Changes in global food consumption increase GHG emissions despite efficiency gainsalongglobal supply chains 全球食品消费的变化增加了温室气体排放，尽管在全球供应链上取得了效率提高
Received: 4 November 2022 收到：2022 年 11 月 4 日
Accepted: 9 May 2023 接受日期：2023 年 5 月 9 日
Published online: 15 June 2023 2023 年 6 月 15 日在线发布
Check for updates 检查更新
Yannan Zhou (1) Klaus Hubacek (1)
}
Abstract 摘要
Greenhouse gas (GHG) emissions related to food consumption complement production-based or territorial accounts by capturing carbon leaked through trade. Here we evaluate global consumption-based food emissions between 2000 and 2019 and underlying drivers using a physical trade flow approach and structural decomposition analysis. In 2019, emissions throughout global food supply chains reached of anthropogenic GHG emissions, largely triggered by beef and dairy consumption in rapidly developing countries-while per capita emissions in developed countries with a high percentage of animal-based food declined. Emissions outsourced through international food trade dominated by beef and oil crops increased by equivalent, mainly driven by increased imports by developing countries. Population growth and per capita demand increase were key drivers to the global emissions increase (+30% and , respectively) while decreasing emissions intensity from land-use activities was the major factor to offset emissions growth ( . Climate change mitigation may depend on incentivizing consumer and producer choices to reduce emissions-intensive food products. 温室气体（GHG）排放与食品消费相关，通过捕捉通过贸易泄漏的碳来补充基于生产或领土的账户。在这里，我们使用物理贸易流动方法和结构分解分析评估了 2000 年至 2019 年间全球基于消费的食品排放及其基础驱动因素。2019 年，全球食品供应链中的排放达到人为 GHG 排放的 ，主要是由于快速发展中的国家牛肉和乳制品消费所引发的，而发达国家的人均排放量却下降，这些国家的动物性食品比例较高。通过国际食品贸易外包的排放主要由牛肉和油料作物主导，增加了 当量，主要是由于发展中国家的进口增加。人口增长和人均需求增加是全球排放增加的关键驱动因素（分别增加了 30%和 ），而土地利用活动的排放强度降低是抵消排放增长的主要因素（ ）。气候变化缓解可能取决于激励消费者和生产者选择减少排放密集型食品产品。
The agrifood system drives global land-use, agricultural and other beyond-farm activities and contributes to about one-third of global anthropogenic greenhouse gas (GHG) emissions . The United Nations projects that an additional of the current food demand will be needed to feed the world's estimated population of 9.1 billion by 2050 (ref. 4). Population growth, expansion of food production and an increase in animal-based diets are likely to further increase emissions and squeeze the global carbon budget . Thus, mitigating emissions at every stage of food supply chains from production to consumption is crucial to limit global warming . 农食系统推动全球土地利用、农业和其他农场以外的活动，并贡献约全球人为温室气体（GHG）排放的三分之一 。联合国预计，到 2050 年，将需要额外 的当前食品需求来满足预计的 91 亿人口（参考文献 4）。人口增长、食品生产的扩张以及动物性饮食的增加可能进一步增加排放量并挤压全球碳预算 。因此，从生产到消费的食品供应链的每个阶段减少排放对于限制全球变暖至关重要 。
Production-based emissions (PBE) or territorial emissions are based on emissions from production (including exports) within a region . Previous studies have quantified global GHG emissions from food production based on global food-related emissions inventories (for example, FAOSTAT, provided by the Food and Agriculture Organization of the United Nations (https://www.fao.org/faostat/ en/#data/); and EDGAR-Food, provided by EDGAR-Emissions Database for Global Atmospheric Research (https://edgar.jrc.ec.europa. eu/edgar_food)). However, food products are increasingly traded internationally through global supply chains, and geographically distant consumer demand may lead to emissions outsourcing to producers . Consumption-based emissions (CBE) accounting allocates emissions from producers to final consumers irrespective of the place of production . CBE is complementary to PBE and allows allocating responsibility and informs emissions mitigation from a consumer perspective. CBE helps to understand to what extent final consumers trigger emissions along the entire global supply chain, allows quantification of virtual flows in trade outsourced to other countries and provides information for additional policy tools for emissions mitigation with a focus on consumption . Therefore, a detailed assessment of global consumption-based GHG emissions throughout food supply chains with a breakdown into the detailed process and product levels is needed to reveal the distant emissions drivers and to facilitate emissions mitigation from a consumer perspective. However, such consumption-based assessments are hampered due to the complexity and variety of processes in which different food products are cultivated, processed and traded through multiple intermediate region and the required degree of data consistency and granularity in terms of processes and products of the global agrifood system. 基于生产的排放（PBE）或领土排放是基于一个地区内生产（包括出口）的排放。先前的研究已经根据全球食品相关排放清单（例如，由联合国粮食和农业组织提供的 FAOSTAT（https://www.fao.org/faostat/ en/#data/）和由全球大气研究排放数据库 EDGAR 提供的 EDGAR-Food（https://edgar.jrc.ec.europa. eu/edgar_food））量化了食品生产的全球温室气体排放。然而，食品产品越来越通过全球供应链进行国际贸易，地理上相距甚远的消费者需求可能导致排放外包给生产者。基于消费的排放（CBE）会将排放从生产者分配给最终消费者，而不考虑生产地点。CBE 是对 PBE 的补充，可以分配责任并从消费者角度指导减排工作。 CBE 有助于了解最终消费者在整个全球供应链中触发排放的程度，允许量化委托给其他国家的贸易中的虚拟流量，并提供有关消费方面排放减少的额外政策工具的信息。因此，需要对整个食品供应链中的全球基于消费的温室气体排放进行详细评估，包括详细的过程和产品级别的细分，以揭示远程排放驱动因素，并促进从消费者角度减排。然而，由于不同食品产品在多个中间地区通过多种不同的培育、加工和贸易过程，以及全球农食系统的过程和产品方面所需的数据一致性和细粒度的复杂性和多样性，这种基于消费的评估受到阻碍。
A number of studies use a bottom-up life-cycle assessment (LCA) to investigate emissions of specific food products during their life cycle . However, these results are not comparable because of differences in scope and oftentimes ignore differences in emissions from different origins along global food supply chains . With the international, time-series input-output databases at high sectoral detail, multi-regional input-output (MRIO) analysis is now widely used for tracing consumption-based emissions . MRIO is applied to quantify emissions induced by food consumption based on inputoutput relations (in monetary values) along supply chains . This approach has been frequently criticized due to its highly aggregated sectors lacking product details . For example, soybean, together with other oilseed crops such as palm oil and rapeseed, is aggregated in the same oil crop sector, ignoring important finer-scale differences in terms of land use, input requirements and associated emissions. Physical trade flow accounting (PTF) provides a more detailed analysis of trade flows for agricultural products based on higher sectoral and product resolution . Some PTF based on bilateral trade approaches uses the difference between production, imports and exports to calculate GHG emissions from food consumption but without consideration of re-export via longer international supply chains. The improved PTF developed by Kastner et al. provides a framework with detailed data to link consumption and associated impacts to the origins of cultivated crops or livestock (on-farm stages) beyond bilateral trade . 一些研究使用自下而上的生命周期评估（LCA）来调查特定食品产品在其生命周期中的排放 。然而，由于范围 的差异以及往往忽视全球食品供应链中不同来源的排放差异 ，这些结果是不可比较的。随着高部门详细信息的国际、时间序列投入产出数据库的出现，多区域投入产出（MRIO）分析现在被广泛用于追踪基于消费的排放 。MRIO 被应用于根据供应链 中的投入产出关系（以货币价值表示）来量化由食品消费引起的排放。由于其高度聚合的部门缺乏产品细节 ，这种方法经常受到批评。例如，大豆与其他油籽作物（如棕榈油和菜籽油）被聚合在同一油料作物部门中，忽视了土地利用、投入需求和相关排放方面的重要细微差异。物理贸易流量核算（PTF）基于更高的部门和产品分辨率提供了对农产品贸易流向的更详细分析 。 基于双边贸易方法的一些 PTF 使用生产、进口和出口之间的差异来计算食品消费的温室气体排放 ，但没有考虑通过更长的国际供应链进行再出口。Kastner 等人开发的改进型 PTF 提供了一个框架，其中包含详细数据，将消费和相关影响与栽培作物或牲畜的起源（农场阶段）联系起来，超越了双边贸易 。
Here we analyse the trend of consumption-based food GHG emissions of 153 products (both animal- and plant-based food) in 181 countries or areas for the years 2000,2005, 2010, 2015 and 2019. Using the PTF approach by Kastner et al. and detailed trade data from FAOSTAT , we reallocate production-based emissions and from agricultural land use and land use change (LULUC), agricultural production and beyond-farm processes (excluding emissions from household and end of life throughout the supply chains of 153 products to final consumers. All emissions are in equivalents ( -eq) using 100-year global warming potentials of and used in the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5). We quantify emissions embodied in food domestic supply and trade (that is, imports and exports) between countries involving re-exports. Finally, structural decomposition analysis is applied to identify the contributions of five driving factors from production to consumption to variations in consumption-based emissions-namely emissions intensity, trade structure, domestic supply ratio, per capita consumption and population. Our study uses the most recent data to attribute emissions across the entire food supply chains at a global scale to final consumers with a consistent and detailed breakdown of processes and products. This allows us to indicate how to reduce food emissions from production to consumption through policy applications for the entire supply chain and final consumers. 在这里，我们分析了 181 个国家或地区的 153 种产品（包括动物和植物食品）在 2000 年、2005 年、2010 年、2015 年和 2019 年的基于消费的食品温室气体排放趋势。利用 Kastner 等人的 PTF 方法 和 FAOSTAT 的详细贸易数据 ，我们重新分配了基于生产的 排放 和 从农业用地利用和土地利用变化（LULUC）、农业生产以及农场以外的过程（不包括家庭和终端 的排放）到 153 种产品的供应链最终消费者。所有排放均以 当量（ -eq）计算，使用了气候变化政府间专门委员会（IPCC）第五次评估报告（AR5）中使用的 100 年全球变暖潜势 和 。我们量化了食品国内供应和贸易（即进口和出口）中包含的排放，涉及再出口。 最后，应用结构分解分析来确定从生产到消费的五个驱动因素对消费排放变化的贡献，即排放强度、贸易结构、国内供应比例、人均消费和人口。我们的研究使用最新数据，将排放归因于全球范围内整个食品供应链上的最终消费者，具有一致且详细的过程和产品细分。这使我们能够指出如何通过针对整个供应链和最终消费者的政策应用来减少从生产到消费的食品排放。
Results 结果
Emissions driven by global and national food consumption 全球和国家食品消费驱动的排放
In 2019, food consumption in the five highest-emitting countries, China (2.0 Gt CO -eq), India (1.3 Gt), Indonesia (1.1 Gt), Brazil and the United States (1.0 Gt), was responsible for more than of global food supply chain emissions (16.0 (95% confidence interval 11.4-20.7) -eq), which cover most of the emissions of the global agrifood system (Fig. 1; details of uncertainty ranges in Supplementary Table 1). Annual global GHG emissions associated with food increased by (that is, -eq) from 2000 to 2019 , which largely owes to consumption rise in populous countries, with China contributing , India and Pakistan to emissions growth. 2019 年，排放量最高的五个国家，中国（2.0 Gt CO2 当量）、印度（1.3 Gt）、印度尼西亚（1.1 Gt）、巴西和美国（各 1.0 Gt），其食品消费占全球食品供应链排放的超过一半（16.0（95%置信区间 11.4-20.7）CO2 当量），这覆盖了全球农业食品系统的大部分排放（图 1；不确定性范围的详细信息请参见附表 1）。与食品相关的全球温室气体排放量从 2000 年至 2019 年增加了（即 CO2 当量），这主要归因于人口众多国家的消费增长，其中中国贡献了，印度，巴基斯坦。
The substantial increase in consumption of animal-based products contributed to of the global emissions rise, reaching almost half of the total food emissions , with 7.9 (5.9-10.1) -eq in 2019. We find that many countries have dominated animal-based emissions, represented by Australia ( ), the United States ( ) and South Asian countries including India (63%). The share of animal-based emissions in total emissions continued increasing in most developing countries/regions (for example, Brazil, East Asia) but remained stable in affluent countries. Beef and dairy contributed and of the increase in global animal-based emissions and reached -eq and -eq, respectively, in 2019 (Supplementary Fig. 1; details of the uncertainty ranges in Supplementary Table 2). Top emitters of beef consumption included Brazil ( -eq), the United States (409 Mt) and Argentina (118 Mt) in 2000 but later included Brazil (409 Mt), China (402 Mt) and the United States (365 Mt). Increased consumption of beef led to of China's growth of animal-based emissions. Beef's contribution is similar to pork, which dominates China's meat market. Emissions from beef consumption constitute of animal-based emissions in Brazil, and over occurred in the rest of Latin America and the Caribbean (LAC), the United States, Japan and Southeast Asia. Emissions from India's dairy consumption increased considerably by 1.2 times, reaching of national animal-based emissions and over of global dairy emissions in 2019 . Dairy consumption in Russia, Oceania and European countries also contributed to over half of national animal-based emissions. 动物产品消费的大幅增加导致全球排放量增加 ，几乎达到总食品排放量的一半 ，2019 年达到 7.9（5.9-10.1） -eq。我们发现许多国家主导着动物产品排放，例如澳大利亚（ ）、美国（ ）和包括印度在内的南亚国家（63%）。动物产品排放在大多数发展中国家/地区（例如巴西、东亚）的总排放中所占比例继续增加，但在富裕国家保持稳定。牛肉和乳制品分别贡献了全球动物产品排放增加的 和 ，分别在 2019 年达到 -eq 和 -eq（附图 1；附表 2 中的不确定性范围详细信息）。牛肉消费的主要排放国包括巴西（ -eq）、美国（409 Mt）和阿根廷（118 Mt）在 2000 年，但后来包括巴西（409 Mt）、中国（402 Mt）和美国（365 Mt）。牛肉消费的增加导致了中国动物产品排放的 增长。牛肉的贡献与猪肉相似，猪肉主导了中国的肉类市场。 牛肉消费排放占巴西动物类排放的 ，其余 发生在拉丁美洲和加勒比地区（LAC）、美国、日本和东南亚。印度奶制品消费排放大幅增加 1.2 倍，达到国家动物类排放的 以上，2019 年全球奶制品排放的 以上。俄罗斯、大洋洲和欧洲国家的奶制品消费也贡献了国家动物类排放的一半以上。
The consumption of grains and oil crops is responsible for (3.4 Gt CO 2 -eq in 2019 and -eq) of global plant-based emissions, respectively (Supplementary Fig. 2; details of uncertainty ranges in Supplementary Table 2). Rice contributes to over half of the global grain-related emissions (1.7 -eq), with Indonesia (20%), China (18%) and India (10%) being the top three contributors. Soybean -eq) and palm oil ( -eq) have the largest shares in global emissions from oil crops with and , respectively. Brazil's demand for soy-related food products generated the largest percentage of the world's soybean-related emissions ( ) in 2000 , but it was replaced by China (32%) after 20 years. Indonesia, the world's leading consumer of palm oil, has the largest emissions from palm oil ( of the global total in 2019), followed by Southeast Asia (13%), Western Europe (10%) and China (9%). 粮食和油料的消耗分别占全球植物基排放的 （2019 年为 3.4 Gt CO2 当量 和 当量）（附图 2；不确定性范围详见附表 2）。稻米贡献了全球粮食相关排放的一半以上（1.7 当量），印度尼西亚（20%）、中国（18%）和印度（10%）是排放量最高的三个国家。大豆（ 当量）和棕榈油（ 当量）在全球油料排放中占比最大，分别为 和 。巴西对大豆相关食品的需求在 2000 年产生了全球大豆相关排放的最大百分比（ ），但在 20 年后被中国（32%）取代。印度尼西亚作为全球最大的棕榈油消费国，棕榈油排放最多（2019 年全球总量的 ），其次是东南亚（13%）、西欧（10%）和中国（9%）。
There are apparent inequalities in per capita emissions induced by food consumption worldwide, but the disparities have been gradually declining (Fig. 2). Consistent with the scope of production-based estimates , global average per capita emissions from food supply chains have increased from 1.8 (95% confidence interval 1.6-3.1) to t -eq during the study period (details of uncertainty ranges in Supplementary Table 5). Australia has the highest average animal-based emissions ( -eq per person in 2019) from consumption, followed by Brazil (3.0 per person), Canada (2.5 per person) and the United States (2.1 t per person) (Supplementary Fig. 3; details of uncertainty ranges in Supplementary Table 6). Although 全球食品消费引起的人均排放存在明显的不平等，但这些差距已逐渐减少（图 2）。与基于生产的估计范围一致，食品供应链的全球平均人均排放量在研究期间从 1.8（95%置信区间 1.6-3.1）增加到 t -eq（补充表 5 中不确定性范围的详细信息）。澳大利亚的人均动物排放量最高（2019 年为 -eq 每人），其次是巴西（每人 3.0 ）、加拿大（每人 2.5 ）和美国（每人 2.1 吨）（补充图 3；补充表 6 中不确定性范围的详细信息）。尽管
Fig. 1 GHG emissions throughout global supply chains from consumption of food products by country in 2000 and 2019. a, b, The background map shows the level of consumption-based emissions at the country scale in 2000 (a) and 2019 (b). The pie chart shows the fraction of consumption-based emissions of animal-based and plant-based food products in 2000 (a) and 2019 (b), and the size represents the total emissions for 18 countries/regions. AUS: Australia; BRA: Brazil; CAN: Canada; CHN: China; ROEA: rest of East Asia; EE: Eastern Europe; IND: India; IDN: Indonesia;JPN: Japan; ROLAC: rest of Latin America and the 图 1 2000 年和 2019 年各国食品产品消费全球供应链温室气体排放。a、b、背景地图显示了 2000 年（a）和 2019 年（b）各国消费为基础的排放水平。饼图显示了 2000 年（a）和 2019 年（b）动物性和植物性食品产品消费为基础的排放占比，大小代表了 18 个国家/地区的总排放量。AUS：澳大利亚；BRA：巴西；CAN：加拿大；CHN：中国；ROEA：东亚其他地区；EE：东欧；IND：印度；IDN：印度尼西亚；JPN：日本；ROLAC：拉丁美洲其他地区
Caribbean; NENA: Near East and North Africa; ROO: rest of Oceania; RUS: Russia; ROSA: rest of South Asia; ROSEA: rest of Southeast Asia; SSA: sub-Saharan Africa; USA:United States; WE: Western Europe. Details for the division and scope of 18 countries/regions are shown in Supplementary Tables 3 and 4. Base map layer: World Countries (http://tapiquen-sig.jimdo.com). Carlos Efraín Porto Tapiquén. Orogénesis Soluciones Geográficas. Porlamar, Venezuela 2015. Based on shapes from Environmental Systems Research Institute. developed countries emit more animal-based emissions per capita ( -eq per person) than the global average, differences exist between these affluent countries. For example, people in Australia, Canada and the United States have higher per capita animal-based emissions than Western Europeans (1.4 t CO -eq per pons ), mainly due to higher red meat consumption. Indonesia ( -eq per person in 2019), Oceania (2.6 per person) and Brazil (2.0 t per person) have the highest level of plant-based emissions per capita despite a downward trend (Supplementary Fig. 4). Canada (1.8 -eq per person) and European countries ( -eq per person) have larger average plant-based emissions than other developed countries, mainly due to large demand for oil crops (for example, palm oil) and stimulants (for example, coffee). Although below the global average of animal- ( -eq per person in 2019) and plant-based emissions (1.1 t per person), per capita GHG emissions of the top two most populous countries, China (1.4 per person) and India (1.0 per person), increased by and , respectively. 加勒比地区; NENA: 中东和北非; ROO: 大洋洲其他地区; RUS: 俄罗斯; ROSA: 南亚其他地区; ROSEA: 东南亚其他地区; SSA: 撒哈拉以南非洲; USA:美国; WE: 西欧。18 个国家/地区的划分和范围详见附表 3 和 4。基础地图层：世界国家（http://tapiquen-sig.jimdo.com）。Carlos Efraín Porto Tapiquén。Orogénesis Soluciones Geográficas。委内瑞拉波尔马尔，2015 年。基于环境系统研究所的形状。发达国家的人均动物排放量（每人 -eq）高于全球平均水平，这些富裕国家之间存在差异。例如，澳大利亚、加拿大和美国的人均动物排放量高于西欧国家（每人 1.4 吨 CO -eq），主要是由于更高的红肉消费。印度尼西亚（2019 年每人 -eq）、大洋洲（每人 2.6 ）和巴西（每人 2.0 吨）的植物排放量人均水平最高，尽管呈下降趋势（附图 4）。加拿大（1.每人每年 8 个（4 -eq）和欧洲国家（每人 5 -eq）的平均植物基排放量比其他发达国家更高，主要是由于对油料作物（例如棕榈油）和刺激物（例如咖啡）的大量需求。尽管动物-（2019 年每人 6-7 -eq）和植物基排放量（每人 1.1 吨）低于全球平均水平，但全球排放量最高的两个人口最多的国家，中国（每人 1.4 个）和印度（每人 1.0 个），分别增加了 10 和 11。
International trade has reshaped food emissions patterns 国际贸易已经改变了食品排放模式
Figure 3 and Supplementary Fig. 5 show the countries with the largest amounts of emissions embodied in food imports and exports and their ratio of domestic emissions to consumption-based emissions in 2019. Emissions from most major exporters are dominated by two categories-oil crops and beef. Indonesia ( -eq in 2019) and Brazil (196 Mt CO -eq) are the world's largest exporters of embodied emissions from oil crops, dominated by palm oil and soybean, respectively. Indonesia's export of oil crop emissions almost tripled during the study period, while Brazil's emissions increased by . Australia (200 Mt CO Ce -eq in 2019) and Brazil ( -eq) export the largest amounts of beef-related emissions, followed by India ( -eq) and the United States ( -eq). We found that major net exporters, excluding Malaysia which highly relies on meat imports (from India, Australia and so on), create over of their food emissions within their national boundaries. As the world's largest net exporter, Brazil's emissions exports reached the highest level in the mid-term of the 图 3 和补充图 5 显示了 2019 年食品进口和出口中排放量最大的国家，以及它们国内排放量与基于消费的排放量之比。大多数主要出口国的排放主要由两类产品主导-油料作物和牛肉。印度尼西亚（2019 年 -eq）和巴西（196 兆吨 CO -eq）是世界上最大的油料作物排放出口国，分别以棕榈油和大豆为主导。印度尼西亚的油料作物排放出口在研究期间几乎翻了三番，而巴西的排放量增加了 。澳大利亚（2019 年 200 兆吨 CO Ce -eq）和巴西（ -eq）出口了最多与牛肉相关的排放量，其次是印度（ -eq）和美国（ -eq）。我们发现，主要净出口国，除了高度依赖肉类进口的马来西亚（来自印度、澳大利亚等），超过 的食品排放在其国内边界内产生。作为世界上最大的净出口国，巴西的排放出口在中期达到了最高水平。
Fig. 2 | Per capita GHG emissions of food consumption by country in 2000 and 2019. a, b, The background map shows the level of per capita consumption-based emissions at the country scale in 2000 (a) and 2019 (b). The pie chart shows the fraction of average consumption-based emissions of animal-based and plant-based food products per person in 2000 (a) and 2019 (b), and the size represents per capita emissions of 18 countries/regions. Abbreviations of 18 countries/regions and the source of the base map are listed in Fig. 1 caption. study period (720 Mt CO-eq in 2010) and declined (to -eq in 2019) in the later period. 图 2 | 2000 年和 2019 年各国人均食品消费温室气体排放量。a、b、背景地图显示了 2000 年（a）和 2019 年（b）各国人均基于消费的排放水平。饼图显示了 2000 年（a）和 2019 年（b）每人动物性和植物性食品的平均基于消费的排放量的比例，大小代表 18 个国家/地区的人均排放量。18 个国家/地区的缩写和基础地图的来源在图 1 标题中列出。研究期间（2010 年为 720 Mt CO -eq）并在后期下降（2019 年为 -eq）。
Overtaking the United States and Japan, China is by far the world's largest importer of embodied emissions (585 Mt CO-eq in 2019). China's imports of embodied emissions are dominated by oil crops (46%) and pork (16%), and both import volumes have quadrupled mainly due to an increase in China's domestic demand for palm oil (+4.6 times), soybean oil ( +1.8 times) and soybean cake for pig feed ( +4.5 times). Beef makes up the largest component of embodied emissions imports of the United States (39% in 2019), Japan (42%), Russia (51%) and the Republic of Korea (43%), while oil crops (mainly palm oil and soy) account for a large share in imports of embodied emissions by India ( ) and the Netherlands (51%). Over this period, of consumption-based food emissions in developed countries were generated overseas. This ratio in developed countries with only a weak degree of self-sufficiency, such as Japan, the Republic of Korea and European countries, reached over . In contrast, developing countries generated of food-related emissions within national boundaries in 2000, although this ratio declined to in 2019. 中国远远超过美国和日本，是世界上最大的体现排放进口国（2019 年达到 585 Mt CO2-eq）。中国的体现排放进口以油料作物（46%）和猪肉（16%）为主导，这两种进口量主要因中国国内对棕榈油（增长 4.6 倍）、大豆油（增长 1.8 倍）和猪饲料用大豆饼（增长 4.5 倍）需求增加而增加了四倍。牛肉占美国（2019 年 39%）、日本（42%）、俄罗斯（51%）和韩国（43%）的体现排放进口的最大组成部分，而油料作物（主要是棕榈油和大豆）占印度和荷兰体现排放进口的很大比例。在这段时间内，发达国家基于消费的食品排放有 是在海外产生的。在只有较弱自给自足程度的发达国家，如日本、韩国和欧洲国家，这一比例达到 。相比之下，发展中国家在 2000 年内产生了 国内食品相关排放，尽管这一比例在 2019 年下降到 。
We observe that the patterns of emissions embodied in international trade of food have changed gradually and that developing countries, especially China, are playing an increasingly important role (Fig. 4). Between 2000 and 2019, the share of emissions embodied in international trade to total consumption-based food emissions increased from to . In of animal-based and of plant-based food emissions were embodied in trade. Over this period, imports of embodied emissions of developed countries kept constant , but its share in global trade declined from to . In 2000, the United States, Japan and Western European countries, which are the world's richest countries, dominated international trade with their imports contributing to nearly half of the total food-related emissions embodied in global trade. By 2019, this share has dropped to , while China has become the largest importer of embodied emissions (22%). For example, the largest embodied emissions flows to China, that is, imports from Brazil ( -eq) and Indonesia (69 Mt -eq), increased around fourfold, respectively, while flows from Brazil ( ) and Indonesia (-33%) to Western Europe, which were the largest in the beginning, decreased. However, emissions transfers within Europe have intensified, such as flows between Western European countries (+53%). Animal-based and plant-based emissions embodied in food exports to developing countries have increased by and 1.5 times. Increased food 我们观察到，国际贸易中食品排放的模式逐渐发生变化，发展中国家，尤其是中国，正在发挥越来越重要的作用（图 4）。2000 年至 2019 年，国际贸易中食品排放占总基于消费的食品排放的比例从 增加到 。 的动物性食品排放和 的植物性食品排放体现在贸易中。在这段时期，发达国家的排放体现进口保持不变 ，但其在全球贸易中的份额从 下降到 。2000 年，美国、日本和西欧国家，这些是世界上最富有的国家，主导了国际贸易，其进口贡献了全球贸易中近一半的总食品相关排放。到 2019 年，这一份额已经下降到 ，而中国已成为排放体现的最大进口国（22%）。 例如，最大的体现排放流向中国，即来自巴西（ -eq）和印度尼西亚（69 Mt -eq）的进口分别增加了大约四倍，而来自巴西（ ）和印度尼西亚（-33%）到西欧的流量，这在一开始是最大的，却减少了。然而，欧洲内部的排放转移加剧，例如西欧国家之间的流量增加了（+53%）。向发展中国家出口的基于动物和植物的排放在食品出口中增加了 和 1.5 倍。食物增加
Fig. 3 | GHG emissions embodied in domestic supply and international trade of food of major countries in 2000, 2010 and 2019. a, Ratio of domestic GHG emissions to total embodied emissions of food consumption by 18 major countries. Domestic GHG emissions refer to the emissions embodied in domestic food supply within a national territory including emissions from all food products, animal-based and plant-based food products (from left to right). b, GHG emissions embodied in food imports and exports of 18 major countries. The circles represent net imports or exports of emissions from food consumption. 图 3 | 2000 年、2010 年和 2019 年主要国家食品国内供应和国际贸易中所包含的温室气体排放量。a，18 个主要国家食品消费中国内温室气体排放量占总体排放量的比例。国内温室气体排放量指的是国内食品供应中所包含的排放量，包括所有食品产品、动物性和植物性食品产品的排放量（从左到右）。b，18 个主要国家食品进口和出口中所包含的温室气体排放量。圆圈代表食品消费中的净进口或出口排放量。
b
c
Plant-based GHG emissions 植物基温室气体排放
Fig. 4 | Patterns of emissions flows embodied in trade. a-c, Patterns of emissions flows embodied in international trade of all types of (a), animal-based (b) and plant-based (c) food product among and within 18 countries/regions in 2000 and 2019 (unit: -eq). Width of the lines represents the volumes of emissions embodied in trade from exporter to importer, and the colour is the same as the exporter. Flows cover more than of total emissions embodied in international bilateral trade annually as small flows are not shown here. Numbers in parentheses in each sub-plot represent the ratio of emissions embodied in trade to total consumption-based emissions. Abbreviations of 18 countries/ regions are shown in Fig. 1 caption. demand in developing countries creates a substantial increase in emissions outsourcing to major food-exporting countries, including Indonesia (+71%), Brazil (+65%), Australia (+34%), Canada (+42%) and the United States (+43%). 图 4 | 贸易中体现的排放流模式。a-c，2000 年和 2019 年 18 个国家/地区之间和内部所有类型（a）、动物性（b）和植物性（c）食品产品国际贸易中体现的排放流模式（单位： -eq）。线条的宽度代表从出口国到进口国贸易中体现的排放量，颜色与出口国相同。由于未显示小流量，因此流量覆盖了每年国际双边贸易中体现的总排放量的 以上。每个子图中括号中的数字表示贸易中体现的排放量与总基于消费的排放量的比率。18 个国家/地区的缩写在图 1 标题中显示。发展中国家的需求大幅增加了向主要食品出口国外包排放的情况，包括印度尼西亚（+71%）、巴西（+65%）、澳大利亚（+34%）、加拿大（+42%）和美国（+43%）。
Drivers of emissions of the global food system 全球食品系统排放的驱动因素
We apply structural decomposition analysis (SDA) to investigate the contributions of different driving factors across the entire food supply chains to the variations of food-consumption emissions globally 我们应用结构分解分析（SDA）来研究不同驱动因素在全球食品供应链中对食品消费排放变化的贡献
and in different regions and countries (Fig. 5 and Supplementary Table 7). Population growth was an important contributor to emissions rise in most countries/regions (except Japan and Russia), which increased global total emissions by during the study period. The greatest emissions increase driven by population was in South Asia , sub-Saharan Africa (SSA) (+64%), Near East and North Africa (NENA) (+59%) and India (+42%). Above countries/regions have a high population growth rate (over 30%) (Supplementary Table 8), with SSA being the highest (71%). The rising per capita consumption level was another important driver of the global emissions increase (+19%) over the period. Per capita consumption drove up food emissions in almost all developing countries, ranging from a modest in LAC to in China. Except for Indonesia and SSA (over are plant based), where farmland expansion leads to extensive land-use change, over of per capita consumption-related emissions increases in developing countries are generated by growing demand for animal-based food, such as China (+60%), India (+87%), NENA (+77%) and LAC (nearly ) (Supplementary Table 7). However, declining demand for animal-based food led to the decline of embodied emissions in Australia (-38%), Japan , the United States ( ) and Canada ( ). These countries' per capita consumption of red meat, such as beef and , respectively), has declined over this period (Supplementary Fig. 6). 在不同的地区和国家（图 5 和附表 7）中。 人口增长是大多数国家/地区（除日本和俄罗斯外）排放增加的重要因素，这在研究期间使全球总排放量增加了 。 由人口推动的最大排放增长发生在南亚 ，撒哈拉以南非洲（SSA）（+64％），中东和北非（NENA）（+59％）以及印度（+42％）。 上述国家/地区人口增长率很高（超过 30％）（附表 8），其中 SSA 最高（71％）。 人均消费水平的上升也是全球排放增加的另一个重要驱动因素（+19％）在此期间。 人均消费推高了几乎所有发展中国家的食品排放，从拉美和加勒比地区的适度 到中国的 。 除了印度尼西亚和 SSA（超过 是植物为基础）外，农田扩张导致广泛的土地利用变化，发展中国家人均消费相关排放增加的 是由对动物性食品需求的增长产生的，例如中国（+60％），印度（+87％），NENA（+77％）和拉美和加勒比地区（几乎 ）（附表 7）。 然而，对基于动物的食品需求的下降导致了澳大利亚（-38%）、日本、美国和加拿大的体现排放量下降。这些国家的人均红肉消费在这一时期也有所下降。
Despite the upward trend of global food emissions by other drivers, emissions intensity, measured by the amount of emissions per unit of weight of food product, was the dominant factor offsetting parts of emissions growth, decreasing global emissions by , avoiding additional -eq emissions globally. Emissions intensity includes three components, that is, the intensity of LULUC, agricultural production and beyond-farm activities. The effect of substantially declining emissions intensity from LULUC activities was responsible for over -eq global emissions decline ( ) with other factors held constant and had a prominent effect on emissions in countries with extensive land-use activities, such as Brazil ( ), SSA (mainly South and Central African regions) (-57%) and Indonesia (Supplementary Table 7). However, the driving effects of emissions intensity related to agricultural production and beyond-farm processes slightly increased the world's emissions by -eq and -eq (+0.5%), respectively. Our decomposition results show that a sharp drop in Brazil's emissions (by -eq) during the period from 2010 to 2015 is attributed to the contribution of decreasing LULUC emissions intensity. The root cause of the decrease in LULUC emissions intensity is shrinking LULUC activities (largely deforestation) and associated emissions. After a series of measures , such as the Forest Code and Amazon Soy Moratorium , for legally limiting deforestation activities in the Amazon, Brazil's deforestation rate reached a historically low level in 2010-2015, with a reduction of 50-80% compared with that in 2004 (ref.37), but this trend has changed considerably under the subsequent political leadership . 尽管全球食品排放呈上升趋势，但以食品产品单位重量排放量衡量的排放强度是抵消部分排放增长的主要因素，全球排放减少了 ，避免了全球额外 -eq 排放。排放强度包括三个组成部分，即土地利用变化排放强度、农业生产和农场以外活动的强度。土地利用变化排放强度大幅下降的效果导致全球排放减少了 -eq（ ），在其他因素保持不变的情况下对拥有广泛土地利用活动的国家的排放产生了显著影响，如巴西（ ）、撒哈拉以南非洲地区（主要是南部和中部非洲地区）（-57%）和印度尼西亚 （附表 7）。然而，与农业生产和农场以外过程相关的排放强度的推动效应略微增加了全球排放 -eq 和 -eq（+0.5%）。 我们的分解结果显示，2010 年至 2015 年期间巴西排放量急剧下降（约 9-eq），这归因于 LULUC 排放强度的降低。LULUC 排放强度下降的根本原因是 LULUC 活动减少（主要是森林砍伐）及相关排放。在一系列措施（如《森林法典》和《亚马逊大豆禁运令》）的影响下，巴西的森林砍伐率在 2010 年至 2015 年达到历史最低水平，较 2004 年减少了 50-80%（参考文献 37），但这一趋势在随后的政治领导下发生了显著变化。
Over this period, changes in the trade structure increased global emissions by (1.1 -eq) through increasing exported products from regions and countries with emissions-intensive production, while a decline in food consumption from domestic supply in importing regions and countries reduced global emissions by -eq). In 2000-2015, food importers became increasingly dependent on exports of emissions-intensive products from agricultural suppliers including Brazil and Indonesia. As a result, international food trade accelerated global emissions. However, international trade tends to reduce emissions of global food consumption after 2015 with the improvement of production productivity in exporting countries. 在这段时间内，贸易结构的变化通过增加来自排放密集型生产地区和国家的出口产品，使全球排放量增加了 （1.1 -eq)，而进口地区和国家从国内供应减少食品消费，减少了全球排放 -eq）。在 2000-2015 年间，食品进口国越来越依赖于包括巴西和印度尼西亚在内的农业供应国的排放密集型产品出口。因此，国际食品贸易加速了全球排放。然而，随着出口国生产生产力的提高，国际贸易在 2015 年后倾向于减少全球食品消费的排放。
Discussion and conclusions 讨论和结论
Our study attributes production-based emissions to final consumers at a product level using physical trade flows that provide complementary information to PBE, thus allowing us to investigate emissions and target mitigation efforts across the whole food supply chain. 我们的研究利用物理贸易流量将基于生产的排放 归因于产品级最终消费者，为 PBE 提供了补充信息，从而使我们能够调查整个食品供应链上的排放并针对减排工作。
Results show considerable differences regarding emissions patterns and effects of drivers between regions and countries, and we could classify them into four groups according to these differences: (1) countries with high per capita food emissions levels and dominant livestock emissions (mainly from red meat) (North America, Australia, LAC); (2) developed countries that rely heavily on imports and outsource substantial amounts of food-related emissions (Japan and Europe); (3) rapidly developing countries with substantial emissions increase driven by rapid population growth or improved living standards (China, South Asia, NENA); and (4) countries with emissions-intensive production, mainly with extensive land-use change activities (Brazil, Indonesia and South and Central African regions). Discussions on comparison with other studies of global food emissions are provided in the Supplementary Discussion. 结果显示，不同地区和国家之间的排放模式和驱动因素效应存在显著差异，我们可以根据这些差异将它们分类为四组：（1）人均食品排放水平高且以畜牧业排放为主（主要来自红肉）的国家（北美、澳大利亚、拉美和加勒比地区）；（2）依赖进口并外包大量与食品相关排放的发达国家（日本和欧洲）；（3）由快速人口增长或改善生活水平驱动的大幅度排放增加的快速发展中国家（中国、南亚、中东北非地区）；以及（4）生产排放密集型的国家，主要进行广泛的土地利用变化活动（巴西、印度尼西亚以及南非和中非地区）。在补充讨论中提供了与其他全球食品排放研究的比较讨论。
Our results show considerable differences in food consumption and associated emissions across countries. Residents in the first group of countries have a livestock-dominated (especially beef) diet and larger associated emissions compared with other groups, while the third group is generating increasing consumption of beef and dairy due to the demand for improving living standards and diet diversity. As for the same protein content, red meat, especially beef, generates more emissions than poultry, fish and plant-based protein products . Thus, the growth of the global population and rising per capita demand for emissions-intensive food are likely to boost emissions further. Diet shifts, including reducing excessive intake of red meat or improving shares of plant-based protein, will not only reduce emissions but avoid health risks such as obesity and cardiovascular disease . However, widespread and lasting diet shifts (for example, the EAT-Lancet planetary health diet ) are very difficult to achieve within a narrow time frame. Therefore, incentives that encourage consumers to reduce red meat or buy products with higher environmental dividends through eco-labelling, adding taxes or subsidies reflecting some of the environmental costs in product prices and education on actual food emissions could help to reduce food emissions . 我们的研究结果显示，不同国家的食物消费和相关排放存在显著差异。第一组国家的居民以畜产品为主（尤其是牛肉）的饮食和相关排放量较其他组更大，而第三组由于对改善生活水平和饮食多样性的需求，牛肉和乳制品的消费正在增加。对于相同的蛋白质含量，红肉，尤其是牛肉，产生的排放比家禽、鱼类和植物蛋白产品更多。因此，全球人口增长和人均排放密集型食品需求的上升可能会进一步推动排放。饮食转变，包括减少过量摄入红肉或提高植物蛋白的份额，不仅可以减少排放，还可以避免肥胖和心血管疾病等健康风险。然而，广泛和持久的饮食转变（例如，EAT-Lancet 行星健康饮食）在狭窄的时间框架内很难实现。 因此，通过生态标识鼓励消费者减少红肉消费或购买具有更高环境回报的产品的激励措施，通过在产品价格中反映部分环境成本的税收或补贴，以及对实际食品排放进行教育，可以帮助减少食品排放。
International food trade policies incorporating environmental externalities that are less covered in production-side policies are urgently needed to avoid possible emissions leakage and realize emissions reduction across supply chains. Emissions outsourced through international food trade increased by -eq over the study period, accelerating the global emissions increase and unequal distribution. Countries in the second and third groups have considerably lower than CBE by outsourcing their domestic food emissions through imports from agricultural suppliers such as Brazil, Indonesia and Oceania. Emissions embodied in these food imports vary considerably depending on the originating country, while the world's main food suppliers are not regions with the highest efficiency. For example, the total emissions intensity of production per kilogramme of beef in Western European countries (range from -eq) is far less than in Brazil (44-46 kg CO -eq) (Supplementary Fig. 7), but the latter is the largest beef exporter for European countries . Countries with high efficiency for domestic production import emissions-intensive products from regions with a large scale of LULUC activities or low agricultural efficiency will tend to increase emissions of the global food system. Although the magnitude of food emissions embodied in global trade is considerable, proposals for measures to avoid carbon leakage such as the European Union's proposed Carbon Border Adjustment Mechanism have rarely been extended to include agricultural or food-related emissions. Key emissions-intensive products that dominate international food trade (for example, beef from Australia, beef and soybean from Brazil, palm oil from Indonesia) could be targets of such taxation policies. Our data and model with information at the product level can help quantify the size of the necessary adjustment. 国际食品贸易政策迫切需要考虑环境外部性，这些外部性在生产端政策中覆盖较少，以避免可能的排放泄漏，并实现整个供应链的减排。通过国际食品贸易外包的排放在研究期间增加了 -eq，加速了全球排放增加和不均等分布。第二组和第三组国家通过从巴西、印度尼西亚和大洋洲等农业供应国进口，将国内食品排放外包，其排放量明显低于 CBE。这些食品进口中所包含的排放量因原产国而异，而世界主要食品供应国并非效率最高的地区。例如，西欧国家每公斤牛肉的生产总排放强度（范围从 -eq）远远低于巴西（44-46 千克 CO -eq）（附图 7），但后者是欧洲国家最大的牛肉出口国 。 高效率国家从具有大规模土地利用变化活动或低农业效率的地区进口排放密集型产品，往往会增加全球食品系统的排放。尽管全球贸易中食品排放的规模相当可观，但很少有提议采取措施以避免碳泄漏，例如欧盟提出的碳边境调整机制 ，很少扩展到包括农业或与食品相关的排放。主导国际食品贸易的排放密集型产品（例如来自澳大利亚的牛肉、来自巴西的牛肉和大豆、来自印度尼西亚的棕榈油）可能成为此类征税政策的目标。我们的数据和模型，结合产品级别的信息，可以帮助量化必要调整的规模。
A series of trade policies are accelerating emissions through increasing food imports from countries/regions with emissionsintensive production. For instance, the European Union's Green 一系列贸易政策通过增加从排放密集型生产国/地区进口食品来加速排放。例如，欧盟的绿色
o
Fig. 5 | Contributions of five driving factors to changes in GHG emissions from food consumption. a-s, Contributions of five driving factors to changes in GHG emissions from food consumption of the global (a) and 18 countries/ regions (b-s) between 2000 and 2019. The grey bars indicate total emissions. 图 5 | 五个驱动因素对食品消费温室气体排放变化的贡献。a-s，五个驱动因素对全球食品消费温室气体排放变化（a）和 2000 年至 2019 年间 18 个国家/地区（b-s）的贡献。灰色条表示总排放量。
The coloured bars represent the absolute contribution (positive or negative) of different driving factors to the changes in global and national/regional emissions in every period. 彩色条形图代表不同驱动因素对每个时期全球和国家/地区排放变化的绝对贡献（正面或负面）。
Deal encourages less intensive agriculture in Europe and increasing imports of agricultural products from countries such as Brazil, the United States, Indonesia and Malaysia . Another example that leads to an emissions increase through trade is the US-China trade war, which led China to import more soybean from Mercosur countries to reduce its dependence on the United States . Above imports from major suppliers induced by demand led to a surge in deforestation and associated emissions. However, trade between diverse international partners provides opportunities to ameliorate emissions by allowing consumers to choose products from places with less emissions-intensive production. Long-term commitments are needed to comprehensively assess emissions embodied in the entire supply chain for trade-offs between domestic production and imports from multiple origins, thereby minimizing global impacts. 交易鼓励欧洲进行更少密集的农业，并增加从巴西、美国、印度尼西亚和马来西亚等国家进口农产品 。通过贸易导致排放增加的另一个例子是美中贸易战，导致中国从南方共同市场国家进口更多大豆以减少对美国的依赖 。由需求引发的主要供应商的进口导致了森林砍伐和相关排放的激增。然而，与多样化的国际合作伙伴之间的贸易为消费者提供了选择来自生产排放较少的地方的产品的机会。需要长期承诺来全面评估整个供应链中体现的排放，以在国内生产和多个来源的进口之间进行权衡，从而最大程度地减少全球影响。
Furthermore, our study traces the origins and emissions intensities of specific products, which ultimately flow to final consumers. Results show that reducing PBE through agricultural intensification with technology improvement or lower levels of resource inputs (reflected in lower emissions intensity), especially for agricultural producers from the fourth group with abundant natural resources (for example, forests, peatland) , which generated vast amounts of emissions from widespread LULUC activities such as deforestation, is vital for mitigating climate effects across food supply chains. Changes in consumer behaviour or trade policies (for example, proposed legislation to eliminate deforestation by European countries ) in the second and third group of countries can trigger deeper impacts via food supply chains and eventually improve production-side efficiency for the fourth group . Altered levels and composition of food consumption (with less emissions-intensive products) could reduce land-use change, relocate production to places with fewer emissions or incentivize food suppliers to decrease emissions intensity and avoid destructive environmental impacts (for example, through the Amazon Soy Moratorium ). However, we find that the fourth group of countries themselves have substantial consumption-based emissions due to the domestic demand for emissions-intensive products (for example, oil crops). Raising awareness and legislation nationally to reduce emissions from food production is needed across these countries; otherwise, the domestic leakage may offset part of the emission reduction brought by supply chain measures . 此外，我们的研究追踪特定产品的起源和排放强度，这些产品最终流向最终消费者。结果显示，通过农业集约化与技术改进或更低水平的资源投入（反映在更低的排放强度中）来减少 PBE，尤其是对于拥有丰富自然资源（例如森林、泥炭地）的第四组农业生产者，这些生产者通过广泛的土地利用变化活动（如森林砍伐）产生了大量排放，对于减缓食品供应链中的气候影响至关重要。消费者行为或贸易政策的变化（例如，欧洲国家提出的消除森林砍伐的立法）可能会在第二和第三组国家引发更深层次的影响，最终提高第四组的生产效率。 改变食品消费水平和构成（使用排放较少的产品）可以减少土地利用变化，将生产转移到排放较少的地方，或者激励食品供应商降低排放强度并避免破坏性环境影响（例如，通过亚马逊大豆禁运 ）。然而，我们发现第四组国家本身由于对排放密集型产品（例如油料作物）的国内需求而产生了相当大量基于消费的排放。需要在这些国家全国范围内提高意识和立法，以减少食品生产中的排放；否则，国内泄漏可能抵消供应链措施带来的部分减排效果 。
Methods 方法
Food-consumption accounting 食品消费会计
We apply the physical trade flow (PTF) approach proposed by Kastner et al. to calculate the consumption of 153 food products (both primary and processed products) (Supplementary Tables 9 and 10) based on the physical trade between 181 countries or areas in five given years (2000/2005/2010/2015/2019) (Supplementary Methods 1.1). We use the criteria proposed by the United Nations to define developed and developing countries (Supplementary Table3). Countries or areas are classified into 18 countries/regions for comparison according to geographical location (Supplementary Fig. 8 and Supplementary Table 4). The PTF approach by Kastner et al. allows for tracing product flows through international supply chains and final consumers to which products ultimately flow based on domestic production and bilateral trade between countries. We use data from the detailed trade matrix of products of FAOSTAT to construct the matrix showing the physical flows between counties. All data are in units of mass (metric tonnes). Detailed data sources used for this study are shown in Supplementary Methods 1.2 and Supplementary Table 11. We mainly use reported import data by assuming that imports are more reliable due to strict customs records . The PTF approach assumes that domestic production and imported products are proportionally distributed between domestic supply and exports. Because of the limited shelf life of food and the relatively small share of agricultural commodities used for food stocks, this study does not include variations in stocks. 我们应用了 Kastner 等人提出的物理贸易流（PTF）方法 来计算 153 种食品产品（原始和加工产品）的消费量（附表 9 和 10），基于 181 个国家或地区在五个给定年份（2000/2005/2010/2015/2019）之间的物理贸易（附表 1.1）。我们使用联合国提出的标准 来定义发达国家和发展中国家（附表 3）。根据地理位置，国家或地区被分类为 18 个国家/地区以进行比较（附图 8 和附表 4）。Kastner 等人的 PTF 方法允许通过国际供应链和最终消费者追踪产品流向，这些产品最终基于国内生产和国家之间的双边贸易流向。我们使用 FAOSTAT 的产品详细贸易矩阵数据 来构建显示国家之间物理流动的矩阵。所有数据以质量单位（公吨）表示。本研究使用的详细数据来源显示在附表 1.2 和附表 11 中。 我们主要使用报告的进口数据，假设进口更可靠，因为严格的海关记录。PTF 方法假定国内生产和进口产品在国内供应和出口之间按比例分配。由于食品的保质期有限，农产品用于食品库存的份额相对较小，本研究不包括库存变化。
The PTF approach by Kastner et al. is suitable for linking consumption and associated environmental impacts to crop cultivation or livestock raising (on-farm stages) at a product level . To investigate the GHG emissions of processed products generated during on-farm processes, we transform the bilateral trade matrix of processed products using the ratio of sources for primary products, which is developed based on the proportion of domestic production and imports of primary products (Supplementary Methods 2.2). We use conversion factors for agricultural commodities from the Food and Agriculture Organization of the United Nations (FAO) to convert the processed products into primary products, and some missing factors are supplemented by using the factors from the Global Trade Analysis Project (GTAP) Data Base with Nutritional Accounts (Supplementary Methods 2.2 and Supplementary Table 9). Therefore, we can obtain the new production and bilateral trade matrix of the processed products in the form of primary equivalents, which trace the sources of raw materials for processed product production and the destination where these processed products are finally consumed (Supplementary Methods 2.3). Here we simplify the calculation by ignoring the difference between inputs during the production of processed products and assuming all primary products used as raw materials are consumed in one place. Furthermore, agricultural products for non-food use are excluded by using data of non-food-use commodities from the food balance sheet of FAOSTAT (Supplementary Methods 2.5). Kastner 等人提出的 PTF 方法适用于将消费与相关环境影响与作物种植或畜牧业（农场阶段）联系起来 在产品水平 。为了调查在农场过程中产生的加工产品的温室气体排放量，我们使用基于国内生产和进口原始产品比例开发的原始产品来源比率转换加工产品的双边贸易矩阵（附录方法 2.2）。我们使用联合国粮食及农业组织（FAO）的农产品转换因子将加工产品转换为原始产品，一些缺失的因子则通过使用全球贸易分析项目（GTAP）数据库与营养账户（附录方法 2.2 和附录表 9）中的因子进行补充。 因此，我们可以以初级等价物的形式获得加工产品的新生产和双边贸易矩阵，追踪加工产品生产的原材料来源以及这些加工产品最终被消费的目的地（补充方法 2.3）。在这里，我们简化计算，忽略了加工产品生产过程中输入之间的差异，并假设所有用作原材料的初级产品在一个地方被消费。此外，通过使用 FAOSTAT 的食品平衡表中的非食用商品数据，排除了用于非食品用途的农产品 （补充方法 2.5）。
Quantification of consumption-based food emissions 基于消费的食品排放量的量化
By combining the emissions intensity (the amount of emissions per unit weight of food product) and the consumption matrix (Supplementary Fig. 9 provides the accounting framework), the consumption-based emissions of each product are calculated as follows : 通过结合排放强度（每单位食品产品重量的排放量）和消费矩阵（附图 9 提供了会计框架），计算每种产品的基于消费的排放量如下 ：
where refers to the consumption-based GHG emission of product . represents the vector of emissions intensity of product from food supply chain process , of which is total emissions generated from supply chain process of product and is the production vector of product is the vector of share of in , of which (domestic material consumption) is the amount of product consumed domestically and (domestic material input) represents total inputs of product in one country; equals plus exports of product (or production plus imports). denotes the trade structure of product , of which is the matrix of export shares in and is the identity matrix with the same dimension as matrix (Supplementary Methods 2.1). 其中 指的是产品 的基于消费的温室气体排放量。 代表产品 从食品供应链过程 的排放强度向量，其中 是产品 供应链过程 产生的总排放量， 是产品 的生产向量是 在 中的份额向量，其中 （国内物质消耗）是国内消费的产品 的数量， （国内物质投入）代表一个国家产品 的总投入； 等于 加上产品 的出口（或生产加进口）。 表示产品 的贸易结构，其中 是 中出口份额的矩阵， 是与矩阵 （补充方法 2.1）具有相同维度的单位矩阵。
To obtain the emissions intensity along supply chain processes, we distribute the annual GHG emissions (including and ) from LULUC, agricultural and beyond-farm activities to plant- and animal-based products using the similar approach performed by Hong et al. and are converted into equivalents using the 100 -year global warming potential values of 28 and 265 from IPCC AR5 (ref. 56). National emissions data are obtained from the FAOSTAT Climate Change dataset (Supplementary Table 11), which provides data of country- and process-specific emissions from the food system based on activity data and IPCC Tier 1 Methodology. Results of consumption-based emissions of and are shown in Supplementary Fig. 10 and Supplementary Dataset 2. Detailed GHG categories and emissions processes are shown in Supplementary Table 12. 为了获得供应链过程中的排放强度，我们将来自土地利用变化、农业和农场以外活动的年度温室气体排放（包括 和 ）分配到基于植物和动物的产品中，使用了洪等人执行的类似方法。 和 被转换为 当量，使用了 IPCC AR5（参考文献 56）中的 100 年全球变暖潜势值 28 和 265。国家排放数据来自 FAOSTAT 气候变化数据集 （附表 11），该数据集提供了基于活动数据和 IPCC 第一层方法的食品系统中特定国家和过程的排放数据。 和 的基于消费的排放结果显示在附表 10 和附录数据集 2 中。详细的温室气体类别和排放过程显示在附表 12 中。
Allocation of LULUC emissions to food products. A top-down approach is applied to allocate production-based LULUC emissions due to the expansion of cropland or pasture to primary products. LULUC emissions include: (1) and from burning (of forests, savannah, humid tropical forests and organic soils), (2) from net 将陆地利用变化引起的生产基础 LULUC 排放分配给食品产品。采用自上而下的方法来分配由于耕地或牧场扩张而产生的生产基础 LULUC 排放到初级产品。LULUC 排放包括：（1）来自燃烧（森林、稀树草原、湿热带森林和有机土壤）的 和 ，（2）来自净
forest conversion and (3) and from the drainage of organic soils. We assume that LULUC emissions are directly related to land-use areas for the production of primary products and distribute the annual LULUC emissions to products according to harvested cropland areas or pasture areas for feeding livestock in a given year. LULUC emissions intensities are calculated using the production and LULUC emissions of primary products (Supplementary Methods 3.1). All data of emissions amounts , land-use areas and production quantity are obtained from FAOSTAT . Legacy emissions cumulated in land due to LULUC activities over time or absorbed emissions by land due to agriculture abandonment are not incorporated. On the basis of the LULUC emissions intensities of each product, we assign LULUC emissions to final consumers using the PTF approach as equation (1). Results of consumption-based LULUC emissions in 181 countries are shown in Supplementary Fig. 11 and Supplementary Dataset 3. 森林转化和（3） 和 来自有机土壤排水。我们假设土地利用变化排放与用于生产初级产品的土地面积直接相关 ，并根据每年收获的耕地面积或放牧地面积来将年度土地利用变化排放分配给产品。土地利用变化排放强度是使用初级产品的生产和土地利用变化排放进行计算的（附录方法 3.1）。所有排放量 、土地利用面积 和生产数量 的数据均来自 FAOSTAT 。由于土地利用变化活动而在土地上积累的传统排放或由于农业放弃而被土地吸收的排放未纳入考虑。根据每种产品的土地利用变化排放强度，我们使用 PTF 方法将土地利用变化排放分配给最终消费者，如方程（1）所示。181 个国家的基于消费的土地利用变化排放结果显示在附图 11 和附录数据集 3 中。
Allocation of agricultural emissions to food products. Emissions from agricultural production for crops are: (1) from crop residues, (2) and from burning crop residues, (3) from synthetic fertilizer, (4) from the use of synthetic fertilizer, (5) from manure applied to soils, (6) from rice cultivation and (7) , and from energy use for crop cultivation . We allocate production-based agricultural emissions to crops and calculate agricultural emissions intensities based on the production of crops from FAOSTAT (Supplementary Methods 3.2). Emissions from crop residues are allocated by nitrogen contents and production of specific crops, while emissions from burning residues are distributed by the amounts of burned biomass of crops. from synthetic fertilizers is allocated to primary crops according to their fertilizer input rate and harvested areas from FAOSTAT . Emissions from manure applied to soils and rice cultivation are distributed by harvested areas of crops and rice production quantity , respectively. In addition, we use the impact coefficient of food products (emissions per unit weight of the product) (Supplementary Table 13) to assign emissions of energy use to products. 将农业排放分配给食品产品。作物农业生产排放为：（1） 来自作物残留物，（2） 和 来自燃烧作物残留物，（3） 来自合成肥料，（4） 来自合成肥料使用，（5） 来自施加于土壤的粪肥，（6） 来自稻田耕作，（7） ， 和 来自用于作物种植的能源使用 。我们将基于生产的农业排放 分配给作物，并根据 FAOSTAT （附录方法 3.2）的作物生产计算农业排放强度。作物残留物的排放 根据氮含量和特定作物的生产分配，而燃烧残留物的排放 则根据作物燃烧生物量的数量分配。合成肥料的排放分配给主要作物，根据它们的肥料投入率 和 FAOSTAT 的收获面积。施加于土壤的粪肥和稻田耕作的排放 分别根据作物和稻米生产数量的收获面积分配。 此外，我们使用食品产品的影响系数（每单位产品重量的排放量） （附表 13）来将能源使用的排放分配给产品。
Emissions from the agricultural production of livestock (meat, dairy and eggs) are generated in five main processes: enteric fermentation, manure management, feed production, manure left on pasture and energy consumption. Country- and animal-specific emissions from enteric fermentation of ruminant animals and manure management based on Tier 1 level are obtained from FAOSTAT and then allocated to livestock products using FAOSTAT statistics of production . FAOSTAT provides data on emissions generated in manure left on pasture as well. Emissions of manure left on pasture are allocated into livestock products according to the pasture areas needed for feeding different animals, and then the emissions intensity is calculated based on production amounts of livestock products. 畜牧业（肉类、奶制品和鸡蛋）的农业生产排放主要由五个过程产生：反刍发酵、粪便管理、饲料生产、牧场上的粪便和能源消耗。来自反刍动物的反刍发酵和粪便管理的国家和动物特定排放根据 FAOSTAT 的 Tier 1 级别获得，然后根据 FAOSTAT 生产统计数据分配给畜产品。FAOSTAT 还提供了牧场上的粪便产生的排放数据。根据喂养不同动物所需的牧场面积，将留在牧场上的粪便排放分配到畜产品中，然后根据畜产品的生产量计算排放强度。
Emissions from feed crops are allocated to the livestock products that consume the feed during production. Emissions from feed crops, including barley, maize, wheat, rapeseed cake and soybean cake for livestock production are allocated to livestock according to the feed conversion ratios specific to each product at the national level . Feed conversion ratios are calculated based on the national feed use quantities and weight factors of each livestock product (Supplementary Methods 2.4). Then we calculate feed emissions per unit weight of animal-based products using the same approaches as crops. Moreover, we use data on production and emissions generated from the energy use of freshwater and marine products to calculate the emissions intensity from fishery production. 饲料作物排放量分配给在生产过程中消耗饲料的畜产品。饲料作物的排放量，包括大麦、玉米、小麦、菜籽饼和大豆饼，用于畜产品生产，根据每种产品在国家水平上的饲料转化比例分配给畜产品。饲料转化比例是根据国家饲料使用量和每种畜产品的重量因子计算的（附录方法 2.4）。然后，我们使用与作物相同的方法计算每单位动物产品重量的饲料排放量。此外，我们使用关于淡水和海产品能源使用产生的生产和排放数据来计算渔业生产的排放强度。
On the basis of the emissions intensity of crops and livestock during agricultural production, we assign agricultural emissions to final consumers of 153 food products using the PTF approach as equation (1). Results of consumption-based agricultural emissions in 181 countries are shown in Supplementary Fig. 11 and Supplementary Dataset 3. 根据农业生产中作物和牲畜的排放强度，我们使用 PTF 方法将农业排放分配给 153 种食品产品的最终消费者，如方程（1）所示。 181 个国家的基于消费的农业排放结果显示在附图 11 和附录数据集 3 中。
Allocation of beyond-farm emissions to food products. Bottom-up aggregation and top-down allocation approaches are combined to distribute beyond-farm emissions to products. Emissions from beyond-farm processes include: and from (1) processing, (2) packaging, (3) retail, (4) transport, (5) and from fertilizer manufacturing and (6) and from industrial wastewater treatment related to food. The statistical data of total national emissions in the above six processes are obtained from FAOSTAT . National emissions from food processing, packaging, retail and industrial wastewater treatment are downscaled to the product level by using the impact coefficient of 153 products (Supplementary Methods 3.3). Because the food-transport emissions are closely related to the transport distance and freight volume, we use the monetary values between transport and food-related sectors from the GTAP database to distinguish emissions from domestic and international transport. Therefore, emissions intensities of specific products at different distances (within or between countries) can be calculated using the impact coefficient for food transport. In addition, emissions of fertilizer manufacturing are allocated according to the same approach of distributing synthetic fertilizer-related emissions in agricultural production. Beyond-farm emissions are attributed to final consumers using the PTF approach shown in equation (1). Results of consumption-based beyond-farm emissions in 181 countries are shown in Supplementary Fig. 11 and Supplementary Dataset 3. 将农场以外的排放分配给食品产品。自下而上的聚合和自上而下的分配方法结合起来，将农场以外的排放分配给产品。来自农场以外过程的排放包括： 和 来自（1）加工、（2）包装、（3）零售、（4）运输、（5） 和 来自化肥制造以及（6） 和 来自与食品相关的工业废水处理。上述六个过程的全国总排放统计数据来自 FAOSTAT 。食品加工、包装、零售和工业废水处理的全国排放 通过使用 153 种产品的影响系数 （附录方法 3.3）下调到产品水平。由于食品运输排放与运输距离和货运量密切相关，我们使用 GTAP 数据库 中运输和与食品相关部门之间的货币价值来区分国内和国际运输的排放。因此，可以使用食品运输的影响系数计算不同距离（国内或国际之间）的特定产品的排放强度。 此外，肥料制造的排放按照在农业生产中分配合成肥料相关排放的相同方法进行分配。超出农场的排放归因于使用方程式（1）中显示的 PTF 方法的最终消费者。181 个国家的基于消费的超出农场排放结果显示在附图 11 和附录数据集 3 中。
Identification of driving factors 驱动因素的识别
To understand the driving forces behind emissions of food consumption, we employ Structural Decomposition Analysis (SDA), a widely adopted approach in energy and emission studies , to decompose the global and regional emissions of 153 products as: 为了了解食品消费排放背后的驱动力，我们采用结构分解分析（SDA），这是能源和排放研究中广泛采用的方法，来分解 153 种产品的全球和区域排放
where refers to the consumption-related emissions of 153 products. The equation includes five factors: emissions intensity of product in process ; trade structure of product defined in equation (1); domestic supply ratio of product , indicating the ratio of locally produced food to total food inputs; per capita consumption of product ; and population . The difference between two time periods can be expressed as: 其中 指的是 153 种产品的与消费相关的排放。该方程包括五个因素：产品 在过程 中的排放强度；方程（1）中定义的产品 的贸易结构；产品 的国内供应比例，表示本地生产食品占总食品投入的比例；人均消费量 ；和人口 。两个时间段之间的差异可以表示为：
Thus the changes in consumption-based emissions during 20002005, 2005-2010, 2010-2015 and 2015-2019 can be decomposed by five factors as: 因此，2000-2005 年、2005-2010 年、2010-2015 年和 2015-2019 年的基于消费的排放变化可以通过五个因素进行分解：
where represents changes in a factor from base year (0) to target year . Each of five terms in equation (4) denotes the contributions to emissions changes that are triggered by one factor if other variables are kept constant. The five factors in the SDA model can result in first-order decompositions, and here we use the solution named the average of two polar decompositions to approximate the average of all possible decompositions. Equation (4) is finally converted as: 其中 表示从基准年（0）到目标年 的因素变化。方程（4）中的五个项分别表示由一个因素触发的排放变化对其他变量保持恒定的贡献。SDA 模型中的五个因素可以导致 一阶分解，这里我们使用名为两个极分解的平均解 来近似所有可能分解的平均值。方程（4）最终转换为：
where represents changes in consumption-based emissions along the supply chains of 153 products; captures the change of emissions intensity of product in supply chain process measures the change in international trade structure of product denotes the change in the ratio of locally produced product to total inputs of product identifies changes in per capita consumption of product ; and measures changes in population. 其中 代表 153 种产品供应链上基于消费的排放变化； 捕捉产品 在供应链过程中的排放强度变化 衡量产品 国际贸易结构的变化 表示本地生产产品 总输入比例的变化 标识产品 人均消费的变化； 衡量人口变化。
Uncertainty assessment 不确定性评估
Our results of global consumption-based emissions during different supply chain processes are generally consistent with global production-based food emissions inventories from FAOSTAT . Similar to the uncertainty analysis performed by Tubiello et al., and Hong et al. , we conduct a Monte Carlo approach (running 10,000 simulations) to assess the uncertainty range of consumption-based emissions by simulating activity data, emission factors and parameters for each process according to the default uncertainty ranges derived from the standard IPCC guidelines and individual uncertainty ranges from previous studies (Supplementary Table 14). Uncertainty ranges of confidence intervals of consumption-based food GHG emissions are adopted. Detailed uncertainty ranges of food emissions are provided in Supplementary Tables 1, 2, 5 and 6 and Supplementary Datasets 8 and 9 . We consider only the uncertainties generated in the production processes and do not include the uncertainties caused by trade because we cannot obtain the uncertainty ranges of original statistical data for reported imports of agricultural products . We recognize that the uncertainties of trade data in this study have an unknown magnitude. 我们在不同供应链过程中全球消费基础排放的结果通常与 FAOSTAT 的全球生产基础食品排放清单一致。与 Tubiello 等人以及 Hong 等人进行的不确定性分析类似，我们采用蒙特卡洛方法（运行 10,000 次模拟）来评估消费基础排放的不确定性范围，通过根据标准 IPCC 指南和先前研究的个别不确定性范围推导的默认不确定性范围模拟每个过程的活动数据、排放因子和参数。采用了消费基础食品温室气体排放的置信区间的不确定性范围。食品排放的详细不确定性范围在补充表 1、2、5 和 6 以及补充数据集 8 和 9 中提供。我们仅考虑生产过程中产生的不确定性，不包括由于贸易而引起的不确定性，因为我们无法获得农产品进口的原始统计数据的不确定性范围。 我们意识到这项研究中贸易数据的不确定性具有未知的幅度。
Limitations 限制
Our study has the following limitations: 我们的研究有以下限制：
First, the PFT approach by Kastner et al. allows us to quantify re-exports to other countries based on conversion matrices but ignores the connections with other sectors within an economy compared to the MRIO-based approach. We do not choose physical MRIO because the Food and Agriculture Biomass Input-Output model is outdated and the updated version is not yet available, and Exiobobase does not have as many countries and product detail as our database. The PTF approach we use is thus very suitable to capture relatively simple food supply chains but may ignore more complex processing and repacking steps in global supply chains and thus introduce some system boundary cut-off errors . A more feasible design in the next step requires integration with models such as MRIO to investigate the entire supply chain considering the heterogeneity of production inputs and connections between food-related and other sectors. 首先，Kastner 等人的 PFT 方法允许我们根据转换矩阵量化向其他国家的再出口，但与基于 MRIO 的方法相比，它忽略了与经济内其他部门的联系。我们没有选择物理 MRIO，因为食品和农业生物质投入产出模型已经过时，更新版本尚未推出，而 Exiobobase 没有我们数据库那么多的国家和产品细节。因此，我们使用的 PFT 方法非常适合捕捉相对简单的食品供应链，但可能忽略了全球供应链中更复杂的加工和重新包装步骤，从而引入一些系统边界截断错误。在下一步中，更可行的设计需要与 MRIO 等模型集成，以考虑生产输入的异质性以及食品相关部门与其他部门之间的联系。
Second, we do not consider heterogeneity within countries. Countries present sub-national differences in land use, agricultural and other activities and related emissions. However, data in terms of production, trade and emissions along the entire food supply chain at the sub-national level are available for a few products and are limited to a range of potential errors with inconsistent data sources. 其次，我们没有考虑国家内部的异质性。各国在土地利用、农业和其他活动以及相关排放方面存在亚国家差异。然而，在亚国家层面上，关于生产、贸易和排放的数据仅适用于少数产品，并且受限于潜在错误范围和数据来源不一致。
Therefore, we focus only on the emissions from consumption and trade at a country level. 因此，我们只关注各国在消费和贸易方面的排放。
Third, our study focuses on upstream emissions along food supply chains before the household level and excludes emissions from household consumption and end of life (that is, waste management) . Above emissions beyond supply chains are difficult to allocate to specific products given the limited data availability and are not part of the international trade flows. Nonetheless, given the large magnitude of these emissions, especially methane emissions from the decay of solid food waste in landfills and open dumps , future studies that explore the mitigation of food emissions from consumers will incorporate such emissions as an extension of findings. 第三，我们的研究重点放在家庭层面之前的食品供应链上游排放，并排除家庭消费和生命周期结束（即，废物管理） 。供应链之外的排放难以分配给特定产品，因为数据可用性有限，并不属于国际贸易流动的一部分。尽管如此，鉴于这些排放的规模巨大，尤其是来自垃圾填埋场和露天垃圾场固体食物废物腐烂的甲烷排放 ，未来研究将探讨消费者减少食品排放的方法，并将这些排放作为研究结果的延伸。
Finally, the data available for this study have some limitations. Data of production for some processed products have the problem of item aggregation in 2000 and 2005, and we separate these products based on their shares in 2010. Meanwhile, because of the lack of a standard distribution approach and harmonized food emissions coefficients at a product level, emissions from different processes are attributed to specific products according to different approaches applied by previous studies that may lead to biased results. Moreover, this study does not account for the legacy emissions or carbon removals from land that are difficult to allocate to specific years or products. With the improvement of data availability (for example, the use of dynamic land-use models), a more consistent and complete accounting framework of the food system in the future will cover these emissions with breakdown into detailed products at global, national and sub-national levels. 最后，本研究可用的数据存在一些限制。一些加工产品的生产数据在 2000 年和 2005 年存在项目聚合问题，我们根据它们在 2010 年的份额将这些产品分开。同时，由于缺乏标准的分配方法和产品级别的协调食品排放系数，不同过程的排放根据先前研究采用的不同方法归因于特定产品，可能导致偏倚的结果。此外，本研究未考虑难以分配到特定年份或产品的土地遗留排放或碳去除。随着数据可用性的提高（例如，使用动态土地利用模型），未来食品系统的更一致和完整的会计框架将覆盖这些排放，并将其细分为全球、国家和次国家级别的详细产品。
Reporting summary 报告摘要
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. 有关研究设计的更多信息可在与本文相关联的《自然出版物组合报告摘要》中找到。
Data availability 数据可用性
The LULUC, agricultural and beyond-farm emissions data are curated by the FAO and freely available from FAOSTAT . Population data used in this study are obtained from World Population Prospects of the United Nations . Data of monetary values for transport and food-related sectors are obtained from the GTAP database . Supplementary methods, discussion, figures, tables and datasets used in the analysis can be found in the Supplementary Information files. More detailed results are available from the corresponding authors on reasonable request. Source data are provided with this paper. LULUC、农业和农场以外的排放数据由联合国粮食及农业组织策划，并可在 FAOSTAT 上免费获取。本研究使用的人口数据来自联合国世界人口展望 。运输和与食品相关的部门的货币价值数据来自 GTAP 数据库 。分析中使用的补充方法、讨论、图表和数据集可在补充信息文件中找到。如需更详细的结果，请向相应作者提出合理请求。本文附带源数据。
Code availability 代码可用性
Code developed for data processing in MATLAB is available in the Supplementary Information files. MATLAB 中用于数据处理的代码可在附加信息文件中找到。
References 参考资料
Crippa, M. et al. Food systems are responsible for a third of global anthropogenic GHG emissions. Nat. Food 2, 198-209 (2021). Crippa, M.等。食品系统负责全球人为温室气体排放的三分之一。自然食品 2, 198-209 (2021)。
Tubiello, F. N. et al. Pre-and post-production processes increasingly dominate greenhouse gas emissions from agri-food systems. Earth Syst. Sci. Data 14, 1795-1809 (2022). Tubiello, F. N.等。前期和后期生产过程越来越主导农业食品系统的温室气体排放。地球系统科学数据 14, 1795-1809 (2022)。
Greenhouse Gas Emissions From Agrifood Systems-Global, Regional and Country Trends, 2000-2020. FAOSTAT Analytical Brief Series No. 50 (FAO, 2022); https://www.fao.org/food-agriculturestatistics/data-release/data-release-detail/en/c/1616127/ 农业食品系统温室气体排放-全球、区域和国家趋势，2000-2020 年。FAOSTAT 分析简报系列第 50 号（联合国粮农组织，2022 年）; https://www.fao.org/food-agriculturestatistics/data-release/data-release-detail/en/c/1616127/
How to Feed the World in 2050 (FAO, 2019); https://www.fao.org/ fileadmin/templates/wsfs/docs/expert_paper/How_to_Feed_the_ World_in_2050.pdf 如何在 2050 年养活世界（联合国粮食及农业组织，2019 年）; https://www.fao.org/fileadmin/templates/wsfs/docs/expert_paper/How_to_Feed_the_World_in_2050.pdf
The Future of Food and Agriculture-Alternative Pathways to 2050 (FAO, 2018); https://www.fao.org/global-perspectives-studies/ resources/detail/en/c/1157074/ 粮食和农业的未来-2050 年的替代路径（联合国粮食及农业组织，2018 年）; https://www.fao.org/global-perspectives-studies/resources/detail/en/c/1157074/
Clark, M. A. et al. Global food system emissions could preclude achieving the and climate change targets. Science 370 , 705-708 (2020). 克拉克，M. A.等。全球食品系统排放可能会妨碍实现 和 气候变化目标。科学 370，705-708（2020 年）。
Bajželj, B. et al. Importance of food-demand management for climate mitigation. Nat. Clim. Change 4, 924-929 (2014). Bajželj, B.等。食品需求管理对气候减缓的重要性。自然气候变化 4，924-929（2014）。
Dhakal, S. et al. in Climate Change 2022: Mitigation of Climate Change (eds Shukla, P. R. et al.) Ch. 2 (Cambridge Univ. Press, 2022). Dhakal, S.等人在《气候变化 2022：气候变化的减缓》（主编 Shukla, P. R.等人）第 2 章（剑桥大学出版社，2022 年）。
Peters, G. P. From production-based to consumption-based national emission inventories. Ecol. Econ. 65, 13-23 (2008). 彼得斯，G. P. 从基于生产到基于消费的国家排放清单。生态经济学。65，13-23（2008）。
Hong, C. et al. Global and regional drivers of land-use emissions in 1961-2017. Nature 589, 554-561 (2021). 洪等人。1961-2017 年土地利用排放的全球和区域驱动因素。自然 589，554-561（2021）。
Xu, X. et al. Global greenhouse gas emissions from animal-based foods are twice those of plant-based foods. Nat. Food 2, 724-732 (2021). 徐等人。动物性食品的全球温室气体排放量是植物性食品的两倍。自然食品 2，724-732（2021）。
Hubacek, K., Feng, K., Minx, J., Pfister, S. & Zhou, N. Teleconnecting consumption to environmental impacts at multiple spatial scales: research frontiers in environmental footprinting. J. Ind. Ecol. 18, 7-9 (2014). Hubacek, K., Feng, K., Minx, J., Pfister, S. & Zhou, N. 将消费与环境影响在多个空间尺度上进行连接：环境足迹研究前沿。工业生态学杂志。18, 7-9 (2014)。
Hubacek, K., Feng, K., Chen, B. & Kagawa, S. Linking local consumption to global impacts. J. Ind. Ecol. 20, 382-386 (2016). Hubacek, K., Feng, K., Chen, B. & Kagawa, S. 将本地消费与全球影响联系起来。工业生态学杂志，20，382-386（2016）。
Wiedmann, T. & Lenzen, M. Environmental and social footprints of international trade. Nat. Geosci. 11, 314-321 (2018). Wiedmann, T. & Lenzen, M. 国际贸易的环境和社会足迹. Nat. Geosci. 11, 314-321 (2018).
Barrett, J. et al. Consumption-based GHG emission accounting: a UK case study. Clim. Policy 13, 451-470 (2013). 巴雷特，J.等。基于消费的温室气体排放核算：英国案例研究。气候政策 13，451-470（2013 年）。
Liu, Z. et al. Four system boundaries for carbon accounts. Ecol. Modell. 318, 118-125 (2015). 刘，Z.等。碳账户的四个系统边界。生态模型。318，118-125（2015 年）。
Davis, S. J. & Caldeira, K. Consumption-based accounting of emissions. Proc. Natl Acad. Sci. USA 107, 5687-5692 (2010). 戴维斯，S.J.和卡尔迪拉，K.基于消费的 排放核算。美国国家科学院院刊 107，5687-5692（2010）。
Feng, K. et al. Outsourcing within China. Proc. Natl Acad. Sci. USA 110, 11654-11659 (2013). 冯，K.等人。在中国内部外包 。美国国家科学院院刊 110，11654-11659（2013）。
Hong, C. et al. Land-use emissions embodied in international trade. Science 376, 597-603 (2022). 洪等人。国际贸易中体现的土地利用排放。科学 376，597-603（2022）。
Sandström, V. et al. The role of trade in the greenhouse gas footprints of EU diets. Glob. Food Secur. 19, 48-55 (2018). Sandström 等人。欧盟饮食温室气体足迹中贸易的作用。全球食品安全。19，48-55（2018）。
Li, M. et al. Global food-miles account for nearly of total food-systems emissions. Nat. Food 3, 445-453 (2022). 李等人。全球食品里程几乎占总食品系统排放的 。自然食品 3，445-453（2022）。
Cucurachi, S., Scherer, L., Guinée, J. & Tukker, A. Life cycle assessment of food systems. One Earth 1, 292-297 (2019). Cucurachi, S., Scherer, L., Guinée, J. & Tukker, A. 食品系统的生命周期评估。One Earth 1, 292-297 (2019).
Hubacek, K. & Feng, K. Comparing apples and oranges: some confusion about using and interpreting physical trade matrices versus multi-regional input-output analysis. Land Use Policy 50, 194-201 (2016). Hubacek, K. & Feng, K. 比较苹果和橙子：关于使用和解释物理贸易矩阵与多区域输入输出分析存在一些混淆。《土地利用政策》50, 194-201 (2016)。
Behrens, P. et al. Evaluating the environmental impacts of dietary recommendations. Proc. Natl Acad. Sci. USA 114, 13412-13417 (2017). Behrens, P.等人。评估饮食建议的环境影响。美国国家科学院院刊 114，13412-13417（2017）。
Kastner, T., Kastner, M. & Nonhebel, S. Tracing distant environmental impacts of agricultural products from a consumer perspective. Ecol. Econ. 70, 1032-1040 (2011). Kastner, T., Kastner, M. & Nonhebel, S. 从消费者角度追踪农产品的远程环境影响。生态经济学。70，1032-1040（2011）。
Kastner, T. et al. Cropland area embodied in international trade: contradictory results from different approaches. Ecol. Econ. 104, 140-144 (2014). Kastner, T.等人。国际贸易中体现的耕地面积：不同方法得出矛盾结果。生态经济学。104, 140-144 (2014)。
Bruckner, M. et al. FABIO-the construction of the food and agriculture biomass input-output model. Environ. Sci. Technol. 53, 11302-11312 (2019). 布鲁克纳等人。FABIO-食品和农业生物质投入产出模型的构建。环境科学技术。2019 年，53，11302-11312。
Foong, A., Pradhan, P., Frör, O. & Kropp, J. P. Adjusting agricultural emissions for trade matters for climate change mitigation. Nat. Commun. 13, 3024 (2022). 冯，A.，普拉丹，P.，弗罗尔，O.和克罗普，J. P. 调整农业排放以应对贸易问题对气候变化缓解至关重要。自然。通讯。13，3024（2022）。
Kim, B. F. et al. Country-specific dietary shifts to mitigate climate and water crises. Glob. Environ. Change 62, 101926 (2020). 金，B. F.等。国家特定的饮食转变以缓解气候和水危机。全球环境变化 62，101926（2020 年）。
Tubiello, F. N. et al. Greenhouse gas emissions from food systems: building the evidence base. Environ. Res. Lett. 16, 065007 (2021). Tubiello, F. N.等。食品系统的温室气体排放：建立证据基础。环境研究通讯。16, 065007（2021）。
Tubiello, F. N. et al. Methods for Estimating Greenhouse Gas Emissions From Food Systems. Part III: Energy Use in Fertilizer Manufacturing, Food Processing, Packaging, Retail and Household Consumption (FAO, 2021); https://doi.org/10.4060/cb7473en Tubiello, F. N.等人。 估算食品系统温室气体排放的方法。 第三部分：肥料制造、食品加工、包装、零售和家庭消费中的能源使用（FAO，2021 年）; https://doi.org/10.4060/cb7473en
Le Tourneau, F.-M. Is Brazil now in control of deforestation in the Amazon? Cybergeo https://doi.org/10.4000/cybergeo. 27484 (2016). 勒图尔诺，F.-M. 巴西现在是否控制了亚马逊的森林砍伐？Cybergeo https://doi.org/10.4000/cybergeo. 27484 (2016).
Soares-Filho, B. et al. Cracking Brazil's forest code. Science 344, 363-364 (2014). Soares-Filho 等人。破解巴西的森林法典。科学 344，363-364（2014）。
Heilmayr, R., Rausch, L. L., Munger, J. & Gibbs, H. K. Brazil's Amazon soy moratorium reduced deforestation. Nat. Food , 801-810 (2020). Heilmayr, R., Rausch, L. L., Munger, J. & Gibbs, H. K. 巴西亚马逊大豆禁运减少了森林砍伐。Nat. Food 801-810 (2020).
Mataveli, G. et al. Science-based planning can support law enforcement actions to curb deforestation in the Brazilian Amazon. Conserv. Lett. https://doi.org/10.1111/conl. 12908 (2022). Mataveli, G.等人。科学规划可以支持执法行动，遏制巴西亚马逊地区的森林砍伐。保护信函。https://doi.org/10.1111/conl. 12908（2022）。
Cohn, A. S. et al. Cattle ranching intensification in Brazil can reduce global greenhouse gas emissions by sparing land from deforestation. Proc. Natl Acad. Sci. USA 111, 7236-7241 (2014). Cohn, A. S.等人。巴西的牛肉养殖强化可以通过节约免于毁林的土地来减少全球温室气体排放。美国国家科学院院刊 111, 7236-7241 (2014)。
Poore, J. & Nemecek, T. Reducing food's environmental impacts through producers and consumers. Science 360, 987-992 (2018). Poore, J. & Nemecek, T. 通过生产者和消费者减少食品的环境影响。科学 360，987-992（2018 年）。
Yip, C. S. C., Lam, W. & Fielding, R. A summary of meat intakes and health burdens. Eur. J. Clin. Nutr. 72, 18-29 (2018). 叶，C. S. C.，林，W. & 费尔丁，R. 肉类摄入量和健康负担摘要。欧洲临床营养学杂志。72，18-29（2018）。
Willett, W. et al. Food in the Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food systems. Lancet 393, 447-492 (2019). 威莱特（Willett，W.）等人。《人类世的食物：EAT-Lancet 委员会关于可持续食品系统中健康饮食的研究》。《柳叶刀》（Lancet）393，447-492 页（2019 年）。
Vinci, C. European Union Beef Sector: Main Features, Challenges and Prospects (European Parliamentary Research Service, 2022); https://www.europarl.europa.eu/RegData/etudes/ BRIE/2022/733676/EPRS_BRI(2022)733676_EN.pdf Vinci, C. 欧盟牛肉行业：主要特点、挑战和前景（欧洲议会研究服务，2022 年）; https://www.europarl.europa.eu/RegData/etudes/ BRIE/2022/733676/EPRS_BRI(2022)733676_EN.pdf
Fuchs, R., Brown, C. & Rounsevell, M. Europe's Green Deal offshores environmental damage to other nations. Nature 586, 671-673 (2020). Fuchs, R., Brown, C. & Rounsevell, M. 欧洲绿色协议将环境破坏转移到其他国家。自然 586, 671-673 (2020).
Fuchs, R. et al. Why the US-China trade war spells disaster for the Amazon. Nature 567, 451-454 (2019). 福克斯，R.等人。为什么美中贸易战对亚马逊意味着灾难。自然 567，451-454（2019 年）。
The State of the World's Forests 2022. Forest Pathways for Green Recovery and Building Inclusive, Resilient and Sustainable Economies (FAO, 2022); https://www.fao.org/3/cb9360en/ cb9360en.pdf 世界森林状况 2022 年。绿色复苏和建设包容性、具韧性和可持续经济的森林路径（联合国粮食及农业组织，2022 年）; https://www.fao.org/3/cb9360en/ cb9360en.pdf
Proposal for a Regulation of the European Parliament and of the Council on the Making Available on the Union Market as Well as Export from the Union of Certain Commodities and Products Associated with Deforestation and Forest Degradation and Repealing Regulation (EU) No. 995/2010 (European Commission, 2021); https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex %3A52021PC0706 欧洲议会和理事会关于在欧盟市场上提供以及从欧盟出口与森林砍伐和森林退化相关的某些商品和产品的条例草案，并废除《欧盟法规》（EU）第 995/2010 号（欧洲委员会，2021 年）；https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A52021PC0706
Moran, D. et al. Quantifying the potential for consumer-oriented policy to reduce European and foreign carbon emissions. Clim. Policy 20, S28-S38 (2020). 莫兰等人。量化消费者导向政策减少欧洲和外国碳排放的潜力。气候政策 20，S28-S38（2020 年）。
Villoria, N., Garrett, R., Gollnow, F. & Carlson, K. Leakage does not fully offset soy supply-chain efforts to reduce deforestation in Brazil. Nat. Commun. 13, 5476 (2022). Villoria, N., Garrett, R., Gollnow, F. & Carlson, K. 泄漏并未完全抵消大豆供应链努力减少巴西森林砍伐。Nat. Commun. 13, 5476 (2022).
Kastner, T., Erb, K.-H. & Haberl, H. Rapid growth in agricultural trade: effects on global area efficiency and the role of management. Environ. Res. Lett. 9, 034015 (2014). Kastner, T., Erb, K.-H. & Haberl, H. 农业贸易迅速增长：对全球面积效率的影响及管理角色。环境研究通讯 9, 034015 (2014)。
World Economic Situation and Prospects 2022 (United Nations, 2022); https://www.un.org/development/desa/dpad/publication/ world-economic-situation-and-prospects-2022/ 2022 年世界经济形势与展望（联合国，2022 年）; https://www.un.org/development/desa/dpad/publication/world-economic-situation-and-prospects-2022/
Dalin, C., Wada, Y., Kastner, T. & Puma, M. J. Groundwater depletion embedded in international food trade. Nature 543, 700-704 (2017). 大林，C.，瓦达，Y.，卡斯特纳，T.和普马，M. J. 地下水过度开采蕴含在国际粮食贸易中。自然 543，700-704（2017 年）。
Chepeliev, M. Incorporating nutritional accounts to the GTAP Data Base. J. Glob. Econ. Anal. https://doi.org/10.21642/JGEA.070101AF (2022). Chepeliev, M. 将营养账户纳入 GTAP 数据库。全球经济分析杂志。https://doi.org/10.21642/JGEA.070101AF (2022)。
Davis, S. J., Burney, J. A., Pongratz, J. & Caldeira, K. Methods for attributing land-use emissions to products. Carbon Manage. 5, 233-245 (2014). 戴维斯，S. J.，伯尼，J. A.，庞格拉茨，J.和卡尔代拉，K.将土地利用排放归因于产品的方法。 碳管理。 5，233-245（2014 年）。
Saikku, L., Soimakallio, S. & Pingoud, K. Attributing land-use change carbon emissions to exported biomass. Environ. Impact Assess. Rev. 37, 47-54 (2012). 赛库，L.，索伊马卡里奥，S.和平古德，K.将土地利用变化的碳排放归因于出口生物质。环境影响评估审查。2012 年，37，47-54。
Land Use-Land, Inputs and Sustainability Dataset, FAOSTAT Online Database (FAO, 2022); https://www.fao.org/faostat/en/#data/RL 土地利用-土地、投入和可持续性数据集，FAOSTAT 在线数据库（FAO，2022）；https://www.fao.org/faostat/en/#data/RL
Fertilizer Use by Crop (FAO, 2006). 作物的肥料使用（FAO，2006 年）。
Conant, R. T., Berdanier, A. B. & Grace, P. R. Patterns and trends in nitrogen use and nitrogen recovery efficiency in world agriculture. Glob. Biogeochem. Cycles 27, 558-566 (2013). Conant, R. T., Berdanier, A. B. & Grace, P. R. 世界农业中氮利用和氮回收效率的模式和趋势。全球生物地球化学循环 27, 558-566 (2013)。
IPCC. Climate Change 2013: The Physical Science Basis (eds Stocker, T. F. et al.) (Cambridge Univ. Press, 2013). IPCC. 气候变化 2013：物理科学基础（主编 Stocker, T. F.等）（剑桥大学出版社，2013 年）。
Manure Left on Pasture, Climate Change Dataset, FAOSTAT Online Database (FAO, 2022); https://www.fao.org/faostat/en/#data/GP 畜粪留在牧场上，气候变化数据集，FAOSTAT 在线数据库（FAO，2022）; https://www.fao.org/faostat/en/#data/GP
Osei-Owusu, A. K., Kastner, T., de Ruiter, H., Thomsen, M. & Caro, D. The global cropland footprint of Denmark's food supply 2000-2013. Glob. Environ. Change 58, 101978 (2019). Osei-Owusu, A. K., Kastner, T., de Ruiter, H., Thomsen, M. & Caro, D. 丹麦食品供应 2000-2013 年的全球耕地占用足迹。全球环境变化 58, 101978 (2019)。
Herrero, M. et al. Biomass use, production, feed efficiencies, and greenhouse gas emissions from global livestock systems. Proc. Natl Acad. Sci. 110, 20888-20893 (2013). 赫雷罗等人。全球畜牧系统的生物质利用、生产、饲料效率和温室气体排放。美国国家科学院院刊。110，20888-20893（2013）。
Kalt, G., Kaufmann, L., Kastner, T. & Krausmann, F. Tracing Austria's biomass consumption to source countries: a product-level comparison between bioenergy, food and material. Ecol. Econ. 188, 107129 (2021). Kalt, G., Kaufmann, L., Kastner, T. & Krausmann, F. 追踪奥地利生物质消耗到来源国家：生物能源、食品和材料之间的产品级比较。生态经济学。188, 107129 (2021)。
de Ruiter, H. et al. Total global agricultural land footprint associated with UK food supply 1986-2011. Glob. Environ. Change 43, 72-81 (2017). de Ruiter, H.等。与英国食品供应相关的 1986-2011 年全球农业用地总占地面积。全球环境变化 43, 72-81 (2017)。
Aguiar, A., Chepeliev, M., Corong, E. L., McDougall, R. & van der Mensbrugghe, D. The GTAP Data Base: version 10. J. Glob. Econ. Anal. https://doi.org/10.21642/JGEA.040101AF (2019). 阿吉亚尔（Aguiar）、切佩列夫（Chepeliev）、科龙（Corong）、麦克道尔（McDougall）和范德门斯布鲁格（van der Mensbrugghe）. GTAP 数据库：第 10 版. 全球经济分析杂志. https://doi.org/10.21642/JGEA.040101AF（2019 年）。
Su, B. & Ang, B. W. Structural decomposition analysis applied to energy and emissions: some methodological developments. Energy Econ. 34, 177-188 (2012). 苏，B. & 安，B. W. 结构分解分析应用于能源和排放：一些方法论发展。能源经济学。34，177-188（2012）。
Muñoz, P. & Hubacek, K. Material implication of Chile's economic growth: combining material flow accounting (MFA) and structural decomposition analysis (SDA). Ecol. Econ. 65, 136-144 (2008). Muñoz, P. & Hubacek, K. 智利经济增长的物质含义：结合物质流量核算（MFA）和结构分解分析（SDA）。生态经济学。65，136-144（2008）。
Tubiello, F. N. et al. The FAOSTAT database of greenhouse gas emissions from agriculture. Environ. Res. Lett. 8, 015009 (2013). Tubiello, F. N.等人。农业温室气体排放的 FAOSTAT 数据库。环境研究通讯。8, 015009 (2013)。
Marques, A. et al. Increasing impacts of land use on biodiversity and carbon sequestration driven by population and economic growth. Nat. Ecol. Evol. 3, 628-637 (2019). Marques, A.等人。由人口和经济增长驱动的土地利用对生物多样性和碳封存的影响日益增加。自然生态与进化。3, 628-637 (2019)。
Wood, R. et al. Global sustainability accounting-developing EXIOBASE for multi-regional footprint analysis. Sustainability 7, 138-163 (2014). Wood, R.等人。全球可持续性会计-开发用于多区域足迹分析的 EXIOBASE。《可持续性》7, 138-163 (2014)。
Stadler, K. et al. EXIOBASE 3: developing a time series of detailed environmentally extended multi-regional input-output tables. J. Ind. Ecol. 22, 502-515 (2018). Stadler, K.等人。EXIOBASE 3：开发详细环境扩展多区域投入产出表的时间序列。工业生态学杂志。22，502-515（2018）。
We thank T. Kastner for providing the code for the PTF approach. We thank the support from Greenpeace Germany for the initial data analysis, modelling and discussions as part of the project 'Outsourced Environmental Degradation of the EU'. This research is also supported by the National Natural Science Foundation of China , the Shandong Natural Science Foundation of China (ZR2O21MG013), the major programme of the National Social Science Foundation of China (21ZDA065), the United Kingdom Research and Innovation (UoB Policy Support Fund PSF-16). For the purpose of open access, a CC BY public copyright licence is applied to any Author Accepted Manuscript arising from this submission. Y.L., Y.H., D.W. and Y.Z. acknowledge the funding support by the China Scholarship Council Ph.D. programme. 我们感谢 T. Kastner 提供 PTF 方法的代码。我们感谢绿色和平德国支持项目“欧盟外包环境恶化”的初步数据分析、建模和讨论。本研究还得到中国国家自然科学基金 的支持，中国山东省自然科学基金(ZR2O21MG013)，中国国家社会科学基金重大项目(21ZDA065)，英国研究与创新(UoB 政策支持基金 PSF-16)的支持。为了实现开放获取，任何由本次提交产生的作者接受的手稿都适用 CC BY 公共版权许可证。Y.L.、Y.H.、D.W.和 Y.Z.感谢中国国家留学基金委员会博士项目的资助。
Author contributions 作者贡献
Y.L., Y.S. and K.H. designed the research. Y.L. performed the analysis with support from Y.H., D.W. and Y.Z. on analytical approaches and visualization. Y.L. led the writing with efforts from H.Z., Y.S. and K.H. Y.S. and K.H. supervised and coordinated the overall research. All co-authors reviewed and commented on the manuscript. Y.L.、Y.S.和 K.H.设计了研究。Y.L.在 Y.H.、D.W.和 Y.Z.的支持下进行了分析，涉及分析方法和可视化。Y.L.主导了写作工作，得到了 H.Z.、Y.S.和 K.H.的努力。Y.S.和 K.H.监督和协调了整个研究。所有合著者都审查并评论了手稿。
Competing interests 竞争利益
The authors declare no competing interests. 作者声明没有竞争利益。
Additional information 额外信息
Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s43016-023-00768-z. 补充信息 在线版本包含的补充资料可在 https://doi.org/10.1038/s43016-023-00768-z 上找到。
Correspondence and requests for materials should be addressed to Yuli Shan or Klaus Hubacek. 函件和资料请求应寄至 Yuli Shan 或 Klaus Hubacek。
Peer review information Nature Food thanks Francesco Tubiello and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. 《自然食品》感谢 Francesco Tubiello 和其他匿名审稿人对本文的审稿工作所做的贡献。
Reprints and permissions information is available at www.nature.com/reprints. 转载和权限信息请访问 www.nature.com/reprints。
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 出版商注意施普林格自然对已发表的地图和机构隶属保持中立。
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Springer Nature 或其许可方（例如学会或其他合作伙伴）根据与作者或其他权利持有人签订的出版协议拥有本文的独家权利；作者对本文的已接受手稿版本进行自行存档，完全受限于该出版协议和适用法律的条款。
(c) The Author(s), under exclusive licence to Springer Nature Limited 2023 （c）作者，独家许可给施普林格自然有限公司 2023
nature portfolio 自然投资组合
Reporting Summary 报告摘要
Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist. 自然出版集团希望提高我们发表的作品的可重复性。这种形式为报告的一致性和透明度提供了结构。有关自然出版集团政策的更多信息，请参阅我们的编辑政策和编辑政策清单。
Statistics 统计
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. 对于所有统计分析，请确认以下项目是否出现在图例、表格图例、正文或方法部分中。
n/a Confirmed n/a 确认
The exact sample size ( ) for each experimental group/condition, given as a discrete number and unit of measurement 每个实验组/条件的确切样本大小（ ），以离散数字和测量单位给出
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly 关于是否从不同样本中进行了测量，还是是否重复测量了同一样本的声明
The statistical test(s) used AND whether they are one- or two-sided 使用的统计检验以及它们是单侧还是双侧
Only common tests should be described solely by name; describe more complex techniques in the Methods section. 只有常见的测试应该仅仅用名称来描述；更复杂的技术应该在方法部分进行描述。
A description of all covariates tested 所有被测试的协变量的描述
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons 对任何假设或更正的描述，例如正态性检验和多重比较校正的调整
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) 统计参数的完整描述，包括中心趋势（例如均值）或其他基本估计（例如回归系数）
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) 变异（例如标准偏差）或相关不确定性估计（例如置信区间）
For null hypothesis testing, the test statistic (e.g. ) with confidence intervals, effect sizes, degrees of freedom and value noted Give values as exact values whenever suitable. 对于零假设检验，测试统计量（例如 ）与置信区间、效应大小、自由度和 值一起给出。在适当时，将 值作为精确值给出。
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings 对于贝叶斯分析，关于先验选择和马尔可夫链蒙特卡洛设置的信息
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes 对于分层和复杂设计，识别适当的测试级别并全面报告结果
Estimates of effect sizes (e.g. Cohen's , Pearson's ), indicating how they were calculated 效应大小的估计值（例如 Cohen's ，Pearson's ），表明它们是如何计算的
Our web collection on statistics for biologists contains articles on many of the points above. 我们针对生物学家的统计数据收集包含了许多上述要点的文章。
Software and code 软件和代码
Policy information about availability of computer code 计算机代码可用性的政策信息
Data collection Data collection is performed in Matlab 2022b and Microsoft Excel. 数据收集 数据收集是在 Matlab 2022b 和 Microsoft Excel 中进行的。
Data analysis Data analysis including physical trade flow analysis and structural decomposition analysis is performed in Matlab 2022b. Code developed for data processing in MATLAB is available in the Supplementary Information files. 数据分析 数据分析包括物理贸易流分析和结构分解分析是在 Matlab 2022b 中进行的。 用于 MATLAB 数据处理的代码可在附加信息文件中找到。
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information. 对于使用自定义算法或软件的手稿，这些算法或软件对研究至关重要，但尚未在已发表的文献中描述的情况，软件必须提供给编辑和审稿人。我们强烈鼓励将代码存储在社区存储库（例如 GitHub）中。有关提交代码和软件的更多信息，请参阅《自然出版集团指南》。
Data 数据
Policy information about availability of data 数据可用性政策信息
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: 所有稿件必须包括数据可用性声明。该声明应提供以下信息（如适用）：
Accession codes, unique identifiers, or web links for publicly available datasets 公开数据集的接入代码、唯一标识符或网络链接
A description of any restrictions on data availability 数据可用性的任何限制描述
For clinical datasets or third party data, please ensure that the statement adheres to our policy 对于临床数据集或第三方数据，请确保陈述符合我们的政策
Human research participants 人类研究参与者
Policy information about studies involving human research participants and Sex and Gender in Research. 涉及人类研究参与者和研究中的性别和性别的政策信息。
Reporting on sex and gender 关于性别和性别的报道
N/A
Population characteristics 人口特征
N/A
Recruitment 招聘
N/A
Ethics oversight 道德监督
N/A
Note that full information on the approval of the study protocol must also be provided in the manuscript. 请注意，研究方案获批准的全部信息也必须在手稿中提供。
Field-specific reporting 领域特定报告
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. 请选择以下最适合您研究的选项。如果您不确定，请在选择之前阅读相应部分。
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences 生命科学 行为与社会科学 生态、进化与环境科学
Ecological, evolutionary & environmental sciences study design 生态、进化和环境科学研究设计
All studies must disclose on these points even when the disclosure is negative. 所有研究都必须在这些方面披露，即使披露是负面的。
Study description 研究描述
This study 1) evaluates global consumption-based food emissions covering land use and land use change, agricultural production and beyond-farm activities in 181 countries/areas between 2000 and 2019; 2) quantifies the emissions embodied in food domestic supply and trade (i.e., imports and exports) between countries; 3) uses structural decomposition analysis to identify the contributions of five driving factors to variations in consumption-based emissions. This study indicates how to reduce food emissions from production to consumption through policy applications for the entire supply chain and final consumers. 本研究评估了 2000 年至 2019 年间 181 个国家/地区的全球基于消费的食品排放，涵盖土地利用和土地利用变化、农业生产以及农场以外活动；量化了食品国内供应和贸易（即进口和出口）中所包含的排放；利用结构分解分析确定了五个驱动因素对消费排放变化的贡献。本研究指出了如何通过政策应用来减少从生产到消费的食品排放，涵盖整个供应链和最终消费者。
Research sample 研究样本
This study covers emissions along entire supply chains from the consumption of 153 food products in 181 countries. Data are extracted from the following sources: 1) Quantify consumption of food products in countries using the physical trade flow approach. Data in terms of production and bilateral trade of 153 products are obtained from FAOSTAT. 2) Calculate emissions of crops and livestock products based on the combination of the emission intensity of each product in processes and the consumption. 3) Country-level emissions generated in supply chain processes (LULUC, agriculture and beyond-farm activities) are collected from FAOSTAT; statistical data related to food products such as land use, feed use, fertilizer inputs etc. are obtained from FAOSTAT. 4) Other socioeconomic statistical data such as population, economic connection and impact coefficient are collected from the database including UN World Population Prospects, GTAP or previous studies. 5) apply SDA to decompose the changes in consumption-based emissions to five driving factors. 6) conduct an uncertainty assessment using a Monte Carlo approach. 本研究涵盖了来自 181 个国家的 153 种食品产品消费的整个供应链排放。数据来自以下来源：1）使用物理贸易流方法量化各国食品产品的消费。从 FAOSTAT 获取了 153 种产品的生产和双边贸易数据。2）根据每种产品在过程中的排放强度和消费量计算作物和畜产品的排放。3）从 FAOSTAT 收集了供应链过程（LULUC、农业和农场以外活动）产生的国家级排放；与食品产品相关的统计数据，如土地利用、饲料使用、化肥投入等，均来自 FAOSTAT。4）其他社会经济统计数据，如人口、经济联系和影响系数，均来自包括联合国世界人口展望、GTAP 或先前研究在内的数据库。5）应用 SDA 将基于消费的排放变化分解为五个驱动因素。6）使用蒙特卡洛方法进行不确定性评估。
Sampling strategy 抽样策略
We apply the physical trade flow approach to quantify consumption-based emissions with the physical flow matrix (production and trade flows) of 153 food products between 181 countries every year. The calculation is based on annual statistical data (in the unit of mass) during 2000 and 2019. This study covers most food categories as well as most countries with data availability to investigate the consumption-based emissions with product details at a global level. 我们应用物理贸易流方法，利用每年 181 个国家之间 153 种食品产品的物理流矩阵（生产和贸易流）来量化基于消费的排放。计算基于 2000 年至 2019 年的年度统计数据（以质量单位）。本研究涵盖了大多数食品类别以及大多数国家的数据可用性，以调查全球范围内的基于消费的排放，并提供产品细节。
Data collection 数据收集
Yanxian Li collected the most recent data required for this study in a bulk download from FAOSTAT (https://www.fao.org/faostat/en/) on December 2021. Population data are obtained from UN World Population Prospects 2022 (https://population.un.org/wpp/ Download/Standard/Population/), economic values for the food transport sector are collected from GTAP 10 (https:// www.gtap.agecon.purdue.edu/databases/v10/index.aspx), and other data such as the environmental impact coefficients are extracted from individual publications cited by this study. 严贤李于 2021 年 12 月从 FAOSTAT（https://www.fao.org/faostat/en/）批量下载了本研究所需的最新数据。人口数据来自联合国 2022 年世界人口展望（https://population.un.org/wpp/Download/Standard/Population/），食品运输部门的经济价值数据来自 GTAP 10（https://www.gtap.agecon.purdue.edu/databases/v10/index.aspx），其他数据如环境影响系数则从本研究引用的各个出版物中提取。
Timing and spatial scale All the statistical data in this study covers the years 2000, 2005, 2010, 2015 and 2019. All data are collected on time from the online database. The data collection starts on 1/1/2022 and ends on 1/4/2022. 时间和空间尺度本研究中的所有统计数据涵盖了 2000 年、2005 年、2010 年、2015 年和 2019 年。所有数据均及时从在线数据库收集。数据收集始于 2022 年 1 月 1 日，结束于 2022 年 1 月 4 日。
Data exclusions 数据排除
No data is excluded in this analysis. 在这个分析中没有排除任何数据。
Reproducibility 可重复性
We describe all detailed methods and data sources, processing and analysis to ensure the reproducibility of the work. For the model simulations, we keep all input files in the record to make sure reproducibility. 我们描述了所有详细的方法和数据来源，处理和分析，以确保工作的可重复性。对于模型模拟，我们保留所有输入文件的记录，以确保可重复性。
Randomization 随机化
This is not relevant to our study because our work is not an "Experimental" study. We used statistical data and published data to perform the calculation. 这与我们的研究无关，因为我们的工作不是一项“实验性”研究。我们使用统计数据和已发表的数据进行计算。
Reporting for specific materials, systems and methods 报告特定材料、系统和方法
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. 我们需要作者提供关于许多研究中使用的某些材料、实验系统和方法的信息。在这里，指出列出的每种材料、系统或方法是否与您的研究相关。如果您不确定列表项是否适用于您的研究，请在选择响应之前阅读相应部分。
Materials & experimental systems Involved in the study 材料和实验系统 参与研究
n/a Involved in the study n/a 参与研究
Antibodies 抗体
Eukaryotic cell lines 真核细胞系
ChIP-seq
Palaeontology and archaeology 古生物学和考古学
Flow cytometry 流式细胞术
Animals and other organisms 动物和其他生物
MRI-based neuroimaging 基于 MRI 的神经影像学
Clinical data 临床数据
Dual use research of concern 关注的双重用途研究
'Integrated Research on Energy, Environment and Society (IREES), Energy and Sustainability Research Institute Groningen (ESRIG), University of Groningen, Groningen, the Netherlands. Academy of Plateau Science and Sustainability, Qinghai Normal University, Xining, China. Institute of Blue and Green Development, Weihai Institute of Interdisciplinary Research, Shandong University, Weihai, China. School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham, UK. College of Economics and Management & Research Centre for Soft Energy Science, Nanjing University of Aeronautics and Astronautics, Nanjing, China. Business School, University of Shanghai for Science and Technology, Shanghai, China. ฤe-mail: y.shan@bham.ac.uk; k.hubacek@rug.nl 能源、环境和社会综合研究（IREES），格罗宁根能源与可持续发展研究所（ESRIG），荷兰格罗宁根大学。 青海师范大学高原科学与可持续发展研究院，中国西宁。 威海跨学科研究院蓝绿发展研究所，山东大学威海研究院，中国威海。 伯明翰大学地理、地球和环境科学学院，英国伯明翰。 南京航空航天大学经济与管理学院及软能源科学研究中心，中国南京。 上海理工大学商学院，中国上海。 电子邮件：y.shan@bham.ac.uk；k.hubacek@rug.nl
