EXECUTIVE SUMMARY

The implementation of the National Policy Statement for Freshwater Management (Ministry for the Environment) has placed a strong imperative for regional authorities to establish and implement limits to protect water quality and ecological values of waterbodies in their region.

Environment Southland (ES) is considering management targets for nutrient load limits for the region's shallow coastal lakes. These tend to be located in highly modified landscapes and as a result are under threat of degradation from land-use intensification.

This study consists of two main parts:

1. A literature review of national and international sources documenting nutrient loading to shallow lakes, and the effects on water quality characteristics (e.g. chlorophyll-a, water clarity) and ecological integrity (aquatic macrophytes, andother communities).

2. A modelling approach to assess nutrient loading to shallow lakes in Southland and the South Island, fitting Vollenweider regression models to nutrient loading and water quality data. We used the national CLUES (Catchment Land Use for Environmental Sustainability) model to predict loads of total phosphorus (TP) and total nitrogen (TN) and this was then related to in-lake water quality and ecological condition data for up to 19 lakes.

Review of the international literature indicates that managing loads to control phosphorus (P) concentrations is the most significant management action for protecting the ecological integrity (EI) of shallow freshwater lakes. In the extensive work conducted in European shallow lakes (UK, Denmark, Netherlands), managing catchment nutrient loads to maintain (or restore) shallow lakes to a clear-water macrophyte-dominated state, in-lake P concentration and sediment P content was nearly always the most immediate management objective. Patterns emerged in the literature around the importance of external and internal sources. For lakes that had experienced historically high loads or point source pollution, internal nutrient loading became a significant management objective, whereas moderately degraded systems responded more rapidly to catchment load reductions.

In European lakes (Denmark, Netherlands, UK), a critical threshold for P occurred around 100 mg TP/m3, with very few lakes above this threshold having macrophytes covering > 50%of the lake bed. In contrast, in the warmer North American lakes (Florida) this threshold was closer to 50 mg/m

3. The nitrogen (N) threshold occurred around 1000 mg TN/m3 in bothEuropean and North American lakes, above which few lakes had macrophyte cover exceeding 50%. From the study data available, lakes that experienced P loading in excess of 80 kg P/ha/y were likely to have accumulated P rich sediments that could result in internal loading.

The modelling work on the South Island shallow coastal lake-set provided significant relationships between N and P loading and several EI indicators. This was particularly evident for CLUES P loading corrected to in-lake TP concentrations using Vollenweider models. The strongest relationships were evident for chlorophyll-a and the trophic level index(TLI), with 79% and 86% of their variation explained by the TP loading variable, respectively.

These relationships were linear, so for management purposes TLI targets for lakes could be directly related back to nutrient load targets for the lake. The strong model performance for this study is possibly related to the relatively short water residence times of the lakes, minimising the potential for in-lake nutrient processing. However there is a risk when using such correlational data, that modifying loads (either by load reductions or allowed increases) to particular lakes would not result in expected changes in retention or in-lake concentration due to factors such as internal loading or short-term blooms causing shifts in aquatic plant dominance (regime shifts).

From our modelling work, there was consistently less variation accounted for in TLI, chlorophyll-a, and other EI variables (e.g. macrophytes, macroinvertebrates) by TN loadingmodels. However in several instances models were still significant suggesting that nitrogen loading is related to chlorophyll-a and TLI in some lakes. In-lake nutrient ratios of TN:TPsuggested that many of the lakes in the data set are likely to be phosphorus limited (16 of 19 for TN:TP), with a smaller number being co-limited by both N and P. This could be possibly related to N-fixation when systems become highly -limited, but it was beyond the scope ofthis study to evaluate. A recent international meta-analysis study looking at the importance of nitrogen to water quality in shallow lakes reported that while chlorophyll-a and trophic statuswas more often controlled by P, macrophyte species composition and (in some cases) cover was related to in-lake TN concentration, with some species clearly having TN tolerances or preferences. This emphasises the importance of co-management of N and P for managing aquatic macrophyte communities in shallow lakes.

External loading estimates for South Island lakes were considerably less than those documented in most overseas studies, being maximum 32 kg P/ha/y. This made interpretation from the literature around risk levels difficult to interpret in the New Zealand context. However, based on negative phosphorus retention coefficients (suggestive of internal loads) for three of the more eutrophic NZ lakes, this indicated some risk when external P loads exceed 17 kg P/ha/y, which could ultimately generate internal loading issues. However further data collection around calculating nutrient retention coefficients (i.e.inflow/outflow monitoring) would provide greater certainty in these findings.

Results from the set of NZ lakes, relating nutrient loading to other key shallow lake EIvariables (such as macrophyte and macroinvertebrate communities), also indicated that P loading was the more proximal variable relating to benthic communities. These results tend to indicate lower thresholds than results from overseas studies, with NZ lakes having loads resulting in mean summer TP concentrations of >50 mg/m3 (and nearly always having little orno macrophyte cover). Macroinvertebrate community richness was even more sensitive tonutrient loading and declined linearly by nearly 42% (from approximately 26 to 15 taxa) compared with reference condition lakes at the 50 mg/m3 TP upper limit for macrophytecover. So if this variable was used as an indicator of shallow lake EI, related to biodiversity values, we would suggest a more conservative target to minimize losses in biodiversity values. We would also suggest the loads, back-calculated using Vollenweider models, be no greater than levels to achieve a TP (and possibly N) concentration in the mid-eutrophic range(Burns et al. 2000), or 32 mg/m3 TP (and possibly 531 mg/m3 TN). This would be based on the combined relationships of P loading with macrophyte cover, macroinvertebrate richness and euphotic depth and be associated with a meaningful scale for in-lake nutrient concentrations, in this case TLI. It is important to recognise, because most of the CLUES loading data could not be thoroughly validated by inflow monitoring data for most lakes, some caution is suggested in directly applying CLUES load statistics, and further collection of inflow data would be useful in validating annual load predictions.

Overall, results from modelling work provide some valuable information in terms of relationships of N and P loading with key variables related to the EI of shallow lakes. The nature of these relationships can be used to help inform the decisions on loading rates for specific lakes to achieve water quality outcomes for the lake. The nature of these outcomes can then be discussed at a community planning level, taking into account the community's aspirations, and with water quality targets developed for specific waterbodies. This investigation was predominantly focused on ecological indicators related to EI, and did not consider other freshwater uses such as recreational and amenity values as they relate to nutrient limits. As such, this is identified as a gap and possibly an area for further work.