallow subtidal areas at three
locations in the study area (Fig. 3a & b). At each site 3 to 5 shovel samples were taken at
fixed intervals down a 20 m transect (for full details of the sampling see Long and Poiner
1993). The seagrass was placed in a divers mesh bag and returned to the surface where it
was sieved over the side of the dinghy in a 10 mm lug basket to remove the sediment
from the rhizomes, labelled and stored on ice for further processing at the CSIRO
marine laboratories, Cleveland, Qld. The location sampled on the foreshore area on the
southern side of Boigu Island was selected as the impact location in the
Before/After/Control/Impact analysis. Forty-three sites were sampled there in an area
of 3.413 km2 from 1986 to 1989 before the reported dieback (Fig. 3a). Two embayments
just south of the study area at the south and east end of Badu Island were selected as
control locations with 38 and 54 sites sampled from 1986 to 1989 in an area of 2.02 and
1.911 km2 respectively (Fig. 3b).
November 1993 survey
Quantitative sampling of seagrass biomass was done at Boigu and Badu Islands to match
up with the historical data sampled at these three locations in the
Before/After/Control/Impact (BACI) experimental design to test for significant changes
in seagrass biomass at these locations (Green 1993). At the location on the foreshore area
of the southern margin of Boigu Island (impact location), 24 core samples were taken
with a 0.07 m2 seagrass grab (Long et al. 1994) along two transects separated
approximately 1 km east-west and 170 m north-south to match up with the samples
taken there from 1986 to 1989 (Fig. 3a). At the east and south end of Badu Island, just
south of the study area (control locations) in sheltered embayments of the island, 21
samples were taken from each location respectively in a grid arrangement with 310 m
separation in both the east-west and north-south directions to match up with the samples
taken there from 1986 to 1989 (Fig. 3b).
Back at CSIRO Marine Laboratories the seagrass samples were placed in a 10% solution
of orthophosphoric acid for 30 min to clean the epiphytes from the shoots (refer to Long
and Poiner, 1993 for full details). The samples were rinsed in freshwater and separated
into species and the above-ground shoots and stems were separated from the below
ground rhizomes. The seagrass was dried at 60°C until a constant weight and weighed to
the nearest 0.1 g.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 8
Data Analysis
Mapping — seagrass presence or absence
Point-to-area spatial data transformations with a Geographic Information System (GIS)
was used to create Voroni maps of seagrass presence or absence for the study area for
the historical (1986–1989) and the November 1993 survey data. The two maps were
overlaid to produce a third which mapped areas where seagrass was present over the
entire eight year study period; seagrass was absent over the study period; seagrass was
present during the survey but was absent before the survey and; seagrass was absent
during the survey but was present before the survey. The last category mapped areas
where seagrass had disappeared.
a). b).
N
Mai River 0 5 10 km
Papua New Guinea
Badu Island
Boigu Island
N
0 2.5 5 km
Boigu Island - south
Badu Island - east
0 500 1000 m
0 500 m 1 km
Badu Island - south
0 1 km
Figure 3. Sites sampled on the foreshore in the impact area at a). the south side of Boigu Island; and
control areas at b). the east and south end of Badu Island. : Historical sites; circle, : November 1993
seagrass dieback survey; dotted line - convex hull around sample points.
BACI analysis
The seagrass biomass sampled at 135 sites on the foreshore and shallow water
embayments of Boigu and Badu Island from 1986 to 1989 were not suitable for a simple
BACI ANOVA because sampling was not random as the sites were spatially clumped
mainly along transects. Therefore large scale spatial trends at each location and each time
on the foreshore areas of Boigu and Badu Island were investigated by fitting a loess
response surface to latitude and longitude. The residuals from the trend surface were
then examined, first to assess the appropriateness of the assumptions that the residuals
were normally distributed with constant variance, and secondly to assess the extent of
T O R R E S S T R A I T S E A G R A S S D I E B A C K 9
any smaller scale spatial correlation. For each site and sample time, a robust estimate of
the semi-variogram was generated (Cressie,1993) and plotted against inter-sample
distance.
Because the location of samples differed between time points, any spatial trends or
dependencies would result in biased estimates of change over time. To accommodate
these effects, the different time points were compared by estimating the mean response
(from the fitted surface and/ or semi variogram) in a fixed polygon (i.e. the same
polygon was used for both times). For this analysis the spatial response was modelled as
a polynomial in latitude and longitude. Whilst less flexible than a loess surface, this yields
simple expressions for the estimated mean response, and for the standard error of the
estimated mean.
In the presence of spatial correlation, the standard error is based on equations given in
Cressie (1993). In the absence of such spatial correlation the standard error is simpler.
Since the estimated mean is simply a linear combination of the regression coefficients for
the fitted surface, it can be obtained directly from the estimated co-variance matrix of the
parameters.
The polygon for each site (Fig. 3) was based on a convex hull around the set of sample
points (aggregated over both sample times). A regular grid of points was generated within
the polygon, and the response integrated by a process of simple averaging. A fifty point
grid was used in each dimension.
Sediment grain size analysis
Sediment samples collected during the historical and November 1993 survey were
analysed for gran size fractions by the method given in Folk 1968 to give percentage
grain size for mud (< 62 µm), sand (62–1,000 µm) and gravel (> 1.0 mm). There were
too few sediment samples (9) taken in the study area for an accurate ANOVA of grain
size fractions so historical and November 1993 samples were matched by geographic
location and a paired sample t-test was used to compare the sediment grain sizes between
the historical and November 1993 surveys. To do this 17 samples taken in 1993 which
were within a 4 km radius of the 9 sites sampled in 1989 were paired up. The three grain
size fractions, gravel, sand and mud, of the November 1993 sediment samples were
averaged for each of the 9 sites to give a paired comparison.
Coral heads were collected from Boigu, Aldai and Mabuiag reef to provide proxt
information about the recent history of seasonal freshwater inputs to the area. The
timing and intensity of fluorescing bands in coral skeletons have been found to correlate
quite well with instrument records of river runoff from adjacent tropical coasts (Isdale,
1996). We used the bands in coral from Boigu to qualitatively reconstruct the freshwater
input history to the study area for the period of interest.
CSIRO scientists researching the tropical rock lobster have been monitoring three sites
in the study area north of Buru Island on a yearly basis from 1989 to 1993. Descriptions
of the epifauna and substrate type were recorded. This data provided ancillary
information on changes of seagrass and epifauna at these three sites over a six year
period.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 10
Results
Seagrass mapping: Inter reefal areas
Seagrass was distributed widely throughout the entire study area from 1986 to 1989
(Fig. 2a). The area of seagrass mapped with GIS, 3,078 km2, was larger than seagrass
absent, 1,188 km2 (Table 1). In contrast, the area of seagrass, 1,871 km2, was lower than
seagrass absent, 2,394 km2 (Fig. 2b). An overlay of the two maps of seagrass presence or
absence of the historical and November 1993 survey data indicated that the largest loss
of seagrass was in the north and north eastern region of the study area between Boigu
and Buru Island, and south east between Buru Island and the top of the Orman Reefs in
region A (Fig. 4). In region A, 1,993 km2, or two thirds (60.2%) of the area had seagrass
before the November 1993 survey and none during the survey. Very little of region A,
3.6%, had no seagrass before the survey and seagrass during the November 1993 survey.
There was a large area, 195 km2, in the north west of the study area which had no
seagrass over the seven year study period from 1986 to 1993. In contrast there was little
evidence to suggest that seagrass had declined in the southern half of the study area
between Buru and Badu Island in region B (Fig. 4). In region B, 2,201 km2, about half the
area (40%) had seagrass before and during the November 1993 survey; 23% had seagrass
before the November survey and no seagrass during; 19% had no seagrass before the
survey and seagrass during the November 1993 survey; and seagrass was absent over the
seven year study period in the remaining 17% of region B.
Table 1. Area analysis of seagrass change (km2) based on the overlay of maps of seagrass presence or
absence from historical 1986-1989 data and the November 1993 seagrass dieback survey data.
Seagrass Region A Region B Total
Absent before, absent now 289 365 653
Absent before, present now 72 412 485
Present before, absent now 1,199 515 1,714
Present before, present now 433 908 1,342
Total 1,993 2,200 4,194
Seagrass change: foreshore and shallow water embayments
The seagrass beds on the shallow water foreshore areas on the southern side of Boigu
Island were lush, mixed species assemblages from 1986 to 1989 numerically dominated
by Cymodocea serrulata, Thalassia hemprichii, Cymodocea rotundata and Enhalus acoroides (Fig. 5).
In contrast the seagrass beds there during the November 1993 survey were mainly
stunted Thalassia hemprichii with scattered Cymodocea serrulata. The detailed survey of east
and south Badu Island indicated that the species composition of seagrass had not
T O R R E S S T R A I T S E A G R A S S D I E B A C K 11
changed and were numerically dominated by Cymodocea rotundata and Thalassia hemprichii at
south Badu and Enhalus acoroides, C. serrulata and T. hemprichii at east Badu Island (Fig. 5).
An analysis of variance for the Loess surface indicated that there were statistically
significant medium-scale (100’s of metres) spatial trends at East Badu before the impact,
and South Boigu after the impact (Table 2).
Table 2. Fit of Loess trend surface by location and sampling occasion. ***: P < 0.0001.
Before After
Location F p Value F p Value
South Badu 0.30 0.966 2.50 0.074
East Badu 3.68 0.000*** 1.26 0.340
South Boigu 1.43 0.217 5.26 0.000***
Small Scale Spatial Dependencies
The Loess response surface was removed from the data before calculation of the semi-
variogram for all three locations, before and after impact (Fig 6a–c). All of the
variograms were essentially flat, which provided little evidence of any spatial dependency,
once the larger scale spatial trends were removed. Thus at the scale of spatial separation
in our study, different samples may be construed as essentially independent.
Fitted Response
There were no significant differences among locations before the reported seagrass
dieback but there were significant differences after the impact (Table 3). The control
versus impact contrast was also statistically significant for the change over time. That is,
the change over time for South Boigu was statistically significantly different from the
change over time for South and East Badu. Furthermore, this difference arose from a
greater decrease in seagrass biomass over time for South Boigu than the control
locations.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 12
Table 3. Fitted response for each location, Before and After Impact and the estimated change in mean
response over time for each location, and a comparison of the control locations (mean of East Badu and
South Badu versus South Boigu).
Pre Impact Post Impact Change
Location Mean s2 Mean s2 Mean s2
East Badu 4.032 0.064 3.503 0.204 0.529 0.268
South Badu 4.298 0.150 2.717 0.152 1.581 0.302
South Boigu 4.646 0.025 1.260 0.057 3.385 0.081
Control vs Impact -0.481 0.078 1.850 0.146 -2.330 0.224
Z-score -1.718 4.849 -4.926
T O R R E S S T R A I T S E A G R A S S D I E B A C K 13
Region A
N
land
reef
Region B AA
AP
PA
PP
0 50 100 km
Figure 4. Torres Strait: Area analysis of change of seagrass produced by overlaying a seagrass presence or
absence map based on data collected from 1986–1989 and a seagrass presence or absence map based on
data collected during the November 1993 seagrass dieback survey. PA: seagrass present before but absent
during the November 1993 survey; AA: seagrass absent before and during the November 1993 survey; AP:
seagrass absent before but present after the November 1993 survey; PP: seagrass present before and after
the November 1993 survey. For the main impact region A, and remaining study areas region B.
2
S e a g r a s s b io m a s s ( g . 0 .0 7 m )
4 .6 5 1 .2 6 4 .0 3 3 .5 0 4 .3 0 2 .7 2
100%
90
80
70
60
50
40
30
20
10
0 PR E PR E PR E
PO ST PO ST PO ST
S BO I E BA D S B AD
Im p a c t C o n tr o ls
Figure 5. Comparison of above-ground biomass broken down by species for east, south and west Badu,
and south Boigu. Sampling occasions — PRE: historical 1984 to 1989 data; DIE: November 1993 seagrass
dieback survey. Locations — EBAD: east Badu; SBAD: south Badu; SBOI: south Boigu and WBAD: west
Badu Island for : Cymodocea rotundata; : Cymodocea serrulata; : Enhalus acoroides; :
Halophila ovalis; : Syringodium isoetifolium; : Thalassia hemprichii; and : Halodule uninervis..
Vertical axis: proportion of total.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 14
a). Control Site: South Badu Control Site: South Badu b). Control Site: East Badu Control Site: East Badu
Pre Impact Post Impact Pre Impact Post Impact
2.5 2.5
3 3
2.0 2.0
1.5 2 1.5
2
gamma gamma gamma gamma
1.0 1.0
1 1
0.5 0.5
0 0.0 0 0.0
0.0 0.002 0.004 0.006 0.008 0.0 0.002 0.006 0.010 0.0 0.002 0.004 0.006 0.008 0.010 0.0 0.002 0.004 0.006 0.008
Distance (m) Distance (m) Distance (m) Distance (m)
T O R R E S S T R A I T S E A G R A S S D I E B A C K 15
c). Impact Site: South Boigu Control Site: South Boigu
Pre Impact Post Impact
2.0
0.6
1.5
0.4
1.0
Gamma
0.2
0.5
0.0 0.0
0.0 0.002 0.006 0.010 0.0 0.004 0.008 0.012
Distance (km) Distance (km)
Figure 6. Torres Strait Before and After Impact (BACI) empirical variogram for Control areas: a). South
Badu; and b). East Badu; and Impact area: c). South Boigu Island.
Environmental data
The sediments in the study area were mainly gravelly sands with mud (Fig. 7). The
average percentage fraction of gravel in region A of the study area in the November 1993
survey, 18.75%, was significantly lower than 1989, (33.37%) based on the paired t-test
(Table 4). The sand fraction in the November 1993 survey, 73.23%, was significantly
higher than 1989, 57.58%. In contrast, there was no significant difference in the
percentage mud between the two years with an pooled average of 8.5% mud. At a couple
of sites, we recorded on our data sheets that there was a thin layer (cm’s) of clean sand
overlying a muddy sand base.
Table 4. T-test for paired comparison of percentage grain size
_
fraction of gravel, sand and mud for sediment samples taken in
1989 and the November 1993 survey. D : mean difference of _
1989 samples - 1993 samples; sD: std. of the difference; sD
stderr. of the difference; ts : t-statistic. *: P < 0.05.
%Gravel %Sand %Mud
1989 33.37 57.58 9.04
1993 18.75 73.23 8.03
_
D 14.62 -15.65 1.02
sD 16.42 16.43 4.54
_
sD 5.47 5.48 1.51
St 2.67* -2.86* 0.67
T O R R E S S T R A I T S E A G R A S S D I E B A C K 16
Ancillary data
Epifauna was dense or sparse throughout much of the study area north of Buru Island in
1989 (Fig. 8a) whereas epifauna was sparse or absent in during the November 1993
survey (Fig. 8b). A visual comparison of the two maps indicated that epifauna had been
affected over a larger area than seagrass. At a couple of sites north of Buru Island we
recorded the presence of white dead corals with little algal growth on them which were
partly covered by clean sand.
% gravel
80
50
25
0 25 50 km
Figure 7. Bubble plots of percentage gravel fraction (> 1.0 mm) at 9 sites sampled in the study area during
1989 (clear bubbles) and 16 sites sampled within a 4 km radius of the 1989 samples during November 1993
(solid bubbles).
The sea urchin, Prionocidaris sp. was sampled only in region A of the study area and where
present, their abundance ranged from 2 to 20 animals per 25m2 (Fig. 9). To test for
significant differences between areas where seagrass was lost and the remaining areas an
ANOVA was done on square-root transformed abundance of urchins. The results
indicated that the abundance of Prionocidaris sp. was significantly higher (P < 0.005) in
areas where seagrass was lost (2.326 animals.25m2; se = 0.503) than remaining areas in
region A (0.365 animals.25m2; se = 0.293).
Examination of the coral heads from Boigu, Aldai and Mabuiag reefs indicated that only
the Porites spp.coral head from Boigu was suitable for providing a seasonal proxy record
of freshwater inputs to the area. The corals taken from Aldai and Mabuiag reef were
unsuitable because the corallite dimensions were coarse, a feature of most of the Favid
species . Analysis of the fluorescence in the coral head from Boigu indicated that there
T O R R E S S T R A I T S E A G R A S S D I E B A C K 17
were four abnormal seasons: 1982/83, 1983/84, 1984/85 and 1990/91 (Fig. 10). The
1982/83 and 1984/85 seasons were wetter than averga and run-off was in a single long
seasonal pulse. In contrast the 1983/84 season was apparently very dry. The 1990/91
season was very unusual. It started normally and run-off occurred as a consistent pulse
until near the end of the season, where reduced somewhat, but then showed a short but
intense run-off period shortly after December 1990.
Discussion
More than 1,400 km2 of seagrass north and north east of Buru Island in region A of the
study area was lost between 1989 and 1993 which represented a 60% reduction in
seagrass in the study area and a 10% reduction of the 13,425 km2 of seagrass estimated
for the whole of Torres Strait (Long and Poiner, 1997). The magnitude of the loss of
seagrass in Torres Strait is the same as a recently reported dieback of > 1000 km2 of
seagrass reported in 1992 for Hervey Bay (Preen et al., 1995).
The results of the BACI analysis indicated that there was a significant reduction of
seagrass on the foreshore areas along the southern margin of Boigu Island in the impact
area than the control locations at South and East Badu Island. Moreover, because there
was clear spatial structure in seagrass biomass for East Badu and South Boigu the analysis
of change of seagrass over time must take into account any differences in the position of
sample points. There were, however, no small scale spatial correlation among samples at
the distances sampled (10’s to 100’s m apart) in this study. Thus the different samples
may be construed as essentially independent. This finding greatly simplified the
estimation of the standard error since the estimated mean is simply a linear combination
of the regression coefficients for the fitted surface and can be obtained directly from the
estimated co-variance matrix of the parameters.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 18
a). b).
Boigu Is
Boigu Is
Buru Is
Epibenthos
Epibenthos dense
dense sparse
sparse
absent
absent
Badu Badu Moa Is
Figure 8. Distribution of epifauna in the study area in (a) 1989 lobster survey and (b) November 1993 survey.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 19
#
##
A #
#
#
##
#
##
### #
# #
#
#
# #
# ##
#
# # ## #
#
Prionocidaris sp.
per 100 m2
B
#
# # 80
40
10
N
0 50 100 km
Figure 9. Bubble plot of the sea urchin, Prionocidaris sp. per m2, estimated from the visual spot dives during
the November 1993 seagrass dieback survey. red areas: Seagrass absent after the impact and present before.
Region A: main area of impact; Region B: no impact.
1 9 9 3 /4 s u m m e r
250
F in e f lu o r e s c e n t b a n d s o o n a f te r D e c e m b e r 1 9 9 0
200
150
100
liv in g tis s u e
50
0 40 80 120 160 200
s te p s ( 0 .5 m m )
Figure 10. Plot of cross-sectional fluorescence of Porites sp. coral head from the reef flat at the east tip of
Boigu Island. Vertical axis: Fluorescence (relative units)
T O R R E S S T R A I T S E A G R A S S D I E B A C K 20
There have been a number of reported seagrass diebacks in Torres Strait in the early
1970’s which are in themselves anecdotal, but when taken together suggests that diebacks
are a natural event in Torres Strait. The two main rivers that empty into Torres Strait are
the Mai and the Fly river. There is no available data on river runoff by the Mai river.
There has been no marked changes in land clearing practises in the mainland catchment
of Papua New Guinea which empties into Torres Strait land and consequently land
degradation with concomitant increased runoff and suspended sediment can not explain
the loss of seagrass in this study. Preen et al. (1995) attributed the large loss of seagrass in
Hervey Bay to a complex array of factors associated with a cyclone and flooding which
was exacerbated by soil erosion due to land clearing in the adjoining catchment area.
Seagrass is critical habitat for many commercial important species of prawns and fish and
is an important food source for dugongs and turtles. The distribution of dugongs are
reliable indicators of seagrass distribution and disproportionately high concentrations of
dugongs are associated with large seagrass beds in Torres Strait and the Starcke region of
north Queensland (Preen et al. 1995). Torres Strait supports the largest reported
population of dugongs in Australia and possibly the world (Marsh 1995) and the main
area where the seagrass had disappeared coincided with the largest concentration of
dugongs in Torres Strait in 1987 (Marsh 1995). A recent survey of dugongs in 1989 in
Torres Strait has indicated that the largest concentrations of dugongs were now south of
Buru Island (H. Marsh pers. comm.) which indicates that the dieback has affected the
distribution of dugongs in northern Torres Strait. The seagrass dieback Hervey Bay was
followed by a reduction in the population of dugongs from 1937 to less than 200
dugongs (Preen et al. 1995).
The maps of seagrass presence or absence indicated a change of seagrass distribution in
inter reefal areas of north western Torres Strait. The biomass and community structure
of seagrass was also affected shown by the decrease in biomass and diversity and change
in species composition on the foreshore and shallow subtidal areas along the south side
of Boigu Island at the northern end of the study area. There was a significantly lower
biomass of seagrass on the foreshore areas and shallow subtidal areas along the south
side of Boigu Island during the November 1993 survey than before and diversity was also
significantly lower. There was a shift in the species composition of seagrass from a
community dominated by C. serrulata, T. hemprichii and E. acoroides to an assemblage of
low diversity dominated by a single species, T. hemprichii. In contrast, there were no
significant differences in biomass at the south and east end of Badu Island at the
southern end of the study area. The results of the BACI test and the seagrass mapping
suggests that the dieback was largely restricted to the northern part of the study area.
Anecdotal evidence provided by CSIRO fisheries doing research in the area on Panulirus
ornatus (tropical rock lobsters) also suggested that seagrass, lobsters and epifauna had
declined in the area (Pitcher et al. 1994). At three sites north of Buru Island in the study
area, descriptions of the epifauna and seagrass were recorded on a yearly basis from 1989
to 1993.
Lobsters were abundant at the three sites from 1989 to 1991 and were absent in 1992
and 1993. The Orman reef area, which is adjacent to the area of interest north of Buru
Island, normally has abundant lobster but has shown a decline in lobsters in 1993 based
on survey data and from changes in fishing effort by the rock lobster freezer boat fleet.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 21
There have been dramatic changes in seagrass abundance at the three sample sites since
1989. Seagrass abundance declined from high in 1990 (higher even than in 1989) to low
in 1991, to zero or near zero abundances in 1992. The decline was greatest for the ovoid
species, H. spinulosa mainly and H. ovalis to a lesser degree, in part because they were the
dominant of the two forms previously (CSIRO, unpublished data).
Sparse epibenthic 'garden' (hard corals, whips, sponges) bottom and Sargassum spp. were
a common feature of consolidated sand bottoms in the area with occasional reef
epibenthic assemblages on rock outcrops. There was a general decline in the amount of
epifauna at these three sites with little or no epifauna recorded there in 1992.
The disappearance of seagrass in this area was also accompanied by a decline in epifauna,
shown by the results of this study. The loss of seagrass could explain the reduction in
epifauna in the northern region of the study area as the removal of seagrass can
destabilise the sediments and often leads to blowout holes and a shift in the sediment
grain size distribution to sandier sediments which could affect epifauna. There were
significant changes in sediment grain size distribution in the northern region of the study
area with sandier sediments during the November 1993 survey than before the survey
which is consistent with this hypothesis. Alternately, the decline could have been caused
by the movement of sand and suspended sediment in the area which smothered epifauna
and reduced seagrass. There was evidence of this at a few sites where we recorded on the
data sheets that there was a layer of clean sand over a muddier sand base and at some
other sites where we recorded that the corals were partially buried by sand. There are
mobile sand waves in many regions of Torres Strait including the west end of Buru
Island and parts of the region are in a tidal bedload parting zone which is characterised
by strong tidal currents and shifting sand waves (Harris 1991).
There was evidence of an unusually large but short-lived run-off event from the Mai
River on the Papuan mainland north of Boigu at the end of the 1990/91 monsoon
season. Reduced salinity was probably not the cause of the seagrass decline because
seagrass was present on the foreshore areas along the south margin of Boigu Island and
also at the east end of Boigu Island which is only a few km from the mouth of the Mai
River. Thus the cause of the decline of seagrass can not be determined with any degree
of certainty, however, the information suggests that a complex interaction of
hydrological and sedimentary factors and river runoff event was responsible for the
dieback These factors resulted in the loss of large areas of seagrass in the subtidal inter-
reefal areas and a significant reduction in the biomass, but not complete loss of seagrass
in the intertidal and shallow subtidal foreshore areas of south Boigu Island.
The abundance of the sea urchin, Prionocidaris sp. was high (up to 1 per m2) in areas
where seagrass had disappeared in the north eastern part of the study area and was
significantly higher than other areas. Although they may not cause the dieback the large
numbers in some areas may interfere with the recovery of seagrass. This aspect of the
dieback requires further research.
The results of this study suggest that run-off pulses may well prove to be an important
factor in structuring the seagrass and epifauna communities near the major rivers on the
Papuan New Guinea coast which empty into Torres Strait.
T O R R E S S T R A I T S E A G R A S S D I E B A C K 22
Acknowledgments
This study has been funded by the Australian Fisheries Management Authority as part of
its Torres Strait Protected Zone fisheries research program. Some results of this project
were based on data collected by CSIRO for their tropical rock lobster research.
References
Admiralty (1973). Australia Pilot. Vol. III. 6th ed. Oxford University Press, 320 pp.
Anon. (1996). Australian National Tide Tables 1996, Australian Hydrographic
Publication 11, Australian Government Publishing Service, Canberra, 353 pp.
Birch W.R. and Birch M. (1984). Succession and pattern of tropical intertidal seagrasses
in Cockle Bay, Queensland, Australia: a decade of observations. Aquatic Botany, 19:
343–367.
Bode L. and Mason L.B. (1994). Numerical modelling of tidal currents in Torres Strait
and the Gulf of Papua. Report to Victorian Institute of Marine Science, Melbourne,
Australia, 65 pp.
Bridges K.W., Phillips R.C. and Young P.C. (1982). Patterns of some seagrass
distributions in the Torres Strait, Queensland. Australian Journal of Marine and Freshwater
Research, 33: 273–283.
Brouns J.J.W.M. (1985). A comparison of the annual production and biomass in three
monospecific stands of the seagrass Thalassia hemprichii (Ehrenb.) Aschers. Aquatic
Botany, 23: 149–175.
Brouns J.J.W.M. (1987). Aspects of production and biomass of four seagrass species
(Cymodoceoideae) from Papua New Guinea. Aquatic Botany, 27: 333–362.
Coles R.G., Lee Long W.J. and Squire L.C. (1985). Seagrass beds and prawn nursery
grounds between Cape York and Cairns. Queensland Department of Primary
Industries Information Series, QI85017, 31 pp.
Cressie N.A. (1993). Statistics for Spatial Analysis. John Wiley and Sons, NY, 900 pp.
den Hartog C. (1970). The seagrasses of the world. North-Holland Publishing;
Amsterdam, 275 pp.
Dennison W.C. (1987). Effects of light on seagrass photosynthesis, growth and depth
distribution. Aquatic Botany, 27: 15–26
Folk R.L. (1968). ‘Petrology of Sedimentary Rocks.’ (University of Texas Press: Austin.)
Green R.H. (1993). Application of repeated measures designs in environmental impact
and monitoring studies. Australian Journal of Ecology, 18: 81–98
T O R R E S S T R A I T S E A G R A S S D I E B A C K 23
Harris P.T. (1991). Reversal of subtidal dune assymetries caused by seasonal reversing
wind-driven currents in Torres Strait, north eastern Australia. Cont. Shelf Res. 1991.
11/7: 655-662
Harris, P.T. (1988). Sediments, bedforms and bedload transport pathways on the
continental shelf adjacent to Torres Strait, Australia-Papua New Guinea, Continental
Shelf Research, 8: 979-1003.
Isdale P.J. (1996). Coral rain gauges: the fluorescence proxy record in reef corals.
Proceedings of the IGBP-PAGES/PEP-II Symposium on Paleoclimate and
Environmental variability in the Austral-Asia Transect during the past 2000 years.
IAHS Nagoya University, Japan. pp51-59.
Lee Long W.J., Coles R.G. and McKenzie L. J. (1996). Deepwater seagrasses in
northeastern Australia — how deep, how meaningful. In: Seagrass Biology [J. Kuo,
R.C. Phillips, D.I. Walker and H. Kirkman eds.], Proceedings of an International
Workshop, Rottnest Island, Western Australia 25–29 January 1996
Lee Long W.J., Mellors J.E. and Coles R.G. (1993). Seagrasses between Cape York and
Hervey Bay, Queensland, Australia. In: Tropical Seagrass Ecosystems; Structure and
Dynamics in the Indo West Pacific. Australian Journal of Marine and Freshwater Research,
44: 19–31
Long B.G. and Poiner I.R. (1993). Distribution and community structure of the
seagrasses of Torres Strait. Final report to the Torres Strait Fisheries Scientific
Committee, April 1993, 110 pp.
Long B.G. and Poiner I.R. (1997). Seagrass communities of Torres Strait. Report to
TSFSAC #27, 1997, 49 pp.
Long B.G. and Skewes T.D. (1994). Use of GIS for research and management of the
marine resources of Torres Strait. (Abstract) 6th Pacific Congress on Marine Science
and Technology, Townsville, Australia 4–8 July, p. 178
Long B.G., Skewes T.D. and Poiner I.R. (1994b). An efficient method for estimating
seagrass biomass. Aquatic Botany, 47: 277– 291
Long B.G., Skewes T.D. and Poiner I.R. (1995). Torres Strait marine geographic
information system. In: Recent advances in marine science and technology ‘94 [O.
Bellwood, H. Choat & N. Saxena (eds.)], 4–8 July, 1994, Townsville, Australia, pp.
231–239
Marsh H. (1995). Torres Strait Dugong, 1994. Stock Assessment Report, edited by the
Torres Strait Fisheries Assessment Group. AFMA, Canberra, pp.200
Pitcher C.R., Dennis D. and Skewes T. (1995). Report to TSFSAC.
Pitcher C.R., T.D. Skewes D.M. Dennis and J.H. Prescott (1992). Distribution of
seagrasses, substratum types and epibenthic macrobiota in Torres Strait, with notes on
pearl oyster abundance. Australian Journal of Marine and Freshwater Research, 43: 409–419
T O R R E S S T R A I T S E A G R A S S D I E B A C K 24
Pitcher, C.R., T.D. Skewes, D.M. Dennis (1994) Research for management of the ornate
rock lobster, Panulirus ornatus, fishery in Torres Strait: final report on CSIRO
research from 1990-1993. CSIRO Division of Fisheries Final Report, 47pp.
Poiner I.R., Staples D.J. and Kenyon R. (1987). Seagrass communities of the Gulf of
Carpentaria. Australian Journal of Marine and Freshwater Research, 38: 121–131
Preen, A.R., Lee Long, W.J. and Coles, R.G. (1995). Flood and cyclone related loss, and
partial recovery, of more than 1000 km2 of seagrass in Hervey Bay, Queensland,
Australia. Aquatic Botany, 52(1–2): 3 – 19.
Underwood A.J. (1993). The mechanics of spatially replicated sampling programmes to
detect environmental impacts in a variable world. Australian Journal of Ecology, 18: 99-
116
Walker, D.I. and McComb, A.J. (1992). Seagrass degradation in Australian coastal waters.
Marine Pollution Bulletin, 25(5-8): 191-195.
Wolanski E., Pickard G.L. and Jupp D.L.B. (1984). River plumes, coral reefs and mixing
in the Gulf of Papua and Northern Great Barrier Reef. Estuarine, Coastal and Shelf
Science, 18: 291–314
Wolanski, E., Ridd, P. and Inoue, M. (1988). Currents through Torres Strait. Journal of
Physical Oceanography, 18: 1535–1545.