Skip to main content.

2. Methods

2.1 Monitoring methods used

The cover of wadeable areas of river beds by potential nuisance periphyton (i.e., mats and filamentous algae) was assessed at monthly intervals whenever flow conditions allowed safe wading at 73 of the 77 NRWQN sites (Fig. 1, Appendix 1) using methods described in Smith et al. (1989). The four sites excluded were: two Auckland region sites (Hoteo AK1 and Rangitopuni AK2), that have silty beds dominated by macrophytes rather than periphyton, and two Rotorua region sites (Tarawera at Awakaponga RO2 & Waikato at Reid’s Farm RO6), that are too deep for wading.

Where possible ten observations of 0.5 m radius patches of riverbed were made at equally-spaced points across a wadeable cross-section of the river using an underwater viewer. However, because the NRWQN rivers are medium to large (mean baseflow width 54 m, range 7-200 m) some rivers were too deep/fast for this and observations were confined to the wadeable margin from one bank or, at 10% of sites, to observations from bridges or cableways. Responses to a 2004 questionnaire of NIWA field team personnel who carry out the monitoring found that on average eight (standard deviation (SD) = 3) observations of the bed are made at the standard points along these cross-sections/partial transects and transects cover an average 37% (SD 35%) of the total width (Appendix 1). Sites that are never suitable for wading (0% wadeable) are DN4, HV5, HM2, HM3, HM4, HM5 and WH3 (Appendix 1), and sites that are often not wadeable are DN1, GS1, WH2, TK4 and TK6 (Appendices 1 and 2). Less than 10% of the width is normally wadeable at AX1-4, CH1, CH3, CH4, GY1, GY2, GY4, TU1, WH4, RO5 and GS4 (Appendix 1). Despite these sampling constraints, the observation methods reflect the way the public would observe periphyton cover so that the methods are considered to be valid for answering the question of whether the aesthetic effects of riverbed cover by periphyton are increasing over time and/or along river systems.

The stalked diatom Didymosphenia geminata, commonly referred to as “Didymo”, appeared at several South Island NRWQN rivers towards the end of the period covered by this report. It was first observed in 2004 at DN9, in 2005 at NN5, GY1, TK4, TK6, AX1, AX4, DN7 and DN8, and in 2006 at DN6. Didymo is unusual in that it often reaches very high biomass at pristine sites. Because it appeared late in the monitoring period and only formed very high biomass at a few sites, the decision was made to exclude from this analysis the data for infected Didymo sites for the years where it caused obvious large increases in cover, to avoid it confounding the general patterns and trends over the rest of the monitoring period. The data removed were for 2004 to 2006 at DN9, and 2006 at AX1 and AX4.

Field staff received field training and were provided with a training video and a methods manual. Methods were assured during biannual field visits for most of the monitoring period. Nevertheless, after a change in crew at one of the field teams, a measurement error occurred with thin brown biofilms recorded as periphyton mats giving erroneous high results until the error was detected and corrected. Because of this assessment error, matted periphyton data for the years 2003 to 2005 have been deleted for the sites WA1-9.

2.2 Trend analysis

Overall site trends of periphyton cover over the 17 years of monitoring were assessed using the annual mean and annual maximum cover data, rather than using monthly data that are highly variable due to the effects of floods. Analysis of variance was used to compare mean annual cover and mean maximum cover amongst sites classified as baseline (n = 23) and impact (n = 42) by Smith at al. (1989). The smaller number of pseudo-baseline sites (8) were excluded from this aspect of the analysis. Non-parametric Spearman’s Rank correlations (rs) of periphyton cover versus time were used to evaluate whether there were statistically significant trends using Zar’s (1984) critical values for two-tailed tests. Trends are presented as statistically significant for individual sites at both the traditional 95% level of confidence (P < 0.05) and at the 90-95% confidence level (0.1 > P > 0.05), with the latter confidence level used to identify sites where there is a weaker trend for early warning. The Type I error rates (i.e., chance of detecting a false positive result) are <1:20 and 1:10 to 1:20, respectively when assessing whether a trend is occurring at an individual site, which is likely to be of interest to river users and regional councils. However, because the analysis involved multiple trend assessments, the risk of Type I statistical errors (i.e., false positives) within the whole dataset is increased by each additional correlation on the same dataset. This was allowed for in the assessment of trends in the whole data by calculating the False Discovery Rate (FDR)(Benjamini & Hochberg 1995) that controls for the number of false positives (set at 5%) when conducting multiple correlations on the same dataset.

Trends in differences in cover between baseline and impact sites on the same river were investigated (also using rs) using the monthly data because observations were usually made at the paired sites on the same day. Mean annual and annual maximum total periphyton covers were also compared between individual upstream and downstream monitoring site pairs along rivers using paired t-tests (P < 0.05). The statistical significance of these trend and t-test comparisons are presented with and without FDR adjustments for multiple analyses.

 

|