water quality and ecosystem monitoring programme reef ...€¦ · thompson a, schaffelke b,...

84
Water Quality and Ecosystem Monitoring Programme Reef Water Quality Protection Plan Synthesis and spatial analysis of inshore monitoring data 2005-08 Angus Thompson Britta Schaffelke Glenn De’ath Edward Cripps Hugh Sweatman Townsville January 2010

Upload: others

Post on 17-Oct-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

Water Quality and Ecosystem Monitoring Programme

Reef Water Quality Protection Plan

Synthesis and spatial analysis of inshore monitoring data 2005-08

Angus Thompson Britta Schaffelke

Glenn De’ath Edward Cripps

Hugh Sweatman

Townsville January 2010

Page 2: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

Australian Institute of Marine Science PMB No 3 Townsville Qld 4810. Report history 07 January 2009 Draft Report submitted to GBRMPA 16 June 2009 Peer review comments received by AIMS 10 July 2009 Revised report submitted to GBRMPA 21 July 2009 Further comments received by AIMS 08 January 2010 Second revised version submitted to GBRMPA Disclaimer This report has been produced for the sole use of the party who requested it. The application or use of this report and of any data or information (including results of experiments, conclusions, and recommendations) contained within it shall be at the sole risk and responsibility of that party. AIMS does not provide any warranty or assurance as to the accuracy or suitability of the whole or any part of the report, for any particular purpose or application. Subject only to any contrary non-excludable statutory obligations neither AIMS nor its personnel will be responsible to the party requesting the report, or any other person claiming through that party, for any consequences of its use or application (whether in whole or part).

This report should be cited as Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme - Reef Water Quality Protection Plan: Synthesis and spatial analysis of inshore monitoring data 2005-08. Report to the Great Barrier Reef Marine Park Authority. Australian Institute of Marine Science, Townsville. 78 pp.

Title page photographs: AIMS From left to right: Coral assemblage at Dunk Island, diver hammering in starpicket to mark the start of a survey transect, coral assemblage at Fitzroy Island.

Page 3: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

i

Contents

Executive Summary ............................................................................................................................................. 1 Introduction to this Report .............................................................................................................................. 5 1. Spatial patterns in inshore coral reef assemblages and their relationship to environmental variables ..................................................................................................................................... 6 1.1 INTRODUCTION ............................................................................................................. 6 1.2 METHODS .................................................................................................................... 7

Sampling locations ........................................................................................................................................ 7 Environmental quality .................................................................................................................................. 8 Coral reef benthos ....................................................................................................................................... 9 Data analysis .................................................................................................................................................. 9

Univariate Analysis ............................................................................................................................... 10 Multivariate Analysis ............................................................................................................................ 11

1.3 RESULTS AND DISCUSSION ......................................................................................... 13 1.3.1 Selection of environmental variables for analysis ................................................................... 13

Spatial patterns in water quality variables ...................................................................................... 13 Sediment quality variables .................................................................................................................. 16 Final selection of explanatory variables ........................................................................................... 18 Do the selected environmental variables vary among the regions? ......................................... 18

1.3.2 Do the selected environmental variables influence the benthic assemblages on inshore coral reefs? ................................................................................................................................... 20

Hard coral assemblages ....................................................................................................................... 21 Soft coral assemblages ......................................................................................................................... 31 Macroalgae ............................................................................................................................................. 38

2. Assessment of the utility of autonomous logging instruments for water quality monitoring ... 44 2.1 INTRODUCTION ........................................................................................................... 44 2.2 METHODS .................................................................................................................. 45

Water quality instrument deployments ............................................................................................... 45 Direct water sampling for instrument validation ............................................................................... 47 Laboratory comparison of instruments ............................................................................................... 47 Data analysis ............................................................................................................................................... 48

2.3 RESULTS AND DISCUSSION ......................................................................................... 49 Direct water sampling .............................................................................................................................. 49 Instrumental water sampling .................................................................................................................. 49 Comparison of chlorophyll values from direct water sampling and instrumental sampling .... 54 Comparison between instruments ....................................................................................................... 58

3. Conclusions ................................................................................................................................................... 61 4. References ..................................................................................................................................................... 64 Appendix 1: Environmental variables .......................................................................................................... 68 Appendix 2: Univariate analyses of coral reef benthos ........................................................................... 72 Appendix 3: Redundancy analyses of coral reef benthos ....................................................................... 75 Appendix 4: Model checking of analyses ................................................................................................... 77

Page 4: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

ii

List of Figures and Tables

Figure 1.1 Twenty-four sampling locations considered in this report that were sampled under the Reef Plan MMP inshore marine water quality and coral monitoring tasks.. ...................... 7 Figure 1.2 Principal Components Analysis (PCA) biplot of water quality variables. ....................... 14 Figure 1.3 Principal Components Analysis (PCA) biplot of the particulate water quality variables. ................................................................................................................................................ 15 Figure 1.4 Principal Components Analysis (PCA) biplot of the dissolved water quality variables ................................................................................................................................................. 16 Figure 1.5 Principal Components Analysis (PCA) biplot of the nutrient sediment variables. ..... 17 Figure 1.6 Principal Components Analysis (PCA) Biplot of the grainsize variables. ...................... 17 Figure 1.7 Regional summaries of environmental variables at the 24 study reefs.. ......................... 19 Figure 1.8 Term plots for the factor region from generalised linear models for each hard coral assemblage attributes. ..................................................................................................................................... 22 Figure 1.9 Redundancy analysis (RDA) biplots for cover (left) and juvenile density (right) of hard coral genera. ....................................................................................................................................... 23 Figure 1.10 Representative coral assemblage types in the Fitzroy Region. ....................................... 24 Figure 1.12 Multivariate regression trees for cover of hard coral genera (a), and density of hard coral genera juveniles (b) ...................................................................................................................... 27 Figure 1.13 Redundancy analysis bi-plot of the relationship between cover (a) and juvenile density (b) of hard coral genera and environmental variables ............................................... 29 Figure 1.14 Term plots for the factor Region from generalised linear models for soft juvenile density (a) and richness of genera (b). ......................................................................................... 32 Figure 1.15 Redundancy analysis biplots for the cover (left), and density of juvenile (right) soft coral genera ................................................................................................................... 33 Figure 1.16 Term plots of significant relationships between soft coral assemblage attributes and environmental variables as indicated by generalised linear models .............................................. 34 Figure 1.17 Multivariate regression trees for adult cover of soft coral genera (a) and density of soft coral genera juveniles (b). ................................................................................................... 35 Figure 1.18 Redundancy analysis bi-plot of the relationship between cover (a) and juvenile densities (b) of soft corals and environmental variables .......................................................... 37 Figure 1.19 Term plots for the factor region from generalised linear models for macroalgal cover (a) and richness of genera (b) ....................................................................................... 38 Figure 1.20 Redundancy analysis biplot for the cover of macroalgal genera .................................... 40 Figure 1.21 Term plots of significant relationships between macroalgae cover and environmental variables as indicated by generalised linear models. ..................................................... 41 Figure 1.22 Multivariate regression tree for cover of macroalgal genera ......................................... 42 Figure 1.23 Redundancy analysis bi-plot of the relationship between the cover of macroalgae genera and environmental variables ...................................................................................... 43 Figure 2.1 FLNTUSB logger deployed at Pelican Island in the Fitzroy NRM Region ...................... 45

Page 5: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

iii

Figure 2.2 Fourteen sampling locations (red dots) where Eco FLNTUSB Combination instruments were deployed and regular direct water sampling was undertaken. ............................ 46 Figure 2.3 Calibration chamber set up used for inter-instrument comparisons of Eco FLNTUSB water quality loggers ....................................................................................................................................... 47 Figure 2.4 Example of a time series of chlorophyll and turbidity from field deployments of WET Labs Eco FLNTU Combination Fluorometer and Turbidity Sensors ....................................... 49 Figure 2.5 Randomly chosen daily profiles (n = 18) of 10-minute spaced measurements by WET Labs Eco FLNTUSB Combination Fluorometer and Turbidity Sensors. ................................. 50 Figure 2.6 Continuous time series plots of chlorophyll concentrations showing highly variable patterns of long and short-term variation across 10 locations. ............................................................ 51 Figure 2.6 continued Continuous time series plots of chlorophyll concentrations showing highly variable patterns of long and short-term variation across 4 locations. ................................... 52 Figure 2.7 Boxplots of means and SDs of randomly chosen daily profiles (n = 24) of 10-minute spaced samples (144 per day) show considerable variation. ............................................ 53 Figure 2.8 Chlorophyll a levels from direct water sampling and Eco FLNTUSB Combination Fluorometer and Turbidity Sensors from 11 locations across varying periods of time. ............... 55 Figure 2.9 Comparison of chlorophyll concentrations obtained from direct water sampling (red points) and Eco FLNTUSB instruments (blue points) from 11 locations and 1-4 times ................. 56 Figure 2.10 Comparison of chlorophyll concentrations obtained from direct water sampling (red points) and Eco FLNTUSB instruments (blue points) from 11 locations and 4-5 times. ................ 57 Figure 2.11 Relationships between the chlorophyll values from eight Eco FLNTUSB instruments and the phytoplankton culture dose added. ....................................................................... 59

Figure A4-1 Model checking for analysis of directly sampled water quality data. ........................... 77 Figure A4-2 Model checking for analysis of inter-instrument comparison data .............................. 78

Table 1.1 Monitoring locations selected for analyses of relationships between environmental quality and coral reef characteristics. ............................................................................................................ 8 Table 1.2 Results of Analysis of variance (ANOVA) comparing values of the three selected environmental response variables among NRM Regions........................................................................ 18 Table 1.3 Proportional difference in assemblage attributes on reefs in the Wet Tropics, Burdekin and Mackay regions relative to values for he Fitzroy region.. ............................................. 21 Table 1.4 Proportional difference (%) in soft coral juvenile richness on reefs in the Fitzroy Region compared to each of the other three regions. ............................................................. 31 Table 1.5 Proportional difference of macroalgae cover and richness on reefs in the Mackay Region compared to each other three regions. ....................................................................... 39 Table 2.1 Locations selected for inshore water quality monitoring by autonomous instruments (Wetlabs FLNTUSB) and deployment and change-over times. ..................................... 46 Table 2.2 Analysis of deviance for the effects of locations and days nested in locations for chlorophyll shows strong variation due to both locations and days.................................................... 49

Page 6: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

iv

Table 2.3 Analysis of deviance for the effects of locations and sampling method (direct water sampling vs. instrument) for chlorophyll a. ....................................................................... 54 Table 2.4 Analysis of deviance for the effects of locations, days nested in locations and sampling method (direct water sampling vs. instrument) for chlorophyll a. ...................................... 54

Table A1-1 Summary data of water quality at the 24 study reefs. ...................................................... 68 Table A2-1 Results from generalised linear model applied to total hard coral cover.. ................ 72 Table A2-2 Results from generalised linear model applied to total hard coral richness.. ............ 72 Table A2-3 Results from generalised linear model applied to total hard coral juvenile density. ................................................................................................................................................................ 72 Table A2-4 Results from generalised linear model applied to total hard coral juvenile genus richness. .................................................................................................................................................. 72 Table A2-5 Results from generalised linear model applied to total soft coral cover.. .................. 72 Table A2-6 Results from generalised linear model applied to total soft coral richness.. ............. 73 Table A2-7 Results from generalised linear model applied to total soft coral juvenile density. ................................................................................................................................................................ 73 Table A2-8 Results from generalised linear model applied to total soft coral juvenile genus richness. .................................................................................................................................................. 73 Table A2-9 Results from generalised linear model applied to total hard macroalgal cover.. ...... 74 Table A2-10 Results from generalised linear model applied to total hard macroalgal richness.. 74 Table A2-11 Results from sequential analysis of deviance. ................................................................... 74 Table A3-1 Results from redundancy analysis of genus-level hard coral cover data. ..................... 75 Table A3-2 Results from redundancy analysis of genus-level hard coral juvenile density data. ... 75

Page 7: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

1

Executive Summary

This report is the output from an additional project under the umbrella of the Reef Plan Marine Monitoring Program (MMP) to provide comprehensive analyses of i) the relationships between the coral reef benthos and the available environmental data (water quality, sediment quality) on 24 survey reefs from three years (2005-2008) of monitoring and ii) assess the utility of using autonomous logging instruments for water quality monitoring.

The survey reefs are located in four regions of the Great Barrier Reef (GBR) and span about seven degrees of latitude, with those in the southern Fitzroy region approaching the southern extent of inshore reefs on the GBR. The ecological attributes included in the analyses represented the abundance of the key benthic organisms (measured by cover of adult corals and algae and density of juvenile corals) and their biodiversity (measured as taxonomic richness).

Our analysis showed clear differences in the coral reef assemblages along latitudinal and environmental gradients. All assemblages in our study were influenced most strongly by the levels of the ‘particulate’ water quality index (particulate nitrogen, phosphorus and organic carbon; dissolved organic carbon; suspended solids; chlorophyll and Secchi depth). Tolerance to low light is related to the capacity of some hard coral taxa to compensate for reductions in photosynthesis in low light conditions by feeding on suspended particles. Phototrophic soft corals are not known to switch between feeding modes and become less abundant in turbid conditions while heterotrophic soft corals were rare on our study reefs. While macroalgae are light-dependent, some can utilise nutrients from particulate organic matter that settle on their thalli.

We found clear evidence that juvenile hard and soft coral assemblages are more sensitive to high levels of the ‘particulate’ water quality index. It has long been assumed that water quality is most likely to shape reef assemblages through affecting coral reproduction and recruitment. Because adult corals can tolerate poorer water quality than recruits and colonies are potentially long-lived, reefs may retain high coral cover even under conditions of declining water quality, but have low resilience.

It has to be acknowledged that the water quality conditions in these analyses are representing the results from a limited number of sampling years at low sampling frequency (two to three times a year for logistical and cost reasons) and may not be typical for the long-term conditions. However, the MMP inshore water quality monitoring has much improved with the deployment of water quality instruments at 14 reef locations since October 2007. High frequency time series of chlorophyll concentrations and turbidity levels are now becoming available and will be an invaluable data source in the near future.

The differences we detected in the GBR inshore reef assemblages provide a useful starting point for the detection of long-term trends in coral reef benthos in future monitoring under the MMP. Our results indicate that the particulate components of water quality (suspended

Page 8: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

2

sediment and particulate nutrients and carbon) are the most important drivers of coral reef communities. Should changes in land management practices in the GBR catchments under the Reef Plan lead to decreased levels of particulates in coastal and inshore waters of the GBR, we expect to be able to detect associated changes in coral reef assemblages, based on the relationships between assemblage composition and water and sediment quality described in this report.

Selection of variables for the analysis of relationships with coral reef organisms

Median values of particulate nitrogen, phosphate, dissolved and particulate organic carbon increased towards the south. Particulate phosphorus and chlorophyll values were very similar among regions. The Burdekin region had the highest values of suspended solids and, correspondingly, the lowest values of Secchi depth. The water quality variables fell into two distinct, highly correlated groups. The first group included variables related to particulate water quality. The second group was made up of all analysed dissolved water quality variables. Indices calculated from each of these groups were selected as environmental variable to analyse their influence on benthic assemblages. There were no obvious differences in values of the particulate and dissolved water quality indices among the NRM regions.

The proportion of sediments with grainsizes below 63 µm was selected as the sediment quality variable for analyses. These fine clay or silt sediments are considered to have the greatest effect on benthic coral reef assemblages and are also likely to have increased in supply to the GBR since European settlement. Reefs in the Mackay Whitsunday Region had higher values of ‘clay silt’ than those in the other three regions.

Spatial patterns in coral reef benthos in relation to water and sediment quality

Hard corals

Hard coral cover on reefs in the Fitzroy Region was significantly higher than on reefs in the Mackay Whitsunday, Burdekin or Wet Tropics regions, while the density of juvenile colonies and the richness of genera of both the adults and the juveniles were lower on reefs in the Fitzroy region. The coral assemblages on three of the four reefs in the Fitzroy region were dominated by large stands of branching Acropora, resulting in high coral cover (77- 99%) but low taxonomic richness. These regional differences were largely independent of environmental variables and reflect a general decline in richness of corals toward the southern end of the GBR.

Hard coral richness was typically higher at 5m than at 2m water depth, while total adult coral cover and the density of juveniles were not different. Assemblage composition

Page 9: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

3

differed between depths with e.g., higher cover of the fast-growing genera Acropora and Pocillopora at 2m, which is likely to be a consequence of steep light gradients on the inshore reefs caused by high turbidity.

Increasing levels of the ‘particulate’ water quality index resulted in lower total coral cover and richness of juvenile coral genera. Some hard coral genera (e.g. Turbinaria and Goniastrea) showed higher cover at higher levels of the ‘particulate’ water quality index, while cover of genera including Pocillopora, Porites and Acropora was higher where levels of particulates were lower and water was clearer. Substantially more genera of juvenile corals were found in less turbid environments, which suggests that these early life stages are more sensitive to high turbidity than established colonies.

Soft corals

Total cover and richness of adult soft corals did not differ between regions. The density of juvenile colonies was higher and the taxonomic composition of the adult and juvenile soft coral assemblages was different on reefs in the Mackay Whitsunday region, and correlated with high ‘clay/silt’ content of the sediments in this region. The richness of juvenile soft corals was much lower on reefs in the Fitzroy region, likely due to the southern location of these reefs.

The cover and richness of adult and the density of juvenile soft coral colonies all declined with increasing levels of the ‘particulate’ water quality index. The ‘dissolved’ water quality index showed no significant relationships to the soft coral assemblages in any of the models. The density of juvenile colonies showed similar relationships to the environmental variables, suggesting that cover and representation of adult genera were at least in part dependent on recruitment patterns.

The total cover, richness, juvenile density or juvenile richness of soft coral assemblages did not differ between 2m and 5m depth. Only the cover of adult ‘gorgonian’ colonies was consistently higher at 5m; this group includes a number of heterotrophic genera that are not affected by decreasing light levels with depth in turbid water.

Macroalgae

The taxonomic composition of macroalgal assemblages was significantly different between the four regions, which has not been described before and its significance is unknown. The current assemblage types could influence whether and what type of macroalgal-dominated assemblage might develop after a disturbance causing coral mortality.

Wet Tropics and Mackay-Whitsunday reefs had the lowest cover of macroalgae and macroalgal cover and richness were significantly lower in the Mackay Whitsunday

Page 10: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

4

region. This was correlated with a significantly higher concentration of particulates in the water column in this region.

The cover and richness of macroalgae initially increased with increasing levels of the ‘particulate’ water quality index, but decreased at high levels. Cover of the genera Sargassum and Dictyota were greater at higher levels of the ‘particulate’ water quality index. These genera are very abundant on inshore reefs of the GBR and are likely to be the main drivers of the overall relationship between total macroalgal cover and the ‘particulate’ water quality index. Reefs with relatively low levels of the ‘particulate index’ also had low macroalgal cover (<4%).

The composition of the macroalgal assemblages did not differ between 2 and 5 m depths; the assemblages at 5m depth have low cover, which is likely due to light limitation, but the same suites of genera are present as in the shallow water assemblages.

Assessment of the utility of logging instruments for water quality monitoring

Automated instruments (Eco FLNTUSB) that measure and record chlorophyll a, turbidity and temperature data every 10 minutes were installed at 14 MMP survey reefs in October 2007. Comparison of instrument data with results from water samples collected by divers at each logger sites showed that the instruments record reliable chlorophyll data.

The frequent sampling (readings every10 minutes) revealed that chlorophyll and turbidity values varied considerably over each day, which is likely to be associated with tidal forcing and phytoplankton dynamics. A better understanding of the factors controlling chlorophyll a concentrations warrants future research and analyses.

It is concluded that (not surprisingly), instrument data provide a more accurate picture of average chlorophyll a concentrations at a deployment site that a single grab water sample.

It is recommended that manual water sampling is replaced by routine use of loggers (and remote sensed data), wherever possible, with continued regular manual collection of water samples to validate logged and remotely sensed data.

Page 11: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

5

Introduction to this Report

Since 2005, the Australian Institute of Marine Science (AIMS) has provided monitoring information to the Great Barrier Reef Marine Park Authority (GBRMPA) as part of the Reef Water Quality Protection Plan Marine Monitoring Programme (Reef Plan MMP). AIMS’ activities under the Reef Plan MMP of comprise two components: • Inshore marine water quality monitoring • Inshore coral reef monitoring Results of this monitoring have been reported annually to the GBRMPA (CRC Consortium 2006, Schaffelke et al. 2007, 2008). These reports were largely descriptive, with limited analysis of spatial and temporal patterns of water quality and spatial patterns in the structure of coral reef assemblages. The 2008 report also contained preliminary analyses of the relationships between some environmental parameters (water and sediment quality) and coral reef benthos. Here we report on an additional project under the Reef Plan MMP umbrella to provide comprehensive analyses of the relationships between the coral reef benthos and the available environmental data (water quality, sediment quality) from the three years of monitoring to date. This is a major step towards answering the question how reef status is related to water quality, the key question under Reef Plan. The here developed statistical models will form the basis for future analysis of Reef Plan MMP water quality and coral reef assemblage data. The project had two main objectives • Analysis of patterns in inshore coral reef assemblages and their relationship to environmental

variables and use of these relationships to describe reef status; • Assessment of the utility of time-series data of water quality parameters (turbidity, chlorophyll,

temperature from automatic loggers. This part of the programme commenced in October 2007 and 9-12 months of data will be available for the analysis) in comparison to the traditional spot sampling that was undertaken so far.

The following report has two main chapters relating to these two objectives.

Page 12: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

6

1. Spatial patterns in inshore coral reef assemblages and their relationship to

environmental variables

1.1 Introduction

Extensive water sampling throughout the Great Barrier Reef (GBR) lagoon over the last 25 years has shown that the typical concentration of nutrients, chlorophyll a and other water quality parameters vary across the GBR province. There are persistent latitudinal, cross-shelf and seasonal variations in these concentrations (summarised in Furnas 2005, Brodie et al. 2007, De’ath and Fabricius 2008). While much of the variation in water quality is natural, land use on the adjacent river catchments has increased the levels of sediment and nutrients entering the coastal areas of the GBR. To address this, the Reef Water Quality Protection Plan (Reef Plan) was announced in October 2003. The Plan was developed jointly by the Commonwealth and Queensland governments to “halt or reverse the decline in water quality entering the reef”. Most of the Reef Plan’s activities are land-based, aiming to improve land use practices and so reduce the amount of nutrients and sediment entering river systems that flow into the GBRWHA. As an integral part of Reef Plan, the Great Barrier Reef Marine Park Authority established the Reef Plan Marine Monitoring Programme (Reef Plan MMP) in 2005. This programme has measured the condition of inshore water quality and of key biological assemblages in the vulnerable inshore region and has established a monitoring system to track trends in the amount of sediment, nutrients and other pollutants entering GBR waters and to provide data to measure change in coral reef assemblages that are related to environmental quality as the Reef Plan takes effect. Coral reefs in general are adversely affected by local and regional anthropogenic activities such as overfishing, pollution, sedimentation, and eutrophication (e.g. Bellwood et al. 2004), in addition to disturbances such as severe storms, outbreaks of the coral-eating crown-of-thorns seastar and diseases. On a global scale, climate change is currently the major threat to coral reefs by increasing thermal stress, ocean acidification, storm activity and disease outbreaks (Hughes et al. 2003, Hoegh-Guldberg et al. 2007). Water quality is an important driver of coral reef health at local (reviewed in Fabricius 2005), regional (van Woesik et al. 1999, Fabricius et al. 2005), and a GBR-wide scales (De’ath and Fabricius 2008). The effects of various water quality constituents on benthic organisms (particularly corals) are manifold ranging from disturbance by sedimentation, light reduction through increased turbidity, reduced calcification rates by excess inorganic nutrients, inhibition of photosynthesis by herbicide exposure and in most cases the impact is greater on early life history stages than on adult corals (e.g., Fabricius 2005, Negri et al. 2005, Cantin et al. 2007). Here we report on a comprehensive analysis of the relationships of a number of the characteristics of inshore coral reef benthos (benthic cover and composition) with concurrently collected environmental data (water quality, sediment quality, temperature) from three years of monitoring under the Reef Plan MMP. This report is a major step towards answering the question how reef status is related to water quality, the key question under Reef Plan.

Page 13: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

7

1.2 Methods

SAMPLING LOCATIONS

Analyses of the relationship between attributes of the coral reef benthos and several environmental variables were based on information from 24 locations in the inshore GBR (Figure 1.1, Table 1.1), spanning seven degrees of latitude. These sites had the most complete datasets of water quality, sediment and biological data.

Figure 1.1 Twenty-four sampling locations considered in this report that were sampled under the Reef Plan MMP inshore marine water quality and coral monitoring tasks. The reef locations had annual coral reef benthos surveys, water (biannual) and sediment (annual) quality sampling.

Page 14: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

8

ENVIRONMENTAL QUALITY

[Note: For more detailed methods refer to CRC Reef Consortium 2005 and Schaffelke et al. 2008] Water samples were collected from the 24 locations twice a year during the wet and dry seasons of 2005/06, 2006/07 and 2007/08. At each sampling occasion, water was collected at two to three depths through the water column with Niskin bottles. Sub-samples taken from the Niskin bottles were analysed for dissolved nutrients and carbon (NH4, NO2, NO3, PO4, Si(OH)4), TDN, TDP, DOC), particulate nutrients and carbon (PN, PP, POC), suspended solids (SS) and plant pigments (chlorophyll a). Sediment samples were collected from the 24 locations during 2006 and 2007. Pooled cores of the upper 1 cm of the sediment at each reef survey site at 5m depth were for analysed for grain size distribution and for content of inorganic carbon, organic carbon and total nitrogen.

Table 1.1 Monitoring locations selected for analyses of relationships between environmental quality and coral reef characteristics.

NRM Region Coral monitoring locations

Wet Tropics

Snapper Island North Fitzroy Island West High Island West King Reef Frankland Group West Frankland Group East North Barnard Group Dunk Island North

Burdekin

Pelorus and Orpheus Island West Pandora Reef Havannah Island Lady Elliot Reef Geoffrey Bay

Mackay

Double Cone Island Daydream Island Pine Island Hook Island Dent Island Seaforth Island Shute and Tancred Islands

Fitzroy

Barren Island Pelican Island Humpy & Halfway Island North Keppel Island

Page 15: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

9

CORAL REEF BENTHOS

[Note: For more detailed methods refer to CRC Reef Consortium 2005 and Schaffelke et al. 2008] Estimates of coral and algal cover were derived from the average of winter samples in 2006 and 2007 at the reefs listed in Table 1. Benthic assemblages were sampled along 120m of transects at both 2m and 5m below Lowest Astronomical Tide (LAT) in two sites at each location. Two sampling strategies are used to provide information on different life stages within the benthic assemblages. 1. Benthic cover: The proportion of the substrate occupied by benthic organisms was estimated from

digital photographic images taken at 50cm intervals along five 20m long transects laid on the 2m and 5m depth contours at each site at each reef. These sites had permanent markers. The benthic organisms lying beneath 5 evenly spaced points on each image were identified to the highest taxonomic level possible (governed by image quality). For the purpose of this report these data are aggregated to the cover of genera of hard corals, soft corals and the larger fleshy algae (macroalgae). As the proportion of points landing on any particular genus will be highly skewed toward larger individuals, estimates of cover (especially for corals) basically represent the adult life stages.

2. Counts of coral colonies <10cm in size: Along the same transects for which cover was estimated the number of hard and soft coral colonies that fell wholly or partially within a 34cm (data slate width) wide belt were recorded. These colonies were identified to the taxonomic level of genus in the field. An effort was made to exclude small colonies that were obviously the result of partial mortality of larger colonies so these counts can be assumed to represent the number of Juvenile corals. As the number of juvenile colonies may be limited by the space available for settlement, count data were corrected by dividing the counts for each genus by the proportion of substrate not occupied by either hard or soft corals (as estimated from photo transect samples). These corrected data represented the density of juveniles in the areas of each reef that were not occupied by adult coral colonies.

DATA ANALYSIS

Values for water quality parameters at each station were calculated as depth-weighted means. Due to changes in methodology at the beginning of the 2007/08 monitoring season, water quality data were only available for 14 core sites from that year. To maximise the number of locations for which water quality data were available, data were pooled over seasons and sampling years. Fourteen water quality variables were selected for analyses: nitrite, nitrate, ammonium (summarised as dissolved inorganic nitrogen , DIN), total dissolved nitrogen, particulate nitrogen, phosphate, total dissolved phosphorus, particulate phosphorus, silicic acid, dissolved organic carbon, particulate organic carbon, chlorophyll a, suspended solids and Secchi depth. Three components of sediment quality and 11 sediment grain size variables were selected for analyses: organic carbon, inorganic carbon and nitrogen content. The grain size variables were the proportion of the sediment samples from each reef corresponding to the following standard grain size categories: Clay, Silt very fine, Silt fine, Silt medium, Silt coarse, Sand very fine, Sand fine, Sand

Page 16: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

10

medium, Sand coarse, Sand very coarse, and granules. As these were proportions, the sum of the grain size fractions at each reef was 100%. To maximise the number of locations for which water quality data were available, data were pooled over seasons and sampling years. Each environmental variable was log transformed and scaled prior to analysis to accommodate different measurement units. Prior to transformation, values of zero (= below the detection limit of the respective analytical method) were set to half the limit of detection for that variable. Since environmental variables were likely to be correlated, exploratory principal components analyses were used to select a set of environmental variables for analysis of relationships with coral reef benthos. We included the following coral reef benthos data types from the 24 study reefs, for depths of 2m and 5m:

1. Percent cover for 57 genera of hard coral, 24 genera of soft coral and 13 genera of macroalgae.

2. Density of juveniles of 59 genera of hard coral and 22 genera of soft coral. Values of each benthic variable were averaged to give a single value for 2m and 5m depths at each reef. Data were fourth root transformed before analyses. Univariate Analysis

Univariate analyse were applied to the summary assemblage attributes of total cover for hard coral, soft coral and macro algae, the richness of genera from each of these cover estimates, the density of juvenile hard and soft corals and richness of these juvenile assemblages. All univariate models were generalised linear models (GLM) (quasipoisson) with a log-link (Venables and Ripley 2002). For each GLM, the response in the given assemblage attribute was tested for a significant relationship with variation in region and depth and environmental variables related to particulate water quality, dissolved water quality and clay-silt content of the reefal sediment. In each case both linear and non-linear (quadratic) terms for the environmental variables were examined. The non-linear environmental terms were not statistically significant for any summary attributes of hard and soft coral assemblages, so these terms were omitted and simpler models including only the linear terms for the environmental variables were included in the final model. The index based on particulate water quality components (see below) had a non-linear (quadratic) relationship with cover and richness of algae and so the quadratic term was retained for that benthic group. Presentation of results from these models includes:

1. Partial effect plots for significant terms. Partial effect plots visually display the effect of explanatory variables on the response (i.e. cover/density or richness) after accounting for the other variables included in the model.

2. Coefficient estimates, standard errors, test statistics and p-values for each of the models.

The coefficient estimates provide meaningful interpretations of the effect that a change in an

Page 17: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

11

explanatory variable has on the mean of the response. The standard errors and test statistics together provide information to determine whether an explanatory variable is statistically significant. This information is summarised by the corresponding p-values. An explanatory variable is deemed significant if has a p-value less than 0.05.

3. A sequential analysis of deviance for each model. The first three variables in the analysis of

deviance are particulate, dissolved and clay silt, in that order. The remaining two variables are region and depth, in that order. The sequential analysis of deviance is included to determine whether differences among regions explain a significant amount of variation in benthic assemblage attributes once the variation due to water quality and sediment variables have been accounted for.

Multivariate Analysis

The use of summary variables such as total coral cover or total juvenile density can mask differential responses by component taxa. For this reason, multivariate analyses of the benthic assemblages were used to identify responses of particular genera to the environmental variables. The multivariate analysis of benthic assemblages employed two techniques.

1. Multivariate regression trees (De’ath 2002). Trees were used to investigate the distribution of genera among the levels of the factors region and depth and along the environmental variables to estimate the hierarchy and levels of variables most closely associated with variation in benthic assemblages. The multivariate regression trees are summarized by the number of leaves in each tree and the cross validation error. The cross validation error ranges from zero to one, zero representing a perfect predictor and the prediction becoming poorer as the cross-validation error approaches one. The size of each tree is selected on the basis of cross-validation and the 1-SE error rule (Brieman et al. 1984).

2. Redundancy analysis. Redundancy analyses (RDA) were used test the strength of relationship

between spatial and environmental variables and each coral and algae assemblage data set. Results of RDAs include the presentation of the percentage of variation in assemblages accounted for by the fit of the RDA model along with marginal ANOVA tables that test the significance of the relationship between each explanatory variable and assemblage composition. Ordinations of the RDA results are presented as biplots to help visualised the relationships between the various benthic assemblages and environmental variables (Jongman et al. 2004). RDA revealed strong regional effects that tended to dominate the relationship between environmental variables and the cover/density of genera in biplots. Confounding between the environmental variables and region may also obscure relationships between genera and the environmental variables. In addition, differences in assemblages between 2m and 5m depth at any given reef cannot be attributed to the environmental variables as these are only available as averages for the whole reef. To maximise the variation in assemblages attributable to the environmental variables partial RDAs were applied to each benthic assemblage. These partial RDAs first accounted for differences in assemblages between depths, then fitted the three environmental variables. Importantly, a term for region was not included. Because of this, any confounding between region and environmental variables

Page 18: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

12

could be attributed to those environmental variables. The biplots resulting from these partial RDAs give the best relationship between the distribution of genera relative to the environmental variables.

All analyses were carried out in the R statistical package (R Development Core Team 2008) using the packages mvpart and vegan.

Page 19: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

13

1.3 Results and Discussion

The analyses of the effect of several environmental variables on adult hard corals, juvenile hard corals, adult soft corals, juvenile soft corals and macroalgae at 24 reefs on the Great Barrier Reef (Figure 1.1, Table 1.1) are grouped into two sections: Section 1 explores the spatial patterns of the water quality and sediment variables obtained in the Reef Plan MMP and correlations between them. The large number of environmental variables was reduced to three representative explanatory variables for the following analyses. These were two combined variables representing a particulate water quality index, a dissolved water quality index and a sediment quality variable, the proportion of clay silt in the sediment, which also predicts the content of organic carbon and nitrogen in sediments. Section 2 is the main part of this chapter and assesses the relationship between the selected environmental explanatory variables and the benthic assemblages found on the 24 study reefs using univariate and multivariate analyses.

1.3.1 SELECTION OF ENVIRONMENTAL VARIABLES FOR ANALYSIS

Spatial patterns in water quality variables

Most water quality variables varied considerably among the sampling regions (Appendix 1: Table A1). The concentrations of dissolved inorganic nitrogen (DIN) were very high and DON values very low in the Mackay Whitsunday Region. This was due to three sampling sites that had extremely high values during the wet season 2007; otherwise data were similar among regions. Median values of particulate nitrogen (PN) increased towards the south, with high variability in the Wet Tropics and Burdekin regions. For the phosphorus species, median values for phosphate (DIP) increased to the south, with high variability in the Fitzroy region, while dissolved organic phosphorus concentrations were high in the Mackay Whitsunday and very variable in the Wet Tropics region (Appendix 1: Table A1). Particulate phosphorus was very similar among regions, with high variability in Burdekin and Wet Tropics region. Silicic acid had the highest median value in the Wet Tropics, with high variability in the Fitzroy region. Dissolved and particulate organic carbon (DOC, POC) values were highest in the Fitzroy region, while the other regions were relatively similar (Appendix 1: Table A1). Chlorophyll values were similar among regions, with greatest variability in the Wet Tropics region. The Burdekin Region had the highest values of suspended solids (SS) and, correspondingly, the lowest values of Secchi depth (Appendix 1: Table A1). It has to be accepted that the water quality conditions are only reflective of the sampling years and may not be typical for the region in the long-term. For example, the wet season 2007 was much more pronounced than in 2006 and the Fitzroy Region had no significant land runoff for many years. The correlations between the water quality variables were explored using principal components analysis (PCA). The variables fall into two distinct groups (Figure 1.2), members of each group being highly correlated. The first group was made up of PN, PP, DOC, POC, chlorophyll, SS and Secchi depth; these are all related to particulate matter, except for DOC. The second group was made up

Page 20: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

14

of all DIN parameters (nitrate, nitrite and ammonium), DIP, Si, and total dissolved nitrogen and phosphorus (TDN, TDP) and are all dissolved water quality variables. The correlation between the particulate and dissolved water quality variables was minimal (Figure 1.2) so they were examined separately.

Figure 1.2 Principal Components Analysis (PCA) biplot of water quality variables. The first principal component explains 44.58% of the variation of the water quality parameters and is positively correlated with all water quality variables. The second principal component explains 19.33%. Note the groupings below and above the horizontal line at zero. In the separate PCA that only included the particulate water quality variables and DOC, the first principal component explained 73.5% of the variation and was negatively correlated with the seven particulate variables (PN, PP, DOC, POC, chlorophyll, SS and Secchi depth; Figure 1.3). The second principal component explained 15.12% of the variation. The inverse of the first principal component from this analysis was used as an explanatory variable in subsequent analyses and is referred to as the “particulate water quality index” or shorter “particulate WQ”. Larger values of “particulate WQ” correspond to higher values of particulate water variables. In following analyses, the negative of the first principal component is scaled to lie between 1 and 100 for ease of interpretation.

Page 21: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

15

Figure 1.3 Principal Components Analysis (PCA) biplot of the particulate water quality variables. The first principal component explains 73.5% of the variation in the particulate water quality variables and is negatively correlated with PN, PP, DOC, POC, chlorophyll, SS and Secchi depth. The second principal component explains 15.12% of the variation in the particulate water quality variables. When the dissolved water quality components were considered alone, the first principal component explained 49.4% of the variation and was positively correlated with the DIN parameters (nitrate, nitrite and ammonium), DIP, Si, and TDN and TDP (Figure 1.4). The second principal component explained 17.99% of the variation. The first principal component was used as an explanatory variable in subsequent analyses and is referred to as the “dissolved water quality index”, or shorter “dissolved WQ”. Larger values of “dissolved WQ” correspond to greater concentrations of the seven dissolved water quality variables. In following analyses, the first principal component was scaled to lie between 1 and 100 for ease of interpretation.

Page 22: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

16

Figure 1.4 Principal Components Analysis (PCA) biplot of the dissolved water quality variables. The first principal component explains 49.4% of the variation in the particulate water quality variables and is positively correlated with DIN parameters (nitrate, nitrite and ammonium), DIP, Si, and TDN and TDP. The second principal component explains 17.99% of the variation in the dissolved water quality variables. Sediment quality variables

The correlations between the transformed sediment nutrient content variables (organic and inorganic carbon and total nitrogen) were also explored using PCA (Figure 1.5). The first principal component explained 74.63% of the variation in the nutrient sediment variables and was positively correlated with inorganic carbon, total nitrogen and organic carbon. The second principal component explained 24.16% of the variation. The sediment grainsizes are proportional data and, the sum of the grainsize factions for each sample was 100. The first principal component explains 66.1%, the second principal component explains 23.74% of the variation (Figure 1.6). Note the tight cluster of smaller grainsize fractions, which correspond to sediments in the size class of clay and silts (< 63 µm). Correlation between these clay silt grainsizes and sediment organic carbon and nitrogen content is very high (0.91; data not shown). These fine clay/silt sediments are thought to have the greatest effect on benthic coral reef assemblages and are also those likely to have increased in supply to the GBR since European settlement (Furnas 2003, Wolanski et al. 2005 and 2008). We therefore selected the proportion of clay/silt-sized sediments at each reef as an explanatory variable in subsequent analyses; this is referred to as “clay silt”.

Page 23: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

17

Figure 1.5 Principal Components Analysis (PCA) biplot of the nutrient sediment variables. The first principal component explains 74.63% of the variation in the nutrient sediment variables and is positively correlated with inorganic carbon, total nitrogen and organic carbon. The second principal component explains 24.16% of the variation in the nutrient sediment variables.

Figure 1.6 Principal Components Analysis (PCA) Biplot of the grainsize variables. The first principal component explains 66.1% of the variation in the nutrient sediment variables. The second principal component explains 23.74% of the variation in the nutrient sediment variables. Note that grainsize data are proportions.

Page 24: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

18

Final selection of explanatory variables

On the basis of the preceding analyses, the following variables were used subsequently as explanatory variables that might influence cover and richness of adult and juvenile hard corals, adult and juvenile soft corals, and algae:

1. The negative of the first principal component, scaled to lie between 1 and 100, for the particulate water quality variables. This variable was denoted the “particulate water quality index”.

2. The first principal component of the dissolved water quality variables, scaled to lie between 1 and 100. This variable was denoted the “dissolved water quality index”.

3. The proportion of sediments with grainsizes below 63 µm. This variable was denoted “clay silt”.

Two additional variables were also included in the analyses. These were:

4. The survey reefs distributed among four broad geographic (NRM) regions: Burdekin (5 reefs), Fitzroy (4 reefs), Mackay (7 reefs) and the Wet Tropics (8 reefs). This variable is referred to as “Region”.

5. At each reef measurements of corals and algae were made at two depths – 2 meters and 5 meters below LAT. This variable is referred to as “Depth”.

Do the selected environmental variables vary among the regions?

There were no obvious differences in values of ‘particulate’ and ‘dissolved’ water quality indices among the NRM regions (Figures 1.7, Table 1.2). Reefs in the Mackay Whitsunday Region had higher values of ‘clay silt’ than those in the other three regions (Figure 1.7, Table 1.2). Table 1.2 Results of Analysis of variance (ANOVA) comparing values of the three selected environmental response variables among NRM Regions. Sum Sq= sum of squares, Mean Sq= mean square Factor DF Sum Sq Mean Sq F Pr (>F) Response variable: particulate water quality Region 3 1100.3 366.8 0.4034 0.7521 Residual 20 18184.3 909.2 Response variable: dissolved water quality Region 3 3862.3 1287.4 2.1836 0.1216 Residual 20 11791.6 589.6 Response variable: sediment clay silt Region 3 5672.8 1890.9 13.174 <0.001 Residual 20 2870.6 143.5

Page 25: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

19

a

b c

Figure 1.7 Regional summaries of environmental variables at the 24 study reefs. a) particulate and b) dissolved water quality, c) sediment clay silt content. Wet= Wet Tropics Region, Bur= Burdekin Region, Mac= Mackay Whitsunday Region, Fit= Fitzroy Region. Boxes represent the interquartile range (i.e., 50% of the data), horizontal black lines are medians and whiskers are 95% confidence intervals.

Page 26: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

20

1.3.2 DO THE SELECTED ENVIRONMENTAL VARIABLES INFLUENCE THE BENTHIC ASSEMBLAGES ON INSHORE CORAL REEFS?

This section analyses the relationship between the selected environmental explanatory variables and the benthic assemblages found on the 24 study reefs. The assemblage attributes used in these analyses were hard coral cover (for 57 genera), juvenile hard coral (60 genera), soft coral cover (33 genera), juvenile soft coral (32 genera) and macroalgal cover (13 genera). Adult cover and juvenile density of hard and soft corals were treated separately in parallel analyses. The presentation and discussion of results first identifies possible spatial variation in assemblages (effects of region and depth) before assessing relationships between the assemblages and their environment as described by water and sediment quality variables. For each benthic group, univariate analyses in the form of generalised linear models (Venables and Ripley 2002) were applied to measures of both total cover/density and richness of hard and soft corals and macroalgae. The aim of these univariate analyses was to identify broad responses of assemblages to the spatial and environmental variables included in the models. Variation in benthic assemblages is likely to result from a combination of the local variation in environmental variables in combination with broader regional variation. It is necessary to consider potential regional signals prior to describing relationships between coral assemblages and local environmental variables. The 24 reefs included in this study span about seven degrees of latitude, with those in the most southern region (Fitzroy region) approaching the southern extent for inshore reefs on the east coast of Australia. A decline in coral assemblages (both in cover and richness) with the transition from tropical to temperate zones is common to reef provinces world wide and largely attributed to the lower temperature and potentially winter sun aspect being suboptimal to most coral species (Veron 1995). However, the modelling of high level assemblage attributes, such as total coral cover or richness as above, can mask differential responses of individual taxa to environmental variables. To assess these individual responses, multivariate regression trees (De’ath 2002) and redundancy analyses (Jongman et al. 2004) were applied to the genus-level assemblage data to explore, define and predict relationships between assemblage composition and environmental variables. The final branches of the regression trees represent the environmental niches that best describe certain assemblage compositions based on the environmental variables included in the analysis. Importantly, the focus was to identify potential relationships between benthic assemblages and environmental variables that represent the effects of possible downstream effects of land use practices with the goal of improving understanding of the environmental limitations imposed on coral assemblages in the inshore environment.

Page 27: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

21

Hard coral assemblages

This section examines the effect of regional location, depth and three environmental variables on the distribution of hard coral assemblages on inshore reefs of the GBR. Fifty-seven genera of Scleractinan corals were recorded in the photo transects while juvenile colonies (<10cm in maximum dimension) of 59 genera were counted. These adults and juvenile assemblages were first analysed in terms of overall abundance (specifically percentage cover of the substrate, and density of juvenile colonies per m2 of transect not already occupied by either hard or soft coral) and richness, as the number of genera present at a sampling location. Regional and depth-related variation in hard coral assemblages

Variation in benthic assemblages is likely to result from a combination of the local variation in environmental variables in combination with broader regional variation. The 24 reefs included in this study span about seven degrees of latitude and reefs in the most southern region (Fitzroy region) approach the southern limit for inshore reefs on the East coast of Australia. Potential regional signals need to be considered before describing relationships between coral assemblages and local environmental variables. Each hard coral assemblage attribute (cover, richness, juvenile density and juvenile richness) varied significantly among regions (P<0.05, Appendix2: Tables A2-1to A2-4), but in each case the average levels for each assemblage attribute observed on reefs in the Fitzroy region differed from those observed elsewhere (Figure 1.8). The estimated differences in levels of each assemblage attribute for reefs in each region relative to values for the Fitzroy region are presented in Table 1.3. On average, the cover of hard corals on reefs in the Fitzroy Region was significantly higher than on reefs in the Mackay, Burdekin or Wet Tropics regions (Figure 1.8a). In contrast, the density of juvenile colonies was lower on Fitzroy Region reefs (Figure 1.8b), although the difference between juveniles densities on reef in the Fitzroy and Mackay Whitsunday regions was not statistically significant (Table 1.3). The richness of genera in both the adult and the juvenile assemblages was comparatively low in the Fitzroy region (Figure 1.8c, d). There were no discernable differences in any of the hard coral assemblage attributes between the Mackay Whitsunday, Burdekin or Wet Tropics regions, with the one exception that the richness of juvenile colonies on Wet Tropics reefs was on average 77% lower than on reefs in the Mackay region (P=0.024, Appendix 2 Table A2-4). Table 1.3 Proportional difference in assemblage attributes on reefs in the Wet Tropics, Burdekin and Mackay regions relative to values for he Fitzroy region. Tabulated values are how much lower (-) or higher (+) estimates of each assemblage attribute are in each region compared to those for the Fitzroy region. (ns) signifies that this difference was not significant at P=0.05. Hard coral assemblage attribute Wet Tropics Region Burdekin Region Mackay Region Cover -49% -46% -43% Richness 94% 116% 153% Juvenile density 118% 106% 89%(ns) Juvenile richness 106% 108% 167%

Page 28: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

22

Figure 1.8 Term plots for the factor region from generalised linear models for each hard coral assemblage attributes. Term plots present the fit of the term, in this case region, to the residuals resulting from the fitting of all other terms in the model to the response variable (assemblage attribute). The y-axis values represent the deviation of the regional values from the overall intercept of the model (rather than mean values). Dotted lines are 95% confidence intervals. Some of the variation among the regions may be due to regional differences in values of the environmental variables. To assess whether regional effects existed over and above any effects of the environmental variables, the environmental terms were added sequentially to statistical models prior to inclusion of the factors ‘Region’ and ‘Depth’. There were significant regional differences in each assemblage attribute despite the prior removal of variation explained by the environmental variables (Appendix 3 Tables A3-1-A3-4). This result suggests that differences between the Fitzroy region and the other three regions were largely independent of the variation in the selected water and sediment quality variables and reflect shift in hard coral assemblages toward the southern end of the GBR and a general latitudinal decline in biodiversity of corals that is steepest in the transition from tropical to temperate waters (Veron 1995). Regional patterns in coral assemblage composition by genus generally reflected the summary attributes presented above. RDA of both genus-level cover and genus-level juvenile density indicated significant regional differences in the assemblages (P<0.05, Appendix 4 Tables A4-1 and A4-2). For both hard coral cover (Figure 1.9a) and juvenile density (Figure 1.9b), assemblages on reefs in the Fitzroy region were clearly distinct from those in other regions. Furthermore, these ordinations clearly indicate lower generic richness on the reefs of the Fitzroy Region, indicated by the majority of vectors for genera pointing away from reefs of the Fitzroy Region. In many instances this lower

-0.5

0.0

0.5

1.0

Hard coral coverP

artia

l effe

ct o

f reg

ion

Wet Bur Mac Fit

a

-1.0

-0.5

0.0

0.5

Hard coral juvenile density

Wet Bur Mac Fit

b

-1.0

-0.5

0.0

0.5

Hard coral richness

Part

ial E

ffect

of R

egio

n

Wet Bur Mac Fit

c -1.0

-0.5

0.0

0.5

Hard coral juvenile richness

Wet Bur Mac Fit

d

Page 29: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

23

richness reflected the absence of genera from the Fitzroy region; 28 of the 57 genera recorded as adults and 31 of the 59 genera observed as juveniles were not recorded from the Fitzroy region. The RDA showed that 13% of the total variation in cover of hard coral genera and 15.6% of the variation in juvenile density of genera was due to the differences between Fitzroy region assemblages and those to the north. Further separating the three northern regions indicated that in total, regional differences accounted for 22.4% of the variation in adult coral cover and 25.9% of the variation in juvenile assemblages.

1.5

-1.0

-0.5

0.0

0.5

1.0

Hard coral cover

RDA1 49.7 %

RD

A2

18.

2 %

1.5

-1.0

-0.5

0.0

0.5

1.0

Hard coral cover

RDA1 49.7 %

RD

A2

18.

2 %

Particulate

Clay-silt

DissolvedBurdekin

-1.0 -0.5 0 0.5 1.0

Hard coral juvenile density

RDA1 50.8 %

RD

A2

19.

4 %

ParticulateDissolved

Clay-silt

-1.0 -0.5 0 0.5 1.0

MackayFitzroy

Wet Tropics

Fitzroy

Mackay

Wet Tropics

Burdekin

Figure 1.9 Redundancy analysis (RDA) biplots for cover (left) and juvenile density (right) of hard coral genera. Represented are the ordinations of assemblages constrained to the environmental variables ‘Region’, ‘Depth’, ‘Particulate’ and ‘Dissolved’ water quality and ‘Clay-silt’ content of the sediments. The proportions of the variance explained by the first and second RDA axes are presented. These five environmental/spatial variables account for 37.9% (hard coral cover) and 43.2% (hard coral juvenile density). Given the marked difference between hard coral assemblages on reefs in the Fitzroy region and the three other regions we provide here a brief description of these assemblages. Four reefs were included from the Fitzroy region, on three of these the coral assemblage was broadly similar and consisted primarily of large stands of branching Acropora, predominantly A. muricata and A. intermedia (Figure 1.10a). On these three reefs the cover of Acropora was between 76.6% and 98.7% of the total cover. This dominance of just a few species explains the discrepancy between high cover and low richness. Further, the substratum available for settlement of coral larvae in such assemblages is primarily composed of either the dead lower branches of otherwise living colonies or mobile coral fragments, which both are sub-optimal substrata for coral recruitment. The lower branches within Acropora thickets die as the colonies grow, presumably in response to the suboptimal conditions of reduced light and possibly water movement. In the Fitzroy Region, dead Acropora skeletons are heavily colonised by the macroalgae Lobophora variegata, making these surfaces also unsuitable for coral recruitment. Mobile coral fragments are moved around by wave action, causing abrasion and smothering of newly recruited corals. This lack of suitable substrate in conjunction with low richness of larval supply may partly explain the discrepancy between high adult coral cover and low density of recruit sized colonies in the Fitzroy Region.

Page 30: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

24

Figure 1.10 Representative coral assemblage types in the Fitzroy Region. (a) Coral assemblage typical of Barren, North Keppel, Humpy and Halfway Islands, dominated by branching Acropora (b) Coral assemblage at 5 m Pelican Is with the hard coral genera Goniastrea and Hydnophora pictured. Photos: J. Davidson, AIMS.

Page 31: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

25

The forth reef in the region, Pelican Island, lies in particularly turbid water in Keppel Bay, at the outer boundary of the estuarine area of the Fitzroy River. While the assemblage at 2m depth is also dominated by Acropora (85%) the assemblage is more diverse. At 5m, however, light levels are often very low and the photophilic genus Acropora largely absent. At this depth, the assemblage is dominated by large Alveopora, Hydnophora, Goniastrea, Psammocora and Turbinaria colonies (Figure 1.10b), an assemblage composition not seen on any other reefs included in this study. Both the richness of genera observed in adult coral cover estimates and densities of juvenile colonies was higher at 5m than at 2m water depth (P<0.05, Appendix 2: Tables A2-2 and A2-4). The higher richness at 5m was also apparent in analyses of assemblage composition (as genus-level cover or juvenile density) where the overall difference between 2m and 5m assemblages accounted for 4.7% (cover) and 3.4% (juvenile density) of the variation in coral assemblages (Appendix 3: TablesA3-1and A3-2). However, these differences in richness and assemblage composition did not translate into differences in overall abundance; both adult coral cover and the overall density of juveniles were not significantly different between depths (P> 0.05, Appendix 2: Tables A2-1 and A2-3). Only the cover of the genera Acropora and Pocillopora were significantly higher at 2m, while the cover of 17 other genera was significantly higher at 5m. Juvenile assemblages showed similar differences between depths, with only Acropora and its subgenus Isopora having significantly higher density at 2m. The genera Acropora and Pocillopora may be more abundant at 2m depth because they are able to compete with macroalgae, which have a much higher cover at 2m depth (see below), due to their faster growth rates (Harriott 1999) and more erect growth forms than most other hard coral genera. Fifteen 15 other hard corak genera had significantly higher densities of juveniles at 5m depth. For ten genera (Ctenactis, Goniopora, Moseleya, Mycedium, Oxypora, Pachyseris, Palauastrea, Pectinia, Podabacia and Turbinaria) both adult cover and density of juveniles was higher at 5m. The strong distinction between coral assemblages at 2m and those at 5m depth is likely to be a consequence of steep environmental gradients on the inshore reefs associated with increasing depths, especially the decreasing light availability in the turbid inshore waters. Corals at 5 m depth would be also exposed to larger amounts of settling particles from the wind-driven resuspension of fine sediments compared to those at 2m, leading to even more pronounced light limitation (Wolanski et al. 2005). A number of the coral genera found on the inshore reefs at 5 m depth are recognised as being low light tolerant as they are found in turbid environments (Done 1982) or as part of deep water assemblages in clearer waters (Titlyanov and Latypov 1991). Relationship between hard coral assemblages and environmental variables

In this analysis, the hard coral assemblage attributes, cover, richness, juvenile density and juvenile richness were tested for their relationship with the selected environmental variables. When regional and depth effects were taken into consideration, only the cover of hard corals showed a significant relationship to any of the three environmental variables (P<0.05, Appendix 2: Table A2-1). The cover of hard coral increased with proportion of the clay-silt content of the sediment; for a one percent increase in the proportion of clay and silt-sized particles in the sediments the cover of hard coral increased at a rate of 2% (Figure 1.11c). There was also some evidence that increasing levels of the ‘particulate’ water quality index resulted in lower coral cover and lower richness of juvenile coral genera (0.05<p<0.1 in both cases, Appendix 2: Tables A2-1 and A2-4, Figure 1.11a, b).

Page 32: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

26

Figure 1.11 Term plots of significant relationships between hard coral assemblage attributes and environmental variables as indicated by generalised linear models. Plotted are the fit of the environmental variable indicated on the x-axis against the residuals from the model after fitting of the remaining environmental variables. Appling multivariate regression trees to both the cover and juvenile density of hard coral genera revealed broadly similar environmental divisions in assemblage composition. However, the predictive power of the trees was greater for the juvenile densities (cross validation error, 0.804) than for cover-based estimates of assemblage composition (0.983), indicating less “noise” in the relationship between juvenile density and the environment. For both cover (Figure 1.12a) and juvenile density (Figure 1.12b) the most substantial difference in assemblage composition was between assemblages in the Fitzroy region and the other three regions. In both cases, a secondary regional split occurred between assemblages in the Mackay region and the assemblages in the Wet Tropics and Burdekin regions. As has been presented above (Figure 1.7c), the proportion of clay-silt in the sediments on reefs in the Mackay Whitsunday Region is high compared to reefs in other regions. Only one reef (Snapper Island North, in the Wet Tropics Region) had a clay-silt proportion within the range of values occurring on Mackay Whitsunday reefs, so the distinctness of the Mackay Whitsunday reefs is likely to reflect assemblage changes in response to higher proportions of fine sediment.

0 20 40 60 80 100

-1.0

-0.5

0.0

0.5

Hard coral cover

Particulate

Parti

al e

ffect

of p

artic

ulat

e

a

0 20 40 60 80 100-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

Hard coral juvenile richness

Particulate

Parti

al e

ffect

of p

artic

ulat

e

b

0 10 20 30 40 50 60

-0.5

0.0

0.5

1.0

1.5Hard coral cover

Clay-silt

Parti

al e

ffect

of c

lay-

silt c

Page 33: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

27

Figure 1.12 Multivariate regression trees for cover of hard coral genera (a), and density of hard coral genera juveniles (b). The number at the terminal branch of the tree is the number of reefs where a particular assemblages type occurred (two depths at each of the 24 survey reefs, total N=48). The assemblages also varied in response to the levels of the ‘particulate’ water quality index. For juvenile densities, the analysis separated the reefs at quite high levels of this parameter indicating that the response of juvenile corals was most pronounced at the more turbid sites within the range included in this study (Figure 1.12b). The differentiating effect of sediment quality/region was more pronounced at sites with lower turbidity. For hard coral cover, the assemblage response to particulates is secondary to the sediment quality/region split, suggesting a potential reversal of the weighting between these two variables, compared to juvenile assemblages.

a

b

Region

Error : 0.672 CV Error : 0.804

10

1614

8

Region Wet Tropics, Burdekin, Mackay

Wet Tropics, BurdekinRegion

< 59.25

Particulate

Hard coral juvenile density

Mackay

> 59.25

Fitzroy

< 62.1

> 29.416 2

14 12

14

Mackay

Wet Tropics, Burdekin, Mackay

RegionDissolved

Particulate

Hard coral cover Error : 0.696 CV Error : 0.983

Fitzroy

Wet Tropics, Burdekin> 62.1

< 29.41

Page 34: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

28

Finally, the analyses separated the assemblage composition of adult hard corals in the Fitzroy region based on the level of ‘dissolved’ water quality. This split simply differentiates between assemblages at Pelican Island and those on the three reefs further offshore in this region. It should be noted, however, that both ‘particulate’ and clay-silt are also higher at this reef and any of the three environmental variables or a combination of the three could be equally implicated in this difference in assemblage composition. To further investigate the responses of individual hard coral genera to environmental variables, RDAs were carried out to maximise the variation in assemblages explained by the selected environmental variables. Regional terms were excluded from the model due to the confounding between regions and environmental variables. The three environmental variables explained 17.3% (cover) and 20.6% (juvenile density) of the total variation in hard coral assemblage composition observed on the reefs included in this study (P<0.05, Appendix 3: Tables A3-1b and A3-2b). However, comparing results from RDA models that fit the environmental variables after first removing variation attributable to region with those models that did not include a term for region indicated that the environmental variables particulate and clay-silt were significantly associated with variation in both juvenile and adult assemblages irrespective of the factor ‘region’ (p<0.05, Appendix 4: Tables A4-1and A4-2). The dissolved water quality variable, however, showed no significant relationships to the hard coral assemblages. The ordinations of the results of the RDA models based only on the water and sediment quality variables help to visualise the responses of different genera to the environmental variables (Figure 1.13a, b). There are some broader generalisations. Cover of a large number of adult hard coral genera is higher at high levels of sediment ‘clay-silt’, indicated by the genera with vectors pointing in the same direction as the clay silt vector on the plot (Figure 1.13a). In contrast, none of the coral genera have higher cover on reefs with lower levels of fine grained particles, indicated by the lack of genus vectors pointing in the opposite direction to the ‘clay-silt’ vector. Similarly, there are some genera (e.g. Turbinaria, Goniastrea and Moseleya) with higher cover at higher levels of the ‘particulate’ water quality index. In contrast, cover of genera such as Pocillopora, Porites, Acropora, Isopora and Echinopora are higher where levels of particulates were lower, i.e. in clearer water locations (Figure 1.13a). The density of juvenile colonies responded to the selected environmental variables in a way similar to the cover of hard corals (Figure 1.13b). A number of genera had higher densities of juveniles associated with higher proportions of clay and silt in the substrate. There were a small number of genera with higher densities of juveniles associated with higher levels of ‘particulates’, but substantially more genera showed the opposite relationship, with vectors pointing in the opposite direction to ‘particulates’, indicating a preference for less turbid environments. However, there is a subtle difference between the relationships observed in the adult cover and juvenile assemblages: many of the adult genera that showed a positive relationship to the proportion of clay and silt in the sediment had vectors at right angles with the ‘particulate’ water quality vector, indicating a negligible association with this parameter. For juveniles, however, most of the genus vectors point away from the ‘particulate’ vector, indicating a strong negative association. This is supported by the vector for ‘particulate’ also being closer to the horizontal dimension in the juvenile coral ordination, in which

Page 35: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

29

RDA1 55.2 %

RD

A2

30

.4 %

RDA1 55.2 %

RD

A2

30

.4 %

Pocillopora

Isopora

Moseleya

HeliofungiaPavona

Favia

Merulina

Herpolitha

Goniastrea

Caulastrea

Galaxea

Echinopora

LobophylliaOxypora

Cyphastrea Mycedium

Pachyseris

Acropora

Goniopora

Turbinaria

Porites

Pectinia

Particulate

Dissolved

Clay-silt

RDA1 61.1 %

RD

A2

26

%

RDA1 61.1 %

RD

A2

26

%

Pseudosiderastrea

Cyphastrea

Seriatopora

GoniastreaCaulastrea

Pavona

Psammocora

Scolymia

Fungia

MerulinaEchinophyllia

Porites

Euphyllia

Galaxea

OxyporaHeliofungia

Echinopora

Podabacia

Pachyseris

Turbinaria

Lobophyllia

Moseleya

PectiniaParticulate

Dissolved

Clay-silt

Figure 1.13 Redundancy analysis bi-plot of the relationship between cover (a) and juvenile density (b) of hard coral genera and environmental variables. The plot is the ordination of the assemblage space constrained to the variation described by the three environmental variables represented by heavy black arrows and bold type. The relationship between this constrained space and each genus are represented by the length and direction of the genus vectors (light grey lines). For clarity only vectors for the 60% of genera relating most to the environmental variables, of which the top 40% are named are presented. Grey dots plot the position of reef assemblages relative to the environmental variables.

a

b

Page 36: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

30

the majority (61%) of the relationship between juvenile density and the environmental variables is explained. We are unable to provide an explanation why a number of coral genera showed higher adult cover and juvenile densities on reefs with high proportions of fine sediments (both indicated by univariate analysis and RDA). However that many of the genera showing the strongest preference for fine sediments have been shown capable of easily removing such sediments from their surface (Stafford-Smith and Ormond 1992) suggests some adaption to such an environmental niche. Similarly part of the ability to exist in low light (turbid settings) is due to the capacity of some coral taxa to partly compensate for reductions in energy yield derived from photosynthesis by feeding on suspended particles (Anthony 1999, Anthony 2000, Anthony and Fabricius 2000) or exhibiting some degree of physiological plasticity (Anthony and Connolly 2004). Although few species have been studied it seems plausible that variability in the expression of these traits will determine limit to which species can persist along turbidity gradients. Indeed of the genera showing higher occurrence in turbid settings such traits in Turbinaria mesenterina result in increased lipid (energy) stores presumably allowing colonies to persist during very turbid periods that would otherwise cause a substantial energy deficit if colonies were reliant on photosynthesis alones (Anthony 2006). Similarly Goniastrea retiformis was shown to have a greater ability to increase energy gain derived through feeding on particulate matter than Porites cylindrica in an experimental setting and hence maintain a energy surplus into more turbid conditions (Anthony and Fabricius 2000). This result is consistent with our findings that Goniastrea showed a higher occurrence at higher levels of particulate water quality vector in contrast to Porites that was negatively associated with this vector (Figure 1.13a and b). Acropora species were also clearly associated with low ‘particulate’, high light environments. However, because Acropora is the most species-rich coral genus on the GBR (Veron 2006), interpretation of this association is hampered by the limited taxonomic resolution. There are fewer Acropora species inshore, representing a subset of the species suite on GBR offshore reefs (DeVantier et al. 2006) and Done (1982) even classified GBR inshore reefs as 'non-Acropora' reefs. Our results show that some inshore reefs with relatively low level of ‘particulate’ have coral assemblages with a very high cover of Acropora spp. A. millepora, a species found across the entire shelf of the GBR, can feed on suspended particles but less efficiently than species with higher turbidity tolerance (Anthony 2000, 2006). A large number of genera in the juvenile hard coral assemblage were associated with less turbid water. This is supported by the laboratory findings that coral recruit survival and juvenile growth are negatively affected by light reduction and sedimentation (Fabricius 2005). The fact that both the juvenile and adult assemblages show similar environmental relationships is consistent with the finding of Baird et al. (2003) that corals settle preferentially into certain habitats and that this preference is reflected in the composition of the adult assemblage.

Page 37: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

31

Soft coral assemblages This section examines the effect of regional location, depth and three environmental variables on the distribution of adult and juvenile soft coral assemblages on inshore reefs of the GBR. Soft coral species can be broadly separated into two groups, those that like most hard corals obtain at least part of there energy requirements from a symbiotic relationship with photosynthetic algae (zooxanthellate octocorals= phototrophs) and those that feed on plankton and particulate matter (azooxanthellate octocorals= heterotrophs). These different modes of energy acquisition determine separate environmental niches (Fabricius and Death 2008). Because the sampling under the Reef Plan MMP was mainly designed to capture hard coral assemblages on inshore reefs (and, hence, focused on shallow reef areas with sampling at 2 and 5 m depth below LAT), the monitored soft coral assemblages predominantly represented phototrophic soft corals that tend to co-occur with hard corals, with only some heterotrophic genera occurring. Photo transects returned cover estimates for 24 taxonomic units, 23 of these are actual soft coral genera while one is a collective taxon we termed ‘gorgonian’, with taxa that could not be reliably identified from photographs and included predominantly heterotrophic genera with sea fan or sea whip-like morphology. However, for simplicity we refer in the report to these taxonomic units as ‘genera’. In situ counts along belt transects returned estimates for juvenile colonies off 22 genera (<10cm in maximum dimension), with juvenile-sized colonies of the ‘gorgonian’ group not included. Similarly, the genus Xenia was not included in the juvenile data set as it forms large aggregations of colonies that are mostly smaller than10cm in diameter but were likely produced by vegetative rather than sexual reproduction. Regional and depth-related variation in soft coral assemblages

The density and richness of juvenile colonies was significantly different between the four sampling regions (Figure 1.14), whereas the adult cover and richness was not (Appendix 2: Tables A2-5 to A2-8). Regional contrasts indicate that the density of juvenile soft coral colonies was higher on reefs in the Mackay Region than on reefs in the Wet Tropics Region (Figure 1.14a). The Mackay reefs had particularly high densities of juvenile Sarcophyton, Sinularia and to a lesser degree Cespitularia compared to the Wet tropics. The richness of juveniles soft corals was much lower on reefs in the Fitzroy Region compared to the other three regions (Figure 1.14b, Table 1.4). Table 1.4 Proportional difference (%) in soft coral juvenile richness on reefs in the Fitzroy Region compared to each of the other three regions. Tabulated values are how much higher estimates of richness were in each region compared to those estimated for the Fitzroy Region. Wet Tropics Region Burdekin Region Mackay Region Juvenile soft coral richness 84 % 87 % 187 %

Page 38: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

32

Figure 1.14 Term plots for the factor Region from generalised linear models for soft juvenile density (a) and richness of genera (b). Term plots present the fit of the term, in this case Region, to the residuals resulting from the fitting of all other terms in the model to the response variable (assemblage attribute). The y-axis values represent the deviation from the overall intercept of the model rather than mean values observed for levels of factor variable. As before for hard coral assemblages, we assessed whether the observed regional differences in soft coral assemblages were independent of the effects of environmental variables by applying models that fit the environmental terms sequentially prior to inclusion of factors for region and depth. The higher densities of juvenile soft corals in the Mackay region were more related to the environmental setting of the reefs in this region than a regional effect per se. This was not the case for the low richness of soft coral juveniles observed in the Fitzroy Region. The regional (southern) setting of these assemblages rather than the local environmental setting was responsible for this low richness, indicated by the analysis result that maintained a significant regional effect despite prior removal of variation attributable to the environmental variables (Appendix 2: Table A2-11). Also, both adult and juvenile soft coral assemblage compositions were clearly different between regions (redundancy analysis, P<0.05, Appendix 3: Tables A3-3 and A3-4). The Mackay Whitsunday Region soft coral assemblages were clearly separated from those observed on reefs in the other regions (Figure 1.15a, b). The assemblages in the Fitzroy Region were also different from the Burdekin and Wet Tropics reefs, which supported the results of the univariate analysis above where lower richness of juvenile genera was apparent on the Fitzroy Region reefs. In total, the factor Region explained 22.1% of the total variation in genus cover and 22.4% of the variation in genus juvenile density.

-1.0

-0.5

0.0

0.5

1.0

1.5

Soft coral juvenile densityPa

rtia

l effe

ct o

f reg

ion

Wet Bur Mac Fit

a

-1.0

-0.5

0.0

0.5

Soft coral juvenile richness

Wet Bur Mac Fit

b

Page 39: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

33

-1.0 -0.5 0.0 0.5

-0.5

0.0

0.5

Soft coral juvenile density

RDA1 51.7 %

RD

A2

27.

1 %

-1.0 -0.5 0.0 0.5

-0.5

0.0

0.5

Soft coral juvenile density

RDA1 51.7 %

RD

A2

27.

1 %

ParticulateDissolved

Clay-silt

Wet Tropics

Mackay Fitzroy

Burdekin

-1.0 -0.5 0.0 0.5 1.0

-1.0

-0.5

0.0

0.5

Soft coral cover

RDA1 47.4 %

RD

A2

26

.1 %

Mackay

Clay-silt

DissolvedParticulate

Wet Tropics Burdekin

Fitzroy

Figure 1.15 Redundancy analysis biplots for the cover (left), and density of juvenile (right) soft coral genera. Represented are the ordinations of assemblages constrained to the five spatial and environmental variables ‘Region’, ‘Depth’, ‘Particulate’ and ‘Dissolved’ water quality, and ‘Clay-silt’ content of the sediments. The proportions of the variance explained by the first and second redundancy analysis axes are presented. The five variables account for 31.8% (cover) and 34% (juvenile density) of the variation in soft coral assemblages. The soft coral assemblage attributes of cover, richness, juvenile density or juvenile richness were not different between the sampling locations at 2m and 5m depths (Appendix 2: Tables A2-5 to A2-8). Only the cover of adult colonies of the composite group ‘gorgonian’ had consistently higher cover at 5m than at 2m, indicated by the examination of the 95% confidence intervals of the proportion of cover found at 5m at each reef at which a genus occurred. The ‘gorgonian’ taxon combines the cover of a number of heterotrophic genera. The increased cover of this group at 5m depth suggests a shift in the balance between energy availability from photosynthesis toward energy derived from heterotrophic feeding, which is likely to be caused by the reduced light availability in turbid water at this depth (Cooper et al. 2008). This pattern was also seen in the density of juvenile colonies, with the phototrophic genera Cladiella and Clavularia being more abundant at 2m depth while Klyxum, Cespitularia, Junceella and Stereonephthya had higher densities at 5m depth. Klyxum species are considered phototrophic while the genera Junceella, Cespitularia and Stereonephthya have both phototrophic and heterotrophic species (van Oppen et al. 2005).

Page 40: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

34

Relationship between soft coral assemblages and environmental variables

In this analysis, the soft coral assemblage attributes, cover, richness, juvenile density and juvenile richness were assessed for their relationship with the three selected environmental variables. The adult cover and richness and the density of juvenile soft coral colonies all declined with increasing levels of the ‘particulate’ water quality index (P<0.05, Appendix 2: Table A2-5, A2-7, A2-8; Figure 1.16).

Figure 1.16 Term plots of significant relationships between soft coral assemblage attributes and environmental variables as indicated by generalised linear models. Plotted are the fit of the environmental variable indicated on the x-axis against the residuals from the model after fitting of the remaining environmental variables. The relationships between adult and juvenile soft coral assemblage composition and the spatial and environmental variables were investigated with multivariate regression trees. For both adult cover and juvenile density the most substantial difference in soft coral assemblage composition occurred between assemblages in the Mackay region and those elsewhere (Figure 1.17 a, b). As we discussed in the hard coral results section, the separation of assemblages in the Mackay Whitsunday Region is almost an exact proxy for a separation of assemblages based on levels of the variable ‘clay-silt’ content of the sediments. The predictive power of the trees was greater for the juvenile densities (cross validation error (0.72) than cover based estimates of assemblage composition (0.91), indicating less “noise” in the relationship between juvenile density and the environment.

0 20 40 60 80 100

-3

-2

-1

0

1

Soft coral cover

Par

tial e

ffect

of p

artic

ulat

e

a

0 20 40 60 80 100-1.0

-0.5

0.0

0.5

Soft coral richness

Particulate

b

0 20 40 60 80 100-3

-2

-1

0

1

Soft coral juvenile density

Particulate

Par

tial E

ffect

of P

artic

ulat

e

c

Page 41: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

35

Soft coral juvenile densityError : 0.497 CV Error : 0.792

< 68.58

>14.3914

RegionWet Tropics, Burdekin, Fitzroy

Region

Dissolved Particulate

Particulate

24

224 2

>11.56

Wet Tropics Burdekin

Soft coral coverError : 0.516 CV Error : 0.91

Wet Tropics Burdekin

Particulate>11.56

>14.39

Particulate

RegionWet Tropics, Burdekin, Fitzroy

< 68.58

Dissolved

> 2.63

Clay-silt Region

22 2

2

24

2

14

Mackay

<14.39

<11.56> 68.58

< 2.63

Mackay

<14.39

<11.56> 68.58

b

a

Soft coral juvenile densityError : 0.497 CV Error : 0.792

< 68.58

>14.3914

RegionWet Tropics, Burdekin, Fitzroy

Region

Dissolved Particulate

Particulate

24

224 2

>11.56

Wet Tropics Burdekin

Soft coral coverError : 0.516 CV Error : 0.91

Soft coral coverError : 0.516 CV Error : 0.91

Wet Tropics Burdekin

Particulate>11.56

>14.39

Particulate

RegionWet Tropics, Burdekin, Fitzroy

< 68.58

Dissolved

> 2.63

Clay-silt Region

22 2

2

24

2

14

Mackay

<14.39

<11.56> 68.58

< 2.63

Mackay

<14.39

<11.56> 68.58

b

a

Figure 1.17 Multivariate regression trees for adult cover of soft coral genera (a) and density of soft coral genera juveniles (b). The number at the terminal branch of the tree is the number of reefs where a particular assemblages type occurred (two depths at each of the 24 survey reefs, total N=48). Following the separation of the Mackay assemblages, the next split separated both the adult and juvenile assemblages on four reefs with low levels of ‘particulate’ water quality (Fitzroy Island and the two reefs in the Franklands Group in the Wet Tropics Region and Pelorus Island in the Burdekin Region, Figure 1.17). High levels of ‘dissolved’ water quality then separated Pelican Island in the Fitzroy region from all remaining assemblages. On this reef, the cover of the heterotrophic soft coral group ‘gorgonian’ is 1.8% at 5m depth, which is more than five times higher than at any other reef

Page 42: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

36

included in this study. And lastly, the variable ‘clay-silt’ separated the adult soft coral at Barren Island in the Fitzroy Region from the remaining reefs (Figure 1.17a). The cover of Xenia at Barren Island is 19.2% and 3.2% at 2m and 5m depth, respectively, while this genus only reaches a maximum of 1% cover at any other reef in this analysis. The responses of individual genera to environmental variables were investigated with redundancy analyses. The three environmental variables explained 16.9% (adult soft coral cover) and 17.2% (juvenile soft coral density) of the total variation in assemblage composition observed on the reefs included in this study, mainly driven by the ‘particulate’ water quality index and/or ‘clay silt’ content of the sediment (Appendix 3: Tables A3-3b and A3-4b, excluding the variable ‘Region’ because of the confounding between regions and environmental variables). The ‘dissolved’ water quality index showed no significant relationships to the soft coral assemblages in any of the models. Earlier we showed the negative relationship between total adult soft coral cover and richness and the environmental variable ‘particulate’ water quality (Figure 1.16a, b). This was mainly due to the genera Sinularia, Briareum, Nephthea, Lobophytum, Clavularia and Rhytisma showing a strong negative relationship to the level of ‘particulate’ (Figure 1.18a). All of these genera are phototrophic (van Oppen et al. 2005, Fabricius and De’ath 2008), and show high abundances on some of the reefs with low levels of ‘particulate’ water quality. Also noteworthy are the strong positive relationships between the phototrophic genera Klyxum and Sarcophyton and the proportion of fine grained particles in the substrate (‘clay silt’ variable). For these genera, the level of turbidity seems more limiting than sedimentation; however, it is possible that these two variables interact. A comprehensive assessment of interactions between environmental variables is beyond the limits of the current data set, as a much larger number of reefs would be required to generate sufficient combinations and levels of environmental variables. The density of juvenile colonies showed similar relationships to the environmental variables (Figure 1.18b), indicating that abundance and representation of adult genera were at least in part dependent on recruitment patterns. Our surveys of juvenile density did not record most ‘gorgonian’ heterotrophic genera, however, their abundance was low on most reefs (A. Thompson, pers. obs.). The gorgonian group was only a minor component also of the adult soft coral assemblages on most reefs. The exception to this was Pelican Island where gorgonians reached a mean cover of 2%; this reef had the highest level of ‘particulate’ water quality and the least pronounced reef development of all 24 study reefs. It is possible that the chronically high concentrations of particulate water quality parameters at Pelican Island exceed levels conducive to sustained reef development (van Woesik et al. 1999, Cooper et al. 2007).

It is noteworthy that, in general, heterotrophic soft corals were not abundant on the study reefs and assemblages do not seem to shift towards heterotrophic taxa with increasing turbidity and ‘particulate’ water quality as shown by Fabricius and McCorry (2006) for reefs in Hong Kong, but rather have less total soft coral cover. This suggests that shallow (2m and 5m) turbid inshore reefs on the GBR may not be suitable habitats for heterotrophic soft corals, which are generally more abundant and diverse in deeper waters on inshore reefs of the GBR, especially in the North (Fabricius and De’ath 2008).

Page 43: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

37

RDA1 58.5 %

RD

A2

25

.8 %

RDA1 58.5 %

RD

A2

25

.8 %

CapnellaSansibia

Gorgonian

Nephthea

Rhystima

Erythropodium

Briareum

Clavularia

Lobophytum

Sarcophyton

Klyxum

Sinularia

Particulate

Dissolved

Clay-silt

RDA1 63.3 %

RD

A2

30

.2 %

RDA1 63.3 %

RD

A2

30

.2 %

Capnella

Scleronephthya

Rhytisma

Cespitularia

Sansibia

Klyxum

Lobophytum

Briareum

Clavularia

Sinularia

Particulate

Dissolved

Clay-silt

Sarcophyton

Figure 1.18 Redundancy analysis bi-plot of the relationship between cover (a) and juvenile densities (b) of soft corals and environmental variables. The plot is the ordination of the assemblage space constrained to the variation described by the three environmental variables represented by heavy black arrows and bold type. The relationship between this constrained space and each genus are represented by the length and direction of the genus vectors (thin grey lines). For clarity only vectors for the 60% of genera relating most to the environmental variables, of which the top 40% are named are presented. Grey dots plot the position of reef assemblages relative to the environmental variables.

b

a

Page 44: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

38

Macroalgae

This section examines the effect of regional location, depth and three environmental variables on the distribution of macroalgae. For this report we focused on macroalgal taxa with upright fleshy thalli, and did not include turf algae or crustose coralline algae. These upright taxa could be reliably identified from photo transects and are considered to be the main competitors with other phototrophic benthic groups, such as hard and soft corals (Fabricius 2005, Birrell et al. 2008). Photo transects returned cover estimates for 13 taxonomic units, 12 of these are actual macroalgal genera while one is a collective taxon we termed ‘calcareous red macroalgae’, with genera that could not be reliably differentiated from photographs. However, for simplicity we refer to these taxonomic units as ‘genera’. Other genera, mainly fleshy red algae, which could not be reliably identified from photographs were excluded from the analysis, e.g. the genus Laurencia that certainly occurs on some study reefs but is morphologically similar to other genera. However, at any given reef the cover estimates for the thirteen genera analysed captured the majority of the total cover of large fleshy macroalgae. Large macroalgae are often fouled by epiphytic algal assemblages, and cover might be underestimated, especially for genera such as Hypnea and Dictyota with relatively small thalli. Regional and depth-related variation in macroalgal assemblages

Macroalgal cover and richness of genera both varied significantly among regions (P<0.05, Appendix 2: Tables A2-9 and A2-10). Regional contrasts indicate that both the cover and richness of macroalgae were significantly lower on reefs in the Mackay Whitsunday Region compared to the other three regions (Figure 1.19a, b, Table 1.5). In addition, the cover on reefs in the Wet Tropics Region was lower than on reefs in either the Burdekin or Fitzroy regions (Figure 1.19a).

Figure 1.19 Term plots for the factor region from generalised linear models for macroalgal cover (a) and richness of genera (b). Term plots present the fit of the term, in this case region, to the residuals resulting from the fitting of all other terms in the model to the response variable (assemblage attribute). As such the y-axis values represent the deviation from the overall intercept of the model rather than mean values observed for levels of factor variable.

-3-2-10123

Macro algae cover

Part

ial e

ffect

of r

egio

n

Wet Bur Mac Fit

a

-1.5

-1.0

-0.5

0.0

0.5

1.0

Macro algae richness

Wet Bur Mac Fit

b

Page 45: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

39

Table 1.5 Proportional difference of macroalgae cover and richness on reefs in the Mackay Region compared to each other three regions. Tabulated values are how much higher estimates of mean cover and richness were in each region compared to those estimated for the Mackay region. For example, the average cover in the Burdekin region was 69.1 times higher than in the Mackay region. Macroalgal assemblage attribute Wet Tropics Region Burdekin Region Fitzroy Region Cover 616% 6910% 4120% Richness 220% 445% 217% As before for hard and soft coral assemblages, we assessed whether the observed regional differences in macroalgal assemblages were independent of the effects of environmental variables by applying models that fit the environmental terms sequentially prior to inclusion of factors for region and depth. A confounding of regional variation with environmental variables is particularly likely in the case of macroalgae, because the main regional difference was the significantly lower macroalgal cover and richness in the Mackay Whitsunday Region. This region has a much higher proportion of clay-silt in the reefal sediments compared to other regions (see Figure 1.7c). For both macroalgal cover and richness, the previously recognised regional differences were confirmed after removal of variation attributable to the environmental variables, suggesting that the regional differences in algal assemblages exist over and above the influence of the local environmental setting (Appendix 2: Table A2-11). The regional differences observed in overall cover and richness of macroalgae were also reflected in the marked differences in assemblage composition between the regions (redundancy analysis, P<0.05, Appendix 3: Table A3-5). However, not only was the assemblage composition in the Mackay Whitsunday Region different, but also the assemblages of other three regions were clearly separated from each other (Figure 1.20). In total, the factor Regions explained 31.4% of the overall variation in macroalgal assemblages on the reefs included in this study. The inshore reefs of the Mackay Whitsunday Region haven’t had an acute disturbance for many years now; it would be interesting to see how the macroalgal cover would change if an acute disturbance opened up substratum for algal colonisation (McCook et al. 2001, Done et al. 2007), The strong regional differences in macroalgae assemblage composition and the observation of large-scale patches of similar assemblages on individual reefs are interesting and have not been described before. The reasons for these patterns are currently unknown. It is possible that mechanisms such as limited propagule dispersal and vegetative reproduction may explain the higher similarity between nearby compared to more distant macroalgal assemblages. Macroalgal propagules primarily settle within metres of the parent plant but may be dispersed over several km by currents (Norton 1992; Kendrick and Walker 1995, Stiger and Payri 1999) but there is no quantitative information about dispersal in coral reef macroalgae. Banks of propagules, microscopic recruits or tissue remnants (as present in temperate macroalgal assemblages; Santelices et al. 1995, Lotze et al. 2000) on substratum not covered by live corals could also facilitate sustained re-growth of the same assemblages, but the importance of these ‘marine seed banks’ in coral reef systems are, again, unknown.

Page 46: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

40

-1.0 -0.5 0.0 0.5

-1.0

-0.5

0.0

0.5

g

RDA1 57.2 %

RD

A2

32.

3 %

-1.0 -0.5 0.0 0.5

-1.0

-0.5

0.0

0.5

g

RDA1 57.2 %

RD

A2

32.

3 %

Particulate

Dissolved

Clay-silt

Wet Tropics

Burdekin

Fitzroy

Mackay

Figure 1.20 Redundancy analysis biplot for the cover of macroalgal genera. Represented is the ordination of the assemblages constrained to the five spatial and environmental variables ‘Region’, ‘Depth’, ‘Particulate’ and ‘Dissolved’ water quality, and ‘Clay-silt’ content of the sediments. The proportions of the variance explained by the first and second redundancy analysis axes are presented. The five variables account for 31.4% of the variation in macroalgal assemblages. The cover of macroalgae was significantly lower at 5m than at 2m, when pooled over all reefs (P=0.041, Appendix 2: Table A2-9). However, this difference was mainly due to the genera Hypnea and Sargassum, which showed generally higher cover at 2m, indicated by examination of the 95% confidence intervals of the proportion of cover found at 5m at each reef at which an algal genus occurred (data not shown). None of the genera included in the analysis had higher cover at 5m. The confidence intervals are, however, relatively broad due to the inclusion of very low covers for most genera at some reefs resulting in large and variable proportional differences among depths. Relationships between macroalgal assemblages and environmental variables

In this analysis, total macroalgal cover and richness were assessed for their relationship with the three selected environmental variables. The cover and richness of macroalgae increased non-linearly with increasing levels of the ‘particulate’ water quality index, then decreased again at high levels, while the cover also showed a positive relationship with the level of ‘clay silt’ content of the sediment (Figure 1.21; P<0.05, Appendix 2: Table A2-9, the response of macroalgal richness is only marginally significant, P= 0.052). We suggest that the positive relationship between ‘clay silt’ and the cover of macroalgae is an artefact of the confounding between the regional effects and the effect of the ‘clay silt’ content. The cover of macroalgae is lowest in the Mackay Whitsunday Region Mackay region, where the ‘clay silt’ content in the sediment is highest. While macroalgae are phototrophic organisms and need sufficient light for photosynthesis, their nutrient demand is generally covered by dissolved nutrients. However, it has been shown that some Sargassum species can utilise particulate nutrients from organic matter deposited on their thalli (Schaffelke 1999), which might explain the positive response of macroalgal cover and richness to

Page 47: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

41

relatively high levels of particulates in the water, before decreasing again when higher turbidity overrides the positive effect of particles as a nutrient source. The genus Sargassum is a dominant component of many GBR inshore reefs, both in terms of biomass and of cover (Schaffelke and Klumpp 1997).

0 20 40 60 80 100

-4-3-2-1012

Macro algae cover

Particulate

Par

tial e

ffect

of p

artic

ulat

e

a

0 20 40 60 80 100

-1.0

-0.5

0.0

0.5

Macro algae richness

Particulate

b

0 10 20 30 40 50 60

-2

-1

0

1

2

3

4Macro algae cover

Clay-silt

Parti

al e

ffect

of c

lay-

silt c

Figure 1.21 Term plots of significant relationships between macroalgae cover and environmental variables as indicated by generalised linear models. Plotted are the fit of the environmental variable indicated on the x-axis to the residuals from the model after fitting of the remaining explanatory variables. The relationships between macroalgal assemblage composition and the spatial and environmental variables were investigated with multivariate regression trees. The greatest difference in macroalgal assemblage composition was associated with the level of the ‘particulate’ water quality index (Figure 1.22). Reefs with relatively low levels of ‘particulate’ all had macroalgal assemblages that were characterised by low cover of any genus. The highest total macroalgal cover in this group was 4.3% at Barren Island at 5m depth. This assemblage type was found on reefs in all four regions, but predominantly on reefs in the Wet Tropics and Mackay Whitsunday regions. The assemblages on reefs with higher levels of ‘particulate’ water quality, and hence higher turbidity, were further separated based on the level of the ‘dissolved’ water quality index (Figure 1.22). Low levels of the variable ‘dissolved’ separated two adjacent reefs in the Burdekin Region, Havannah Island and Pandora Reef, from the other turbid water reefs. Algal assemblages on these reefs had very high total macroalgal cover (>30%), dominated by the genera Lobophora and Sargassum. Finally, the reefs with high ‘particulate’ and ‘dissolved’ water quality were distinguished by region. The Burdekin and Fitzroy regions had reefs with generally high macroalgal cover, with a high proportion of the genus

Page 48: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

42

Lobophora. The reefs in the Wet Tropics and Mackay Whitsunday regions had generally lower macroalgal cover (<14%) with a variety of genera.

14

4

20

10

Region

Dissolved

< 17> 17

Particulate> 29< 29

Wet Tropics, Mackay Burdekin, Fitzroy

Error : 0.399 CV Error : 0.788

14

4

20

10

Region

Dissolved

< 17> 17

Particulate> 29< 29

Wet Tropics, Mackay Burdekin, Fitzroy

Error : 0.399 CV Error : 0.788

Figure 1.22 Multivariate regression tree for cover of macroalgal genera. The number at the terminal branch of the tree is the number of reefs where a particular assemblages type occurred (two depths at each of the 24 survey reefs, total N=48). The responses of individual genera to environmental variables were investigated with redundancy analyses. The three environmental variables explained 15.8% of the total variation in macroalgal assemblages on the study reefs, mainly driven by ‘particulate’ water quality (Appendix 3: Table A3-5b, excluding the variable ‘Region’ from the model because of a confounding between regions and environmental variables). The ‘clay silt’ content of the sediment and ‘dissolved’ water quality did not relate significantly to the algal assemblages. Total macroalgal cover showed a relationship with ‘particulate’ water quality that approximated an optimum curve with highest values at medium levels of ‘particulate’ (Figure 1.21a). The importance of the variable ‘particulate’ is supported here, as it is the main component of the first dimension in the redundancy analysis, which accounts for 85% of the variation in algal assemblages (Figure 1.23). Especially the genera Sargassum and Dictyota clearly show higher cover at higher levels of ‘particulate’ and are likely to be the main drivers of the overall relationship of total macroalgal cover with ‘particulate’, as these are very abundant genera on GBR inshore reefs (Schaffelke and Klumpp 1997). Several genera (especially the brown algae Lobophora, Padina and Colpomenia) showed a negative relationship to ‘clay silt’ content of the sediment. The reason for this is unclear but may be due these genera being less tolerant to sedimentation as they have a fan or bubble-shaped morphology and are anchored close to the substratum.

Page 49: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

43

Macroalgal assemblages were not different between the two depths, even though total macroalgal cover was significantly different (see above). Depth explained only 1.7% of the total variation in the assemblages. Algal assemblages at 5m depth have low cover but are mostly composed of the same suite of genera as shallow water assemblages. g

RDA1 85.5 %

RD

A2

13

.5 %

g

RDA1 85.5 %

RD

A2

13

.5 %

Peyssonnelia

Halimeda

Asparagopsis

Calcareous.red.macroalgae

Colpomenia

Padina

Lobophora

Sargassum

Particulate

Dissolved

Clay-silt

Dictyota

Figure 1.23 Redundancy analysis bi-plot of the relationship between the cover of macroalgae genera and environmental variables. The plot is the ordination of the assemblage space constrained to the variation described by the three environmental variables represented by heavy black arrows and bold type. The relationship between this constrained space and each genus are represented by the length and direction of the genus vectors (thin grey lines). For clarity only vectors for the 80% of genera relating most to the environmental variables are labelled. Grey dots plot the position of reef assemblages relative to the environmental variables.

Page 50: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

44

2. Assessment of the utility of autonomous logging instruments for water quality monitoring

2.1 Introduction

The biological productivity of the Great Barrier Reef is supported by nutrients (e.g. nitrogen, phosphorus, silicate, iron), which are supplied by a number of processes and sources (Furnas et al. 1997; Furnas 2003). These include upwelling of nutrient-enriched subsurface water from the Coral Sea, rainwater, fixation of gaseous nitrogen by cyanobacteria and freshwater runoff from the adjacent catchment. Land runoff is the largest source of new nutrients to the Reef (Furnas 2003). However, most of the inorganic nutrients used by marine plants and bacteria on a day-to-day basis come from recycling of nutrients already within the Great Barrier Reef ecosystem (Furnas et al. 2005). Extensive water sampling throughout the Great Barrier Reef over the last 25 years has established the typical concentration range of nutrients, chlorophyll a and other water quality parameters and the occurrence of persistent latitudinal, cross-shelf and seasonal variations in these concentrations (summarised in Furnas 2005, De’ath and Fabricius 2008). While concentrations of most nutrients, suspended particles and chlorophyll a are normally low, water quality conditions can change abruptly and nutrient levels increase dramatically for short periods following disturbance events (wind-driven re-suspension, cyclonic mixing, river flood plumes). However, nutrients introduced, released or mineralised into Great Barrier Reef lagoon waters during these events are generally rapidly taken up by pelagic and benthic algae and microbial assemblages (Alongi and McKinnon 2005), sometimes fuelling short-lived phytoplankton blooms and high levels of organic production (Furnas et al. 2005). These short-term events (flood plumes, resuspension) are recognised as driving factors for the resilience of coastal coral assemblages (Fabricius 2005). The current design of the lagoon water quality monitoring under the Reef Plan MMP is based upon manual, ship-based sampling three times a year (from 2008/09 onwards, semi-annually in previous years). Such a sampling approach is unsuitable to resolve the frequency and magnitude of such short-term events and only useful to discern large-scale (e.g., De’ath and Fabricius 2008) and/or long-term patterns. For example, the longest and most detailed time series of a suite of water quality parameters in the Great Barrier Reef has been sampled by AIMS at 11 coastal stations between Cape Tribulation and Cairns since 1989 (2-4 times per year) and has been continued under Reef Plan MMP. In this long-term time series, all parameters (dissolved organic and particulate nitrogen and phosphorus, respectively, suspended solids), except chlorophyll a, showed significant long-term patterns, with generally decreasing values since the early 2000s (Schaffelke et al. 2008). Water quality monitoring using autonomous instruments, which was fully implemented as a routine component of Reef Plan MMP in 2007/08, has immensely improved our capacity to measure key water quality parameters in close proximity (less than ca. 1 m) to corals and benthic assemblages on coastal reefs and to record short-term variability in water quality associated with flood plumes and wind-driven resuspension events, albeit at only a small number of locations (14 reefs).

Page 51: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

45

In this report we assess the utility of the instrumental measurement of water quality parameters (focusing on chlorophyll) in comparison to the ‘traditional’ direct water sampling undertaken so far under the Reef Plan MMP. This will inform the continued improvement and rationalisation of the Reef Plan MMP design and procedures.

2.2 Methods

[Note: For more detailed methods refer to CRC Reef Consortium 2005 and Schaffelke et al. 2008]

WATER QUALITY INSTRUMENT DEPLOYMENTS

The Eco FLNTUSB Combination instruments perform simultaneous in situ measurements of chlorophyll fluorescence, turbidity and temperature. The fluorometer monitors chlorophyll concentration by directly measuring the amount of chlorophyll a fluorescence emission, using blue LEDs (centred at 455 nm and modulated at 1 kHz) as the excitation source. A blue interference filter is used to reject the small amount of red light emitted by the LEDs. The blue light from the sources enters the water at an angle of approximately 55–60 degrees with respect to the end face of the unit. The red fluorescence (683 nm) emitted is detected by a silicon photodiode positioned where the acceptance angle forms a 140-degree intersection with the source beam. A red interference filter discriminates against the scattered blue excitation light. Turbidity is measured simultaneously by detecting the scattered light from a red (700 nm) LED at 140 degrees to the same detector used for fluorescence. In field deployments, the instruments were used in ‘logging’ mode and recorded a data point every 10 minutes for each of the three parameters, which was a mean of 50 instantaneous readings. The instruments were deployed at all 14 water quality monitoring sites (Table 2.1, Figure 2.2) from October 2007, as close as possible to the inshore reef surveys sites at 5 m depth (LAT), and changed-over every 3-5 months. To decrease fouling of the instrument, the loggers were wrapped with plastic and electrical tape and the lower part of the instrument additionally with copper tape. Underwater, the instruments were attached with a custom-made clamp to a star picket with the measurement window pointing downward (Figure 2.1).

Figure 2.1 FLNTUSB logger deployed at Pelican Island in the Fitzroy NRM Region in October 2007. Note the co-deployed passive sampler for pesticide monitoring.

Page 52: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

46

Table 2.1 Locations selected for inshore water quality monitoring by autonomous instruments (Wetlabs FLNTUSB) and deployment and change-over times. Data from locations highlighted by grey shading were included in the present analyses.

NRM Region FLNTUSB deployment locations

Wet Tropics Snapper Island North Fitzroy Island West High Island West Frankland Group West Dunk Island North Burdekin Pelorus & Orpheus Is West Pandora Reef Geoffrey Bay Mackay Whitsunday Double Cone Island Daydream Island Pine Island Fitzroy Barren Island Pelican Island Humpy & Halfway Island

Figure 2.2 Fourteen sampling locations (red dots) where Eco FLNTUSB Combination instruments were deployed and regular direct water sampling was undertaken. See Table 2.1 for location names.

Page 53: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

47

In the following analyses, instrument data for chlorophyll a concentrations were compared to chlorophyll a concentration data, obtained by direct water sampling. Instrument data included in the analyses were from 11 of the 14 locations (see Table 2.1), selecting those time series that had no instrument failures that led to long periods of data losses.

DIRECT WATER SAMPLING FOR INSTRUMENT VALIDATION

Water samples were collected at each instrument deployment/ retrieval subtidally by diver-operated Niskin bottle sampling close to the autonomous water quality instruments. These samples were processed and analysed for chlorophyll and suspended solids concentrations using standard laboratory techniques (see Schaffelke et al. 2008 for details). Chlorophyll results collected from 11 locations (Table 2.1) on between 2 and 4 occasions were included in the analyses. On each occasion, 2 samples each with 2 replicates were analysed. There were a few missing replicates in the dataset.

LABORATORY COMPARISON OF INSTRUMENTS

Measurements of chlorophyll a concentrations in the laboratory were used to compare individual Eco FLNTUSB instruments. These were carried out in a calibration chamber made of black PVC pipe, placed on a stirrer. The chamber was filed with 3.3L of filtered seawater. Each instrument in turn was immersed in the seawater and a series of measurements (every 10 seconds, 5 samples averaged per measurement) was taken by each instrument over a period of about five minutes. Then, a volume of a pure phytoplankton culture in growing phase, usually of Isochrysis sp. (a prymnesiophyte widely used as a larval feed in the aquaculture industry), was added to the filtered sea water as a chlorophyll source. The plankton was kept in suspension and uniformly mixed by the stirrer. The resulting chlorophyll concentration in the chamber was again measured over a period of about five minutes, by one instrument after another. Then, more plankton culture was added and the chlorophyll concentration measured successively by each of the eight loggers to create a range of increasing chlorophyll concentrations (in the data set used for this report 0, 3, 8 and 20 ml of Isochrysis culture were added for the testing of eight individual instruments).

Figure 2.3 Calibration chamber set up used for inter-instrument comparisons of Eco FLNTUSB water quality loggers. A clear chamber is shown for illustration purposes, actual measurements were carried out in a black PVC chamber to reduce potential interference of ambient light

Page 54: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

48

DATA ANALYSIS

To assess differences between locations and dates within locations were analysed using generalized linear models with a log link and standard deviation proportional to the mean. The latter results in the standard errors of estimates that are proportional to the estimate. Assumptions of the model were verified using standard tests (Appendix 4). Chlorophyll a data from direct water sampling were averaged across duplicates and tests were based on F-ratios using the variation of samples within sampling times as the measure of error. Two data points (replicates of a sample) were excluded since their variation was abnormally large. Chlorophyll data obtained by ECO FLNTUSB loggers were analysed at two levels: (a) the original 10-minute samples and (b) the daily averages of the 10 minute samples. Chlorophyll concentrations obtained from direct water sampling were compared to concentrations obtained from instrumental records. Two comparisons were carried out: • Direct water samples vs. data from instruments, averaged over 3 h around the time the direct

water samples were taken; • Direct water samples vs. data from instruments, averaged over the day the direct water samples

were taken. The period of 3 h was selected to maximise the number of readings that could be compared with the instantaneous values from the direct water sampling without running the risk of using variable records if the time period was selected as too short. After deployment the instrument is generally heated up by in-air transport exposed to the weather and the first six records (1 h) are usually discarded. For comparisons between individual loggers, linear regression models were used to select a best model for the chlorophyll measurements obtained under laboratory conditions. Three models were selected: • Model 1: quadratic response curves and intercepts for each logger

[count ~ poly(dose, 2) * logger] • Model 2: linear responses and intercepts for each logger [count ~ dose * logger] • Model 3: linear responses for each logger and a single intercept [count ~ dose:logger] All analyses were conducted using the package R (R Development Core Team, 2008).

Page 55: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

49

2.3 Results and Discussion

DIRECT WATER SAMPLING

There were strong significant differences between locations and between times within locations (Table 2.2). The precision of estimated chlorophyll values for each time at each location was 10.3%. Thus for the mean observed chlorophyll of 0.370 µgL-1 (95% CI = (0.335, 0.408) and the standard error was 5.15%. Table 2.2 Analysis of deviance for the effects of locations and days nested in locations for chlorophyll shows strong variation due to both locations and days.

Effect df Deviance F P Locations 10 13.79 259.8 <0.001

Locations/Day 33 17.02 97.4 <0.001

Error 44 0.235

INSTRUMENTAL WATER SAMPLING

The times series obtained using FLNTU loggers delivered high-frequency, location-specific data records. As an example the 12 month long time series of daily means of chlorophyll a and turbidity from Dunk Island is given in Figure 2.4. In the following only the chlorophyll data are analysed because results derived from instruments and from direct water sampling can be directly compared, whereas the optical measurement of turbidity and the direct measurement of total suspended solids (from filters) measure related, but different parameters and their relationship can change depending on the nature of the suspended solids at each location.

Oct-07 Apr-08 Jul-08 Oct-08 Jan-08

Chl

orop

hyll

(ug

L-1)

0.01

0.1

1

Turb

idity

(NTU

)

0.1

1

10

Chlorophyll (ug L-1)Turbidity (NTU)

Figure 2.4 Example of a time series of chlorophyll (µg L-1, green line) and turbidity (NTU, black line) from field deployments of WET Labs Eco FLNTU Combination Fluorometer and Turbidity Sensors at Dunk Island in the Wet Tropics NRM Region over a 12 month period (October 2007 to October 2008). Missing data in March 2008 were due to a technical problem with the instrument.

Page 56: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

50

Figure 2.5 Randomly chosen daily profiles (n = 18) of 10-minute spaced measurements (144 per day) by WET Labs Eco FLNTUSB Combination Fluorometer and Turbidity Sensors. Scales are on log base two, and typical ranges cover a ratio of approximately 2.5 : 1 and the average standard deviation is ~ 30% over one day.

Page 57: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

51

Figure 2.6 Continuous time series plots of chlorophyll concentrations showing highly variable patterns of long and short-term variation across 10 locations.

Page 58: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

52

Figure 2.6 continued Continuous time series plots of chlorophyll concentrations showing highly variable patterns of long and short-term variation across 4 locations.

Page 59: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

53

The daily averages over longer time periods (month to years) give a good representation of mean values and event-driven changes in chlorophyll a concentrations (Figures 2.4 and 2.6). However, inspection of the data original instrument data, i.e. from readings obtained every 10 minutes, indicates very high variability of chlorophyll concentrations throughout a single 24 h period (Figures 2.5 and 2.7). This high variability has not previously been described. Past sampling was generally based on collection of direct water samples that only represent snapshots in time and space and could not resolve the high temporal variability. The causes of these major fluctuations in chlorophyll a concentrations over a 24 period are currently not resolved; influences of the tidal cycle and diurnal differences in plankton density are all likely explanations that need to be explored in the future.

Figure 2.7 Boxplots of means and SDs of randomly chosen daily profiles (n = 24) of 10-minute spaced samples (144 per day) show considerable variation. Scales are on log base two, and typically means ranged from 0.2 to 1.2, with an overall mean of 0.55. The average standard deviation of observation within a day ~ 0.18 (log base 2) equivalent to a SD of ~13% of the mean.

Page 60: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

54

COMPARISON OF CHLOROPHYLL A VALUES FROM DIRECT WATER SAMPLING AND INSTRUMENTAL SAMPLING

Direct water samples were collected and analysed for chlorophyll concentrations to compare these with instrument data acquired at the time of manual sampling (Figure 2.8). The match-up of these data showed good agreement in most instances for the comparison of direct water sampling data with instrument readings averaged over the three hours around the time of direct water sampling (Figure 2.9). There were significant differences in chlorophyll concentrations between locations, which could be excepted as the instruments are deployed on reefs with a variety of environmental and water quality conditions (Table 2.3). However, chlorophyll concentrations obtained from direct water samples were not significantly different to concentrations obtained from instrument readings (Table 2.3). The instruments gave values that were on average 5.9% (SE=11.8%) lower than those from direct water sampling. The second analysis compared chlorophyll concentrations from instrument data averaged over a day and the respective data from direct water sampling on the same day. The match-up of these data showed less agreement (Figure 2.10) than the analysis above using instrument data averaged over there hours. There were strong effects of locations and days within locations on direct water sampling and logger-derived chlorophyll a values (Table 2.4). This analysis showed a significant difference between estimates of chlorophyll from direct water samples and the daily averages from instrument readings (Table 2.4). The instruments gave values that were on average 16.4% (SE=5.5%) higher than those from direct water sampling. We included this second analysis to illustrate that the direct water sampling results are just representative of the actual time they were collected (see also Figures 2.5 and 2.7 showing the high variability of chlorophyll values throughout a day). However, in the literature limited numbers of direct water sampling results are sometimes used to represent conditions for a month or a season, which may be misleading. Table 2.3 Analysis of deviance for the effects of locations and sampling method (direct water sampling vs. instrument) on chlorophyll a. Instrument data were averaged over 3h around the time direct water samples were taken.

Effect df Deviance F P Locations 10 4.33 5.50 <0.001

Direct/Instrument 1 0.024 0.302 0.585

Error 42 4.06

Table 2.4 Analysis of deviance for the effects of locations, days nested in locations and sampling method (method (direct water sampling vs. instrument) on chlorophyll a. Instrument data were daily averages of the day direct water samples were taken.

Effect df Deviance F P Locations 10 10.99 17.6 <0.001

Locations/Days 21 17.22 13.1 <0.001

Direct/Instrument 1 0.48 7.7 0.007

Error 63 4.06

Page 61: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

55

Figure 2.8 Chlorophyll a levels (µgL-1) from direct water sampling (red points) and Eco FLNTUSB Combination Fluorometer and Turbidity Sensors (continuous trace) from 11 locations across varying periods of time.

Page 62: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

56

Figure 2.9 Comparison of chlorophyll concentrations obtained from direct water sampling (red points) and Eco FLNTUSB instruments (blue points) from 11 locations and 1-4 times. Data from instruments were averaged over three hours around the time the direct water samples were taken.

Page 63: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

57

Figure 2.10 Comparison of chlorophyll concentrations obtained from direct water sampling (red points) and Eco FLNTUSB instruments (blue points) from 11 locations and 4-5 times. Data from instruments were averaged over the day (24 h) when the direct water samples were taken.

Page 64: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

58

The Eco FLNTUSB instruments returned reliable chlorophyll data compared to traditional direct sampling and analysis methods. Daily averages based on the high frequent instrument readings reflect the actual water column chlorophyll concentrations and their high inherent variability. While this was not modelled as part of this present analysis we argue that chlorophyll values from direct water sampling just represent the actual concentration at the time of sampling and are unlikely to be suited to predict or represent monthly, seasonal or annual chlorophyll concentrations. Future long-term monitoring of water quality should increasingly employ instrumental techniques such as ocean colour remote sensing and autonomous instruments deployed in situ to record high-frequency data series. While these instrument-based techniques are currently limited to chlorophyll a concentrations, various measures of turbidity and coloured dissolved organic matter, the former two of these parameters are known to sufficiently predict a range of other water quality constituents (see first part of this report, De’ath 2007, De’ath and Fabricius 2008). However, ongoing ‘snap-shot’ water sampling has to continue to provide data for high quality validation of the instrument and remote sensing data. The future of monitoring by direct manual sampling of those variables that cannot be measured by in situ instruments or satellites (e.g., organic matter, nutrients and pesticides) needs to be discussed in light of scientific value and operational cost.

COMPARISON BETWEEN INSTRUMENTS

Another important aspect to assess the utility of the Eco FLNTUSB instruments for ongoing water quality monitoring is the comparison of the measurements derived from individual instruments. The instrument responses (logger counts) were converted to chlorophyll units (µg L-1) and averaged for each combination of logger and dose, thus there were 32 data cases. The loggers responded linearly to the changes in added phytoplankton dose but the slopes varied substantially (Figure 2.11, Tables 2.5 and 2.6). The intercepts (response at dose ‘0’) were consistent, i.e. all linear responses go through the same origin (Figure 2.11). Table 2.5 Analysis of variance for the effects of Dose (volume of phytoplankton culture added), slopes of instrument response (Dose: Instrument), and the intercept of the instrument response (Instrument) on chlorophyll a. Df Sum Sq Mean Sq F value Pr (>F) Dose 1 96.194 96.194 33552.9077 <0.0001 Dose: instrument 7 3.151 0.450 157.0051 <0.0001 Instrument 7 0.009 0.001 0.44620.8585 0.8585 Residuals 16 0.046 0.003

Page 65: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

59

Plankton culture dose (ml)

Chl

orop

hyll

a (µ

g L-1

)

Plankton culture dose (ml)

Chl

orop

hyll

a (µ

g L-1

)

Plankton culture dose (ml)

Chl

orop

hyll

a (µ

g L-1

)

Plankton culture dose (ml)

Chl

orop

hyll

a (µ

g L-1

)

Figure 2.11 Relationships between the chlorophyll values from eight Eco FLNTUSB instruments and the phytoplankton culture dose added (a). Adequacy of the fits for the individual loggers (b).

a

b

Page 66: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

60

The best model for chlorophyll a responses from Eco FLNTUSB instruments has a single intercept and a different slope for each instrument (based on significance tests testing three models: quadratic response, linear response with varying intercepts, linear response with single intercept; data not shown). The single intercept for this dataset was 0.136; slopes for each instrument are listed in Table 2.6. Table 2.5 Slopes for linear responses of chlorophyll a measurements of eight Eco FLNTUSB instruments in a inter-calibration test.

Instrument number Slope 819 0.221 821 0.236 822 0.231 823 0.224 824 0.172 825 0.204 826 0.273 828 0.257

The significant variability between response slopes of individual instruments is an issue that could affect the reliability of field measurements as time series are generally created by sequential data records from different instruments (field instruments are exchanged every 4 to 5 months). Calculation of instrumental readings are converted into chlorophyll units (µg L-1) using calibration coefficients supplied by the manufacturer. Response of loggers has been tested against field samples (see above) and in laboratory calibration exercises (Schaffelke et al. 2007) and was generally satisfactory. Further testing needs to be carried out to see if individual instruments consistently respond differently, and whether drift or ageing of the sensors has occurred that might have affected the response. A number of pre- and post deployment check s have been routinely carried out for more than 12 months of deployments of Eco FLNTUSB instruments under Reef Plan MMP. So far, no significant drift of instruments has been observed. However, there is no reference material available that could be used for unequivocal and chlorophyll-specific checks of drift and sensor degeneration. Such a ‘solid standard’, for example, is used by AIMS to assure quality in laboratory based chlorophyll analysis that are also based on fluorometry. We will explore whether AIMS could develop such as reference for use with the Eco FLNTUSB instruments. The issue of inter-instrument variability will be also followed up with the manufacturer, WetLabs Inc., who are still actively improving and developing the Eco FLNTUSB instruments, which are a relatively new product. In the past, AIMS has been in regular and productive dialogue with the company, because we have been the only organisation testing the Eco FLNTUSB instruments in tropical waters. The manufacturer has been extremely interested in our experiences and our feedback about problems and the majority of the instruments have been upgraded and serviced since they were purchased in 2007 under Reef Plan MMP.

Page 67: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

61

3. Conclusions

The coral reef assemblages showed clear differences along latitudinal and environmental gradients. Hard coral assemblages in the Fitzroy region were clearly different from those in all other regions, whereas soft coral assemblages in the Mackay Whitsunday region were most distinct and the macroalgal assemblages were distinctly different in each region. The lower hard coral richness on Fitzroy region reefs is not unexpected given that these reefs are located at the southern end of the GBR. The GBR has a well-known gradient of declining richness towards the higher latitudes (Veron 2005, DeVantier et al. 2006). Mackay Whitsunday reefs, in contrast, were characterised by a high proportion of clay/silt sized sediments at the reefs, which is the likely cause of differences from the other three regions. Sedimentation and associated turbidity varies on a local reef scale, controlled by local hydrodynamics (wind, tides, and exposure) and the proximity to river mouths, which provide new suspended sediment and organic matter (Wolanski et al 2008). In contrast, water quality, especially the levels of dissolved nutrients, is more likely to act on a regional scale as the coastal and inshore water body is generally well-mixed but with clear dilution gradients away from river mouths and the mainland coast (Cooper et al. 2007, De’ath and Fabricius 2008, Schaffelke et al. 2008). Our study is the first in the GBR to compare biological and water quality data sets obtained from exactly the same sites. All assemblages in our study were influenced most strongly by the levels of the ‘particulate’ water quality index. Hard and soft corals generally showed lower cover and richness at high levels of ‘particulate’, while the cover and richness of macroalgae increased with increasing levels of the ‘particulate’ water quality index but decreased at high levels. A small number of taxa in all three benthic groups were apparently tolerant to high levels of the ‘particulate’ water quality index. The strong effect of particulate water quality is also responsible for clear differences between assemblages at the two depths, as higher particulate levels in the water lead to more rapid light attenuation with depth. Our results confirm the outcomes of a previous larger-scale study in the GBR, which found that richness of both hard and soft coral decreased with increasing turbidity and chlorophyll (both parameters are components of our ‘particulate’ water quality index), while the cover of macroalgae increased (De’ath and Fabricius 2008). Another previous analysis, however, showed only weak associations between hard coral and macroalgal cover and chlorophyll and water clarity (Delean and De’ath 2008). The interpretation of the observed assemblage patterns is difficult since there is extremely limited information about the biology of benthic hard and soft coral and macroalgal species. New fundamental research is required to fill this knowledge gap. Tolerance to low light reflects the capacity of some hard coral taxa to compensate for reductions in photosynthesis in low light conditions by feeding on suspended particles. Phototrophic soft corals are not known to switch between feeding modes and become less abundant in turbid conditions, while heterotrophic soft corals were rare on our study reefs. Though clearly light-dependent, some macroalgae (Sargassum species) can use organic matter deposited on their thalli as a nutrient source (Schaffelke 1999).

Page 68: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

62

We found clear evidence that juvenile hard and soft corals are more sensitive to high levels of the ‘particulate’ water quality index than are adult colonies. It has long been assumed that water quality is most likely to shape reef assemblages through affecting coral reproduction and recruitment. Elevated concentrations of nutrients, agrichemicals and suspended sediment can affect in some coral species gametogenesis, fertilisation, planulation, egg size and embryonic development and high levels of sedimentation can affect larval settlement or reduce net recruitment (Fabricius, 2005). Similar levels of these factors may have sub-lethal effects on established adult colonies. Because adult corals can tolerate poorer water quality than can recruits and coral colonies are potentially long-lived, reefs may retain high coral cover even under conditions of poor water quality, but have low resilience. Some high-cover coral assemblages may be relic assemblages formed by adult colonies that became established under more favourable conditions. Such relic assemblages would persist until a major disturbance affects them, but subsequent recovery will be slow if poor water quality results in reduced recruitment. This would lead to the long-term degradation of reefs, since extended recovery time increases the likelihood that further disturbances will occur before recovery is complete (McCook et al. 2001, Done et al. 2007). Additionally, the impact of disturbance events can vary greatly over small spatial scales (Cheal et al. 2002) and affect coral species differentially (Baird and Marshall 2002). The result is that the disturbance history of an assemblage can obscure relationships between environmental variables and coral communities. To date, reefs in the Fitzroy region have been resilient to disturbance, with hard coral cover recovering rapidly following past disturbance events (Sweatman et al. 2005, Diaz-Pulido et al. 2009). However, this rapid recovery has been mainly due to re-growth of surviving fragments of just a few species of Acropora. Our data indicate that the density of juvenile corals is very low on these reefs, even though larval settlement rates are high, which suggests limited potential for recovery from disturbances that kill whole colonies over large areas. We found clear differences in overall coral reef benthos assemblage characteristics (cover and richness) as well as taxonomic assemblage composition. However, our analysis was hampered by the limited availability of the directly samples environmental data. Water quality data were only available for three years, from two to three sampling occasions per year. High frequency time series for chlorophyll concentrations and turbidity levels from the Eco FLNTUSB instruments deployed at 14 reef locations under the Reef Plan MMP will be available in the near future and are likely to offer a much improved characterisation of the water quality setting at each of the survey reefs. Comparison of data from the Eco FLNTUSB instruments for water quality monitoring with traditional direct sampling and analysis methods showed that the instrument records gave reliable chlorophyll data. The frequent sampling (readings every10 minutes) showed that chlorophyll and turbidity values varied over each day, often doubling or halving in a 24h period. A comparison of the chlorophyll concentrations from traditional direct water sampling with daily means of the instrument readings showed an average difference of more than 16%, indicating that traditional ‘snap-shot’ water sampling does not provide a good estimate of the average water quality conditions. While some technical issues with the Eco FLNTUSB instruments still need to be resolved in collaboration with the manufacturer, we believe that these techniques will greatly improve the power of future analysis of long-term trends of both water quality and of coral reef benthos responses to water quality.

Page 69: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

63

The differences in assemblages in the current analysis provide a useful starting point for the detection of long-term trends in coral reef benthos in future monitoring under Reef Plan MMP. Our results indicate that the particulate components of water quality (suspended sediment and particulate nutrients and carbon) are the most important drivers of coral reef communities. If changes in land management practices in the GBR catchments under the Reef Plan lead to decreased levels of particulates in coastal and inshore waters of the GBR, we expect to be able to detect associated changes in coral reef assemblages, based on the relationships between assemblage composition and water and sediment quality described in this report.

Page 70: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

64

4. References

Alongi DM, McKinnon AD (2005) The cycling and fate of terrestrially-derived sediments and nutrients in the coastal zone of the Great Barrier Reef shelf. Marine Pollution Bulletin 51:239-252.

Anthony KRN (1999) Coral suspension feeding on fine particulate matter. Journal of Experimental Marine Biology and Ecology 232:85-106

Anthony KRN (2000) Enhanced particle-feeding capacity of corals on turbid reefs (Great Barrier Reef, Australia). Coral reefs 19:59-67

Anthony KRN (2006) Enhanced energy status of corals on coastal, high-turbidity reefs. Marine Ecology Progress Series 319:111-116

Anthony KRN, Connolly SR (2004) Environmental limits to growth: physiological niche boundaries of corals along turbidity-light gradients. Oecologia 141:373-384

Anthony KRN, Fabricius KE (2000) Shifting roles of heterotrophy and autotrophy in coral energetics under varying turbidity. Journal of Experimental Marine Biology and Ecology 252:221-253

Baird AH, Babcock RC and Mundy CP (2003) Habitat selection by larvae influences the depth distribution of six common coral species. Marine Ecology Progress Series 252:289-293

Baird AH, Marshall PA (2002) Mortality, growth and reproduction in scleractinian corals following bleaching on the Great Barrier Reef. Marine Ecology Progress Series 237:133-141.

Bellwood DR, Hughes TP, Folke C, Nyström M (2004) Confronting the coral reef crisis. Nature 429, 827-833.

Birrell CL, McCook LJ, Willis BL, Harrington L (2008) Chemical effects of macroalgae on larval settlement of the broadcast spawning coral Acropora millepora. Marine Ecology Progress Series 362:129-137

Brieman L, Friedman JH, Olshen RA, Stone CG. (1984) Classification and regression trees. Wadsworth International Group, Belmont, California, USA.

Brodie J, De’ath G, Devlin M, Furnas MJ, Wright M (2007) Spatial and temporal patterns of near-surface chlorophyll a in the Great Barrier Reef lagoon. Marine and Freshwater Research 58: 342-353.

Cantin NE, Negri AP, Willis BL (2007) Photoinhibition from chronic herbicide exposure reduces reproductive output of reef-building corals. Marine Ecology Progress Series 344: 81-93

Cheal AJ, Coleman G, Delean S, Miller I, Osborne K, Sweatman H (2002) Response of coral and fish assemblages to a severe but short-lived tropical cyclone on the Great Barrier Reef, Australia. Coral Reefs 21: 131-142

Cooper TF, Ridd PV, Ulstrup KE, Humphrey C, Slivkoff M, Fabricius KE (2008) Temporal dynamics in coral bioindicators for water quality on coastal coral reefs of the Great Barrier Reef. Marine and Freshwater Research 59: 703-716.

Cooper TF, Uthicke S, Humphrey C, Fabricius KE (2007) Gradients in water column nutrients, sediment parameters, irradiance and coral reef development in the Whitsunday Region, central Great Barrier Reef. Estuarine, Coastal and Shelf Science 74:458-470.

Page 71: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

65

CRC Reef Consortium (2005) Water Quality and Ecosystem Monitoring Programs - Reef Water Quality Protection Plan. Methods and Quality Assurance/Quality Control Procedures August 2005. An unpublished report to the Great Barrier Reef Marine Park Authority, CRC Reef Research, Townsville. 67 p. (Attachments 187 p.)

CRC Reef Consortium (2006) Water Quality and Ecosystem Monitoring Programs–Reef Water Quality Protection Plan. Final Report August 2006 (revised November 2006). Unpublished report to the Great Barrier Reef Marine Park Authority, CRC Reef Research, Townsville. 361 p. (Appendix 138 p.)

De’ath G (2002) Multivariate regression trees: a new technique for modelling genera-environment relationships.

De'ath G (2007) The spatial, temporal and structural composition of water quality of the Great Barrier Reef, and indicators of water quality risk mapping. Unpublished Report to the Reef and Rainforest Research Centre. Australian Institute of Marine Science, Townsville, 66 p.

De'ath G, Fabricius KE (2008) Water Quality of the Great Barrier Reef: Distributions, Effects on Reef Biota and Trigger Values for the Protection of Ecosystem Health. Research Publication No. 89. Great Barrier Marine Park Authority, Townsville, p. 104 p.

Delean S, De’ath G (2008) Spatial and temporal patterns of indicators of reef health on the Great Barrier Reef. Final report to the Reef and rainforest Research Centre. Australian Institute of Marine Science, Townsville. 107 p.

DeVantier L, De’ath G, Turak E, Done T, Fabricius K (2006) Species richness and assemblage structure of reef-building corals on the nearshore Great Barrier Reef. Coral Reefs 25: 329-340.

Diaz-Pulido G, McCook LJ, Dove S, Berkelmans R, Roff G, Kline DI, Weeks S, Evans RD, Williamson DH, Hoegh-Guldberg O (2009) Doom and Boom on a Resilient Reef: Climate Change, Algal Overgrowth and Coral Recovery. PLoS ONE 4:e5239.

Done T, Turak E, Wakeford M, DeVantier L, McDonald A and Fisk D (2007) Decadal changes in turbid-water coral assemblages at Pandora Reef: loss of resilience or too soon to tell? Coral reefs 26: 789–805

Done TJ (1982) Patterns in the distribution of coral communities across the Central Great Barrier Reef. Coral Reefs 1:95-107

Dufrene M, Legendre P. (1997) Genera assemblages and indicator genera: the need for a flexible asymmetrical approach. Ecological Monographs 67: 345-366.

Fabricius KE (2005) Effects of terrestrial runoff on the ecology of corals and coral reefs: Review and synthesis. Marine Pollution Bulletin 50:125-146.

Fabricius KE, De’ath G, McCook LJ, Turak E, Williams DM (2005) Changes in coral, algae and fish assemblages along water quality gradients on the inshore of the Great Barrier Reef. Marine Pollution Bulletin 51:384-398

Fabricius KE, De'ath G (2008) Photosynthetic symbionts and energy supply determine octocoral biodiversity in coral reefs. Ecology 89:3163-3173

Fabricius KE, McCorry D (2006) Changes in octocoral communities and benthic cover along a water quality gradient in the reefs of Hong Kong. Marine Pollution Bulletin 52:22-33

Page 72: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

66

Furnas MJ (2003) Catchments and Corals: Terrestrial Runoff to the Great Barrier Reef. Australian Institute of Marine Science and Reef CRC, Townsville. 353 p.

Furnas MJ (2005) Water quality in the Great Barrier Reef Lagoon: A summary of current knowledge. Chapter 3. In: Schaffelke B, Furnas M (eds) Status and Trends of Water Quality and Ecosystem Health in the Great Barrier Reef World Heritage Area. (CRC Reef, AIMS, Townsville). Unpublished Report to GBRMPA pp. 32-53.

Furnas MJ, Mitchell AW, Skuza M (1997) Shelf-scale nitrogen and phosphorus budgets fro the central Great Barrier Reef (16-19ºS). Proceedings of the 8th International Coral Reef Symposium, Panama 1997; Vol. 1:809-814.

Furnas MJ, Mitchell AW, Skuza M, Brodie J (2005) In the other 90%: Phytoplankton responses to enhanced nutrient availability in the Great Barrier Reef lagoon. Marine Pollution Bulletin 51: 253-256.

Harriot VJ (1999) Coral growth in subtropical eastern Australia. Coral Reefs 18:281-291

Hoegh-Guldberg O, Mumby PJ, Hooten AJ, Steneck RS, Greenfield P, Gomez E, Harvell CD, Sale PF, Edwards AJ, Caldeira K, Knowlton N, Eakin CM, Iglesias-Prieto R, Muthiga N, Bradbury RH, Dubi A, Hatziolos ME (2007) Coral Reefs Under Rapid Climate Change and Ocean Acidification. Science 318:1737-1742

Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR, Folke C, Grosberg R, Hoegh-Guldberg O, Jackson JBC, Kleypas J, Lough JM, Marshall P, Nystroem M, Palumbi SR, Pandolfi JM, Rosen B, Roughgarden J (2003) Climate Change, Human Impacts, and the Resilience of Coral Reefs. Science 301:929-933

Jongman DHG, Ter Braak CJF, Van Tongeren OFR (2004) Data analysis in assemblage and landscape ecology. Cambridge University Press.

Kendrick GA, Walker DI (1995) Dispersal of propagules of Sargassum spp. (Sargassaceae: Phaeophyta): Observations of local patterns of dispersal and consequences for recruitment and population structure. Journal of Experimental Marine Biology and Ecology 192:273-288

Lotze HK, Worm B, Sommer U (2000) Propagule banks, herbivory and nutrient supply control population development and dominance patterns in macroalgal blooms. Oikos 89:46-58

McCook LJ, Jompa J, Diaz-Pulido G (2001) Competition between corals and algae on coral reefs: a review of evidence and mechanisms. Coral Reefs 19: 400-417.

Negri A, Vollhardt C, Humphrey C, Heyward A, Jones R, Eaglesham G, Fabricius K (2005) Effects of the herbicide diuron on the early life history stages of coral. Marine Pollution Bulletin 51:370-383

Norton TA (1992) Dispersal by macroalgae. British Phycological Journal 27:293-301

Pichon M (1976) Comparative Analysis of Coral Reef Community Structure in vicinity of Lizard Island. National Geographic Society Research Report: 711-719.

R Development Core Team (2008) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.

Santelices B, Hoffmann AJ, Aedo D, Bobadilla M, Otaiza R (1995) A bank of microscopic forms on disturbed boulders and stones in tide pools. Marine Ecology Progress Series 129:215-228

Page 73: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

67

Schaffelke B (1999) Particulate nutrients as a novel nutrient source for tropical Sargassum species. Journal of Phycology 35:1150-1157

Schaffelke B, Thompson A, Carleton C, De’ath G, Doyle J, Feather G, Furnas M, Neale S, Skuza M, Thomson D, Sweatman H, Wright M, Zagorskis I (2007) Water Quality and Ecosystem Monitoring Programme – Reef Water Quality Protection Plan. Final Report to GBRMPA. Australian Institute of Marine Science, Townsville. 197 pp.

Schaffelke B, Thompson A, Carleton J, Cripps E, Davidson J, Doyle J, Furnas M, Gunn K, Neale S, Skuza M, Uthicke S, Wright M, Zagorskis I (2008) Water Quality and Ecosystem Monitoring Programme – Reef Water Quality Protection Plan. Final Report August 2008. Report to the Great Barrier Reef Marine Park Authority. Australian Institute of Marine Science, Townsville. 154 pp.

Stafford-Smith MG, Ormond RFG (1992) Sediment-rejection mechanisms of 42 species of Australian scleractinian corals. Australian Journal of Marine and Freshwater Research 43:683-705

Stiger V, Payri CE (1999) Spatial and temporal patterns of settlement of the brown macroalgae Turbinaria ornata and Sargassum mangarevense in a coral reef on Tahiti. Marine Ecology Progress Series 191:91-100

Sweatman H, Burgess S, Cheal AJ, Coleman G, Delean S, Emslie M, Miller I, Osborne K, McDonald A, Thompson A (2005) Long-term monitoring of the Great Barrier Reef. Status Report No.7:CD-Rom AIMS Townsville.

Titlyanov EA, Latypov YY (1991) Light-dependence in scleractinian distribution in the sublittoral zone of South China Sea Islands. Coral Reefs 10:133–138

Van Oppen MJH, Mieog JC, Sanchez CA, Fabricius KE (2005) Diversity of algal endosymbionts (zooxanthellae) in octocorals: the roles of geography and host relationships. Molecular Ecology 14:2403-2417

Van Woesik R, Tomascik T, Blake S (1999) Coral assemblages and physio-chemical characteristics of the Whitsunday Islands: evidence of recent community changes. Marine and Freshwater Research 50:427-440

Venables B and Ripley B (2002) Modern Applied Statistics with S. 4th edition. Springer.

Veron JEN (1995) Corals in space and time: the biogeography and evolution of the Scleractinia. University of New South Wales.

Wolanski E, Fabricius K, Spagnol S, Brinkman R (2005) Fine sediment budget on an inner-shelf coral-fringed island, Great Barrier Reef of Australia. Estuarine, Costal and Shelf Science 65:153-158.

Wolanski E, Fabricius KE, Cooper TF, Humphrey C (2008) Wet season fine sediment dynamics on the inner shelf of the Great Barrier Reef. Estuarine, Coastal and Shelf Science 77, 755-762.

Page 74: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

68

Appendix 1: Environmental variables

Table A1-1 Summary data of water quality variables at the 24 study reefs that were used to calculate the environmental variables (indices) for the analyses (see main text 1.3.1). NH4= ammonium, NO2= nitrite, NO3= nitrate, TDN= total dissolved nitrogen, PN= particulate nitrogen, PO4=phosphate, TDP= total dissolved phosphorus, PP= particulate phosphorus, N= number of sampling occasions NRM Region Coral monitoring locations NH4 N NO2 N NO3 N TDN N PN N PO4 N TDP N PP N µgL-1 µgL-1 µgL-1 µgL-1 µgL-1 µgL-1 µgL-1 µgL-1

Wet Tropics

Snapper Island North 1.616 8 0.086 8 0.837 8 79.525 8 12.014 8 2.660 8 4.795 8 2.401 8 Fitzroy Island West 1.224 8 0.030 8 0.831 8 74.569 8 11.406 8 2.439 8 4.122 8 2.188 8 High Island West 1.351 8 0.069 8 0.651 8 69.867 8 13.824 8 2.213 8 7.836 8 2.534 8 King Reef 1.953 4 0.274 4 0.571 4 76.777 4 17.802 4 2.646 4 9.868 4 3.968 4 Frankland Group West 1.475 4 0.070 4 0.453 4 68.714 4 11.059 4 2.373 4 7.380 4 1.847 4 Frankland Group East 1.191 4 0.146 4 0.122 4 76.620 4 12.548 4 2.226 4 10.478 4 1.841 4 North Barnard Group 0.812 2 0.267 2 0.738 2 97.886 2 24.555 2 3.524 2 5.102 2 6.146 2 Dunk Island North 1.274 12 0.173 12 0.807 12 76.029 12 17.740 12 1.966 12 8.298 12 3.998 12

mean 1.362 0.139 0.626 77.498 15.119 2.506 7.235 3.115 median 1.312 0.116 0.694 76.324 13.186 2.406 7.608 2.467

Burdekin

Pelorus and Orpheus Island West 1.006 8 0.000 8 0.233 8 64.751 8 11.122 8 2.660 8 9.952 8 1.822 8 Pandora Reef 1.101 6 0.010 6 0.325 6 63.946 6 12.965 6 2.533 6 7.808 6 2.649 6 Havannah Island 0.608 4 0.027 4 0.074 4 68.736 4 13.429 4 2.858 4 8.807 4 2.679 4 Lady Elliot Reef 1.462 4 0.207 4 0.599 4 84.352 4 21.970 4 2.808 4 11.836 4 5.747 4 Geoffrey Bay 1.461 6 0.417 6 1.611 6 69.113 6 20.725 6 3.134 6 8.455 6 4.552 6

mean 1.128 0.132 0.568 70.180 16.042 2.799 9.372 3.490 median 1.101 0.027 0.325 68.736 13.429 2.808 8.807 2.679

Page 75: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

69

NRM Region Coral monitoring locations NH4 N NO2 N NO3 N TDN N PN N PO4 N TDP N PP N

Mackay Whitsunday

Double Cone Island 1.507 5 0.061 6 0.251 6 60.461 6 12.925 6 3.060 6 8.816 6 2.150 6 Daydream Island 2.007 6 0.175 6 0.364 6 74.509 6 14.027 6 3.113 6 10.012 6 2.084 6 Pine Island 1.750 2 0.248 2 0.783 2 52.355 2 14.218 2 3.328 2 4.333 2 2.578 2 Hook Island 1.465 4 0.164 4 0.017 4 75.942 4 19.102 4 3.396 4 12.685 4 3.417 4 Dent Island 6.512 4 1.138 4 2.911 4 109.758 4 12.745 4 3.977 4 13.107 4 2.008 4 Seaforth Island 5.486 5 0.982 5 3.839 5 82.450 5 16.924 5 4.217 5 11.189 5 3.050 5 Shute & Tancred Is. 3.436 4 0.511 4 1.946 4 92.096 4 15.468 4 3.854 4 13.053 4 2.586 4

mean 3.166 0.469 1.444 78.224 15.058 3.563 10.456 2.553 median 2.007 0.248 0.783 75.942 14.218 3.396 11.189 2.578

Fitzroy

Barren Island 1.294 5 0.096 5 0.368 5 79.356 6 14.640 6 2.963 6 9.414 6 2.043 6 Pelican Island 1.335 6 0.361 6 1.412 6 90.805 6 19.955 6 6.351 6 11.931 6 4.864 6 Humpy & Halfway Is. 1.523 6 0.091 6 0.249 6 80.167 6 15.050 6 3.505 6 9.488 6 2.640 6 North Keppel Island 0.824 4 0.071 4 0.287 4 82.162 4 15.254 4 2.331 4 9.988 4 2.263 4

mean 1.244 0.155 0.579 83.122 16.225 3.788 10.205 2.952 median 1.315 0.094 0.328 81.165 15.152 3.234 9.738 2.452

Page 76: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

70

Table A1-1 continued. Si= silicic acid, DOC= dissolved organic carbon, POC= particulate organic carbon, Secchi= Secchi depth, SS= suspended solids, Chl= chlorophyll a. NRM Region Coral monitoring locations Si N DOC N POC N Secchi N SS N Chl N

µgL-1 µgL-1 µgL-1 m mgL-1 µgL-1

Wet Tropics

Snapper Island North 168.666 8 726.694 8 112.152 8 6.75 4 1.191 8 0.306 8 Fitzroy Island West 133.662 8 696.704 8 122.352 8 8.75 4 0.954 8 0.389 8 High Island West 205.744 8 700.245 8 130.057 8 8.58 6 1.599 8 0.490 8 King Reef 250.458 4 726.051 4 178.875 4 4.88 4 3.863 4 0.730 4 Frankland Group West 194.431 4 688.997 4 100.195 4 10.40 5 0.963 4 0.317 4 Frankland Group East 145.657 4 673.281 4 122.963 4 9.33 3 1.190 4 0.389 4 North Barnard Group 208.079 2 773.654 2 272.362 2 3.00 2 5.548 2 1.153 2 Dunk Island North 369.708 12 800.283 12 206.439 12 5.20 5 3.384 12 0.692 12

mean 209.551 723.239 155.674 7.11 2.336 0.558 median 200.087 713.148 126.510 7.67 1.395 0.439

Burdekin

Pelorus and Orpheus Island West 59.644 8 675.193 8 107.993 8 7.88 4 1.219 8 0.335 8 Pandora Reef 115.440 6 707.433 6 129.536 6 6.00 5 2.119 6 0.474 6 Havannah Island 46.791 4 699.570 4 167.170 4 5.33 3 1.974 4 0.492 4 Lady Elliot Reef 128.727 4 836.614 4 250.415 4 3.17 3 6.738 4 0.800 4 Geoffrey Bay 126.406 6 799.453 6 236.019 6 4.00 6 4.524 6 0.877 6

mean 95.401 743.653 178.227 5.28 3.315 0.596 median 115.440 707.433 167.170 5.33 2.119 0.492

Page 77: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

71

NRM Region Coral monitoring locations Si N DOC N POC N Secchi N SS N Chl N

Mackay

Double Cone Island 69.797 6 640.733 6 123.275 6 8.30 5 1.127 5 0.495 6 Daydream Island 129.856 6 666.521 6 115.322 6 9.42 6 1.473 6 0.425 6 Pine Island 111.672 2 711.890 2 207.582 2 9.00 1 1.320 2 0.606 2 Hook Island 148.217 4 652.234 4 172.858 4 7.00 4 9.660 3 0.535 4 Dent Island 149.084 4 678.324 4 108.531 4 6.38 4 1.491 3 0.403 4 Seaforth Island 279.531 5 771.707 5 148.683 5 5.00 5 2.112 5 0.653 5 Shute and Tancred Islands 186.578 4 662.487 4 129.022 4 5.50 4 2.111 4 0.457 4

mean 153.533 683.414 143.611 7.23 2.756 0.510 median 148.217 666.521 129.022 7.00 1.491 0.495

Fitzroy

Barren Island 80.945 6 767.117 6 248.067 6 11.80 4 0.573 5 0.336 6 Pelican Island 350.877 6 996.301 6 297.687 6 6.10 5 4.597 5 0.809 6 Humpy & Halfway Island 114.568 6 810.958 6 250.281 6 9.80 5 0.989 5 0.689 6 North Keppel Island 67.884 4 741.282 4 296.748 4 8.75 2 1.228 4 0.323 4

mean 153.568 828.915 273.196 9.11 1.847 0.539 median 97.756 789.038 273.514 9.28 1.108 0.512

Page 78: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

72

Appendix 2: Univariate analyses of coral reef benthos

Table A2-1 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard coral cover. Dispersion parameter for quasipoisson family taken to be 9.21365. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 3.082 0.2579 11.95 8.95e-15 Particulate -0.0074 0.4082 -1.815 0.0771 Clay-silt 0.0177 0.0072 2.475 0.0177 Dissolved 8.497e-5 0.0043 0.020 0.9842 Burdekin -0.0764 0.3072 -0.249 0.8049 Mackay -0.1371 0.2890 -0.474 0.6378 Fitzroy 0.7094 0.2475 2.866 0.0066 Depth 5m 0.0807 0.1637 0.493 0.6244 Table A2-2 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard coral richness. Dispersion parameter for quasipoisson family taken to be 1.857145. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 2.940 0.1265 23.23 <2e-16 Particulate -0.0011 0.0019 -0.577 0.567 Clay-silt -0.0011 0.0040 -0.288 0.775 Dissolved 0.0016 0.0022 0.713 0.480 Burdekin 0.1061 0.1260 0.842 0.405 Mackay 0.2641 0.1516 1.741 0.089 Fitzroy -0.6643 0.1645 -4.039 0.0002 Depth 5m 0.2304 0.0845 2.728 0.0094 Table A2-3 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard coral juvenile density. Dispersion parameter for quasipoisson family taken to be 1.473593. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 1.965 0.2092 9.395 1.13e-11 Particulate -0.0044 0.0033 -1.333 0.190 Clay-silt -4.901e-5 0.0067 -0.007 0.994 Dissolved 0.0039 0.0040 0.972 0.337 Burdekin -0.0567 0.2193 -0.259 0.797 Mackay -0.1439 0.2594 -0.555 0.582 Fitzroy -0.7818 0.2768 -2.824 0.007 Depth 5m -0.0856 0.1450 -0.590 0.558 Table A2-4 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard coral juvenile genus richness. Dispersion parameter for quasipoisson family taken to be 1.298253. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 3.346 0.0921 36.33 <2e-16 Particulate -0.0028 0.0014 -1.957 0.057 Clay-silt -0.0030 0.0029 -1.038 0.305 Dissolved 0.0016 0.0016 0.970 0.338 Burdekin 0.0102 0.0942 0.108 0.914 Mackay 0.2602 0.1108 2.347 0.024 Fitzroy -0.7229 0.1219 -5.931 5.91e-07 Depth 5m 0.1788 0.0624 2.867 0.0066 Table A2-5 Results from generalised linear model (quasipoisson) with a log-link l applied to total soft coral cover. Dispersion parameter for quasipoisson family taken to be 11.49187.

Page 79: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

73

Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 3.009 0.4834 6.224 2.29e-07 Particulate -0.0238 0.0098 -2.422 0.020 Clay-silt -0.0207 0.0168 -1.233 0.225 Dissolved 0.002 0.0094 0.315 0.755 Burdekin 0.1655 0.5625 0.294 0.770 Mackay 0.9847 0.6016 1.637 0.109 Fitzroy 0.0124 0.5879 0.021 0.983 Depth 5m -0.3952 0.3521 -1.122 0.268 Table A2-6 Results from generalised linear model (quasipoisson) with a log-link l applied to total soft coral richness. Dispersion parameter for quasipoisson family taken to be 1.561910. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 1.575 0.2353 6.693 5.02e-08 Particulate -0.0066 0.0037 -1.779 0.083 Clay-silt 0.0033 0.0073 0.453 0.653 Dissolved 0.0041 0.0042 0.988 0.329 Burdekin 0.1270 0.2555 0.497 0.622 Mackay 0.0601 0.2844 0.211 0.834 Fitzroy 0.1385 0.2484 0.558 0.580 Depth 5m 0.0549 0.1566 0.351 0.728 Table A2-7 Results from generalised linear model (quasipoisson) with a log-link l applied to total soft coral juvenile density. Dispersion parameter for quasipoisson family taken to be 2.358268. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 1.0252 0.4921 2.083 0.044 Particulate -0.0233 0.0092 -2.532 0.015 Clay-silt -0.0174 0.0153 -1.142 0.260 Dissolved 0.0035 0.0080 0.436 0.666 Burdekin 0.4773 0.5599 0.852 0.399 Mackay 1.376 0.5794 2.376 0.022 Fitzroy 0.5041 0.5586 0.902 0.372 Depth 5m -0.0853 0.3220 -0.265 0.792 Table A2-8 Results from generalised linear model (quasipoisson) with a log-link l applied to total soft coral juvenile genus richness. Dispersion parameter for quasipoisson family taken to be 1.262443. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 1.861 0.2004 9.285 1.57e-11 Particulate -0.0047 0.0032 -1.485 0.145 Clay-silt -0.0091 0.0066 -1.375 0.177 Dissolved 0.0028 0.0036 0.780 0.440 Burdekin 0.0173 0.2100 0.083 0.935 Mackay 0.4450 0.2473 1.799 0.079 Fitzroy -0.6103 0.2582 -2.364 0.023 Depth 5m 0.0896 0.1374 0.652 0.518

Page 80: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

74

Table A2-9 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard macroalgal cover. Dispersion parameter for quasipoisson family taken to be 5.72469. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 0.515 0.7315 0.704 0.485 Particulate (Linear) 2.219 1.793 1.238 0.223 Particulate (Quadratic) -8.017 1.725 -4.648 3.8e-05 Clay-silt 0.0423 0.0204 2.076 0.045 Dissolved -0.0018 0.0108 -0.171 0.865 Burdekin 2.281 0.5228 4.363 9.1e-05 Mackay -1.967 0.9363 -2.103 0.042 Fitzroy 1.774 0.4325 4.101 0.0002 Depth 5m -0.4725 0.2236 -2.113 0.041 Table A2-10 Results from generalised linear model (quasipoisson) with a log-link l applied to total hard macroalgal richness. Dispersion parameter for quasipoisson family taken to be 1.538513. Coefficients: Estimate Std Error t value Pr(>|t|) Intercept 1.123 0.3036 3.700 0.0007 Particulate (Linear) 0.7755 0.9017 0.860 0.395 Particulate (Quadratic) -1.513 0.7564 -2.000 0.052 Clay-silt 0.0124 0.0102 1.221 0.229 Dissolved 0.0058 0.0061 0.944 0.351 Burdekin 0.5318 0.2635 2.018 0.050 Mackay -1.164 0.4522 -2.574 0.014 Fitzroy -0.0085 0.2738 -0.031 0.975 Depth 5m -0.2540 0.1814 -1.400 0.169 Table A2-11 Results from sequential analysis of deviance. Pr(>F) values are the P values for differences between regions after sequential removal of deviance associated with the three environmental variables, ‘particulate’ and ‘dissolved’ water quality and ‘clay silt’ content of the sediment. Assemblage Attribute Pr(>F) Hard coral cover 0.023 Hard coral richness 7.3e-05 Hard coral juvenile density 0.033 Hard coral juvenile richness 1.4e-07 Soft coral cover 0.439 Soft coral richness 0.934 Soft coral juvenile density 0.119 Soft coral juvenile richness 0.017 Macro algae cover 2.4e-06 Macro algae richness 0.017

Page 81: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

75

Appendix 3: Redundancy analyses of coral reef benthos

Tables A3-1 to A3-5 summarise the results from redundancy analyses of genus-level benthic assemblage data. Two tables are presented for each attribute of the benthic assemblages: • Table a includes the results from redundancy analysis including terms for Region, Depth,

Particulate, Clay-silt and Dissolved. Permutation tests are based on the variance attributable to each term given the inclusion of the remaining four terms in the model.

• Table b explicitly maximises the variance attributable to the environmental variables by first conditioning on depth effects (as environmental variables are collected at the reef level and so do not vary with depth) and then applying a model that does not include a term for region.

It can be assumed that the minimum proportion of variation attributable to region is the difference in the proportion of variation attributed to the residuals (unconstrained) between the two models, a and b. Table A3-1 Results from redundancy analysis of genus-level hard coral cover data. Variable Df Var F Pr(>F) % Variation a Particulate 1 0.200 2.1741 0.005

37.9 Dissolved 1 0.11 1.2160 0.190 Clay-silt 1 0.19 2.0391 0.020 Depth 1 0.28 3.0533 0.005 Region 3 0.94 3.4016 0.005 Residual 40 3.67 62.1 b Particulate 1 0.24 2.204 0.010

17.3 Dissolved 1 0.16 1.456 0.140 Clay-silt 1 0.42 3.957 0.005 Residual 43 4.61 78 Condition(Depth) 4.7 Table A3-2 Results from redundancy analysis of genus-level hard coral juvenile density data. Variable Df Var F Pr(>F) % Variation a Particulate 1 0.15 3.7101 0.005

43.2 Dissolved 1 0.07 1.6985 0.110 Clay-silt 1 0.09 2.3765 0.032 Depth 1 0.09 2.3866 0.027 Region 3 0.53 4.5000 0.005 Residual 40 1.57 56.8 b Particulate 1 0.15 2.9655 0.020

20.6 Dissolved 1 0.09 1.7563 0.074 Clay-silt 1 0.18 3.6365 0.005 Residual 43 2.11 76 Condition(Depth) 3.4

Page 82: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

76

Table A3-3 Results from redundancy analysis of genus-level soft coral cover data. Variable Df Var F Pr(>F) % Variation a Particulate 1 0.10 1.5184 0.028

31.8 Dissolved 1 0.05 0.7548 0.230 Clay-silt 1 0.03 0.4152 0.710 Depth 1 0.03 0.4910 0.560 Region 3 0.32 1.5826 0.005 Residual 24 1.61 68.2 b Particulate 1 0.10 1.2714 0.039

16.9 Dissolved 1 0.07 0.9088 0.140 Clay-silt 1 0.12 1.4688 0.032 Residual 24 1.92 81.7 Condition(Depth) 1.4 Table A3-4 Results from redundancy analysis of genus-level soft coral juvenile density data. Variable Df Var F Pr(>F) % Variation a Particulate 1 0.051 1.5605 0.010

34 Dissolved 1 0.021 0.6371 0.250 Clay-silt 1 0.020 0.6192 0.230 Depth 1 0.018 0.5531 0.370 Region 3 0.165 1.6829 0.005 Residual 22 0.718 66 b Particulate 1 0.04 1.0007 0.090

17.2 Dissolved 1 0.02 0.5389 0.380 Clay-silt 1 0.05 1.193 0.034 Residual 22 0.88 81.1 Condition(Depth) 1.7 Table A3-5 Results from redundancy analysis of genus-level macroalgal cover data. Variable Df Var F Pr(>F) % Variation a Particulate 1 0.1 0.8409 0.052

42.6 Dissolved 1 0.03065 0.2795 0.510 Clay-silt 1 0.01219 0.1112 0.930 Depth 1 0.04253 0.3877 0.290 Region 3 0.6 1.8900 0.005 Residual 13 1.4 57.4 b Particulate 1 0.207 1.3118 0.015

15.8 Dissolved 1 0.034 0.2180 0.540 Clay-silt 1 0.017 0.1101 0.830 Residual 13 2.048 82.4 Condition(Depth) 1.7

Page 83: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

77

Appendix 4: Model checking of analyses of water quality from instrumental records and direct water sampling

Figure A4-1 Model checking for analysis of directly sampled water quality data. The model diagnostics show the model assumptions (see Methods) are well-satisfied.

Page 84: Water Quality and Ecosystem Monitoring Programme Reef ...€¦ · Thompson A, Schaffelke B, De’ath G, Cripps E, Sweatman H (2010) Water Quality and Ecosystem Monitoring Programme

REEF PLAN MMP SPATIAL ANALYSIS OF INSHORE MONITORING DATA 2005-08

78

Figure A4-2 Model checking for analysis of inter-instrument comparison data. The model diagnostics show the model assumptions (see Methods) are well-satisfied.