Dataset for E-GRAFS: France, 33 regions, 1852-2014

The dataset is based on the one built by Le Noë (2018). In this dataset, mainland France is divided in 33 agricultural regions, aggregating adjacent French administrative departments (NUTS 3) having relatively close agricultural systems. The proximity of agricultural systems is captured by the distribution of Utilized Agricultural Area (UAA) between cropland and grassland and the ratio of agricultural imports to exports measured in terms of nitrogen flows (Le Noë 2018, 29). In addition, the dataset includes mainland France as an aggregate of those 33 regions. The dataset spans from 1852 to 2014, with annual records for 22 non-consecutive years. These records are spread out before 1965, covering 1852, 1885, 1906, 1929, 1946 and 1955. From 1965 to 2004, we have records every four to five years, and every two years from 2004 to 2014. This dataset was originally used in the GRAFS model, predecessor of E-GRAFS. As such, the input data has been reworked to match the E-GRAFS input format. In this format, we have 36 cultures compartments, 6 livestock compartments, 2 populations compartments, 57 products compartments, and 18 excretion compartments. The full list of compartments can be accessed in the project file. While GRAFS computed livestock excretion directly from animal headcounts using excretion factors, E-GRAFS externalized this step to allow for more flexibility in the input data. As such, data from Le Noë (2018) was reworked to estimate livestock units (LU) from headcount data, as well as the nitrogen excretion by LU. A LU being defined as a dairy cow producing 3000 kg of milk per year (Eurostat definition), we estimated the nitrogen excretion per LU using the average yearly milk production of dairy cows, and the nitrogen excretion per dairy cow provided by Le Noë (2018, 39–44). We then estimated the LU for each livestock type from the livestock headcount, the nitrogen excretion per head, and the estimated nitrogen excretion per LU. Some incomplete time series in the dataset from Le Noë (2018) were completed. This includes especially poultry production and headcount data, that were completely absent before 1970. While this is explained by a lack of available data – poultry production data was included in French agricultural statistics only from 1967-on – this creates a substantial missing dimension when it comes to studying past agricultural systems relying on local- or self-consumption of some agricultural products – especially eggs and poultry. This also creates inconsistencies when egg production is included in the dataset, but not the corresponding livestock headcount, that originated this production and consumed the corresponding feed. This issue becomes even more visible with E-GRAFS, that disaggregates nitrogen flows that were previously aggregated, creating inconsistencies in the accounting structure. To solve this issue, poultry production data before 1967 was estimated using headcount data provided in agricultural statistics (Ministère de l’agriculture 1936) and production to headcount ratios computed for each region in 1967 (Ministère de l’agriculture 1968), assuming those ratios stayed relatively constant before that. The fertilizer use time series were completed using data from UNIFA (2016) from 1973 to 2014, and – for the mainland France aggregates only – Insee (1966) from 1886 to 1965, and Ministère de l’agriculture (1971) for 1970. In contrast with GRAFS, E-GRAFS disaggregates compartments and the flows between them. The optimization model distributes local agricultural production and imports to their human and animal consumers coherently, including through exports when production exceeds needs. To ensure that this distribution is plausible and not just fulfilling total nitrogen needs, this optimization model requires base diets to use as guidelines. The optimization model dynamically adjusts those base diets depending on resources and consumption, building realistic diets adapted to the historical and region-specific contexts. This adaptability ensures that the model remains grounded in observed constraints while responding to the spatial and temporal variability of agricultural systems. A full description of the allocation model of E-GRAFS is provided in the E-GRAFS documentation. Each of a 6 livestock types has a base diet applied to every territory and time period. In order to build those base livestock diets, we use data from a wide range of authoritative sources, including agricultural guidelines, technical reports, and scientific studies (Steinfeld et al. 2006; Jarrige 1988; Jousseins and Tchakerian 2014; Interbev, interprofession bétail et viande n.d.; Devun, Brunschwig, and Guinot 2012; Arranz 2019; Bossis and Jost 2016; Martin-Rosset 2012). In contrast with base livestock diet, the base human diet is historicized and spatialized to reflect the major changes that happened over the studied time period regarding the animal or vegetal origin of proteins, the average protein consumption per capita, and heterogeneity between regions. We have two fixed baseline diets, for vegetal and animal products respectively. Those fixed base diets were historicized and spatialized by weighting them using the ratio of vegetal or animal nitrogen consumption over total consumption. We use the regimes reconstructed by Le Noë (2018, 30–33) as the source for these weights. The livestock and human diets are available in the data file. References Arranz, Jean-Marc. 2019. “La conduite des brebis laitières en lots d’alimentation.” AUTELO, AUtonomie alimentaire dans les ELevages Ovins laitiers. https://idele.fr/autelo/?eID=cmis_download&oID=workspace%3A%2F%2FSpacesStore%2F7a11fafb-9813-49db-9d07-49d104bb661e&cHash=8f4369abe6dce3c0cb9da216fc454ede. Bossis, Nicole, and Jérémie Jost. 2016. “Observatoire de l’alimentation des chèvres laitières françaises.” France: Institut de l’Élevage (IDELE). https://idele.fr/detail-article/observatoire-de-lalimentation-des-chevres-laitieres-francaises. Devun, Jean, Philippe Brunschwig, and Caroline Guinot. 2012. “Alimentation des bovins : Rations moyennes et autonomie alimentaire.” Collection Résultats. Institut de l’Élevage (IDELE). https://www.la-viande.fr/sites/default/files/documentheque/alimentation-des-bovins-rations-moyennes-autonomie-alimentaire.pdf. Insee. 1966. Annuaire statistique de la France 1966 : résumé rétrospectif. Annuaire statistique de la France 72. Paris: Imprimerie nationale. https://www.bnsp.insee.fr/ark:/12148/bpt6k6424052p. Interbev, interprofession bétail et viande. n.d. “L’alimentation des ovins.” la-viande.fr. Accessed November 18, 2024. https://www.la-viande.fr/animal-elevage/agneau/alimentation-ovins. Jarrige, Robert. 1988. Alimentation des bovins, ovins et caprins. Paris: INRA Editions. https://hal.inrae.fr/hal-02854912. Jousseins, Carole, and Edmond Tchakerian. 2014. “Que mangent les brebis ? Régimes alimentaires et niveaux d’autonomie des élevages.” Collection Théma. Edited by INOSYS Réseaux d’Élevage. France: Institut de l’Élevage (IDELE). https://www.inn-ovin.fr/wp-content/uploads/2017/04/INOSYS-RESEAUX-ELEVAGE_THEMA-Que-mangent-les-brebis.pdf. Le Noë, Julia. 2018. “Biogeochemical Functioning and Trajectories of French Territorial Agricultural Systems : Carbon, Nitrogen and Phosphorus Fluxes (1852-2014).” PhD thesis, Paris: Sorbonne Université. https://tel.archives-ouvertes.fr/tel-02355318. Martin-Rosset, William. 2012. Nutrition et alimentation des chevaux : nouvelles recommandations alimentaires de l’INRA. Savoir faire. Versailles: Quae. https://www.quae.com/produit/1083/9782759221349/nutrition-et-alimentation-des-chevaux. Ministère de l’agriculture. 1936. Statistique agricole de la France : résultats généraux de l’enquête de 1929. Paris: Imprimerie nationale. https://www.bnsp.insee.fr/ark:/12148/bpt6k6209868t. ———. 1968. Statistique agricole 1968 : résultats de 1967. Paris: Imprimerie nationale. https://www.bnsp.insee.fr/ark:/12148/bc6p06z4ddf. ———. 1971. Annuaire 1971 de statistique agricole : résultats de 1970. Vol. 1. 2 vols. Paris: Imprimerie nationale. https://www.bnsp.insee.fr/ark:/12148/bc6p07djjdb. Steinfeld, Henning, Pierre Gerber, Tom D. Wassenaar, Vincent Castel, M. Rosales, and de Haan, C. 2006. Livestock’s Long Shadow: Environmental Issues and Options. Rome: FAO. https://openknowledge.fao.org/handle/20.500.14283/a0701e. UNIFA. 2016. “Historique des livraisons d’engrais en France par département 1972-2015.” Dataset. Personal communication. https://www.unifa.fr/statistiques-du-secteur/les-statistiques-de-campagne-retrouvez-lhistorique-des-campagnes-de.