Agriculture must increase food production to support a growing global population; however, this required increase in production is potentially restricted by freshwater supply for irrigation. Consequently, relying on irrigation to enhance agricultural production is only a partial solution and increasing water use efficiency (WUE) is an important priority. Increasing pastoral diversity has been shown to increase pasture production, especially during warm and dry growing conditions. The positive effect of increasing diversity is typically linked to complementarity in plant traits such as rooting depth and facilitation among species or the inclusion of plants with divergent life histories that use water more efficiently. However, studies of differences in evaporation (E) and WUE between pastures of contrasting diversity at the ecosystem scale are scarce. The objective of this thesis was to contrast seasonal WUE and production strategies of a traditional ryegrass-clover and a more diverse pasture at the ecosystem scale. This study was conducted on an intensively managed commercial dairy farm in the Waikato region of New Zealand. The first objective of this thesis (Chapter 3) was to quantify spatial and temporal variation in E from traditional ryegrass-clover pastures. These baseline measurements of E from intensively managed pastures are important for informing water resource decision making and validation of hydrologic models and remote sensing methods. Evaporation was measured simultaneously over existing ryegrass-clover pasture at 3 sites on the same farm using the eddy covariance (EC) technique. At an annual timescale spatial (770 – 783 mm) and temporal (759 – 776 mm) variations in E were less than 3%. This low variation largely occurred because E was strongly controlled by net radiation (r² = 0.81, p < 0.01, daytime, half-hourly), which did not vary much between sites and years. However, seasonally E was strongly limited when volumetric moisture content (VMC) declined below permanent wilting point. Grazing events, that removed about 55% of leaf material, had no effect on E during autumn and winter but reduced E by up to 5% during summer and spring and it was likely soil water E was compensating for reductions in transpiration. Agreement between E measured by eddy covariance (EEC) and FAO-56 reference crop modelled E (E₀) was good when soil moisture limitation was not occurring. However, during periods of soil moisture limitation, Eo exceeded EEC and a correction factor was needed. The water stress coefficient (Kₛ) and a simple three bin VMC correction factor (KVMC) was trialled and both approaches improved agreement between modelled and measured E but each method had limitations. Further study is needed to determine a simple and robust routine to model E from temperate pastures under soil moisture limitation. The objective of Chapter 4 was to compare, evaporation, gross primary production (GPP), and ecosystem WUE (EWUE) between a traditional ryegrass-clover pasture and a more diverse pasture which included multiple grasses, legumes, and herbs. It was hypothesised that the more diverse pasture, which included deeper rooting species, would be more productive during dry and warm periods because of increased access to soil water and increased WUE associated with the inclusion of more legumes. Carbon exchange and E was measured between September 2012 and June 2015 using a paired EC experimental design with a pre-treatment period (September 2012 to April 2013) to identify any pre-existing site differences. A new ryegrass-clover (New Rye) and a new diverse (New Mix) pasture were established in April 2013 following herbicide application by the direct drill method. Between June 2013 and June 2015 above ground harvestable dry matter (DM) production was also measured. Post-treatment, GPP was higher at New Mix during both dry (4.0%) and wet (8.8%) summer conditions and these increases were supported by DM production measurements. Evaporation rates were not significantly different and consequently both EWUE (GPP/E) and harvest WUE (HWUE, DM/E) were higher at New Mix during summer conditions. No differences in production (GPP and DM) were found during shoulder season conditions while E was significantly lower at New Mix (5.8%) resulting in higher shoulder season EWUE. Both GPP and DM production were lower at New Mix during cool winter conditions while E was not different resulting in lower cool season EWUE and HWUE at New Mix. At an annual scale both production and EWUE were similar between treatments because summer increases at New Mix were compensated for by winter increases at New Rye. Consequently the strategic integration of both ryegrass-clover and more diverse swards on different parts of a farm would likely maintain more even year-round productivity. Increasing plant diversity was shown to increase production during warm dry and warm wet conditions and WUE during warm wet and shoulder season conditions (Chapter 4). However, differences between treatments were small (~ 5%) and this likely occurred because ryegrass was a dominant species at both sites. The optimal mix of species is expected to vary spatially dependent on climate, soil type, and plant water requirements. Consequently, a rapid and cost effective method to screen for productive pasture plant species and mixes with high WUE in situ at farm scale is needed. The objective of Chapter 5 was to test the correlation between WUE calculated from bulk leaf ∆¹³C (a measure of intrinsic WUE, WUEi) and EWUE by comparing the seasonal progression of bulk leaf ∆¹³C and EWUE measured at the paddock scale using EC. Mixed species bulk leaf biomass samples were harvested pre-grazing, dried, sub-sampled, ground, and the ratio of ¹³C to ¹²C was measured. After accounting for the seasonal changes in the atmospheric vapour pressure deficit (VPD) on WUEi, following Farquhar et al. (1982), strong positive correlations were found between WUE calculated from ∆¹³C (WUE∆¹³C) and EWUE (r² > 0.79, p < 0.01) at both New Rye and New Mix. Additional seasonal measurements of production and ∆¹³C on individual plant species grown together at New Mix found important differences in WUE∆¹³C and production among co-existing pasture species. These results indicated ∆¹³C was a suitable tool for comparing WUE between different pasture swards and, importantly, differences in WUE∆¹³C between co-existing pasture species indicated pasture mixtures could be manipulated to optimise WUE. Through this PhD some of the first replicated field-scale measurements of E from intensively grazed pastures were published and spatial and temporal variation in E was low because of the dominant control by net radiation (Chapter 3). Chapter 4 demonstrated increasing pasture diversity had a small (~5%) but important positive effect on warm season EWUE and production and shoulder season EWUE. Ryegrass was a dominant species in both treatments and it is possible larger improvements in EWUE could have been achieved by further optimising pasture mixtures in the sward. Chapter 5 examined the correlations between EWUE and WUE∆¹³C with the goal of developing a rapid and cost effective method to compare WUE between pasture swards, and thereby optimise species mixtures. Strong correlations were shown in addition to important differences in WUE∆¹³C between co-existing pasture species. Combined the findings of Chapters 4 and 5 strongly indicated that mixtures of pasture species within a sward could be manipulated to increase EWUE. However, it is expected that optimal species mixtures are site specific depending on soil, climate and plant water requirements. Following further ecosystem and plant level study to confirm results found at this site, I envisage ∆¹³C measurements could be used by farm advisors, alongside production monitoring, to optimise species selection within a continuously varying spatial context to maximise WUE and farm production.