New Zealand is unique in that half of its national greenhouse gas (GHG) inventory derives from agriculture--predominantly as methane (CH4) and nitrous oxide (N2O), in a 2:1 ratio. The remaining GHG emissions predominantly comprise carbon dioxide (CO2) deriving from energy and industry sources. Proposed strategies to mitigate emissions of CH4 and N2O from pastoral agriculture in New Zealand are: (1) utilising extensive and riparian afforestation of pasture to achieve CO2 uptake (carbon sequestration); (2) management of nitrogen through budgeting and/or the use of nitrification inhibitors, and minimizing soil anoxia to reduce N2O emissions; and (3) utilisation of alternative waste treatment technologies to minimise emissions of CH4. These mitigation measures have associated co-benefits and co-costs (disadvantages) for rivers, streams and lakes because they affect land use, runoff loads, and receiving water and habitat quality. Extensive afforestation results in lower specific yields (exports) of nitrogen (N), phosphorus (P), suspended sediment (SS) and faecal matter and also has benefits for stream habitat quality by improving stream temperature, dissolved oxygen and pH regimes through greater shading, and the supply of woody debris and terrestrial food resources. Riparian afforestation does not achieve the same reductions in exports as extensive afforestation but can achieve reductions in concentrations of N, P, SS and faecal organisms. Extensive afforestation of pasture leads to reduced water yields and stream flows. Both afforestation measures produce intermittent disturbances to waterways during forestry operations (logging and thinning), resulting in sediment release from channel re-stabilisation and localised flooding, including formation of debris dams at culverts. Soil and fertiliser management benefits aquatic ecosystems by reducing N exports but the use of nitrification inhibitors, viz. dicyandiamide (DCD), to achieve this may under some circumstances impair wetland function to intercept and remove nitrate from drainage water, or even add to the overall N loading to waterways. DCD is water soluble and degrades rapidly in warm soil conditions. The recommended application rate of 10 kg DCD/ha corresponds to 6 kg N/ha and may be exceeded in warm climates. Of the N2O produced by agricultural systems, approximately 30% is emitted from indirect sources, which are waterways draining agriculture. It is important therefore to focus strategies for managing N inputs to agricultural systems generally to reduce inputs to wetlands and streams where these might be reduced to N2O. Waste management options include utilizing the CH4 resource produced in farm waste treatment ponds as a source of energy, with conversion to CO2 via combustion achieving a 21-fold reduction in GHG emissions. Both of these have co-benefits for waterways as a result of reduced loadings. A conceptual model derived showing the linkages between key land management practices for greenhouse gas mitigation and key waterway values and ecosystem attributes is derived to aid resource managers making decisions affecting waterways and atmospheric GHG emissions.

A parsimonious nutrient-periphyton model quantifying uptake and recycling below a steady nutrient point source was tested in a shallow, stony-bed river during summer low flow. Close to the source DIN and DRP concentrations decreased linearly with distance, the DIN:DRP uptake ratio was constant (although twice the Redfield ratio), and rates of photosynthesis and algal biomass were high-consistent with model predictions. Further downstream, DIN concentrations were near the detection limit (but DRP was measurable), and rates of photosynthesis and algal biomass were low-also consistent with model predictions. However, two model features were not supported by observations. First, concentrations of organic N and P remained constant, whereas the model predicts increasing concentrations. Second, observations showed that N and P recycling were decoupled, whereas the model assumes close coupling. Although the model was reliable close to the source, it over-simplified recycling in the test stream. The model could be improved by separating dissolved from particulate organics and including recycling from settled particulates and pH-mediated P release from sediment, although this would increase complexity and reduce parsimony.

Instream processes alter the concentration and bioavailability of nutrients as they are transported downstream. By relating primary production and periphyton composition to changes in nutrient concentration in a gravel-bed river this study made inferences about recycling and attenuation. Where dissolved inorganic nitrogen (DIN) was abundant, concentrations decreased linearly with distance but by less than required to meet the nitrogen demand of primary production. Where DIN was barely measurable photosynthesis was reduced but only by 50%. We infer that recycling sustained primary production even when DIN concentrations were negligibly small. One implication is that DIN removal underestimates attenuation. Further experimental research on recycling and improved modelling is required to better quantify the length of streams adversely affected by nutrients.

Phosphorus (P) stores in gravel-bed rivers are released for uptake by periphyton when pH levels exceed 8.5. The Tukituki River has low alkalinity water and frequently experiences periphyton blooms, and daytime pH > 9 during summer low-flows. We measured dissolved reactive P (DRP) and EPC0, the water concentration of DRP at which no net release or sorption from the river bed occurs, in sediment samples from the Tukituki River subject to controlled pH levels before (2014) and after (2017) changes to two wastewater discharges that reduced P release to the river by 95%. DRP released from 2014 sediments at pH 8.5-10 were 30 +/- 10 mg/m(3) above background (pH 8) whereas those released from 2017 sediments were 5 +/- 3 mg/m(3) above background. EPC0 levels in 2014 and 2017 were 11 +/- 6 and 7 +/- 2 mg/m(3), respectively. Field estimates of released DRP calculated from continuous pH and the Redfield equation suggested that most of the readily available DRP released from sediments at elevated pH is derived from material attached to recently deposited sediment. Subsequently, a reduction in wastewater inputs or agricultural runoff should reduce sediment DRP stores, and hence sediment fluxes, within a few years and mitigate periphyton blooms in addition to directly lowering water column concentrations.

Five streams in catchments with pastoral dairy farming as the dominant land use were monitored for periods up to 13 years after adoption of mitigation practices (MPs) to detect changes in stream macroinvertebrate communities and water quality. Water quality and macroinvertebrate community condition was degraded at the start with respect to N, P, suspended solids (SS), Escherichia coli concentrations, and a number of macroinvertebrate metrics, and were typical of catchments with intensive pastoral land use. Trend analysis showed a decrease in SS concentrations for all streams, generally increasing water clarity, and lower E. coli concentrations in two streams. The improvements, attributable to improved stream fencing and effluent disposal, were only sometimes associated with improved macroinvertebrate metrics, and suggest constraints of connectivity, habitat, time, and/or MP effectiveness. Macroinvertebrate species composition had the strongest correlation with total nitrogen at two sites, but changes probably reflected indirect factors associated with TN concentrations were responsible.

The management of streams and rivers can be aided by knowledge of reference conditions. Data from >1000 sites across New Zealand was used to develop a technique to estimate median ammoniacal-N, clarity, Escherichia coli, filterable reactive phosphorus, nitrate-N, suspended solids, and total nitrogen and phosphorus values under reference conditions for streams and rivers as classified by the River Environment Classification (REC). The REC enabled us to account for natural variation in climate, topography and geology when estimating reference conditions. Values for minimally disturbed sites (i.e. <5% in intensive agriculture) were generally within the confidence limits for estimated reference values. Metrics that described: (1) the percentage of anthropogenic contribution to analyte values; and (2) the degree of enrichment beyond the reference conditions, showed that lowland sites classified as warm-wet, warm-dry or cool-dry exhibited the greatest anthropogenic input and enrichment. The consideration of natural variation by REC class informs the setting of water quality objectives through avoiding water quality limits or targets that are either too restrictive, and impossible to meet (e.g. below reference conditions), or too high, such that they have little ecological benefit. We recommend reference conditions be considered by regulatory authorities when assessing water quality impacts, objectives and limits.

Nutrient ratios have been related to nutrient limitation of algal growth in lakes. Retention of nutrients in lakes, by sedimentation and by denitrification, reduces the nutrient concentrations in the water column, thereby enhancing nutrient limitation. Differential retention of nitrogen and phosphorus alters their ratios in lakes and thereby contributes to determine whether nitrogen or phosphorus limits algal growth. We examined the relationships between differential nutrient retention, nutrient ratios, and nutrient limitation in Lake Brunner, a deep oligotrophic lake. The observed retention of nitrogen (20%) and phosphorus (47%) agreed with predictions by empirical equations from literature. As a result of differential retention with a much larger proportion of phosphorus retained than that of nitrogen, the nitrogen:phosphorus ratio was higher in the lake (69) than in the inflows (46). While the mean ratio in the inflows suggested no or only moderate phosphorus limitation, the lake appeared to be severely phosphorus limited. Combining empirical equations from literature that predict nitrogen and phosphorus retention suggests that the nitrogen:phosphorus ratio is enhanced by greater retention of phosphorus compared to nitrogen only in deep lakes with relatively short residence times, such as Lake Brunner. In contrast, in most lakes differential retention is expected to result in lower nitrogen:phosphorus ratios.

A study (2004–11) of a dairy catchment stream entering an oligotrophic lake in an area of very high rainfall (~5 m year–1) yielded median concentrations of total nitrogen (TN), total phosphorus (TP), suspended sediment (SS) and Escherichia coli (E. coli) of 0.584, 0.074 and 3.7 g m–3, and 405/100 mL (most probable number method), respectively. Trend analysis indicated significant (P < 0.01) decreases for TN (–0.08 ± 0.02 g m–3 year–1), TP (–0.01 ± 0.005 g m–3 year–1) and SS (–0.45 ± 0.14 g m–3 year–1) and were partly attributable to improved exclusion of cattle from the stream. Water balance calculations indicated that approximately one-half the rainfall left as deep drainage that by-passed catchment outlet flow recorders. Estimates of catchment yields for TN were improved by taking into account groundwater hydrology and concentrations from well samples. Storm-flow monitoring inflows exceeding the 97.5th percentile contributed ~40% of total loads leaving the catchment so that specific yields for SS, TN and TP augmented by groundwater inputs and storm flows were ~960, 45 and 7 kg ha–1 year–1, respectively. These compared well with modelled results for losses from dairy farms in the catchment of 40–60 kg N ha–1 year–1 and 5–6 kg P ha–1 year–1 and indicated that attenuation losses were relatively small.

Lake Brunner, an oligotrophic monomictic lake on the West Coast of the South Island of New Zealand, is under pressure from urban expansion and increased farming activity, which has led to concern for the effects on water quality in the lake. Epilimnetic nitrogen, phosphorus and chlorophyll a concentrations have increased since 1992, and Secchi depth decreased. This suggests an increased algal productivity caused by increased nutrient inputs, further supported by increased hypolimnetic oxygen depletion since 1992. These observations are likely to have resulted from enhancement of pasture drainage and effluent inputs from expanding dairy farms. The Vollenweider model predicted a mean phosphorus concentration in the lake close to that observed, from estimated catchment loading, suggesting that the Vollenweider model adequately estimated the retention of phosphorous. With the Vollenweider model the effects of potential future loading scenarios were explored. Modelling suggested that a 70% increase in phosphorus loading could turn the lake into a mesotrophic state. Trend analysis of total phosphorus suggests that, with present land uses in the catchment (intensive dairy farming) continuing to develop at the same rate using the same land management practises, this transition to a mesotrophic state will occur by 2040.

Five streams in catchments with pastoral dairy farming as the dominant land use were monitored for periods of 7–16 years to detect changes in response to adoption of best management practices (BMPs). Stream water quality was degraded at the start with respect to N, P, suspended solids (SS) and E. coli concentrations, and was typical of catchments with intensive pastoral agriculture land use. Trend analysis showed a decrease in SS concentration for all streams, generally increasing water clarity, and lower E. coli concentrations in three of the streams. These are attributed to improved stream fencing (cattle exclusion) and greater use of irrigation for treated effluent disposal with less reliance on pond systems discharging to streams. Linkages between water quality and farm actions based on survey data were used to develop BMPs that were discussed at stakeholder workshops. Generic and specific BMPs were developed for the five catchments. The 3–7 year periodicity of major climate cycles, as well as market forces and a slow rate of farmer adoption of simple BMPs mean that monitoring programs in New Zealand need to be much longer than 10 years to detect changes caused by farmer actions. Long-term monitoring is also needed to detect responses to newly legislated requirements for improved water quality.

Increasing demand for global food production is leading to greater use of irrigation to supplement rainfall and enable more intensive use of land. Minimizing adverse impacts of this intensification on surface water and groundwater resources is of critical importance for the achievement of sustainable land use. In this paper we examine the linkages between irrigation runoff and resulting changes in quality of receiving surface waters and groundwaters in Australia and New Zealand. Case studies are used to illustrate impacts under different irrigation techniques (notably flood and sprinkler systems) and land uses, particularly where irrigation has led to intensification of land use. For flood irrigation, changes in surface water contaminant concentrations are directly influenced by the amount of runoff, and the intensity and kind of land use. Mitigation for flood irrigation is best achieved by optimizing irrigation efficiency. For sprinkler irrigation, leaching to groundwater is the main transport path for contaminants, notably nitrate. Mitigation measures for sprinkler irrigation should take into account irrigation efficiency and the proximity of intensive land uses to sensitive waters. Relating contaminant concentrations in receiving groundwaters to their dominant causes is often complicated by uncertainty about the subsurface flow paths and the possible pollutant sources, viz. drainage from irrigated land. This highlights the need for identification of the patterns and dynamics of surface and subsurface waters to identify such sources of contaminants and minimize their impacts on the receiving environments.

Water quantity and quality were monitored for 3 years in a 360-m-long wetland with riparian fences and plants in a pastoral dairy farming catchment. Concentrations of total nitrogen (TN), total phosphorus (TP) and Escherichia coli were 210–75,200 g N m−3, 12–58,200 g P m−3 and 2–20,000 most probable number (MPN)/100 ml, respectively. Average retentions (±standard error) for the wetland over 3 years were 5 ± 1%, 93 ± 13% and 65 ± 9% for TN, TP and E. coli, respectively. Retentions for nitrate–N, ammonium–N, filterable reactive P and particulate C were respectively −29 ± 5%, 32 ± 10%, −53 ± 24% and 96 ± 19%. Aerobic conditions within the wetland supported nitrification but not denitrification and it is likely that there was a high conversion rate from dissolved inputs of N and P in groundwater, to particulate N and P and refractory dissolved forms in the wetland. The wetland was notable for its capacity to promote the formation of particulate forms and retain them or to provide conditions suitable for retention (e.g. binding of phosphate to cations). Nitrogen retention was generally low because about 60% was in dissolved forms (DON and NOX–N) that were not readily trapped or removed. Specific yields for N, P and E. coli were c. 10–11 kg N ha−1 year−1, 0.2 kg P ha−1 year−1 and ≤109 MPN ha−1 year−1, respectively, and generally much less than ranges for typical dairy pasture catchments in New Zealand. Further mitigation of catchment runoff losses might be achieved if the upland wetland was coupled with a downslope wetland in which anoxic conditions would promote denitrification.

Abstract Poor water quality (high concentrations of nitrogen (N), phosphorus (P), suspended solids (SS), and faecal bacteria) in Waiokura Stream, southern Taranaki, New Zealand, is attributed to diffuse and point source (PS) inputs from dairy farming. Trend analysis of concentration time‐series data (2001–2008) and annual yields (i.e., stream load divided by catchment area) showed that significant improvements occurring since 2001 may be attributed to changes in farming practices and riparian management. Yields of filterable reactive P, total P and SS declined by 25–40% as a result of increased riparian protection, a reduction in dairy shed effluent (DSE) pond discharges from 8 to 6 with conversion to land irrigation, and a 25% reduction in the average application rate of P fertiliser. Median annual Escherichia coli concentrations declined at a rate of 116 per 100 ml per year, as a result of fewer PS discharges and improved riparian management. Thus, improvements in stream water quality were attributed to adoption of on‐farm best management practices, fewer DSE discharges and riparian management involving permanent livestock exclusion from stream banks and riparian planting to mitigate runoff from pasture. During 2001–06, N fertiliser use increased by 30% and, with a 130% increase in supplementary cattle feed during 2003–08, led to an increase in average milk solids production 1021 to 1262 kg ha−1 during 2001–06 with the increased production likely associated with increased N leaching losses. Total N and nitrate‐N concentrations and yields increased during 2001–07 as a result of the intensification in land use and increased N cycling. Stream invertebrate surveys using the macroinvertebrate community index (MCI) metric showed little improvement in MCI during 2002–07, probably because of the relatively short timeframe of this study and because water temperatures were not a limiting factor for invertebrate communities. The absence of native forest streams in the proximity of Waiokura Stream that might act as sources of sensitive species to recolonise the restored stream should also be considered as a constraint to improvements in biological community structure.

Little is known about seepage wetlands, located within agricultural landscapes, with respect to removing nitrate (NO3−) from agricultural catchments, mainly through gaseous emissions of nitrous oxide (N2O) and dinitrogen (N2) via denitrification. These variables were quantified using a push–pull technique where we introduced a subsurface water plume spiked with 15N-enriched NO3− and 2 conservative tracers [bromide (Br−) and sulfur hexafluoride (SF6)] into each of 4 piezometers and extracted the plume from the same piezometers throughout a 48-h period. To minimise advective and dispersive flux, we placed each of these push–pull piezometers within a confined lysimeter (0.5 m diameter) installed around undisturbed wetland soil and vegetation. Although minimal dilution of the subsurface water plumes occurred, NO3−-N concentration dropped sharply in the first 4 h following dosing, such that NO3−-limiting conditions (<2 mg/L of NO3-N) for denitrification prevailed over the final 44 h of the experiment. Mean subsurface water NO3− removal rates during non-limiting conditions were 15.7 mg/L.day. Denitrification (based on the generation of isotopically enriched N2O plus N2) accounted for only 7% (1.1 mg/L.day) of the observed groundwater NO3− removal, suggesting that other transformation processes, such as plant uptake, were responsible for most of the NO3− removal. Although considerable increases in 15N-enriched N2O levels were initially observed following NO3− dosing, no net emissions were generated over the 48-h study. Our results suggest that this wetland may be a source of N2O emissions when NO3− concentrations are elevated (non-limited), but can readily remove N2O (function as a N2O sink) when NO3− levels are low. These results argue for the use of engineered bypass flow designs to regulate NO3− loading to wetland denitrification buffers during high flow events and thus enhance retention time and the potential for NO3−-limiting conditions and N2O removal. Although this type of management may reduce the full potential for wetland NO3− removal, it provides a balance between water quality goals and greenhouse gas emissions.

This paper describes key linkages between land management activities and stream water quality for a 5230ha catchment used for intensive pastoral agriculture in southern New Zealand. Due to low annual rainfall and the wide coverage of soils with low available water-holding capacities, flood irrigation of the 2400ha of flat land within the catchment is an important feature impacting on farm business profitability and stream health. Water quality variables and nutrient and sediment yield estimates are reported for a four-year period. This monitoring shows that some improvement in farm environmental performances would generally be desirable, with stream concentrations of nutrients (N and P), sediment and faecal bacteria regularly exceeding guidelines recommended for surface waters. Field measurements, farm management surveys and farm systems modelling have identified some land management practices that appear to be key sources of many of these pollutants. Border dyke irrigation runoff has a potentially large effect on a range of water quality parameters, due to both the excessive stream flows generated by over-watering and the entrainment of P, N and faecal bacteria in this flow as it passes from land to stream. Stock access to some of the remaining un-fenced lengths of the stream was also recognised as an important land management practice that needed to be addressed if some of the key catchment values identified by stakeholders, such as maintaining a healthy trout fishery and a stream suitable for recreational use, were to be protected. Assessments of the effectiveness and cost-effectiveness of a number of potential mitigation practices identified that managements which targeted reducing irrigation runoff (e.g. by installing bunds or using appropriate watering times) and fencing and planting riparian margins showed the greatest potential to meet these key values with least cost to farm businesses. Other farm practices were also identified that incurred nil or minimal cost while also delivering small or moderate benefits to stream water quality.

Water quality in agricultural catchments tends to be worse than in forested (native or exotic) catchments.Reduced water quality tends to have significant effects on the ecosystem of streams, including increased nuisance algal and plant growth (eutrophication) associated with nutrient input, toxicity to aquatic life due to ammonia, faecal contamination, and loss of habitat or spawning areas due to sedimentation.An analysis of catchment contaminant loads from 38 studies conducted since 1975 was carried out to determine if there were differences in loads between land uses under different livestock (dairy, sheep, sheep-and-beef (mixed), deer) and non-agricultural.Significantly more N was lost from dairy catchments than catchments with other land uses, and more sediment lost from deer catchments than other catchments.Median loads of N were greatest from dairy > deer = mixed > sheep > non-agricultural; while loads of P were greatest for deer = mixed > dairy > sheep > non-agricultural; and for sediment, deer > sheep > mixed > dairy > non-agricultural.This information should be considered in catchments of mixed stock types to target the most pertinent mitigation practice for improving water quality. For example, if a stream in a catchment with deer and sheep present is choked with algae and limited by inputs of P, mitigation should focus on deer rather than sheep.

Lentic wetlands are usually regarded as the most important natural freshwater sources of methane (CH4) and nitrous oxide (N2O) to the atmosphere, and very few studies have quantified the importance of lowland streams in trace gas emissions. In this study, we estimated fluxes of CH4 and N2O in three macrophyte-rich, lowland agricultural streams in New Zealand, to place their trace gas emissions in context with other sources and investigate the value of minimising their emissions from agricultural land. All three streams were net sources of both gases, with emission of CH4 ranging from <1 to 500 μmol m−2 h−1 and of N2O ranging from <1 to 100 μmol m−2 h−1 during mid-summer. For CH4, both turbulent diffusion across the surface and ebullition of sediment gas bubbles were important transport processes, with ebullition accounting for 20–60% of the emissions at different sites. The emissions were similar on a per area basis to other major global sources of CH4 and N2O. Although small on a catchment scale compared to emissions from intensively grazed pastures, they were significant relative to low-intensity pastures and other agricultural land uses. Because hydraulic variables (viz. depth, velocity and slope) strongly influence turbulent diffusion, complete denitrification can best proceed to N2 as the dominant end-product (rather than N2O) in riparian wetlands, rather than in open stream channels where N2O fluxes are sometimes very large.

This chapter aims to provide information on how land uses in general, and agriculture in particular, impact waterways. It first discusses the relationships between land use, hydrology and runoff, and then examines the pathways and mechanisms connecting land with water, and the nature and impacts of the pollutants and habitat destruction.