Territorial nitrogen budget and metabolic typologies of EU Agri-Food Systems, 1990–2019

This dataset was developed to implement the GRAFS (Generalized Representation of Agri-Food Systems) approach and provides harmonized nitrogen (N) fluxes and budgets across agri-food systems for 121 territories across the European Union, mainly at the subnational level, covering the period 1990-2019. It builds on previous EU-scale studies by offering consistent and spatially explicit estimates of N inputs, outputs, and internal flows within arable land, permanent cropland, and permanent grassland systems. The dataset includes annual estimates of N content in harvested crops and the areas they cover, categorized by crop type, as well as N inputs from synthetic fertilizers, manure application, symbiotic nitrogen fixation, atmospheric deposition, and ammonia volatilization. It also provides estimates of feed and food imports, animal production, and the allocation of N from cropland and grassland to local human and livestock consumption, as well as exports beyond the territory. All data are based on publicly available sources, primarily Eurostat, and have been integrated using standardized correction and gap-filling procedures to ensure consistency and temporal continuity. Each data point includes a confidence flag indicating whether it is raw, corrected, or gap-filled, with full documentation of the methods used.

The harmonized dataset covers a total of 121 territories over a 30-year period. It is organized as follows: (i) for each of the 16 aggregated crop categories, we report the following three variables: total harvested area (Mha), N harvested (Gg N), and N yield (kg N/ha). (ii) We include total areas (in Mha) for the following major land use types: permanent grassland, temporary grassland, permanent crops, arable land and total cropland. (iii) N deposition rates to cropland are reported in kg N/ha. (iv) Symbiotic N fixation inputs are reported for pulses, temporary grasslands, forage legumes, and soybeans (all in Gg N). (v) Total synthetic fertilizer application to cropland is disaggregated into applications to arable land, permanent grassland and permanent crops (Gg N). (vi) Total N excretion is estimated per animal category (8 categories; Gg N). (vii) We report the N content (Gg N) for animal products (milk, eggs and meat, disaggregated by species and by edible and non-edible parts), total excretion, local feed use, and imported feed. (viii) We include estimates of total excretion, housing excretion, storage and housing losses, grazing excretion on permanent and temporary grassland, and manure application to different land use types (permanent grassland, temporary grassland, permanent crops and other arable land), all expressed in Gg N.

The resulting dataset is provided in a tidy format to facilitate efficient access to and processing of the data. Each file includes a "confidence" column documenting the treatment applied to each data point, referencing the methods detailed in the supplementary material. The "unprocessed" flag indicates that the value corresponds to raw data extracted directly from the sources (mainly Eurostat) at the same spatial resolution. However, this flag does not necessarily imply greater reliability, given that numerous inconsistencies were identified in Eurostat datasets. The "filled" flag denotes that missing values were estimated using national-level data distributed across subnational territories. This flag is always accompanied by a letter referring to the specific filling procedure described in the supplementary material. The "interpolated" flag indicates that the value was obtained through linear interpolation or extrapolation. Finally, the "corrected" flag refers to values that were adjusted because the original source was deemed unreliable. These confidence flags enable users to extract data selectively based on the level of reliability required for their specific application.