Agricultural intensification on the Canterbury Plains in New Zealand has lead to the degradation of natural streams and rivers through lowering of water quality and significant reduction of surface flows from the use of ground and surface water resources. However, this same agricultural expansion has led to the development of a network of permanently flowing open water races to supply stock water to farms across the Canterbury Plains. Stock water races form an extensive network, with approximately 6,500 km of races. Initially I surveyed 62 water races and compared habitat characteristics, water quality, benthic invertebrate and fish communities with nearby natural streams. Races are characterised physically by straight, narrow and shallow channels, and small, uniform substrate. Water races are more turbid than natural streams, and can have high summer temperatures. The benthic macroinvertebrate communities of water races contained a range of taxa, including some not found in natural streams, but communities were less diverse than communities in natural streams, and tended to be dominated by a limited set of generalist taxa. A longitudinal study of three water races showed gradients in physical characteristics of races, including a downstream decrease in channel width, water depth, current velocity and substrate size. However, few strong longitudinal changes to community structure were found, as the generalist taxa commonly occurring in water races were able to tolerate conditions throughout the race network. To test if macroinvertebrate communities were limited by the homogeneous habitat of water races, I conducted a substrate manipulation experiment, where large cobbles and small boulders were added to reaches in five water races. Despite an increase in substrate and current heterogeneity, there were few significant changes to the macroinvertebrate communities over the four months of the manipulation. This outcome does not eliminate low habitat heterogeneity as a limiting factor for water race communities. Rather, the benthic invertebrate community throughout the water race network is a product of the homogeneous habitat, which limits the availability of colonists of taxa that would benefit from increased habitat diversity. A survey of the fish assemblages of water races found races had a depauperate fish community. Only two species were commonly found in water races, and the average species richness of races was 1.5. By contrast natural streams had a higher diversity of fish species (mean 4 three species), and contained representatives of a greater number of species that are typical of streams and rivers on the Canterbury Plains. My research has shown that stock water races provide an important source of aquatic biodiversity on the plains, both in addition to natural streams and in their own right. However, the biodiversity value of stock water races could be improved with enhancement of in-stream habitat.

Agricultural intensification on the Canterbury Plains in New Zealand has lead to the degradation of natural streams and rivers through lowering of water quality and significant reduction of surface flows from the use of ground and surface water resources. However, this same agricultural expansion has led to the development of a network of permanently flowing open water races to supply stock water to farms across the Canterbury Plains. Stock water races form an extensive network, with approximately 6,500 km of races. Initially I surveyed 62 water races and compared habitat characteristics, water quality, benthic invertebrate and fish communities with nearby natural streams. Races are characterised physically by straight, narrow and shallow channels, and small, uniform substrate. Water races are more turbid than natural streams, and can have high summer temperatures. The benthic macroinvertebrate communities of water races contained a range of taxa, including some not found in natural streams, but communities were less diverse than communities in natural streams, and tended to be dominated by a limited set of generalist taxa. A longitudinal study of three water races showed gradients in physical characteristics of races, including a downstream decrease in channel width, water depth, current velocity and substrate size. However, few strong longitudinal changes to community structure were found, as the generalist taxa commonly occurring in water races were able to tolerate conditions throughout the race network. To test if macroinvertebrate communities were limited by the homogeneous habitat of water races, I conducted a substrate manipulation experiment, where large cobbles and small boulders were added to reaches in five water races. Despite an increase in substrate and current heterogeneity, there were few significant changes to the macroinvertebrate communities over the four months of the manipulation. This outcome does not eliminate low habitat heterogeneity as a limiting factor for water race communities. Rather, the benthic invertebrate community throughout the water race network is a product of the homogeneous habitat, which limits the availability of colonists of taxa that would benefit from increased habitat diversity. A survey of the fish assemblages of water races found races had a depauperate fish community. Only two species were commonly found in water races, and the average species richness of races was 1.5. By contrast natural streams had a higher diversity of fish species (mean 4 three species), and contained representatives of a greater number of species that are typical of streams and rivers on the Canterbury Plains. My research has shown that stock water races provide an important source of aquatic biodiversity on the plains, both in addition to natural streams and in their own right. However, the biodiversity value of stock water races could be improved with enhancement of in-stream habitat.

Aquatic ecosystems are especially vulnerable to human impacts associated with agricultural land-use, which provide multiple stressors altering community composition, important ecosystem functions and human valued properties of freshwaters. However, the increased occurrence of excessive levels of nitrate-nitrogen has raised major concerns about toxicity and stress on aquatic life, especially in regions such as the Canterbury Plains, New Zealand. The aims of this thesis were to identify nitrate-nitrogen effects on stream communities, and additionally provide field data to inform proposed national bottom lines for nutrients in New Zealand streams. A field survey was conducted on 41 small streams on the Canterbury Plains spanning a nitrate-nitrogen gradient (mean 0.4 – 11.3 mg/L). Spot nitrate-nitrogen was collected during and after the field survey to measure temporal variation in stream nitrate-nitrogen concentration for six months. This showed nitrate-nitrogen concentration varied between season and sub-region, where concentrations increased in winter and Ashburton had higher nitrate-nitrogen than Rangiora and Lincoln, respectively. These regimes of nitrate-nitrogen showed similar patterns in mean, median and maximum concentrations. To be confident my spot nitrate-nitrogen provided a true representation of long-term water chemistry, I compared Environment Canterbury 12 monthly data with my six monthly data in a sub-set of 15 sites. This comparison showed similar nitrate-nitrogen patterns and range of values between the two datasets. I then compared 12 common benthic invertebrate biotic metrics with my nitrate-nitrogen data and found none were correlated with this contaminant. For example, the Macroinvertebrate Community Index and quantitative variant (QMCI) derived to measure the response to organic pollution provided inconsistent results when applied to my streams. Nevertheless, gut content stoichiometry of the common mayfly grazer Deleatidium spp. indicated improvement in food quality (lower C:N ratio) with higher nitrate-nitrogen concentrations. These results indicated either nitrate-nitrogen does not alter invertebrate structural metrics across this nitrate-nitrogen gradient, or that these biotic metrics measure community structure aspects not affected by nitrate-nitrogen. I then investigated possible community composition patterns across the nitrate-nitrogen gradient. Unconstrained ordination (on presence/absence data) showed invertebrate communities at my sites were influenced primarily by discharge and shade, with the next most important driver being nitrate-nitrogen. A constrained ordination (on the same data) testing the singular effect of nitrate-nitrogen showed a marginally non-significant change in composition, with higher variability in community composition at higher nitrate-nitrogen concentrations. A further aim of my study was to test the draft nitrate-nitrogen bands proposed by Hickey (2013). These nitrate-nitrogen bands may advise guidelines to protect aquatic organisms as required by the National Policy Statement on Freshwater. Analysis of my invertebrate communities showed differences in composition, particularly at < 1 and > 6.9 mg/L bands. Several predatory caddisfly taxa: Triplectides, Neurochorema and Oeconesus were identified as potential indicator species of communities associated with low nitrate-nitrogen. These findings show that nitrate-nitrogen effects are difficult to detect, and that it is not the main driver of community composition in Canterbury streams. However, nitrate-nitrogen may be an important stressor for sensitive benthic invertebrate communities, as effects were observed on pollution tolerant organisms in this study. Therefore, this research has implications for freshwater ecologists and environmental managers striving to improve the health of streams on the Canterbury Plains.

With the increasing degradation of rivers worldwide, an understanding of spatial and temporal patterns in freshwater quality is important. Water quality is highly variable in space and time, yet this is largely overlooked at the scale of stream catchments. I employed a landscape ecology approach to examine the spatial patterning of water quality in complex, impacted stream networks on the Canterbury Plains of the South Island of New Zealand, with the goal of understanding how land-use effects proliferate through stream systems. In particular, I used “snapshot” sampling events in conjunction with spatial modelling and longitudinal profiles to investigate the ways in which spatial and environmental factors influence the variability of water quality in stream networks. Spatial eigenfunction analyses showed that distance measures, which took into account variable connectivity by flow and distance along the stream between sites, explained more spatial variance in water quality than traditional distance metrics. Small upstream reaches were more spatially and temporally variable than main stems (under summer base-flow conditions). The extent of spatial variation in water quality differed between stream networks, potentially depending on linkages to groundwater and the surrounding landscape. My results indicated that the water quality of headwater streams can have a disproportionate influence over water quality throughout an entire network. I investigated spatio-temporal patterns in water quality more intensively in one stream network, the Cam River, in which I found consistent spatial pattern through time. The relative balance between nutrient inputs (pollution and groundwater) and in-stream conditions influenced the spatial pattern of water quality, as well as that of several ecosystem processes which I measured simultaneously. The spatially intensive and explicit approach has allowed identification of key factors controlling water quality and ecosystem processes throughout the Cam River. This research highlights the importance of taking a spatially explicit approach when studying stream water quality and that such an approach could be insightful and will contribute to solving current stream management problems.

By far the greatest proportion of the 1,150,000 acres of light land in Canterbury is found on the Canterbury Plain. This plain, originally covered by "low tussock" and of easy contour, was enticing to the early pioneers and became one of the earliest areas in New Zealand to be settled and farmed. Over the years it has developed into one of the most intensively farmed and productive areas of New Zealand. The dominant characteristic of the light land of Canterbury is undoubtedly the climate. The rainfall is reasonably evenly distributed over the year, but because of the low humidity, high temperatures and warm winds experienced over the summer in association with a free draining soil, the effectiveness of the rainfall over this period is drastically reduced. Consequently active plant growth is severely restricted for several months over the summer, and occasionally extends into the spring and/or autumn periods. The uncertainty as to the length and severity of this restricted growth period and the associated problem of equating the variable feed supply to the stock requirements, both within and between years, is the basic problem confronting the light land farmer. In spite of the environmental difficulties the productivity of the light land has increased several fold since early settlement. The original holdings on the Canterbury Plain were large with their boundaries on the rivers so as to provide access to water. Fine wool sheep were extensively grazed. However the introduction of refrigerated shipping and the extension of the water race system in the 1880's brought about a reduction in the size of holdings and a change in the pattern of farming. Dual purpose sheep were run and by the 1930's in response to favourable crop prices the system of diversified farming was firmly established. Unfortunately this system placed excessive emphasis on cash cropping, particularly on the light soils. Soil fertility was drained, structure severely damaged and subsequent pasture establishment and survival poor. In the late 1940s and early 1950s, with declining crop yields and with more favourable prices being obtained for fat lambs and particularly wool, the emphasis shifted from cropping to livestock farming. The carrying capacity however, was restricted by the reduced soil fertility and poor quality pastures and an environment in which climatic uncertainty tended to inhibit the rapid expansion of stock numbers. The results of research work carried out at the various institutions in Canterbury over the years have undoubtedly promoted a greater understanding and appreciation of the problems confronting the farmer and the limitations of the particular environment in which he must operate. For example, pasture species more suited to the low fertility conditions and climate were introduced with spectacular results. The most significant of these was subterranean clover (introduced in the 1930s), noted for its ability to withstand the summer droughts, to regenerate in the autumn, and to provide an increased bulk of feed in the spring. Research work had shown that both lime and phosphate were necessary on the light land soils, if high pasture production and persistency was to be expected. Soil fertility increased subsequent to a reduction in the emphasis on cropping and with the higher levels of fertiliser application. High fertility pasture species (e.g. white clover and lucerne), were introduced and not only gave higher and more reliable total production but exhibited improved seasonal spread of production. Investigations into pasture diseases and stock health provided answers to specific problems. Research into flock management generally and in comparisons between the productivity of various sheep breeds indicated the most suitable type of flock and breed for the light land farmer. From this and other research work (in conjunction with the observations of leading farmers in the area), an efficient system of light land farming has gradually evolved in which many of the basic problems have been overcome and which has resulted in a raising of the carrying capacity of the light land from ¾ stock unit per acre in the 1930's to 3½-4 stock units per acre at present. A central feature of this system (particularly at high stocking rates), is the high degree of flexibility incorporated in both the stock policies and feed supplies. Where the objective function is to maximise productivity over a period of years, it is essential to utilise the available spring feed efficiently while maintaining the ability to destock when confronted with feed shortages in the spring and early summer. Because of the fluctuating feed supply, which is characteristic of light land, the need to maintain feed reserves and to incorporate a high degree of flexibility in the stock policy is evident if the feed supply and demand are to be equated. In summary, the increased productivity can be attributed to two factors: (1) The ability to grow a greatly increased quantity of herbage per acre with an improved seasonal pattern of production. (2) A more efficient utilisation of the herbage produced. Unlike his counterpart in more reliable farming districts, the light land farmer operates in an environment of uncertainty. Yield uncertainty, particularly at high stocking rates, is the major problem to be overcome and this dictates very largely the system of farming adopted. Price uncertainty is also a significant aspect of light land farming because of the reliance on a limited range of products and the inability to diversify. In an analysis of physical and financial data collected from a sample of light land farms in Canterbury (1) there was no evidence to suggest that any one particular pattern of output was superior to all others. This result was surprising, but may reflect the uncertainty inherent in the environment. (1) For a full discussion on this, see Section 3.3.2(a). Alternatively it may infer that the actual patterns of production are less important than the managerial skill with which they are implemented. These results pointed to the need to explore more fully the following facets of light land management: (1) Given a developed farm, is there any one optimal pattern of production which (a) generates increased profit under average seasonal and price conditions, and (b) is subject to only small variations in profit under changing seasonal and price conditions? (2) Given the potential for the development and expansion of light land farming, how profitable is this from the individual farmer's viewpoint? If, in an evaluation of the first problem, high levels of productivity are shown to be profitable on existing well developed farms, then a reallocation of resources to obtain the desired combination should be recommended. An optimum combination of enterprises shown by such an analysis might well serve as the goal where an undeveloped potential still exists on a farm and where a reallocation and intensification in the use of resources is necessary if productivity is to be increased. In this study of light land farming two case farms have been used and although the results refer specifically to these particular farms, some conclusions of a general nature are possible. In Chapter II the physical characteristics of the area are described. In Chapter III a review of the research into specific problems relating to the management of light land is presented. This is followed in Chapter IV by an explanation of the technical principles of light land farming which have evolved. Chapter V is devoted to the comparison of some of the production possibilities open to the light land farmer using linear programming. An analysis of light land development 1s presented in Chapter VI, while Chapter VII presents the conclusions and summary of the study.

By far the greatest proportion of the 1,150,000 acres of light land in Canterbury is found on the Canterbury Plain. This plain, originally covered by "low tussock" and of easy contour, was enticing to the early pioneers and became one of the earliest areas in New Zealand to be settled and farmed. Over the years it has developed into one of the most intensively farmed and productive areas of New Zealand. The dominant characteristic of the light land of Canterbury is undoubtedly the climate. The rainfall is reasonably evenly distributed over the year, but because of the low humidity, high temperatures and warm winds experienced over the summer in association with a free draining soil, the effectiveness of the rainfall over this period is drastically reduced. Consequently active plant growth is severely restricted for several months over the summer, and occasionally extends into the spring and/or autumn periods. The uncertainty as to the length and severity of this restricted growth period and the associated problem of equating the variable feed supply to the stock requirements, both within and between years, is the basic problem confronting the light land farmer. In spite of the environmental difficulties the productivity of the light land has increased several fold since early settlement. The original holdings on the Canterbury Plain were large with their boundaries on the rivers so as to provide access to water. Fine wool sheep were extensively grazed. However the introduction of refrigerated shipping and the extension of the water race system in the 1880's brought about a reduction in the size of holdings and a change in the pattern of farming. Dual purpose sheep were run and by the 1930's in response to favourable crop prices the system of diversified farming was firmly established. Unfortunately this system placed excessive emphasis on cash cropping, particularly on the light soils. Soil fertility was drained, structure severely damaged and subsequent pasture establishment and survival poor. In the late 1940s and early 1950s, with declining crop yields and with more favourable prices being obtained for fat lambs and particularly wool, the emphasis shifted from cropping to livestock farming. The carrying capacity however, was restricted by the reduced soil fertility and poor quality pastures and an environment in which climatic uncertainty tended to inhibit the rapid expansion of stock numbers. The results of research work carried out at the various institutions in Canterbury over the years have undoubtedly promoted a greater understanding and appreciation of the problems confronting the farmer and the limitations of the particular environment in which he must operate. For example, pasture species more suited to the low fertility conditions and climate were introduced with spectacular results. The most significant of these was subterranean clover (introduced in the 1930s), noted for its ability to withstand the summer droughts, to regenerate in the autumn, and to provide an increased bulk of feed in the spring. Research work had shown that both lime and phosphate were necessary on the light land soils, if high pasture production and persistency was to be expected. Soil fertility increased subsequent to a reduction in the emphasis on cropping and with the higher levels of fertiliser application. High fertility pasture species (e.g. white clover and lucerne), were introduced and not only gave higher and more reliable total production but exhibited improved seasonal spread of production. Investigations into pasture diseases and stock health provided answers to specific problems. Research into flock management generally and in comparisons between the productivity of various sheep breeds indicated the most suitable type of flock and breed for the light land farmer. From this and other research work (in conjunction with the observations of leading farmers in the area), an efficient system of light land farming has gradually evolved in which many of the basic problems have been overcome and which has resulted in a raising of the carrying capacity of the light land from ¾ stock unit per acre in the 1930's to 3½-4 stock units per acre at present. A central feature of this system (particularly at high stocking rates), is the high degree of flexibility incorporated in both the stock policies and feed supplies. Where the objective function is to maximise productivity over a period of years, it is essential to utilise the available spring feed efficiently while maintaining the ability to destock when confronted with feed shortages in the spring and early summer. Because of the fluctuating feed supply, which is characteristic of light land, the need to maintain feed reserves and to incorporate a high degree of flexibility in the stock policy is evident if the feed supply and demand are to be equated. In summary, the increased productivity can be attributed to two factors: (1) The ability to grow a greatly increased quantity of herbage per acre with an improved seasonal pattern of production. (2) A more efficient utilisation of the herbage produced. Unlike his counterpart in more reliable farming districts, the light land farmer operates in an environment of uncertainty. Yield uncertainty, particularly at high stocking rates, is the major problem to be overcome and this dictates very largely the system of farming adopted. Price uncertainty is also a significant aspect of light land farming because of the reliance on a limited range of products and the inability to diversify. In an analysis of physical and financial data collected from a sample of light land farms in Canterbury (1) there was no evidence to suggest that any one particular pattern of output was superior to all others. This result was surprising, but may reflect the uncertainty inherent in the environment. (1) For a full discussion on this, see Section 3.3.2(a). Alternatively it may infer that the actual patterns of production are less important than the managerial skill with which they are implemented. These results pointed to the need to explore more fully the following facets of light land management: (1) Given a developed farm, is there any one optimal pattern of production which (a) generates increased profit under average seasonal and price conditions, and (b) is subject to only small variations in profit under changing seasonal and price conditions? (2) Given the potential for the development and expansion of light land farming, how profitable is this from the individual farmer's viewpoint? If, in an evaluation of the first problem, high levels of productivity are shown to be profitable on existing well developed farms, then a reallocation of resources to obtain the desired combination should be recommended. An optimum combination of enterprises shown by such an analysis might well serve as the goal where an undeveloped potential still exists on a farm and where a reallocation and intensification in the use of resources is necessary if productivity is to be increased. In this study of light land farming two case farms have been used and although the results refer specifically to these particular farms, some conclusions of a general nature are possible. In Chapter II the physical characteristics of the area are described. In Chapter III a review of the research into specific problems relating to the management of light land is presented. This is followed in Chapter IV by an explanation of the technical principles of light land farming which have evolved. Chapter V is devoted to the comparison of some of the production possibilities open to the light land farmer using linear programming. An analysis of light land development 1s presented in Chapter VI, while Chapter VII presents the conclusions and summary of the study.