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

Nationwide and regional state and trends in river water quality 1996-2002

Executive Summary

1. Introduction

2. Methods

3. Results

4. Discussion

5. Acknowledgements

6. References

Appendix 1: Appended Table 1

Appendix 2: Plots of water quality in land-cover classes, grouped by climate classes

3. Results

3.1. Objective 1 - Assessments of water quality state and comparisons among REC classes

3.1.1. Climate level

Examination of the plots for climate classes indicates that the WD climate had the lowest overall water quality (Fig. 3). Median concentrations of DRP, NO_X, NH₄ and E. coli were at least twice as high as the guideline values, and median water clarity was about one-third of the guideline value. In our dataset, the WD climate class consists solely of lowland sites belonging to the pastoral and urban land-cover classes. The poor water quality associated with the WD climate class may in part reflect the condition of low-elevation streams in agricultural and urban watersheds (See Results, Section 3.1.3). In the WW and CD classes, median nutrient and E. coli concentrations exceeded the guideline values, and median clarity was below, but average conditions of at least 25% of the sites that make up the classes were within guideline levels for nutrients and for clarity (Fig. 3, 25th - 75th percentile boxes). These observations suggest that water quality in the WW and CD classes is marginally better than in the WD class. The CX class had the best overall water quality; median values for nutrients and clarity were within the guideline values, and the median E. coli concentration was at the guideline. In our dataset, the CX climate class is dominated by sites in the natural land-cover class (87%), and the high water quality state of this climate class may in part reflect the state of streams in undeveloped catchments.

Comparisons among REC climate classes are summarized in Table 7. For each parameter, classes are ranked from highest median to lowest. Among-class differences were significant for each parameter (Kruskal-Wallis tests, P < 0.05). Pair-wise comparisons indicated that the WD, WW and CD classes had significantly higher nutrient and E. coli concentrations, and lower clarity, than the CW and CX classes. The CW class had higher nutrient and E. coli concentrations and lower clarity than the CX class. The WX class is not included because there were too few monitoring sites.

Table 7: Ranks of water quality parameters in climate classes.

Results of pairwise comparisons are indicated by superscript letters. Differences in median parameter levels in classes with different superscripts are statistically significant (P < 0.05). Sample sizes in parentheses.

3.1.2. Source of flow level

Examination of the plots for 16 source-of-flow classes suggests that sites in three low-elevation classes, CD/L, WD/L, and WW/L had the worst overall water quality (Fig. 4). Median nutrient and E. coli concentrations in these classes are 2 - 8 times higher than the guidelines (with the exception of NO_X in WW/L, 30% over the guideline). All sites in the WD/L and CD/L classes are in pastoral and urban land-cover classes. Some WW/L sites are in natural and exotic forest land-cover classes, but most are in pastoral classes (10% of sites natural, 3% exotic forest, 82% pastoral, 5% urban).

Six source-of-flow classes had consistently high water quality: CW/H, CW/Lk, CW/M, CX/H, CX/Lk, and CX/M. In these classes, median values of all parameters met the recommended guidelines. It is noteworthy that no low-elevation source-of-flow classes were among the classes with high water quality. Three additional classes had low median nutrient concentrations and intermediate clarity, but high E. coli concentrations: CX/L, WX/L and CD/H. The CX/L class is made up predominately of coastal streams and rivers in the Westland Region that drain indigenous forest and pastoral catchments. The WX/L class is composed of sites in the pastoral land-cover class in the Hawkes Bay and Waikato Regions, and the CD/H class is dominated by pastoral sites (90%) in Otago and Canterbury regions, including the mid-elevation portions of most of the large braided rivers of the eastern South Island.

Spatial patterns in water quality state at the source-of-flow level are shown in Figures 5a to 5d. These maps show all fourth-order or larger rivers for each class considered. Water quality state has been extrapolated from the sites in our dataset to all of the rivers those sites represent. Water quality in some of the rivers shown in Figure 4 meet a recommended guideline, although the class as a whole does not. For example, streams on Stewart Island (none of which are included in our dataset) are primarily in the CW/L class, for which median DRP, NO_X, NH₄ and E. coli concentrations did not meet the guideline values (Fig. 4). Since Stewart Island is largely undeveloped, streams there are likely to have high water quality, despite belonging to a class with generally poor water quality. Conversely, some rivers may exceed a guideline when the class as a whole does not. Therefore, the maps should be viewed as indicating risks of degradation, not the current state of each river shown

Comparisons among source-of-flow classes confirm the pattern suggested by Figure 3: water quality in low-elevation classes is poor compared with other classes (Table 8). The low-elevation classes had significantly higher nutrient and E. coli concentrations than the hill, lake and mountain classes in the CD and CW climate classes. The CX/L class had higher DRP and E. coli concentrations than the CX/Lk class, and higher NH₄ concentrations than the CX/M class. The CW/L class had significantly lower clarity than the CW/H and CW/M classes. The general pattern of poor water quality in low-elevation classes may be related to factors that operate at the source-of-flow scale (10² - 10³ km²), such as flood frequency and sediment yield, or to the fact that low-elevation streams are dominated by agricultural land uses: of the total low-elevation river length in New Zealand (~ 195,200 km), the REC classifies 73% as pastoral, 19% natural, 6% exotic forest and 2% urban.

3.1.3. Land-cover level

Nutrient and E. coli concentrations and water clarity in each land-cover class are shown graphically in Appendix 2. The ANZECC and MfE guideline values from Table 4 are indicated on each plot. REC classes in which median values of the water quality parameters exceed the guideline values are listed in Table 9. In total, 61% of the monitoring sites had median values that exceeded the guideline value for DRP, 43% exceeded the guideline for NO_X, 59% exceeded the guideline for NH₄, 73% exceeded the guideline for E. coli, and 40% did not reach the guideline for clarity (Table 10). When exceedance is examined at the land-cover level, several patterns are evident (Table 10) (Note: we did not assess whether median values were significantly different from the guideline). A greater proportion of sites in the urban land-cover category failed to meet the guideline for each parameter than in the other three

Table 8: Comparisons of water quality in source-of-flow classes.

For comparisons between 2 classes, statistically significant differences are shown as inequalities. For comparisons of 3 - 4 classes, classes are ordered from highest to lowest rank, and statistically significant differences are indicated by different superscript letters. NS: Difference not statistically significant. Sample sizes in parentheses. See Figure 3 for parameter medians and comparisons with guidelines.

categories. Median nutrient and E. coli concentrations exceeded the guideline values at almost all urban sites. Over 50% of sites in the pastoral class failed to meet the guideline values for each parameter. Less than 50% of sites in the natural classes exceeded the guidelines for any parameter; in particular, very few sites in the natural classes had excessive median NO_X concentrations. Proportions of sites in the exotic forest classes that exceeded each guideline were generally similar to those of natural sites, with the exception of DRP and NO_X, which exceeded the guideline value more frequently at exotic forest sites

Comparisons of fully classified streams (i.e., climate/source-of-flow/geology/land-cover) were made among classes represented by six or more monitoring sites in each group (Table 11). Five pairs of natural and pastoral land-cover classes within the same climate/source-of-flow/geology class were compared. These comparisons included the most common REC classes in the dataset, with the exception of the CD/L/HS/P class. There were 58 sites in this class, but no CD/L/HS/N sites for comparison. Results of the comparisons indicate that pastoral streams in each class had higher E. coli concentrations and lower clarity that the corresponding natural streams. Pastoral streams in three classes had higher NO_X and NH₄ concentrations than the natural streams, and pastoral streams in two classes had higher DRP concentrations.

To increase the spatial scope of comparisons, additional comparisons were made using pooled source-of-flow and geology classes, and pooled climate classes (Table 12). These comparisons have higher within-group replication, but lower spatial resolution, than the preceding comparison of fully classified streams. Most urban and exotic forest classes were represented by fewer than six sites, so comparisons were again limited to pastoral and natural classes. CW and WW are the climate classes with most sites.

Table 9: River classes in which median parameter values exceed ANZECC and MfE guidelines

View river classes in which median parameter values exceed ANZECC and MfE guidelines (large table)

Table 10: Proportions of monitoring sites in the state dataset that exceeded ANZECC and MfE guidelines.

First number is the number of sites in exceedance, second number is the total number of sites in the dataset.

Table 11: Nonparametric comparisons of pastoral and natural streams at the geology level. Inequalities indicate statistically significant differences (P < 0.05)

NS difference not statistically significant. IS insufficient sample size. Sample sizes in parentheses. See Appendix 2 for parameter medians and comparisons with guidelines.

Table 12: Nonparametric comparisons of pastoral (P) and natural (N) sites within and across climate classes

Inequalities indicate statistically significant difference (P < 0.05). NS: difference not statistically significant. IS: insufficient sample size. Sample sizes in parentheses. See Appendix 2 for parameter medians and comparisons with guidelines.

CX sites were less common, but the numbers of CX pastoral and natural sites were similar. CW and CX climate regions occur throughout the South Island and in the central and southern North Island. The WW climate region is restricted to the North Island. Results of these comparisons corroborated the results for fully classified streams: E. coli concentrations were significantly higher in pastoral streams than in natural streams in each comparison, and nutrient concentrations were significantly higher and clarity lower in pastoral streams in the CW and WW climate classes, and across all climate classes, than in the corresponding natural streams.

To examine water quality state within the land-cover classes making up the "natural" class, as well as in the non-pastoral developed classes (urban and exotic forest), a third set of comparisons were made using land-cover classes pooled over climate, source-of-flow and geology. Among-class differences were significant for each parameter. Land cover classes are arranged in Table 13 from highest median to lowest. The bare land cover class is not included because there were only three sites; our cutoff was six sites. DRP, NO_X and E. coli concentrations were higher in the urban and pastoral classes than in any natural stream classes, and the urban class had higher E. coli concentrations than the pastoral class. No other differences between these classes were significant. The exotic forest class had high DRP and intermediate NO_X concentrations, but low NH4 and E. coli concentrations and high clarity.

Table 13: Ranks of water quality parameters in land cover classes

Results of pairwise comparisons are indicated by superscript letters. Differences in median parameter levels in classes with different superscripts are statistically significant (P < 0.05). Sample sizes in parentheses.

3.2. Objective 2 - Recent trends in water quality

3.2.1. Trends at the national level and across pastoral and natural classes

Over the 1996-2002 period, there was a significant negative trend in river flows across New Zealand (Table 14). As a result of these changing flow conditions, the flow-adjustment of water quality parameters is justified, and only flow-adjusted data are reported in subsequent sections. With respect to flow-adjusted data, there were significant negative trends in NH₄ and NO_X concentrations, and a significant positive trend in water temperature across New Zealand.

Table 14: Trends (SKSSE as % of data median) in flow and water quality parameters at the national scale

Values in bold correspond to statistically significant trends (P < 0.05). N: number of monitoring sites.

Natural and pastoral land-cover classes both had significant negative trends in flow and significant positive trends in clarity and temperature (Table 15). There were also significant negative trends in NO_X concentration in the natural class, and in NH₄ concentration in the pastoral class. NH₄ was the only parameter that showed significant variation between the two land cover classes; the median slope for the pastoral class was lower than in the natural class (Table 15).

Table 15: Trends (SKSSE as % of data median) in flow and water quality parameters (flow-adjusted) in pastoral and natural land cover classes

Values in bold correspond to statistically significant trends (P < 0.05). P-values in table refer to results of Mann-Whitney tests comparing the two land cover classes. Statistically significant differences are in bold. N: number of monitoring sites.

3.2.2. Trends in source-of-flow classes

Trends in flow-adjusted water quality parameter values were generally consistent across source-of-flow classes (Table 16). All significant trends in flow were negative, and all trends in clarity and temperature were positive. Single classes showed negative trends for NO_X (CW/Lk), and NH₄ concentrations (WW/L). There were no significant trends in DRP or E. coli concentrations. Comparison of trends among the classes indicated no significant differences (Kruskal-Wallis tests, P > 0.05). Table 16 only shows classes with six or more sites; classes with fewer sites were not tested.

Table 16: Trends (SKSSE as % of data median) in flow and water quality parameters (flow-adjusted) across source-of-flow classes

View trends (SKSSE as % of data median) in flow and water quality parameters (flow-adjusted) across source-of-flow classes (large table)

3.2.3. Trends in land-cover classes

Statistically significant trends in flow and water quality parameters were consistent across REC classes at the land cover level (Table 17). All trends in flow were negative, and trends in temperature and clarity were positive. There were no significant trends for DRP or E. coli. Three negative trends in ammonium concentrations were also significant, all in pastoral classes.

Due to limited replication, only four comparisons of trends could be made among fully-classified land-cover classes, i.e., among land-cover classes within each climate, source-of-flow, and geology class. Each comparison was between a pastoral class and the corresponding natural class (Table 18). Two comparisons indicated significant differences in temperature trends. In both cases, the positive trend in temperature was significantly greater in the pastoral class than in the natural class. Comparisons also indicated a more strongly negative median slope for ammonium, and more strongly positive clarity slopes at pastoral sites compared with natural land cover sites.

Table 17: Trends (SKSSE as % of data median) in flow and flow-adjusted water quality parameters

View trends (SKSSE as % of data median) in flow and flow-adjusted water quality parameters (large table)

Table 18: Comparison of median slopes in pastoral and natural land cover classes

View comparison of median slopes in pastoral and natural land cover classes (large table)