• About this site|
• Site index|
• Contact us|
• Jobs

Water Quality in Selected Dairy Farming Catchments: A baseline to support future water-quality trend assessments

Executive Summary

Introduction

Methods

Catchment Characteristics

Water-quality Baseline

Summary and Recommendations

References and Information Sources

Appendix 1: Monitoring Catchment Details

Appendix 2: Water-quality Summary Data

Appendix 3: Regression Analyses

Appendix 4: Current Monitoring Programmes

Appendix 5: Recreational Values

Methods

This national baseline report presents a summary of the data received from councils and other agencies. Some of the data reported in this report is previously unpublished, while other data has already been published in council technical reports and in scientific papers. All data sources are listed in the references section of this report, and readers are referred to these sources for specific details about the monitoring and reporting methods in each monitored catchment.

The monitored catchments

There are five Tier 1 catchments and nine Tier 2 catchments situated in 12 of the 16 regions in New Zealand (see table 1 and figure 1).

Table 1: Monitored catchments and monitoring agencies

* The Rhodes Stream and Petrie Creek are small, neighbouring waterways that are part of the same general catchment area. However, they are unconnected, so the water-quality results for each of these streams are generally presented separately later in the report.

Figure 1: Location of Tier 1 and Tier 2 monitoring catchments in New Zealand

View graph|View text description

Map of New Zealand showing the location of all 14 monitoring catchments, split into Tier 1 (five) and Tier 2 (nine). Map shows they are distributed across the country in every climatic and broad geographic zone

Data collection

Land-use and farm information

Information about Tier 1 and 2 catchment features, including land use, was provided by regional councils, mainly from their resource consents and environmental monitoring databases, as well as from other data sets such as the Land Cover Database 2 and AgriBase.

Information about the progress made towards Accord targets in each Tier 1 and 2 catchment was provided by Fonterra Co-operative Group Ltd from its annual on-farm survey results. Survey results for individual farms were aggregated to obtain the catchment figures presented in this report. In a limited number of cases regional councils were able to supplement the Fonterra information with their own farm survey findings (eg, Environment Canterbury), and this supplementary information has been included where appropriate.

Water-quality data

Water-quality baselines have been defined in this report by the median (and range of) values for a number of water-quality measurements, including biological measurements. The measurements considered most appropriate for indicating the effects of land use – in this case, dairying – on water quality are listed in Table 2. Non-dairy farming activities in monitored catchments can also substantially influence the results for the parameters listed in table 2, and this is discussed further in the next section ‘Data limitations and constraints’.

Table 2: Water-quality parameters used in this report

In most cases the water-quality samples used to provide data for this report were collected by regional councils from the monitored catchments and processed in accredited laboratories. National Institute of Water and Atmospheric Research (NIWA) field teams also collect data from the Tier 1 catchments.

The design of the monitoring programmes and frequency of sampling varied between catchments and regions. Monthly sampling in some Tier 1 catchments began in the mid-1990s and continues today, although there have been periods of more frequent data collection
(eg, weekly and fortnightly). The water quality median data presented for Tier 1 catchments in this report is drawn from the monthly sampling programme for the period 2001–2006.

Of the nine Tier 2 catchments, five were sampled on a monthly basis for a single year ending in mid-2007 for the purpose of deriving a water-quality baseline. However, some of these catchments had data for some aspects of water quality that had been collected before 2006/07. The remaining four Tier 2 catchments had a longer time series of data available because the sample sites are part of wider council state-of-environment monitoring programmes. Where these longer time series of data are available, they have been included in the derivation of Tier 2 catchment statistics.

More detail on the number of samples taken at each site is given in the water-quality data tables in appendix 2 (tables A2-A to A2-N).

All of the catchments continue to be monitored at the time of writing of this report and the current programmes are summarised in appendix 4 (table A4-A). However, the intensity and range of sampling differs between catchments, as does the extent to which monitoring agencies are able to commit resources. This is the subject of some discussion in the ‘Summary’ section of this report.

Data analysis and presentation

The analysis of water-quality data presented in this report was undertaken both by councils and by research agencies (eg, AgResearch, NIWA and the Cawthron Institute). In particular, the research agencies were closely involved with the analysis of water-quality data from the Tier 1 best practice dairy catchments. The Ministry for the Environment was provided with, or accessed through published reports, summary data for each catchment (ie, not raw sample data). As a result, the analysis of data in this report is restricted to the presentation and interpretation of descriptive statistics (ie, medians and ranges – minimum and maximum sample results).

Some linear regression of water-quality results against predictor variables (eg, land use, rainfall) has been attempted in order to define apparent patterns more quantitatively, but in most cases the correlations are poor (see appendix 3, table A3-A). This is likely to be a consequence of data limitations (ie, the number of catchments) and the inability of relatively simple statistical tests to account for the effect of multiple catchment factors on water quality.

Percentiles such as the 5th, 25th, 75th and 95th have not been presented because in some cases catchment statistics were derived from only 12 data points, which is insufficient to examine the distribution about the median in any meaningful way. One of the recommendations of this report (see ‘Recommended next steps‘ section) is that more in-depth, standardised statistical analysis be presented in future reports as the data sets for each catchment increase in size.

Use of guidelines

Water-quality data has been compared with national water-quality guidelines where appropriate. There are no national (binding) standards for ambient water quality in New Zealand, and the guidelines that do exist (eg, ANZECC 2000) are generally intended to be used for management and policy purposes rather than as reporting benchmarks. These guidelines do still provide a useful context for water-quality results, but readers are urged to take note of the specific comments about guidelines in each of the relevant results sections.

Data limitations and constraints

Farm survey data

The data provided by Fonterra is described in more detail in the most recent Accord Snapshot of Progress report (Ministry of Agriculture and Forestry, 2009). With regard to the limitations of the data, it is important to recognise that the on-farm surveys are completed by farmers, who may interpret the Accord wording and how it relates to their land in different ways. This leads to differing views on the reliability of data and different perceptions about progress towards Accord targets (eg, Deans and Hackwell, 2008; Jensen and Harcombe, 2008).

For the authors of this report it remains difficult to judge how well the survey data on progress towards Accord actions in each of the monitored catchments reflects the actual extent of actions (and the effectiveness of those actions). Furthermore, the Fonterra data only applies to waterways that meet Accord criteria. There is no standardised data available that describes the extent of fencing and livestock crossing removal for smaller headwater streams.

Data scale, resolution and time lag issues

Relating relatively coarse-scale land-use data to water-quality results and drawing conclusions about the effects of particular management actions on water quality (eg, the fencing of waterways) is very difficult. This was emphasised in the monitoring and reporting strategy for measuring the environmental outcomes of the Accord (Ministry for the Environment, 2006), and has been reinforced during the collection of data in the preparation of this report.

There are many land-use factors that can affect water quality and/or contribute to fluctuations in water quality over time, for which little or no consistent information can be obtained from the existing monitoring regimes. These factors are often very localised, as opposed to being distributed throughout the catchment, and include (Ministry for the Environment, 2006):

• naturally occurring events such as floods, which can have major effects on suspended sediment, turbidity and in-stream biota

• changes in farm personnel, such as the sharemilker, with subsequent changes in environmental performance

• changes in land use or land-use intensity, such as changing from sheep to deer farming, bringing more beef cattle on to a particular property or increasing dairy herd size

• the amount of fertiliser used by all farmers (not just dairy farmers), the timing of its use and how it is applied – aerial topdressing of hill country may, for instance, have significant impacts on phosphate concentrations and algal biomass in streams, and if there is heavy rainfall within several days of fertiliser application, much of that fertiliser may run off to waterways

• stock access to streams – not just by dairy cattle on smaller non-Accord streams but also by beef cattle, sheep and deer

• fences that are too close to streams, or that are broken or poorly maintained and allow stock access to streams

• non-dairy discharges to water

• wintering dairy stock off-site – this has potential major benefits for water quality during these months

• run-off from land irrigation of dairy effluent entering water courses from mole-and-tile drains

• natural inputs, such as geothermal, soil and geological inputs of phosphorus or arsenic.

A further confounding factor is the time lag between an action being taken (or a past farming activity) and the consequential effects of this action on water quality. The exact nature of this time lag will vary depending on location, size of catchment and activity, and is often difficult to quantify. However, it is generally accepted that applying nutrients to land, for example, does not have an immediate effect on water quality (unless the nutrients directly wash into the waterway), and that the lag time before changes are seen in the stream can span many years. This can have significant implications for monitoring programmes – including the programme presently under discussion – which aim to quantify actual water quality improvements that result from land management changes.

Statistical constraints

Following are brief descriptions of some of the methodological and statistical constraints on assessing the state of water quality and the significance of any changes over time or differences between catchments. These mainly relate to the influence of flow on water quality and the limitations of sampling regimes, and are important to consider when interpreting results in this report, and subsequent related reports.

• To include comparable results from as many of the selected catchments as possible, this report presents statistics (eg, medians) calculated for the whole period of monitoring in each catchment. Although this approach broadly characterises water quality, it ‘averages out’ seasonal changes that may be of importance for measuring the effect of relevant land-use activities. For example, during summer low flows, the effect of removing cattle from streams may be more apparent in the E .coli results for that period than in winter because of the overriding influence of flow during winter on E. coli concentrations washed in from the landscape during high rainfall events. It may be possible in the next report to explore changes in seasonal data patterns by requesting data from monitoring agencies in a different format and undertaking flow-related analyses. (Data in this report is generally not flow-adjusted).

• As well as fluctuating across seasons, some water-quality measurements (particularly dissolved oxygen and water temperature) can fluctuate considerably during the course of a normal day. For example, dissolved oxygen concentrations and water temperature normally fall to a minimum in the early hours of the morning (pre-dawn) and reach a maximum in the early afternoon. It is these extremes that are most likely to (a) have adverse consequences for aquatic ecosystems and (b) exhibit the greatest changes in response to changing farm practice. However, the statistics presented in this report relate to repeated single measurements made by monitoring staff during daylight hours only, a limitation that should be considered when judging the long-term value of these indicators. In some catchments, continuous (ie, automatic data-logger) data is available to augment the monthly statistics.

• Some of the water-quality variables exhibit large ranges in monthly measurements, and some catchments have as few as 12 data points from which to derive the medians presented in this report. Although these variations introduce some uncertainty when making initial assessments and comparisons of catchments (ie, in this report), they should be mitigated over time as data sets expand and appropriate statistical tests are used to analyse changes.

• It is difficult to know whether the water quality statistics represent typical or atypical climatic periods in each of the catchments. However, the assessment of long-term rainfall patterns (appendix 1, tables A1-A to A1-M) provides some confidence that baseline monitoring did not occur in particularly unusual rainfall years in most catchments. As a further note, catchments with longer time series are considerably more likely to have representative statistics than those with just one year of monitoring data.

[ Previous| Next