grl52923-sup-0001-supinfo

advertisement
[Geophysical Research Letters]
Supporting Information for
[Coral Reef Metabolism and Carbon Chemistry Dynamics of a Coral Reef Flat]
[Rebecca Albright1,2, Jessica Benthuysen1, Neal Cantin1, Ken Caldeira2, Ken Anthony1]
[1Australian Institute of Marine Science, Townsville MC, Townsville, QLD 4810, Australia
2Department
of Global Ecology, Carnegie Institution for Science, Stanford, California, USA]
Contents of this file
Text S1
Figures S1 to S4
Tables S1 and S2
Additional Supporting Information (Files uploaded separately)
Captions for Datasets S1 to S2
Introduction
[The supplemental information includes a detailed description of the methods (Text S1).
Text S1 is complemented by a schematic diagram of the approach used to generate
bathymetry maps, provided in Figure S1. We also show the variability in the reef flat
carbonate chemistry relative to the adjacent Wistari Channel and atmospheric conditions
(Figure S2); calibration of light data (Figure S3); relationships between net community
calcification (ncc) and light (PAR) and temperature (Figure S4); chemical and physical
conditions of the Heron Island Reef flat during the course of the study (Table S1); and a
table comparing corrected Akaike Information Criteria (cAIC values) that were used for
model selection (Table S2). Captions are included for datasets S1 and S2; these datasets
include data for the Lagrangian surveys and water sampler time-series (i.e., all data that
were used in the analyses for the manuscript).]
1
Text S1.
S1. Material and Methods
S1.1. Study Site
A diversity of coral species (predominantly Acropora) dominate the fore reef,
crest and parts of the outer reef-flat zone at Heron Island. Pavements of crustose coralline
algae (CCA) are also abundant on the wave-impacted front of the crest. Mixed zones of
branching corals, fleshy and calcareous macroalgae (e.g., Halimeda) dominate the 50–
100m belt behind the crest (outer reef flat). Patches of sand and turfs (reef matrix with
assemblages of cropped macroalgae and benthic microalgae) and scattered macroalgae
increase in abundance towards the lagoon [Anthony et al., 2013].
S1.2. Mapping of bathymetry and benthic community structure
Benthic surveys were conducted to characterize the bathymetry and underlying
community structure of the reef flat site. These surveys were conducted over the course
of 5 days (March 5, 6, 9, 22, 23) using transects spaced approximately 20 m apart and
oriented perpendicular to the reef crest. Each transect was surveyed by 2 divers
approximately 3m apart equipped with an underwater camera (Canon S100) and a surface
GPS (Garmin GPSmap 76CS x) attached to a weighted depth gauge (UWATEC Aladin
Tec 2G, calibrated to 0.1m). The GPS location of each benthic photo and the depth
bathymetry along each transect was recorded every 5 seconds. The camera and dive
computer clocks were synchronized to the GPS clock to geo-reference benthic photos and
depth profiles for each transect. Geo-referencing of each photo was conducted using
RoboGEO software (Pretek Inc.).
2
Data from the depth profiles and benthic photographs were used to generate
bathymetry and benthic community composition maps of the study site using Matlab. The
bathymetry map was used to calculate the average water depth for Lagrangian transects,
and the benthic community composition map was used to assess the community
composition in the study site.
S1.2.1. Bathymetry Mapping
A total of 1691 depth measurements were recorded within the study site. The
average horizontal spacing ranged from 1 to 2 (m), and we identified 47 individual
transects identified by points separated by more than 10 m. To account for differences in
tidal height between start and end times of transects (and between transects conducted on
separate days), depths were ‘de-tided’ by correcting for the change in tidal height
between the start and end of each transect. De-tided depth values greater than zero are
areas submerged at the lowest tide, and values less than zero are exposed at low tide (i.e.,
drying heights) (Figure S1). For each transect, we interpolated the de-tided values onto a
regularly spaced grid. For transects with more than 6 points, we applied a low-pass
Butterworth filter with an order from 2 to a maximum 8, depending on the number of
points, and a cut-off wavelength of 30 m. In order to determine a measure of the local
bathymetric non-uniformities in the study site, the low-pass filtered depths, hfilt,detided,
were subtracted from the unfiltered depths, hdetided. Then, the total cross-shore depth
variance is σh2 (standard deviation σh) e.g.[Feddersen and Guza, 2003] and
h
2


N  47
n 1

L
x 0
(hfilt,detided ( x)  hdetided ( x)) 2 dx

N  47
n 1
,
L
3
where x is the cross-shore distance from the first point to L, the transect length. From all
transects, the cross-shore depth standard deviation is σh = 0.09 m, which is a measure of
the bathymetric non-uniformities.
Next, we applied an interpolation of the irregularly scattered hfilt,detided data to a
finer grid, using a Delaunay triangulation with a natural neighbour method. We used this
interpolated grid to determine the mean depth for each Lagrangian drift (according to the
start and end coordinates for each drift) by calculating the average of 50 evenly spaced
points along each individual transect. We then calculated the mean tidal height
contribution (htide) to the water depth for each Lagrangian drift using WXTides32. The
total mean water depth is then
htotal  hfilt,detided  htide
Over each time period we determined the height variance associated with the varying
tidal height and defined it as σt2 (standard deviation is σt). From the 32 traverses, the
mean standard deviation owing to tidal height variations is σt = 0.05 m. The total
standard deviation for each drift is the square of the sum of the variances:
 total   h2   t2
or 0.1 m. Thus, htotal + σtotal describes the mean water depth over each traverse, when the
chemical water sampling was performed, and the uncertainty associated with the spread
in depth values owing to tidal variations and bathymetry irregularities.
S1.2.2. Benthic Community Structure
A total of 890 benthic photographs were analysed using Coral Point Count
software with Excel extensions (CPCe) with 20 random points per photograph. The
benthos was assigned to one of the following categories: (1) live coral; (2) macroalgae;
4
(3) turf algae; (4) crustose coralline algae (CCA); (5) sand; (6) cyanobacteria; (7) dead
coral and/or rubble; (8) zooanthids; (9) and “other” (including invertebrates, sponges,
etc.). Where morphologic forms of CaCO3 (e.g., rubble, CaCO3 rock) were covered with
biologically active groups (e.g., turf, coralline algae, cyanobacteria), the biologically
active group was scored.
S1.3. Lagrangian Sampling
The net rate at which the community altered the chemistry of the overlying water column
was estimated from AT and CT measurements, taking into consideration transit time and
water depth (Equations 1, 2). Lagrangian transport was measured by following water
parcels using a kayak and/or small boat (dependent on water depth) and a hand-held GPS
to record the location of the parcel as it traversed the reef flat (according to Albright et al.
[2013]). (All GPS measurements have an accuracy of 2 to 5 m.) Water parcels were
identified using fluorescein dye during the day and drifters (40 cm tall and 50 cm wide) at
night. Five transects were conducted using both drifters and dye during the day to allow
for comparisons between the two techniques. Because drifters often overestimate currents
at high wind speeds [Albright et al., 2013; Falter et al., 2008], current speeds for
transects that were conducted using a drifter (n=13, 8 night-time transects and 5 daytime
transects) were corrected by an average of 18% using the relationship defined in
[Albright et al., 2013]. Discrete surface samples were taken in duplicate at the beginning,
middle, and end of each Lagrangian transect. Samples were taken in 250 ml borosilicate
bottles and immediately poisoned with 125 µl HgCl2 (0.05% by volume) to inhibit
biological activity. Average water depth for each transect was calculated using the
bathymetric map via the transect location (start and end coordinates) and a tidal signal
5
corresponding to the transect time and duration (refer to Supplemental Information for
details). Discrete depth measurements (N=13) were taken for comparison with calculated
depths; for discrete measurements, water depth was measured using a hand-held depth
sounder alongside of the dye patch/drifter at regular intervals along each transect
(typically five measurements per transect) and averaged. Discrete measurements and
calculated depths were highly correlated (R2=0.91).
S1.4. Chemical Analyses
Water samples were analysed for total alkalinity (AT) and total dissolved inorganic
carbon (CT) using a VINDTA 3C® (Versatile INstrument for the Determination of Total
dissolved inorganic carbon and Alkalinity, Marianda, Kiel, Germany) and a UIC CO2
coulometer (UIC Inc., Joliet, USA). Instrumental precision from 97 certified reference
material (CRM) analyses over the course of the study was 2.6 µmol kg-1 (1SD) for AT and
1.5 µmol kg-1 (1SD) for CT. Immediate duplicate analyses of samples usually yielded
instrumental precision of < 1 µmol kg-1 for AT and ~1 µmol kg-1 for CT. Measurements
and calculations were performed according to Albright et al. [2013] and were consistent
with “best practices” recommendations [Riebesell et al., 2010].
S1.5. Physical conditions
Temperature and depth were measured continuously at the crest and lagoon water
sampler sites using RBR tide gauges (Model XR-420 TG). Discrete water samples were
taken during each Lagrangian drift and measured for salinity using a Guildline Portasal
Salinometer (Model 8410A). Wind data were obtained from the AIMS Weather Station at
Heron Reef (http://data.aims.gov.au/aimsrtds/station.xhtml?station=130). Light was
continuously measured at the study site using an Odyssey® photosynthetically active
6
radiation (PAR) sensor that was calibrated to PAR data from the AIMS Weather Station
at One Tree Island (Figure S3), approximately 19 km southeast of Heron Island. (The
AIMS Weather Station at Heron Island did not have a functioning sensor during the time
of our study).
Noontime irradiance ranged from 306 to 2248 μmol m−2 s−1, averaging 1150±545
μmol m−2 s−1. Winds were predominantly from the east/southeast and averaged 8 ± 3 m
s−1 (mean±SD), ranging from 0 to 20 m s−1. The tidal height ranged from < 0.1 to 2.8 m.
S1.6. Calculations
Changes in seawater carbon chemistry between upstream and downstream sampling
points can be used to calculate rates of net community calcification and net community
production. In most coral reef systems, changes in total alkalinity (AT) are caused
primarily by calcification and dissolution of CaCO3 whereby two moles of AT are
consumed (produced) for every mole of CaCO3 produced (dissolved) (e.g., [Kleypas and
Langdon, 2006]). Accordingly, net community calcification, ncc, (mmol CaCO3 m-2 h-1)
can be calculated as:
ncc  0.5    h 
AT
t
(1)
where ∆AT is the change in total alkalinity between the upstream and downstream
locations (mmol kg-1), ρ is the seawater density (kg m-3), h is the water depth (m), and ∆t
is the duration of the transect (h).
CT is affected by calcification, dissolution, photosynthesis, and respiration. Net
community production, ncp, (mmol C m-2 h-1) was calculated using changes in CT after
taking into account ncc and gas exchange:
7
ncp  h   
(CT  0.5  AT )
 k  s  ( pCO2 water  pCO2 air )
t
(2)
where ΔCT is the change in dissolved inorganic carbon between the upstream and
downstream locations (mmol kg-1), and the term [ k  s  ( pCO2 water  pCO2 air ) ]
approximates gas exchange where k is the gas transfer velocity, s is the solubility of CO2
calculated as a function of salinity and temperature, and ( pCO2 water  pCO2 air ) is the
difference in pCO2 between the surface ocean and the atmosphere. All other parameters
are defined as above. The wind speed parameterization of Ho et al. (2006) was used to
calculate k, which ranged 0.9 to 28.7 cm h-1, with a mean of 11.9 cm h-1. Atmospheric
pCO2 data were obtained from the PMEL CO2 buoy in Wistari Channel
(www.pmel.noaa.gov/co2/story/Heron+Island) and averaged 387 ppm. Resulting gas
exchange estimates were small in comparison to ncp, ranging from -0.02 to 0.03 mmol C
m-2 h-1. Errors for ncc and ncp measurements (provided in Supplemental Data Set S1)
were calculated by propagating standard deviations from depth (0.1 m, see S1.2.1.
Bathymetry Mapping), AT (<1 µmol kg-1 for immediate duplicate analysis), and CT
measurements (~1 µmol kg-1 for immediate duplicate analysis - see S1.4. Chemical
Analyses).
8
Figure Legends
Figure S1. Schematic of methodology used to create the bathymetry map of the Heron
Reef flat. Measured depths were ‘de-tided’ by subtracting the tidal height at the time of
each survey (and correcting for differences in tidal height between the beginning and end
of each survey). These ‘de-tided’ depths were used to create a bathymetric map using
interpolation and mapping functions in Matlab. De-tided depths that are > 0 represent
areas of the reef flat that are submerged by water at lowest astronomical tide or LAT (i.e.,
‘charted depths’), whereas de-tided depths < 0 are areas that are exposed to air (i.e.,
drying heights) at LAT. The resulting map was used to determine water depths for each
of the Lagrangian drift by querying the map for the de-tided depth corresponding to the
mean depth between the start and end coordinates of a transect, and superimposing a tidal
component corresponding to the day and time the drift was conducted. See text below for
details.
Figure S2. Temporal variability in pCO2 of the atmosphere, and the seawater of the
Wistari Channel and the Heron Reef flat over the course of the study. Atmospheric pCO2
and pCO2 of the seawater in the Wistari Channel were measured by the PMEL buoy
(www.pmel.noaa.gov/co2/story/Heron+Island). Mean surface water pCO2 (452 ppm) was
elevated in reef waters relative to the atmosphere (386 ppm) and the open ocean/Wistari
Channel (399 ppm).Reef flat chemistry showed a higher degree of variability than
Wistari Channel, demonstrating the capacity of reef metabolism to alter open surface
water conditions. Gaps in the reef flat data correspond to periods when the instruments
were removed from the reef for maintenance.
Figure S3. (a) Relationship between raw (uncalibrated) photsynthetically active radiation
(PAR) data from the Heron study site and PAR data from the One Tree Island weather
station. This relationship was used to calibrate the Heron PAR sensor (b). One Tree
Island light data were sourced from the Integrated Marine Observing System (IMOS).
IMOS is supported by the Australian Government through the National Collaborative
Research Infrastructure Strategy and the Super Science Initiative.
Figure S4. Linear relationships between (a) ncc and light (photosynthetically active
radiation, PAR) and (b) ncc and temperature. The weak correlation between ncc and
temperature is likely due to the small range of temperatures over which ncc data were
collected (25.9-27.6°C). Open circles indicate daytime measurements, and closed circles
indicate nighttime measurements.
9
Parameter (units)
Salinity*
Temperature (°C)*
AT (µmol kg-1)*
CT (µmol kg-1)*
pHtotal
pCO2 (µatm)
HCO3- (µmol kg-1)
CO32- (µmol kg-1)
CO2 (µmol kg-1)
Ωcalc
Ωarag
Combined
(n=158)
35.4 ± 0.005
35.2-35.5
26.6 ± 0.06
24.2-30.4
2258 ± 2
2176-2307
1969 ± 4
1834-2059
7.99 ± 0.004
7.83-8.14
449 ± 6
281-669
1752 ± 5
1569-1872
205 ± 2
144-261
12 ± 0.2
8-18
4.9 ± 0.04
3.5-6.3
3.3 ± 0.03
2.3-4.2
Crest
(n=79)
35.4 ± 0.007
35.2-35.5
26.7 ± 0.08
25.1-30.0
2258 ± 3
2197-2307
1973 ± 5
1837-2046
7.98 ± 0.006
7.83-8.14
457 ± 8
281-669
1759 ± 7
1569-1869
202 ± 2
144-261
12 ± 0.2
8-18
4.9 ± 0.06
3.5-6.3
3.2 ± 0.04
2.3-4.2
Lagoon
(n=79)
35.4 ± 0.007
35.2-35.5
26.5 ± 0.09
24.2-30.4
2257 ± 3
2176-2304
1965 ± 6
1834-2059
7.99 ± 0.006
7.88-8.11
441 ± 8
299-601
1746 ± 8
1571-1872
207 ± 3
164-256
12 ± 0.2
8-16
5.0 ± 0.07
3.9-6.2
3.3 ± 0.04
2.6-4.1
Table S1. Chemical and physical conditions of the Heron Island Reef flat (mean ± SE,
range). Asterisks denote measured parameters; all other parameters are derived from CT,
AT, salinity and temperature using CO2SYS. Calculations were performed according to
Albright et al. [2013].
10
ncp
ncp + Ωarag
ncp + Ωarag + PAR
ncp + Ωarag + PAR + Temp
Ωarag
AICc
165.55
166.06
168.45
168.89
172.16
Table S2. Corrected Akaike Information Criteria (AICc) values for the relationship
between net community calcification (ncc) and various drivers (ranked from lowest to
highest). AICc values are for linear models generated in R and fit using generalized least
squares by maximizing the log-likelihood.
11
Data Set S1. Data from Lagrangian transects, used to calculate ncc and ncp
Data Set S2. Data from water samplers, used to evaluate diel variability in carbonate chemistry
12
References
Albright, R., C. Langdon, and K. R. N. Anthony (2013), Dynamics of seawater carbonate
chemistry, production, and calcification of a coral reef flat, central Great Barrier Reef,
Biogeosciences, 10(10), 6747-6758, doi: 10.5194/bg-10-6747-2013.
Anthony, K. R. N., G. Diaz-Pulido, N. Verlinden, B. Tilbrook, and A. J. Andersson
(2013), Benthic buffers and boosters of ocean acidification on coral reefs,
Biogeosciences, 10(7), 4897-4909, doi: DOI 10.5194/bg-10-4897-2013.
Falter, J. L., R. J. Lowe, M. J. Atkinson, S. G. Monismith, and D. W. Schar (2008),
Continuous measurements of net production over a shallow reef community using a
modified Eulerian approach, Journal of Geophysical Research-Oceans, 113(C7), doi:
10.1029/2007jc004663.
Feddersen, F., and R. T. Guza (2003), Observations of nearshore circulation: Alongshore
uniformity J. Geophys. Res., 108(C1), 3006, doi: doi:10.1029/2001JC001293.
Kleypas, J. A., and C. Langdon (2006), Coral Reefs and Changing Seawater Carbonate
ChemistryRep.
Riebesell, U., V. J. Fabry, L. Hansson, and J. P. Gattuso (2010), Guide to best practices
for ocean acidification research and data reporting.
13
Download