Variability in nutrient loading to lake ecosystems and associated impacts on water quality

Abstract

Globally, the accelerated eutrophication of lake ecosystems due to excess inputs of nitrogen (N) and phosphorus (P) is a significant problem. The rate of loading of N and P to lakes varies both in space and time; consequently, developing good understanding of such spatial and temporal variability is critical for developing integrated approaches to managing lake water quality. This study aimed to improve understanding of spatial and temporal variations in N and P loading to lakes, and, to examine how this variability affects water quality. The topic was considered at global, national and catchment scales.

Analysis of an extensive global dataset was undertaken to examine relationships between N, P and chlorophyll a (chl a) in lakes along a gradient of latitude inclusive of tropical, temperate and polar regions. The ratio of total nitrogen (TN) to total phosphorus (TP) was positively correlated with latitude, reflecting global–scale variation in nutrient cycling processes and/or nutrient sources. Relative to temperate lakes, the statistical capability of concentrations of TN and TP to predict chl a concentration (i.e. indicating bottom–up control by nutrients) was shown to be poor for both tropical and polar lakes. These differences reflected latitudinal variation in lake ecosystem functioning, and highlighted the potential unsuitability of applying relationships derived for temperate lakes elsewhere. Quantile regression was used to derive theoretical chl a near–maxima as a function of either TN or TP concentrations. Consequently, chl a:TN and chl a:TP yields (by mass) of 0.046:1 and 0.87:1 were determined to approximate the maximum possible yields of chl a under optimal conditions as a proxy for phytoplankton biomass potential.

Nationally–significant relationships between landscape characteristics and in–lake TN and TP concentrations were quantified for a representative sample of 101 lakes in New Zealand. Geographical Information Systems were used to analyse data from a range of sources that related to both lake–specific and wider landscape characteristics. Inferential statistical methods were then used to quantify relationships between in–lake nutrients and both land use and naturally–occurring soil P. National–scale variability in mean catchment soil P was found to be unrelated to in–lake TP concentrations, reflecting the dominant influence of human–related sources of P on TP concentrations in New Zealand lakes. The extent of intensive pastoral agriculture was the best land use predictor of TN and TP concentrations, accounting for 38.6% and 41.0% of variation respectively. Exotic forestry accounted for a further 18.8% of variation in TP concentrations. A sub–sample of lakes for which intensive pastoral agriculture was the dominant catchment land use was then considered to test hypotheses regarding potential interactive effects of eight landscape characteristics on the positive relationship between intensively managed pasture and in–lake nutrient concentrations. Both maximum lake depth and the ratio of catchment to lake area had significant interactive effects, exerting a negative and a positive influence, respectively. In addition, an indicator of hydrological connectivity (lake order) also had a positive interactive effect on the relationship between this land use type and in lake TP (but not TN) concentrations.

To examine these broad relationships at finer spatial resolution, and also to quantify temporal variations in nutrient loading, an extensive field programme was conducted in the catchment of Lake Rotorua (Bay of Plenty, New Zealand); a large (80.5 km2), relatively shallow (zmean = 10.8 m) lake that has experienced eutrophication. The Ngongotaha and Puarenga streams are two major inflows to the lake that were sampled at high frequency throughout a wide range of stream discharge coinciding with rainfall events. Both streams had different catchment characteristics (e.g. land use and hydrogeomorphology) which enabled spatial variations in nutrient loading between sub–catchments located upstream of a common lake ecosystem to be examined. Streams were sampled during a total of 17 hydrological events, including three during which both streams were simultaneously sampled to compare differences in pollutant transport between the streams during similar hydrological conditions. Relationships between nutrient concentrations and stream discharge were broadly similar for the two catchments, and quantification of relationships permitted nutrient loading to be estimated continuously over annual periods. Key findings included the dominance of event loads by dissolved inorganic N and the strong positive correlation between discharge and particulate P concentrations. Quantification of hysteresis in relationships between nutrient concentrations and discharge provided information about the relative importance of near– versus far–channel sources during individual events. For example, elevated concentrations of dissolved inorganic N during recessing hydrograph limbs for the Puarenga Stream suggested diffuse delivery of N from an upstream source. Temporal inequality in estimated loading over a two–year period was high for TP as, for example, 50% of estimated cumulative two–year loads of TP were calculated to have been transported during 10–17% of the two–year time period.

The effects of storm flow discharges on water quality in Lake Rotorua were studied at fine spatial and temporal resolution for a five–day period with high rainfall in summer. An intensive programme of lake and stream sampling was paired with application of a three–dimensional hydrodynamic–ecological model (ELCOM–CAEDYM) to specifically study how dynamic fluxes in water, sediment, N and P transport in the Ngongotaha Stream inflow influenced water quality and phytoplankton nutrient limitation in the transition zone present where the stream enters the lake. Wind–driven basin–scale horizontal circulations in the lake caused deflection of the inflowing stream which strongly influenced water quality in the littoral zone for a distance of up to 1 km from the stream mouth, thus highlighting the potential importance of basin–scale horizontal transport processes in mediating the effects of storm flow discharges on lake water quality. The nutrient limitation status of phytoplankton varied both spatially and temporally within the lake in relation to nutrient transport processes, emphasising the relatively fine spatial and temporal scales at which key processes that affect phytoplankton ecology can occur. Dilution of lake water by the stream inflow strongly affected the spatial distribution of chl a, although the highly spatially resolved sampling identified ‘hot spots’ within the nutrient–rich plume which contributed to fine scale (≈10–30 m) patchiness in the transition zone. The results of nutrient enrichment experiments indicated that such patchiness was consistent with a scenario of relative stimulation in the growth of lentic phytoplankton due to high nutrient availability in the spreading plume.

To further examine the issue of nutrient bioavailability, chemical fractionation techniques and batch culture experiments were conducted to investigate spatial (between streams) and temporal (between periods of varying stream discharge) variations in the bioavailability of particulate P transported in storm flow for the two study streams. Bicarbonate–dithionate extraction indicated that 25–100% of particulate P transported in stream water samples collected during storm flow was potentially bioavailable if exposed to anoxia, e.g. in the lake hypolimnion during calm summer periods. Somewhat paradoxically though, bioassays indicated that, under oxic conditions in the laboratory, bioavailable P was actually higher in filtered samples (particulate P removed) than in unfiltered samples (higher TP concentrations, particulate P present). This result was attributed to net adsorption of dissolved inorganic P to the sediments present in the unfiltered treatments, and therefore highlights the importance of considering physicochemical characteristics of receiving environments when assessing bioavailability of P sorbed to sediments.

Hence, by examining a range of spatial scales and integrating understanding gained using a range of research methods, this study has provided knowledge of underlying drivers of spatial and temporal variability in nutrient loading to lake ecosystems at scales ranging from global to a few metres. Furthermore, it has provided insight into how such variability can affect lake water quality. This knowledge can guide actions that are increasingly required to safeguard the services provided by lake ecosystems in a future with increasing global and local pressures on freshwaters.