A national-scale food production and nutrition dataset for New Zealand

Clemence Vannier^1*Michelle Barnes²

• ¹Landcare Research New Zealand, Lincoln, New Zealand
• ²Landcare Research New Zealand, Palmerston North, New Zealand

Introduction Food production, trade, and nutrient deficiencies are increasingly critical global research areas, given their implications for national food security, public health, and environmental sustainability (Oluwole et al., 2023; Tian et al., 2016). As a major food-exporting nation, New Zealand (NZ) plays an important role in global food supply chains, yet domestic food insecurity remains a concern, particularly among vulnerable populations (Graham et al., 2022; Hobbs et al., 2019). NZ produces and exports food and fibre for 354% of its own population requirement, yet relies heavily on imports when it comes to feeding New Zealanders (64%) (Smith et al., 2021). In NZ, the lack of short-term food security has become increasingly recognised in the face of recent major market disruptions (Snow et al., 2021). Trade and supply chain vulnerabilities further affect food prices and accessibility. In addition, the prevalence of micronutrient deficiencies and diet-related diseases, including obesity and type 2 diabetes, continues to rise, reflecting shifts in dietary patterns and socioeconomic disparities (Mann et al., 2024; World Health Organisation, 2018). For example, iron deficiency, especially among premenopausal women, remains prevalent and contributes to fatigue, reduced cognitive function, and impaired immune response (Calje and Skinner, 2017; Jin et al., 2020). Also, food insecurity persists among vulnerable populations, affecting performance, health, and wellbeing (Graham et al., 2022). Policy measures, including food fortification programs and calls for a national food strategy, aim to mitigate these nutritional challenges and improve population health outcomes. Given the critical role of nutrition in overall wellbeing, ongoing research and policy interventions are essential to addressing micronutrient deficiencies and ensuring equitable access to nutritious food in NZ. Several food production and nutrition databases already provide insights for research, policymaking, and public health. FAOSTAT, managed by the Food and Agriculture Organization of the United Nations (FAO), offers comprehensive data on global food production, trade, and security. The World Health Organization's Nutrition and Food Safety Databases track nutrition-related health indicators, while the Global Dietary Database (GDD) focuses on worldwide dietary intake patterns (GDD, 2025). The FAO also provides country profiles that present country-specific information on agriculture, food security, and nutrition. These databases are essential tools for understanding the relationship between food systems and nutrition. However, none of them provide spatial information and links between the agricultural food system and production and the nutrient content in the food produced. This information is critical to understanding and addressing the gaps between a national food system and nutrient deficiencies in a population. However, limited research has examined whether the types and quantities of food produced in NZ translate into sufficient nutrient availability for domestic consumption (New Zealand Institute of Economic Research, 2025). This gap underscores the importance of linking production data with nutritional indicators to inform policy and improve national food system resilience (Kidd et al., 2025). The objective of our project was to develop the New Zealand Food Production and Nutrition Data Stack (NZFPNDS) as a comprehensive resource comprising 19 spatial layers that provide spatially explicit nutritional indicators for energy, macronutrients, vitamins, and minerals. This data report delivers a detailed methodology and usability guidelines, thereby providing a crucial tool for researchers, policymakers and stakeholders aiming to bridge the gap between food production and national nutritional outcomes. Material and Methods To develop a national-scale dataset linking food production with nutrient availability, we integrated multiple agricultural and nutritional data sources and applied a combination of database preparation, spatial and statistical mapping. This process involved identifying which foods are produced in NZ, estimating their nutrient content using international composition tables, and mapping this information onto a national spatial grid. By combining geospatial data (such as farm types and land cover) with national production statistics and food composition values, we created a set of 19 nutrient-specific maps that represent the potential annual availability of key nutrients across the country. The following sections describe the data collection and processing steps used to construct this dataset. Data Collection Four datasets were employed as modelling inputs for a national-level analysis (Table 1, Figure 1). These were accessed online, manually downloaded, and securely stored during August and September 2024. Three of the datasets were publicly available during this time, with AgriBase® (a product of AsureQuality) being the only dataset licensed to Landcare Research New Zealand Ltd and privately accessed. AgriBase is NZ's national geospatial database of rural properties (farms, orchards, vineyards, forestry blocks, conservation areas), each assigned a unique Farm_ID maintained by AsureQuality. It contains farm boundaries (linked to LINZ land parcels), farm centroids, property roles and contacts, predominant land use, livestock numbers by class, crop/orchard areas, and spatial geometry for use in GIS systems. Data is verified through frequent updates from LINZ parcel records (quarterly, moving to monthly), GPS and orthophoto cross-checks, and error detection from widespread operational use, with AsureQuality acting as custodian to ensure accuracy, security, and traceability. Coverage of the NZ mainland, its near-shore Islands, and the Chatham Islands was included for all layers. An extract of FAOSTAT was downloaded from the "Food and Diet Domain: Availability" (based on supply utilization accounts) (Table 1). We selected 'New Zealand', all food groups and indicators, and the years 2018 to 2022 for output to csv format. The entire geospatial and tabular extent of both geospatial datasets and the whole FAO Nutrient Conversion Table (FAO NCT) were downloaded for modelling purposes (Table 1). Table 1. Input database characteristics Type Name Spatial scale Description Original variable/ table name used Date/period covered by the data Source Quanti tative FAOSTAT National, according to M49 code: 554, ISO2 code: NZ, ISO3 code: NZL Internationally available annual data relevant to NZ was downloaded from the Food and Diet domain: Availability (based on supply utilization accounts). All relevant indicators and food groups were included. FAOSTAT_2 018_2022_ 2018 to 2022 FAO: https://www. fao.org/faost at/en/#data Quanti tative FAO NCT International. Partially informed by the New Zealand Food Composition Database. Internationally available food compositional data tables. Food component, macronutrient, energy, mineral, vitamin, food item and edible portions data were downloaded and matched to corresponding FAOSTAT item codes. FAO_NCT_ EDITED_24 09 Accessed August 2024 FAO: https://doi.or g/10.4060/cc 9678en Spatial AgriBase® National (including Chatham Islands). A nominal accuracy of 0.1– 1 m in urban areas and 1–100 m in rural areas AgriBase contains animal type and numbers by stock class and planted area for different orchard or crop types. Relevant farm enterprise figures were used for modelling input. AGRIBASE_ SHP September 2020 Privately licensed and supplied to Manaaki Whenua – Landcare Research Spatial LCDBv5 National (including Chatham Islands). Classification scheme has a nominal 1 ha minimum mapping unit. A thematic classification of New Zealand's land cover. 'Manuka and/or Kanuka' at 'Class_2018' was extracted. The Chatham Islands dataset was not incorporated due to lack of relevant vegetation. LCDB5_SH P Field 'Class_2018' relevant to summer 2018/19 https://lris.sc info.org.nz/la yer/104400-lcdb-v50-land-cover-database- version-50-mainland- new-zealand/ Data Processing Collaborative Review Initially, manual steps were undertaken following the download of all raw dataset inputs (Figure 1, step 1). Relevant pre-processed/whole food items from the FAOSTAT database download were reviewed and matched with those from the FAO NCT via food item codes (Figure 1, step 2). Selected codes were reviewed and matched to relevant, available AgriBase aggregated farm enterprise detail descriptions via enterprise type names. Enterprise types included both hectare and animal number groupings (Figure 1, step 3). LCDBv5 classes were reviewed against item codes unaccounted for by AgriBase enterprise types. Ultimately, 'Manuka and/or Kanuka' geometries were chosen to inform locations possibly contributing to honey production during the 2018/19 summer. In this case, honey production was based on potential available resources for honeybees. The finalised tables were reformatted and uploaded into a Structured Query Language (SQL) server database prior to executing joins (Figure 1, step 4). Scripted Processing The module 'arcpy' was used for geoprocessing within a Python 3 environment, and SQL was largely used for tabular processing on an SQL server. Processing commenced with the removal of AgriBase farms containing irrelevant enterprise types and combinations, or types with no corresponding FAOSTAT data (Figure 1, step 5). For example, mountainous national park land solely associated with natural forest was excluded. In preparation for final grid outputs, AgriBase farms containing overlapping vector geometries were effectively split into both overlapping and remaining 'enterprise areas' (hectares) (Figure 1, step 5). A recalculation of enterprise-type values for new enterprise areas was achieved via multiplication of original enterprise-type values by a coefficient, whereby enterprise areas were divided by the total original area of a farm to produce the coefficients. New enterprise-type values were also summed per enterprise type, per enterprise area (Figure 1, step 6). 'Manuka and/or Kanuka' geometries were extracted from LCDBv5 via the 'Class_2018' field and combined with AgriBase enterprise areas (Figure 1, step 7). Hectares of 'Manuka and/or Kanuka' were calculated, covering three requirements for its addition to AgriBase and FAOSTAT data-derived tables: total hectares, hectares per enterprise area, and hectares outside of enterprise areas (Figure 1, step 8). AgriBase enterprise-type data was normalized into two new enterprise name and number fields, streamlining enterprise-centric joins and calculations. AgriBase enterprise types 'vegetables' and 'other animal numbers' were summed and added via union to the FAOSTAT table to represent absent total hectares of 'mushrooms and truffles' and 'other meat' animal numbers (Figure 1, step 9). FAOSTAT curation began with 'Unit' conversion and the calculation of 5-year averages for tonnes, hectares, and animal number figures (Figure 1). Item codes were assigned AgriBase enterprise type classes in a new field containing 'item enterprise' types. Item code values were summed across the new item enterprise types to produce 'item enterprise sums', while unusable item codes were excluded (Figure 1, step 10). Item code contribution (per item enterprise type) percentage figures were calculated by dividing item code values by item enterprise sums. Five-year 'item code contribution averages', summing to unity, were produced from resulting item code contribution percentage figures (Figure 1, step 11). Weighted yield values were generated for each item enterprise class, providing values for both hectares and animal number scenarios. This was achieved by dividing mean tonnes by mean hectares, and mean tonnes by mean animal numbers for each item code value. These new mean item code yields were multiplied by the item code contribution averages, then summed across each item enterprise class and unit (Figure 1, step 11). Subsequently, amended nutrient values based on item code contributions to each item enterprise class were produced. Nutrient values were multiplied by the item code contribution averages, then summed across each item enterprise class and unit. To implement this step, FAO NCT data was joined to the finalised FAOSTAT table via the item codes (Figure 1, step 11). Nutrient values were adjusted further, accounting for each FAO NCT item code's sum of proximate components and edible portions. Nutrient values were converted from 100 g to 1 tonne representation and lastly multiplied by the hectare and animal number weighted yield values for each item enterprise class. This produced the finalised set of nutrient values representing each item enterprise class (Figure 1, step 12). The revised AgriBase and FAO tables were joined. Nutrient figures per enterprise type for both hectares and animal numbers were generated, where enterprise numbers were multiplied by the new nutrient values. Nutrient values per enterprise area were produced by summing all enterprise type nutrient values per enterprise area (Figure 1, step 13). The finalised table and geometries were joined via enterprise area IDs. Enterprise areas with null identifiers were deleted, which were indicative of a lack of relevant or usable FAOSTAT data (Figure 1, step 14). A 1 km-square grid was produced from the enterprise area IDs (Figure 1, step 15). For quality assurance, the grid was exported to polygon geometries and joined to tables created in earlier processing steps. All final nutrient values relevant to each enterprise area ID were divided by the grid cell 'count', producing an estimation of nutrient values per cell (Figure 1, step 16). Finally, batch processing was performed on the geometries, creating grids originating from each of the nutrient value fields (Figure 1, step 17). Figure 1. Flowchart of the overall data processing. Geographic Coverage The NZFPNDS covers the NZ mainland, its near-shore Islands, and the Chatham Islands, at a 1 km-square pixel size spatial resolution for all grids. Reusing the Data The NZFPNDS consists of 19 spatial data layers, each representing an annual estimate of a specific nutrient's presence across New Zealand's agricultural landscape. These maps show the estimated amount of nutrients produced by the animals and crops being raised or cultivated in each location, expressed as nutrient mass per square kilometre. Users may view and download all NZFPNDS outputs via the LRIS Portal. All NZFPNDS data and metadata is managed Manaaki Whenua – Landcare Research and is published for use under the Creative Commons by Attribution 4.0 International licence (https://creativecommons.org/licenses/by/4.0/). An overview of the dataset is presented in Table 2. The purpose of the NZFPNDS is to provide information on annual nutrition indicators for food components, macronutrients, energy, minerals, and vitamins (Table 2). Table 2. LRIS Portal grid name and description, and map overview. Each grid represents estimated nutrient mass per square kilometre, produced by agricultural enterprises and calculated from FAO nutrient tables joined to AgriBase farm data. Users can visualise, download, and manipulate these layers interactively through the LRIS Portal. Large areas appear as non-productive in the maps due to New Zealand's mountainous terrain, extensive natural forests, and conservation lands, which limit agricultural use and therefore nutrient production. Macronutrients components NZFPNDS ASH g/km2: The component ash, expressed in grams. This layer displays the estimated mass of mineral residue (ash) produced per square kilometre of agricultural area, derived from FAO nutrient tables joined to AgriBase farm enterprises. NZFPNDS CHOAVLDF g/km2: The macronutrient component available carbohydrate, expressed in grams. This layer displays the available carbohydrate production (grams/km²), indicating potential energy-yielding carbohydrate supply from crops and livestock. NZFPNDS FAT g/km2: NZFPNDS FIBTG g/km2: The macronutrient component fat, expressed in grams. This layer displays fat content in grams per km², calculated from production-weighted yields and nutrient conversion factors. The component total dietary fibre, expressed in grams. This layer displays the estimated dietary fibre production (grams/km²), highlighting areas contributing most to fibre availability. NZFPNDS PROTCNT g/km2: The macronutrient component protein, expressed in grams. This layer displays protein production (grams/km²) from animal and plant enterprises, central to dietary macronutrient supply. Metabolizable energy NZFPNDS ENERC kcal/km2: Metabolizable energy, expressed in kilocalories. This layer displays the total metabolizable energy (kcal/km²), integrating all food production into a common energy metric. Minerals NZFPNDS CA mg/km2: The mineral calcium, expressed in milligrams. This layer displays the calcium production (mg/km²), an indicator of dairy-and crop-based calcium supply. NZFPNDS FE mg/km2: The mineral iron, expressed in milligrams. This layer displays iron availability (mg/km²), relevant for assessing contributions to population iron needs. NZFPNDS K mg/km2: The mineral potassium, expressed in milligrams. This layer displays the potassium production (mg/km²), often derived from fruit, vegetable, and pastoral enterprises. NZFPNDS MG mg/km2: The mineral magnesium, expressed in milligrams. This layer displays the magnesium availability (mg/km²), spatially distributed across enterprise types. NZFPNDS P mg/km2: The mineral phosphorus, expressed in milligrams. This layer displays the phosphorus production (mg/km²), a nutrient closely linked to protein-rich foods. NZFPNDS ZN mg/km2: The mineral zinc, expressed in milligrams. This layer displays the zinc supply (mg/km²), reflecting contributions from both livestock and crops. Beta-carotene & vitamins (micrograms) NZFPNDS CARTBEQ mcg/km2: Beta-carotene equivalents, expressed in micrograms. This layer displays the beta-carotene equivalent production, highlighting areas with potential vitamin A precursors. NZFPNDS RETOL mcg/km2: Vitamin A1/retinol, expressed in micrograms. This layer displays the retinol (vitamin A1) production, largely associated with animal-derived foods. NZFPNDS VITA mcg/km2: Vitamin A (expressed in retinol equivalents) in micrograms. This layer displays the vitamin A (retinol equivalents) availability from combined sources. NZFPNDS VITA RAE mcg/km2: Vitamin A (expressed in retinol activity equivalents), in micrograms. This layer displays the vitamin A in retinol activity equivalents, the standard unit for dietary requirements. Vitamins (milligrams) NZFPNDS RIBF mg/km2: Vitamin B2/riboflavin, expressed in milligrams. This layer displays the riboflavin (vitamin B2) production (mg/km²), concentrated in dairy-producing regions. NZFPNDS THIA mg/km2: Vitamin B1/thiamin, expressed in milligrams. This layer displays the thiamin (vitamin B1) production, often associated with cereals and meat. NZFPNDS VITC mg/km2: Vitamin C/L-ascorbic acid, expressed in milligrams. This layer displays the vitamin C (ascorbic acid) production (mg/km²), highlighting fruit and vegetable regions. Best Usage Information The NZFPNDS contains 19 spatial layers, providing information on annual nutrition indicators for food components, macronutrients, energy, minerals, and vitamins. The NZFPNDS has been modelled employing data from the FAO, AgriBase, and LCDBv5. It has been produced for analyses including aspects of health and wellbeing. Limitations As highlighted in international food composition research, micronutrients such as vitamin C, thiamin, and certain minerals (e.g., iron and zinc) are particularly sensitive to agroecological conditions, varietal differences, and post-harvest handling, leading to greater variability and less accuracy when estimated from global data sources (Deharveng et al., 1999; Greenfield and Southgate, 2003). In contrast, macronutrient values like protein, fat, and energy are generally more consistent across regions. One limitation arises from the exclusion of certain data during processing. Enterprise areas without relevant or usable FAOSTAT item codes were removed, potentially omitting some minor or less-documented food production activities from the dataset. Similarly, unusable FAOSTAT item codes that could not be reliably matched or aggregated were excluded, which may lead to slight underrepresentation of niche food types or less commonly reported nutrient sources. Another key limitation of the NZFPNDS is the spatial and temporal resolution of the data layers. The database is built from a national agricultural database and an international nutritional dataset, neither of which is uniformly updated. As a result, the NZFPNDS represents the year 2020 with averages for food production (yields) in 2018 to 2022. The data availability was not sufficient to capture regional or finer-scale variations in food production and nutrient distribution. Another key limitation of the NZFPNDS is its reliance on the use of the international FAO NCT to estimate the nutritional content of produced foods. Nutrient composition varies according to soil quality, climate conditions, food-processing methods, and post-harvest handling. The database assumes average nutrient values rather than accounting for variations due to production conditions that were not possible to capture. Future improvements could focus on enhancing data resolution. Finally, while the NZFPNDS maps nutrient production locations, its utility for assessing local food access during shocks is limited by New Zealand's centralised food system, where production does not necessarily reflect local availability. Potential Research Pathways By providing spatially explicit data on nutrient availability, the NZFPNDS allows investigation of critical gaps in the food system's ability to meet NZ national nutritional needs (Smith et al., 2021). Climate change The NZFPNDS can support studies on climate change and food system vulnerabilities, particularly in the context of increasing risks from droughts, flooding, and soil degradation. Understanding how environmental stressors affect nutrient yields, crop productivity, and food availability is critical for developing climate-resilient food policies that ensure sustainable nutrition access for all New Zealanders (Barnsley et al., 2021; Drew et al., 2020; Lawrence et al., 2024). Geospatial analysis The NZFPNDS is a geospatial database, which allows for the creation of nutritional risk maps that identify disparities in food access, agricultural productivity, and health outcomes. As the NZFPNDS provides estimates of nutrient production rather than direct measures of nutrient consumption, developing nutritional risk maps would require further information, including: (i) population distribution and nutrient requirements, (ii) domestic consumption versus export allocations, (iii) transport, storage, and processing pathways that affect where food is available and in what form, and (iv) patterns of retail distribution and dietary intake across different population groups. Such integration would allow the estimation of risk and account for the fact that much of New Zealand's food production is exported. Integrating GIS-based layers of agricultural data with demographic and health datasets enables a better understanding of the challenges between food production and nutrient intake at the regional to national scale. This could provide key insights for local and national governments when designing targeted interventions to address food security. In addition, the NZFPNDS could facilitate regional food system analysis by exploring the implications of urbanisation, land-use change, and agricultural diversification on food availability and nutritional security. International nutrient trade Beyond its national applications, the NZFPNDS also offers opportunities for global and comparative research, allowing for benchmarking against international food systems and nutrient trade (Silvestrini et al., 2024). By integrating the database with global datasets such as FAOSTAT and WHO nutritional data, we can conduct cross-country comparisons on food system resilience, agricultural sustainability, and nutritional security (Fletcher et al., 2024). This international perspective is particularly valuable for identifying best practices and policy lessons that could inform the future evolution of NZ's food system. Furthermore, as a FAIR-aligned database (Findable, Accessible, Interoperable and Reusable), the NZFPNDS serves as a model for how spatially explicit nutritional data can be structured, fostering collaboration with global researchers, policymakers, and organisations. Given its broad applicability across multiple disciplines, the NZFPNDS stands as a transformative research tool for understanding and addressing the challenges of food production, trade, and nutrient security. The ability to link agricultural production with national dietary needs through spatial data represents a critical step forward in evidence-based food system research. As NZ and the global community continue to face complex food security challenges, the insights derived from the NZFPNDS may support future efforts to guide policy, promote more sustainable food practices, and contribute to better alignment between national food production and nutritional needs, with potential implications for improving population health over time.