The Marine and Freshwater Research Laboratory (MAFRL, Environmental Science, Murdoch University) monitored water quality in the Peel Harvey Estuary for the past decade. In order to compare pre and post Channel water quality in a meaningful manner, years of similar rainfall were chosen (Table 1).
 

Table 1: Annual rainfall at Mandurah (Bureau of Meteorology) of pre and post Channel years included in this study. Data is included from the 1^st of July in the named year, until the 30^th of June of the following year.
 
 

Pre Channel Year	Annual Rainfall (mm)	Post Channel Year	Annual Rainfall (mm)
1985/86	857.8	1994/95	708.6
1986/87	783.7	1995/96	882.4
1987/88	703.6	1996/97	836.6
1988/89	857.6	1997/98	780.8
1990/91	804.8	1998/99	872.8

 

Sampling was undertaken at the same six sites (Figure 1) during all years. However, the frequency of sampling has varied considerably from weekly to fortnightly and monthly, as directed by the Water and Rivers Commission. There were considerably more sampling occasions in pre as opposed to post Channel years (Table 2).

Table 2: Number of sampling occasions during winter, spring and summer of each year included in this study. Sampling dates are given in Appendix 1.
 
 

Year	Winter	Spring	Summer
1985/86	13	13	13
1986/87	13	13	9
1987/88	11	13	6
1988/89	13	12	7
1990/91	11	8	4
Pre Total	61	59	39
1994/95	10	8	6
1995/96	11	10	6
1996/97	11	13	5
1997/98	8	3	3
1998/99	7	3	3
Post Total	47	37	23

 

The field parameters measured included; Secchi depth, surface and bottom pH (Hanna HI 8424); and profiles for salinity (± 0.1 ppt, Hamon Salinity-Temperature Bridge, Model 602), temperature (± 0.1^oC, Hamon Salinity-Temperature Bridge, Model 602) and dissolved oxygen (± 0.2 mg L^-1,Yeo-Kal Dissolved Oxygen-Temperature Meter, Model 603). Chemical parameters measured were surface and bottom total nitrogen (± 200 µg N L^-1), total phosphorus (± 15 µg P L^-1), ammonium (± 3 µg L^-1), nitrate plus nitrite (± 2 µg L^-1), filterable reactive phosphate (± 2 µg L^-1), silicate (± 2 µg L^-1), chlorophyll a (± 0.1 µg L^-1) and phaeophytin (± 0.1 µg L^-1).

Attenuation coefficients were calculated from the regression of the logarithm of irradiance values against various depths (Kirk, 1977). MAFRL has established a regression analysis for the Peel-Harvey system between attenuation coefficients calculated from irradiance values and secchi depth. Irradiance levels were measured in the water column using a Licor Quantum Meter, with submersible probe, along with secchi depth. Secchi depths were used to calculate the attenuation coefficient, where the Licor Quantum Meter was not suitable (i.e. during overcast conditions).

Surface and bottom samples were subsampled for total nitrogen and total phosphorus. Additional water was filtered through GF/C filter paper (pore size 1.2 µm, Whatman Ltd, England) to determine chlorophyll a concentration and then 0.45 µm cellulose nitrate filter paper for ammonium, nitrate-nitrite, filterable reactive phosphate and silicate. Samples were retained in 'Whirlpaks' (NASCO) and frozen until analysis.

All analyses were carried out by the Marine and Freshwater Research Laboratory. Chlorophyll a samples were extracted in 90% acetone over 24 hours in the dark at 4^oC. This method does not detect picoplankton which is less than 0.8µm in diameter. The chlorophyll a concentrations were measured spectrophotometrically (Varian DMS 90 Spectrophotometer, Varian Techron Pty Ltd, Springvale, Australia) according to the method of Strickland and Parsons (1972). The concentration of all inorganic nutrients were determined by a Lachat Automated Flow Injection Analyser (Lachat Instruments, Milwaukee, USA). Filterable reactive phosphate was analysed by the single solution method (Lachat Instruments, 1996a); nitrate plus nitrite after copper-cadmium reduction (Lachat Instruments, 1996b); determination of silicates is based on the reduction of a silicomolybdic acid by stannous chloride to molybdenum blue (Lachat Instruments, 1996); and ammonia plus ammonium by the phenol-prusside method (Lachat Instruments, 1994). Total Kjeldahl nitrogen (TKN) was determined from sulphuric acid digests followed by analysis for ammonia using the Technicon Autoanalyser II method of salicylate-nitroprusside. Total nitrogen was calculated by adding nitrate plus nitrite to TKN. Total phosphorus was determined from perchloric acid digests followed by analysis for orthophosphate using the manual spectroscopic method of orthophosphate by the single solution method (Major et al.,1972).

Data was divided into seasons and collated into two groups to give means for pre Channel winter, spring and summer as well as post Channel winter, spring and summer. Wilson et al. (1997) in a 3 year pre and post Channel investigation, focussed on the summer and winter seasons. Spring has been included in this report as it was a time of high phytoplankton productivity and therefore, considered important for comparisons.

There has been much debate in recent times on the appropriate form of statistical analysis for water quality data. The data for this report was tested for discrepancies in distribution (skewness and heterogeneity of variance) to determine if they met the preconditions for parametric statistical analysis. Unfortunately, approximately 90% of the data was either skewed or suffered from unequal variances (or both). However, it is considered that the t-test is robust enough to withstand large deviations from normality and considerable heterogeneity of variance (Zar, 1974). In addition, the ranking nature of non-parametric analyses makes them resistant to extreme events such as algal blooms (Helsel and Hirsch, 1992). As it was these extreme events that have historically caused the problems within the Peel-Harvey Estuary, it was considered essential that they should be included in the analyses. The results quoted in this document as being significantly different are those with a p-value <0.05 from t-tests performed using the computer package StatView.

In previous reports (Wilson et al., 1997) graphs of nutrient and chlorophyll a concentrations have been presented as means ± standard error. Standard error is a measure of the accuracy of the mean, but gives little indication of the spread of the data. Therefore, in this report, graphical representations show mean ± standard deviation, indicating the differences in variance between pre and post Channel groups.