Agricultural Systems最新文献

CONTEXT

Yield forecasting, the science of predicting agricultural productivity before the crop harvest occurs, helps a wide range of stakeholders make better decisions around agricultural planning.

OBJECTIVE

This study aims to investigate whether machine learning-based yield prediction models can capably predict Kharif season rice yields at the district level in India several months before the rice harvest takes place.

METHODOLOGY

The methodology involved training 19 machine learning models such as CatBoost, LightGBM, Orthogonal Matching Pursuit, and Extremely Randomized Trees on 20 years of climate, satellite, and rice yield data across 247 of India's rice-producing districts. In addition to model-building, a dynamic dashboard was built understand how the reliability of rice yield predictions varies across district.

RESULTS AND CONCLUSIONS

The results of the proof-of-concept machine learning pipeline demonstrated that rice yields can be predicted with a reasonable degree of accuracy, with out-of-sample R2, MAE, and MAPE performance of up to 0.82, 0.29, and 0.16 respectively. This performance outperformed test set performance reported in related literature on rice yield modelling in other contexts and countries. In addition, SHAP value analysis was conducted to infer both the importance and directional impact of the climate and remote sensing variables included in the model. Important features driving rice yields included temperature, soil water volume, and leaf area index. In particular, higher temperatures in August correlate with increased rice yields, particularly when the leaf area index in August is also high. Building on the results, a proof-of-concept dashboard was developed to allow users to easily explore which districts may experience a rise or fall in yield relative to the previous year. The dashboard show that the model may perform better in some regions than in others. For instance, the absolute percentage error for predicted versus actual yields ranged from an average of 7.1 % in districts in Uttarakhand to an average of 14.7 % in Uttar Pradesh.

SIGNIFICANCE

This study underscores the potential for policymakers to consider scaling and operationalizing machine learning approaches to rice yield prediction in the context of agricultural early warning systems to deliver timely crop yield forecasts on a rolling basis throughout the season, thereby equipping agricultural decision-makers with the ability to make informed choices on irrigation scheduling, fertilizer application, and harvest planning to optimize crop output and resource use.

CONTEXT

The Iberian Peninsula is the world's largest olive (Olea europaea subsp. europaea L.) producing region due to its high environmental suitability for olive growing, consistently accounting for about half of the global share. Moreover, it includes a range of olive-producing regions with Protected Designation of Origin (PDO), aimed to safeguard and promote the distinctive geographical status of agricultural products linked to unique environmental characteristics. Despite the olive industry's economic importance, the impact of climate change on the environmental suitability and the environmental distinctiveness of olive-producing regions is still far from being understood.

OBJECTIVE

The objective of our work was twofold. First, to evaluate changes in the spatial distribution patterns of environmental suitability for olive growing both within and outside PDOs across the Iberian Peninsula under two climate change scenarios within a 2050 time horizon. Second, to evaluate the ability of PDOs to retain their distinctive environmental characteristics in response to new climate regimes.

METHODS

The study area was framed using 1 × 1 km square plots. We used an Ecological Niche Modelling approach, firstly, to model the environmental correlates of environmental suitability for olive growing and, secondly, to forecast their relative change within and outside PDOs. The estimated change in environmental suitability for olive growing was calculated as the percentage variation between the present and each climate change scenario. Additionally, a Random Forests Modelling approach was employed, firstly, to model the environmental correlates of PDOs and, secondly, to evaluate their environmental distinctiveness based on the probability of belonging to a given PDO. The estimated change in environmental distinctiveness of PDOs was calculated as the percentage variation between present and future in the probability of belonging to the same PDO.

RESULTS AND CONCLUSIONS

Our results suggest significant climate-driven range shifts of environmental suitability toward northern latitudes, leading to widespread reductions in southern latitudes both within and outside PDO olive-growing regions. Climate change will also severely affect the idiosyncratic environmental envelope of most PDOs, leading to the loss of their environmental distinctiveness.

SIGNIFICANCE

Our study demonstrates that climate change's impact on olive growing in the Iberian Peninsula might be stronger than previously thought. We propose exploiting the existing genotypic and phenotypic diversity related to climate - or climate diversity - as a way to adapt O. europaea crops to shifting climates and, in turn, allow olive growers to continue to grow in their current location for many years to come.

CONTEXT

Livestock are an important component of livelihoods in smallholder dairy systems in Africa, but are characterized by low animal productivity and large environmental impacts per unit of animal product (e.g. greenhouse gas emissions (GHG) intensities). Governments in African countries have set ambitious targets for dairy systems, but development of climate-smart strategies has been hindered by a scarcity of baseline data and local intervention trials.

OBJECTIVE

We use a rich dataset from smallholder mixed dairy systems in Kenya to determine whether national climate and development goals for 2030 can be met using locally appropriate interventions. Interventions considered included improved herd management and feed interventions.

METHODS

We conducted a yield gap analysis to determine the scope of the existing milk yield gaps, then evaluated the extent to which yield gaps could be closed using interventions in a second step. We outscaled our results to the national level to determine the potential impact of adopting our interventions on national dairy production and GHG emission goals using the FAO Global Livestock Environmental Assessment Model – interactive (GLEAM-i) tool.

RESULTS AND CONCLUSIONS

Our analysis showed that substantial yield gaps exist in Kenyan dairy systems (39 to 49% of attainable yields). These gaps could be closed by intervention packages but not by individual interventions alone. Our outscaled scenarios showed interventions can reduce milk GHG emission intensities (−6.5 to −27.4%), while absolute emissions would increase in most scenarios (−3.9 to +25.9%). To meet national milk production goals, we estimated that a large increase in animal numbers is needed by 2030 compared to 2010 (from ∼2.7 M to 4.5–7.1 M heads of cattle). However, most scenarios fell short of the emissions target (−4% to +48%) by 2030. It may be possible to narrowly meet Kenyan national milk production and GHG emission goals by 2030.

SIGNIFICANCE

National goals for milk production and reducing GHG emissions were only marginally compatible in Kenya. Other sectors of the economy will need to reduce emissions to ensure that food and nutrition security objectives are not jeopardized. In order to achieve national milk goals, there will be need to be a consummate increase in animal numbers even with the adoption of multiple interventions. To meet Kenya's national emissions goals, widespread adoption of several locally appropriate interventions will be required. International support will be needed to meet Kenya's conditional Nationally Determined Contributions under the 2015 Paris Agreement, as well as food and nutrition security goals.

CONTEXT

National and international agendas are focusing on reducing pesticides due to their detrimental effects on flora, fauna, and human health, which has led to the introduction of agri-environmental programmes aimed at reducing the risk of pesticides. Pesticide reduction in agriculture can have an impact on labour time requirements and profitability.

OBJECTIVE

We used winter wheat, sugar beet, and potatoes as examples to analyse the changes in profitability and working time requirements, including management tasks.

METHODS

For the calculations, we used five different production schemes for each crop: reference; (A) reduction of herbicides; (B) reduction of growth regulators, fungicides, and insecticides; combination of schemes (A) and (B); and organic production. The working time requirements for fieldwork and farm management work were modelled for each scheme and crop. The respective partial costs and benefits of the schemes were calculated for each crop.

RESULTS AND CONCLUSIONS

Based on the model assumptions, scheme (B) appears favourable in terms of working time requirements, and profitability of winter wheat and sugar beet. Scheme (A) offers synergies between the same parameters for potato production. Economic analysis shows that crop production with reduced pesticide use may even experience an increase in financial viability if the yield is not severely jeopardised, and farmers can be compensated through premiums and direct payments.

SIGNIFICANCE

Our results can support policy-making, since the labour time requirement and profitability of pesticide-reduced crop production can affect the success of voluntary agri-environmental programmes for the reduction of the risks from pesticide use in agriculture.

CONTEXT

Sustainable dairy production is included in the policy agenda of many countries in the tropics to address, among others, their commitment to the Paris Agreement. To the best of our knowledge, however, a study to assess the impact of the proposed interventions for sustainable dairy production is still lacking for most of those countries. Using policy goals as entry points to develop scenarios can provide insight into the impact of policy interventions on dairy farming practices.

OBJECTIVE

This study aimed to evaluate the implication of interventions towards sustainable dairy development identified by the governments of Indonesia and Costa Rica.

METHODS

Information about current farming practices (i.e. the baseline) were collected on 32 smallholder dairy farms in Indonesia and 24 dairy farms in Costa Rica. Scenarios were designed based on policy goals for dairy development and climate change mitigation in each country. The scenarios for Indonesia encompassed relocation of the dairy sector to Sumatra to allow coupling of livestock to land combined with a restriction on manure production to ensure all manure to be applied to grow forage, and a restriction on the amount of purchased feeds, at two levels: maximally 100% and 50% of the baseline. The scenarios for Costa Rica included a silvopastoral system and a reduction in the amount of purchased feeds, at two levels: 50% and 80% lower than the baseline. We estimated greenhouse gas (GHG) emissions at chain level and carbon (C) stocks at farm level.

RESULTS AND CONCLUSIONS

The scenarios for Indonesia increased herd size and milk output by 240–360%, and GHG emissions per farm by 269–455%, while decreased GHG emissions per kg milk by 1–10%, compared to the baseline. C stocks per farm were higher in the scenarios than in the baseline, but compared to natural vegetation much more C is lost under the scenarios because more land is being used. The scenarios for Costa Rica reduced herd size and milk output by 5–25% and GHG emissions per farm by 17–35%, while GHG emissions per kg milk decreased by 10%, compared to the baseline. C stocks per farm were comparable.

SIGNIFICANCE

To achieve the multiple policy goals for sustainable dairy development, the governments need to consider the trade-off between increasing milk production and reducing GHG emissions. In Indonesia, relocation of the dairy sector needs a strict policy to avoid the expansion of dairy farms into tropical forest land. Furthermore, the Costa Rican government needs to incentivise dairy farmers to implement a silvopastoral system to reduce GHG emissions and land use. This, however, will be at the expense of milk output.

CONTEXT

Estimates of conservation agriculture (CA) adoption vary worldwide because of a lack of a standardized methodology to quantify the simultaneous utilization of its core principles of minimum soil disturbance, permanent soil organic cover and crop diversification. Comparisons of CA adoption among farms across regions requires estimation of the farm area and cropping season where CA principles are applied.

OBJECTIVE

To develop the Conservation Agriculture Appraisal Index (CAAI) as a standardized conceptual framework with defined thresholds that indicates the intensity and frequency of use of each CA core principle. CAAI was subsequently applied to quantify CA adoption on farms across four wheat (triticum aestivum) growing regions, both with and without livestock, including dryland and irrigated systems in Australia and Mexico, respectively.

METHODS

CAAI is a continuous scoring system that estimates the intensity and frequency of application of the core principles and their concurrent utilization to assess the extent of CA adoption. CAAI score is the sum of the scores of each core principle, accounting for the percentage of the farm area and cropping season where CA is applied. CAAI emerged from semi-structured interviews, questionnaires, and farm visits that captured underlying patterns of CA use in regional-specific contexts.

RESULTS AND CONCLUSIONS

CAAI assessed annual CA adoption on 100 farms in four wheat growing regions with different environments and farming systems. The adoption of CA was higher in Australia than Mexico, where partial adoption was more prevalent, especially for summer crops. ‘No adoption’ of CA occurred when one of the core principles consistently scored zero within a year.

SIGNIFICANCE

The CAAI can be used as a benchmarking research tool at the farm level to standardize units for comparisons and identify levels of CA adoption by farm area and cropping seasons between and across regions.

CONTEXT

Life Cycle Assessment (LCA) remains a method of choice for assessing the environmental performance of agricultural systems. However, it is rarely applied to multifunctional extensive production systems, in which livestock use, apart from animal production, maintains a continuous disturbance that sustains the diversity of habitats and species.

OBJECTIVE

This study aims to assess the environmental impact and biodiversity of extensive ruminant production on semi-natural grasslands (SNG), that is, High Nature Value (HNV) farming across Europe. We collected data from a total of 41 HNV farms in five countries (Finland, Estonia, Spain, Greece, and France) that produce beef, sheep, and goats, and that incorporate (to a varied degree) semi-natural and permanent pastures into production.

METHODS

We used LCA to assess the potential environmental impact of HNV farms according to global warming potential (GWP₁₀₀), fossil resource scarcity (FRS), water scarcity (WS) and land use (LU), by using the Solagro Carbon Calculator and OpenLCA software. We assessed biodiversity based on the expert scoring system of SALCA-BD. We compared impacts on per area and per product basis across the farms, and related them to the productivity.

RESULTS AND CONCLUSIONS

Results revealed a considerable variation in all environmental impacts among HNV farms, explained mostly by the type of ruminants, main product (meat or milk) and the production level. GWP₁₀₀ per unit in beef product in France was almost twice as high as that in boreal and 3 times more than in Spain, while sheep systems in Greece varied 7-fold for meat. Sheep systems consistently had the highest GWP₁₀₀, while goat systems used the most land, fossil fuel and water. Small ruminant production in Spain had both the highest land occupation and biodiversity values. Biodiversity was at its highest on farms utilising only SNG for production, which, however, related negatively to the farms' production output. Enteric fermentation accounted for 32% of overall emissions.

SIGNIFICANCE

This study makes a novel contribution towards a better understanding of the environmental performance and production capacity of HNV farming systems that are often used as examples of multifunctional and sustainable ruminant-based production.

Context

On pioneer fronts, the new spatial-temporal evolution of agriculture needs to be understood to help farmers find their way to conciliate food production and forest conservation. Analyzing farm trajectories is consequently critical for designing such futures and to assess their commitments with agroecology principles.

Objective

Based on the analysis of the literature on farm trajectories and pathways we proposed a renewed analytical framework to analyze farm trajectories in pioneer fronts and support the identification of desirable strategies for the future.

Methods

A systematic review adapted from the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology was used. From an initial record of 246 papers, 81 were selected as eligible for the review. The articles were classified in six categories according to three criteria: i) the retrospective or prospective analysis of farm trajectories, ii) the consideration or not of the territorial scale (drawing lessons at territorial scale), iii) the use or not of modeling tools. We also explored whether off-farm factors (such as existing infrastructure at territorial scale or access to credit) and intra-farm factors (such as the organization of family work and the role of women within this organization) were considered since these factors affect farms trajectories in pioneer fronts.

Results and conclusions

Results indicate that the concept of trajectory is mainly associated with retrospective analyses while the concept of pathway is mostly associated with prospective studies generally using simulation tools for the design of future scenarios. The link between trajectories and agroecological principles also has been little explored in the literature. Both retrospective and prospective studies fail to pay sufficient attention to the roles of women and family organization. Lastly, most of the methodologies studied do not fully consider the effects of off-farm territorial factors and public policies on these trajectories.

We propose an analytical framework that would address these limitations.

Significance

This framework is currently used in Brazilian and Colombian Amazon and will help defining sustainable farm trajectories limiting deforestation. Such a framework is needed to support farm development on pioneer fronts and broadly in territories that must deal with highly critical environmental agendas.

CONTEXT

Current agricultural systems in the Netherlands and other parts of Europe depend on large quantities of nutrients from imported feed and artificial fertilizers. This may result in unwanted nutrient accumulation and losses and negative environmental impacts. One way of making these systems more sustainable is to balance livestock numbers with the local feed supply and nutrient requirements for crop production. However, there is currently limited quantitative- and spatially-explicit insight into the effects of reduced imports on livestock numbers and nutrient balances.

OBJECTIVE

The aim of this study is to quantify the number of livestock that can be sustained when minimizing nitrogen (N) imports of feed and artificial fertilizers into Dutch agriculture.

METHODS

We developed a spatially-explicit mathematical programming model to optimize livestock numbers in different scenarios of allowed N imports. The model takes into account current flows of nutrients, related to nutrient supply from livestock manure, feed and fodder, and nutrient requirements in agricultural production.

RESULTS AND CONCLUSIONS

Minimizing N feed imports and artificial N fertilizer use under current land use resulted in an overall reduction of livestock density of 57.4% compared to the current livestock density based on 2020 activity data. In addition, it led to an increase in artificial N fertilizer inputs of 20.5% to maintain the N requirements of agricultural land. In contrast, focusing on minimizing artificial fertilizer use led to a reduction in livestock density of 25.8% compared to the current livestock levels. Depending on the scenario, we found strong regional variation in nutrient balances and livestock numbers. For instance, while currently the application rates of artificial fertilizers and livestock manure are relatively constant in space, this was no longer the case when N import in the form of feed was minimized. In that case, the crop-dominated areas showed a deficit in livestock manure and a substantial increase of artificial fertilizers.

SIGNIFICANCE

The presented model allows for assessing spatially explicit impacts of reduced N imports of feed and artificial fertilizers on livestock density. Using the Netherlands as case study, we show that reducing feed imports is a more effective leverage point to reduce N surplus and external N inputs into agriculture as compared to when reducing artificial fertilizer use. Our findings can inform the development of area-specific strategies that are aimed at reconnecting livestock and agricultural land.