Dynamic, deterministic agricultural models, and current machine learning technologies based on sensor data, enable and support decision making for on-farm management. However, their predictions are subject to various sources of uncertainty. Hybrid analytics that leverage both modelled and sensor data provide predictive information that makes the best of both approaches in a timely fashion to inform operational decision making and enable inclusive uncertainty quantification. We describe and evaluate a probabilistic Bayesian data assimilation tool that combines the state variables from the Sirius wheat (Triticum aestivum L.) development model with time-series environmental and leaf count data. Additionally, the uncertainty associated with input parameters is quantified via expert opinion. The Bayesian approach obtained point estimates through time that were accompanied by inclusive, probabilistic, 95% credible intervals. At the end of simulation, a typical model predicted a final leaf number of 6.6 leaves, Sirius alone predicted seven leaves and the mean of the observed data was 6.7 leaves. The 95% credible interval was estimated as 5.1-8.4 leaves. Importantly, the tool was able to "redirect" simulated outputs if input parameters such as minimum leaf number or base phyllochron were incorrectly specified, with the implication that on-farm decision makers would have advance warning of variation in expected harvest date. Relatively few plants with time-intensive data were sufficient to fit the model, however, more plants would be desirable to reduce the rather wide range of credible intervals. Nevertheless, the tool shows potential and could be readily implemented with low resource requirements, providing more finely tuned harvest date information, with probabilistic uncertainty quantification built-in, for on-farm decisions.

Abstract The world needs to produce more food, more sustainably, on a planet with scarce resources and under changing climate. The advancement of technologies, computing power and analytics offers the possibility that ‘digitalisation of agriculture’ can provide new solutions to these complex challenges. The role of science is to evidence and support the design and use of digital technologies to realise these beneficial outcomes and avoid unintended consequences. This requires consideration of data governance design to enable the benefits of digital agriculture to be shared equitably and how digital agriculture could change agricultural business models; that is, farm structures, the value chain and stakeholder roles, networks and power relations, and governance. We argue that this requires transdisciplinary research (at pace), including explicit consideration of the aforementioned socio‐ethical issues, data governance and business models, alongside addressing technical issues, as we now have to simultaneously deal with multiple interacting outcomes in complex technical, social, economic and governance systems. The exciting prospect is that digitalisation of science can enable this new, and more effective, way of working. The question then becomes: how can we effectively accelerate this shift to a new way of working in agricultural science? As well as identifying key research areas, we suggest organisational changes will be required: new research business models, agile project management; new skills and capabilities; and collaborations with new partners to develop ‘technology ecosystems’. © 2018 The Authors. © 2018 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Management of freshwater quality requires modelling tools for rapid evaluation of land use and management scenarios. This paper presents the CLUES (Catchment Land Use and Environmental Sustain ability) model to address this need. CLUES provides steady state, spatially distributed, integrated catchment models tightly coupled to GIS software to predict mean annual loads of total nitrogen, total phosphorus, sediments and E. co/i, and concentration of nutrients throughout New Zealand (268,000 km(2)) with a subcatchment resolution of 0.5 km(2). CLUES also estimates potential nutrient concentrations for estuaries and provides key farm socio-economic indicators. The model includes a user interface for study area selection, scenario creation, data geo-visualisation, and export of results. It is pre populated with spatial data and parameter values for New Zealand. Evaluation of the model and a summary of applications demonstrate the tractability and utility of national-scale rapid scenario assessment tools within a GIS framework. (C) 2016 Elsevier Ltd. All rights reserved.

Land users and managers require decision support tools (DSTs) that enable them to estimate losses of contaminants from land to freshwater. MitAgator is a DST that estimates losses of nitrogen (N), phosphorus (P), sediment, and fecal indicator bacteria (E. coli) and the cost-effectiveness of different strategies to mitigate losses so that a water quality target can be met at the least cost. Some of the algorithms present within Overseer (a standard DST used in New Zealand for N and P management) have been modified and appended to include spatial analysis in MitAgator. Outputs from MitAgator showed good (R-2 > 0.77; p < 0.001) prediction of measured N and P losses across a range of land uses, but accuracy decreased at larger (catchment) scales. Analysis for P outputs indicated that the most sensitive inputs were hydrological characteristics, followed by soil characteristics and P inputs. Although national databases are used for many of these inputs, if better local data are available, then they should be used. Furthermore, while MitAgator is easy to use by a novice, MitAgator outputs should only be interpreted in collaboration with an experienced user so that limitations concerning cost-effectiveness estimates and spatial and temporal scales are not exceeded.

A meta‐analysis of three national databases determined the potential linkage between soil and surface and groundwater enrichment with phosphorus (P). Soil P was enriched especially under dairying commensurate with an increase in cow numbers and the tonnage of P‐fertilizers sold. Median P concentrations were enriched in surface waters receiving runoff from industrial and dairy land uses, and in groundwater beneath dairying especially in those aquifers with gravel or sand lithology, irrespective of groundwater redox status. After geographically pairing surface and groundwater sites to maximize the chance of connectivity, a subset of sites dominated by aquifers with gravel and sand lithology showed increasing P concentrations with as little as 10 years data. These data raise the possibility that groundwater could contribute much P to surface water if: there is good connectivity between surface and groundwater, intensive land use occurs on soils prone to leaching, and leached‐P is not attenuated through aquifers. While strategies are available to mitigate P loss from intensive farming systems in the short‐term, factors such as enriched soils and slow groundwater may mean that despite their use, there will be a long‐term input ( viz . legacy), that may sustain surface water P enrichment. To avoid poor surface water quality, management and planning may need to consider the connectivity and characteristics of P in soil‐groundwater‐surface water systems.

OVERSEER® Nutrient Budgets (Overseer) is an agricultural management support tool that examines the flow of nutrients in a farming system. There is increasing pressure from a range of users for transparency of the way Overseer functions, particularly the modelling of nitrogen (N) and phosphorus (P) loss to water. The aim of this paper is to provide a conceptual description of the way Overseer models the distribution and fate of N and P in a pastoral system, to support user understanding and correct model use. The core of Overseer is a nutrient budget, which accounts for the flow of nutrient into, around and off the farm. The key strength of Overseer is its ability to model these nutrient transfers around the farm, identifying how much, where and when nutrients move. Other parts of the model then estimate the fate of these nutrients. Nitrogen and P cycle differently around the farm, which is reflected in the way they are modelled. This paper is intended to be a support document for understanding the way Overseer models N and P, and where more detailed information is required, it may be found on the Overseer website (www.overseer.org.nz). Keywords: Nutrient budget, model, leaching, run-off

The user inputs to OVERSEER® Nutrient Budgets (Overseer) allow farm-specific greenhouse gas (GHG) emissions to be estimated. Since the development of the original model, life cycle assessment standards (e.g. PAS 2050) have been proposed and adopted for determining GHG or carbon footprints, which are usually reported as emissions per unit of product, for example, per kg milk, meat or wool. New Zealand pastoral farms frequently generate a range of products with different management practices. A robust system is required to allocate the individual sources of GHGs (e.g. methane, nitrous oxide, direct carbon dioxide and embodied carbon dioxide emissions for inputs used on the farm) to each product from a farm. This paper describes a method for allocating emissions to co-products from New Zealand farms. The method requires allocating the emissions, first, to an animal enterprise, separating the emissions between breeding and trading animals, and then allocating to a specific product to give product (e.g. milk, meat, wool, velvet) footprints from the 'cradle-to-farm-gate'. The meat product was based on live-weight gain. Procedures were adopted so that emissions associated with rearing of young stock used in live-weight gain systems, both as a by-product or a primary product could be estimated. This allows the possibility of total emissions for a meat product to be built up from contributing farms along the production chain.

A transfer function (TF) was developed to assist with the estimation of nitrogen (N) leaching from urine-affected areas in grazed pastures. The proposed TF uses a simple function to describe the likely breakthrough curve for urine-N deposited in different months and in various climates and soils in New Zealand. The TF was designed to be integrated into the OVERSEER® Nutrient budgets model to increase the sensitivity of N leaching to the month of urine deposition, but could also be used in any other model that estimates the water balance and plant N uptake on a monthly basis. The inputs required for the TF are typically readily accessible (e.g. soil texture data) and thus do not add any significant complications when added to OVERSEER. The TF retains OVERSEER as the arbitrator of the main items of N-balance in the farm system, but adds functionality by giving a better temporal discrimination of leaching from the farm system. The procedure for parameterising the TF from a comprehensive set of APSIM (Agricultural Production Systems Simulator) simulations is described. Validation of the leaching estimated by the TF was achieved through a combination of testing against an independent set of APSIM simulations and testing against experimental data. The testing of the TF showed very promising performance. The TF explained 75% of the variability of N leaching simulated by an independent APSIM dataset and agreed well with the experimental data.

The effect of a surface application of lime (5000 kg/ha initially then 2500 kg/ha 1 year later) on soil properties in four soil layers was measured over 15 years on a yellow-grey earth (Duric Palic soil). The maximum increase in soil pH occurred about 2 years after lime was initially applied in the 0-50 mm soil layer, after 5 years at 50-100 mm, after about 12 years at 100-150 mm, and was still increasing at 150-200 mm. The average rate at which soil pH increased until a maximum difference occurred was 0.57, 0.15, 0.04, and 0.009 pH units/year in the 0-50, 50-100, 100-150, and 150-200 mm soil layer, respectively. The average rates of soil re-acidification (rate of decrease after reaching a maximum) were 0.075 and 0.02 pH units/year in the 0-50 and 50-100 mm layer, respectively. For exchangeable calcium (Ca), the average rate of decrease after reaching a maximum was 0.82 and 0.17 cmol(+) Ca/kg per year in the 0-50 and 50-100 mm soil layer, respectively. At this rate of decrease, lime should increase soil pH in the 0-50 mm layer until about 17 years after application. Lime significantly decreased exchangeable magnesium (Mg) for only 5 years, with the maximum decrease between 0 and 100 mm occurring about 3 years after lime was applied. Below 100 mm, exchangeable Mg was about 0.2 cmol(+)/kg lower in the lime treatment from 5 years after lime was applied. Lime decreased Olsen P at an average rate of 0.53 and 0.27 mu g/ml per year in the 0-50 and 50-100 mm layer, respectively up to 6 years after lime was applied. This decrease was partly attributable to higher plant phosphorus (P) uptake in the lime treatment.

A mowing trial was conducted on the Mangatea clay loam (a yellow-grey earth or Umbric Dystrochrept fine hallositic mesic) to examine the mechanisms causing lime responses in a grass/ clover pasture. The trial site was located on a sheep and beef farm 8 km south of Te Kuiti, New Zealand. The site generally had cool winters and warm moist summers, with an annual rainfall of about 1400 mm. The treatments consisted of 3 lime rates (0, 5000, 10 000 kg ha(-1)) in the absence of P and 6 lime rates (0, 1250, 2500, 5000, 7500, 10 000 kg ha(-1)) in the presence of P (50 kg P ha(-1) yr(-1)), all in the presence of sodium molybdate (Mo), and an additional treatment where P was applied in the absence of lime and Mo. Lime increased grass yield but decreased clover yield for the first 2 years after application. in Years 3 and 4, the responsiveness of the grass decreased and that of clover increased. Lime increased net N mineralisation by up to 58 kg ha(-1), resulting in increased grass growth. Plant P uptake in both grass and clover also increased when lime was applied. However, yield responses were only seen in summer in clover when plant P concentrations in clover were less than 3.0 mg g(-1) The lime-induced increase in Mo availability would contribute to lime responses if basal Mo was not applied. Subsoil Al toxicity, particularly in summer, was limiting growth in some parts of the trial. However, there was no evidence that Mn toxicity was limiting yield, nor of any lime-induced trace nutrient deficiencies. Results from the application of P and Mo indicate that there was at least a 6-month delay in the transfer of N from clover to grass, with about 30% of the extra clover N uptake in Year 1 being transferred to extra N uptake in grass in Year 2.

In New Zealand and worldwide, growers are increasingly being required to demonstrate that their nutrient management does not have adverse effects on the environment. Since routine direct measurements of nitrogen (N) losses at an appropriate scale are currently unfeasible, a modelling tool that can be used to assess losses, both actual and potential, from farming systems is therefore needed. We present the upgraded N balance module for cropping systems (OVCrop) developed to be incorporated into the OVERSEER (R) Nutrient Budgets model. It presents an easy-to-use interface, requiring few but meaningful inputs, with the minimum flexibility needed to describe management of real farming systems. Following the developmental guidelines of OVERSEER (R) Nutrient Budget, OVCrop is designed to provide the long-term annual average of N leaching from any user-defined cropping rotation. OVCrop has been parameterized based on process-based model simulations for typical farm systems and environments of New Zealand. The intended use of the OVCrop is to help on-farm nutrient management and to demonstrate future compliance to N leaching regulations. OVCrop is a robust tool, simple enough to be used by people with a low to medium level of expertise, such as farmers and regional councils.

The amount of plant nitrogen (N) uptake from the soil and fixed N in the herbage of clover species were measured using a N-15 dilution technique on mowing trials at 3 lime rates (0, 5000, and 10 000 kg/ha) at a low rainfall site (Matapiro soil) near Hastings over 5 years and at a high rainfall site (Mangatea soil) near Te Kuiti over 4 years. At the low rainfall site, measurements were also made over 5 years on a grazing trial at 2 lime rates (0, 7500 kg/ha). The dominant clover species was subterranean clover on the low rainfall site and white clover on the high rainfall site. Overall, the average proportion of clover N that was fixed from the atmosphere (P-Nfix) was about 0.8 of total N in clover. The average annual amount of fixed N in clover herbage was 35 kg/ha in the grazing trial on the Matapiro soil, 65 kg/ha on the mowing trial and 105 kg/ha in the trial on the Mangatea soil. Over all harvests and lime treatments, the amount of N fixed could be estimated as 0.04 x clover yield (kg DM/ha) on the high rainfall site and 0.046 x clover yield on the low rainfall site. The average annual amount of plant N uptake from the soil by grass, clover, and weeds was 150 kg/ha in the grazing trial and 160 kg/ha in the mowing trial on the Matapiro soil, and 240 kg/ha in the trial on the Mangatea soil. In the first 3 years, lime increased grass N uptake by an average of 24 kg/ha/yr over all trials. In the mowing trials, grass N uptake increased during the first 3 years by an average of 16 kg/ha when 5000 kg/ha of lime was applied, and by 33 kg/ha when 10 000 kg/ha was applied. This increase was attributed to an increase in net N mineralisation. However, the increase in net N mineralisation due to liming was a short term effect as lime had no significant effect on grass N uptake after 3 years.

Background:  Reports on air pollution and asthma exacerbations have been inconsistent, although effects of airborne allergen can be spectacular. With no generalized test for allergen in air, it is not known how far allergen is responsible for nonepidemic exacerbations of the disease.

The OVERSEER nutrient budget model is a farm-scale nutrient reporting and greenhouse gas (GHG) emission accounting tool used extensively throughout New Zealand (NZ) by farmers, farm consultants and fertiliser representatives. The model is increasingly being used as a tool for implementing regional council resource management requirements to limit nitrogen (N) and phosphorus losses to waterways. NZ's main dairy company, Fonterra, also requires dairy farmers to have a nutrient budget as part of the national Clean Streams Accord. This means that a high proportion of NZ farmers can obtain reports of their on-farm GHG emission profile. The GHG emission model is based on models and algorithms used for the NZ GHG national inventory, modified to include a wide range of on-farm management practices. The model estimates methane, nitrous oxide and carbon dioxide (CO2) emissions, and presents the results as CO2 equivalents. This paper describes the model and the benefits of combining nutrients budgets and GHG emissions into a single model. It also demonstrates the effects of management practices on a range of outputs, including N leaching and GHG emissions.

The effect of varying solution calcium (Ca) and magnesium (Mg) concentrations in the absence or presence of 10 mu M aluminum (Al) was investigated in several experiments using a low ionic strength (2.7 x 10(-3) M) solution culture technique. Aluminium-tolerant and Al-sensitive lines of wheat (Triticum aestivum L.) were grown. In the absence of Al, top yields decreased when solution Ca concentrations were <50 mu M or plant Ca concentrations were <2.0 mg/g. Top and root yields decreased when solution Mg concentrations were <50 mu M or plant Mg concentrations were <1.5 mg/g. There were no differences between the lines in solution or plant concentrations at which yield declined. Increasing solution Ca concentrations decreased plant Mg concentrations in the tops (competitive ion effect) but increased plant Mg concentrations in the roots of wheat. This suggests that Ca is competing with Mg when Mg is transported from the roots. Increasing solution Mg concentrations decreased plant Ca concentrations in the tops and the roots (competitive ion effect). In the roots, increasing solution Mg concentrations decreased plant Ca concentrations at a lower solution Ca concentration in the Al-sensitive line than the Al-tolerant line. In the presence of Al, increasing solution Ca and Mg concentrations increased yield (Ca and Mg ameliorating Al toxicity). Yield increased until the sum of the solution concentrations of the divalent cations (Ca+Mg) was 2,000 mu M for the Al-tolerant line or 4,000 mu M for the Al-sensitive line. The exception was that yield decreased when solution Mg concentrations were >1,500 mu M and the solution Ca concentration was 100 mu M (Mg exacerbating Al toxicity). The ameliorative effects of solution Ca or Mg on Al tolerance were not related to plant Ca or Mg concentrations per se.

The conservation and restoration of soil organic matter are often advocated because of the generally beneficial effects on soil attributes for plant growth and crop production. More recently, organic matter has become important as a terrestrial sink and store for C and N. We have attempted to derive a monetary value of soil organic matter for crop production and storage functions in three contrasting New Zealand soil orders (Gley, Melanic, and Granular Soils). Soil chemical and physical characteristics of real-life examples of three pairs of matched soils with low organic matter contents (after long-term continuous cropping for vegetables or maize) or high organic matter content (continuous pasture) were used as input data for a pasture (grass-clover) production model. The differences in pasture dry matter yields (non-irrigated) were calculated for three climate scenarios (wet, dry, and average years) and the yields converted to an equivalent weight and financial value of milk solids. We also estimated the hypothetical value of the C and N sequestered during the recovery phase of the low organic matter content soils assuming trading with C and N credits. For all three soil orders, and for the three climate scenarios, pasture dry matter yields were decreased in the soils with lower organic matter contents. The extra organic matter in the high C soils was estimated to be worth NZ$27 to NZ$150 ha(-1) yr(-1) in terms of increased milk solids production. The decreased yields from the previously cropped soils were predicted to persist for 36 to 125 yr, but with declining effect as organic matter gradually recovered, giving an accumulated loss in pastoral production worth around NZ$518 to NZ$1239 ha(-1). This was 42 to 73 times lower than the hypothetical value of the organic matter as a sequestering agent for C and N, which varied between NZ$22,963 to NZ$90,849 depending on the soil, region, discount rates, and values used for carbon and nitrogen credits.

A model was developed for incorporation into the computer model "Overseer (R) nutrient budgets 2" to estimate phosphorus loss from pastoral farms to surface waters. The model sums potential risk of phosphorus loss from soil (background) with incidental losses attributed to the application of fertiliser or effluent to fields or waterways. The overall risk is estimated as the amount of P lost annually (kg P ha(-1) yr(-1)). Data for 23 sites of grazed pasture (sheep, beef, and dairying) ranging from small plot trials to catchments were used to validate each component of the model. A good relationship (R-2 = 0.97***) was obtained between measured and estimated soil P losses for systems without much incidental P loss. Similarly, a good relationship was observed between measured and estimated overall (soil + incidental) P losses for all systems (R-2 = 0.96***). The estimated loss of P is categorised into low, medium, high or extreme risk to the environment (< 1, 1-2, 2-4, and > 4, respectively) and best management strategies are suggested accordingly. The model shows promise in directing responsible P management for New Zealand pastoral farms. However, inadequacies are likely when scaling up the model for use in large catchments where land use and soils vary greatly.

To establish long-term trends in soil fertility in New Zealand pasture lands, data for Olsen phosphate (Olsen P), pH, and QuickTest (QT) calcium (Ca), magnesium (Mg), and potassium (K) from 246 000 soil samples submitted to a commercial soil sampling laboratory from 1988 to 2001 were analysed. Each record was categorised according to year, land use (dairy, sheep[beef including deer), soil "group" (sedimentary, volcanic, pumice, peats, rest) and regions. Olsen P values were, on average, higher on dairy farms than sheep/beef farms and lower on sedimentary soils than other soils (P > 0.05). Sedimentary and volcanic soils had higher QT K than pumice and peat soils (P > 0.05). Regional differences were largely explained by differences between land use and soil group distribution. However, given this distribution, average Olsen P values were higher in the Waikato, Bay of Plenty/Taupo and Taranaki regions, the east coast of both islands generally had higher average pH, K, and Mg soil test values, and the West Coast region had lower average soil test values than most other regions. Over the 14-year period, Olsen P followed an asymptotic increase, reaching the asymptote between 1994 and 1997 depending on land use and soil group. For the period 1997-2001, 30-65% of samples from dairy farms had Olsen P levels above the range required for near-maximum pasture production. QT Mg decreased consistently by about 0.35 QT units year(-1) when QT Mg was 25. The database provides a too] for looking at soil test trends over time on a national basis to assess the sustainability of soil fertility management practices on pastoral farms.