Introduction

The demographic landscape of China is undergoing a profound transformation, featuring an increasing proportion of elderly citizens. Although China has not been ranked among the top ten countries for population aging in the past decade, considering its vast population, this demographic shift is reshaping the socioeconomic fabric of the nation and raising increasing concerns in the recent several years. According to the National Bureau of Statistics of China (NBSC, http://www.stats.gov.cn/), the population aged 60+ accounted for 19.8% of the total population in China in 2022, reaching 280 million. It has been projected that, from 2022 to 2050, China’s total population will decrease by about 8%, but the population aged 60+ will rise by about 92% (https://population.un.org/wpp/). Population aging would profoundly alter economic growth, energy demand, urbanization, goods consumption patterns, and traveling1,2. Efforts have also been made to link population aging and age structure to GHG emissions in developed and developing countries3,4,5,6,7,8,9,10. Many of these studies were devoted to elucidating the contributions of aging societies to GHG emissions via household and energy consumption. These studies collectively suggest that population aging could have a complex relationship with GHG emissions, potentially leading to emission reductions in some contexts due to changes in economic activities, consumption patterns, and industrial structures. The food consumption and demand of the aged population were also a focus of these studies. However, in most cases, it accounted for only a small portion of total household consumption. Likewise, food consumption-induced GHG emission is a small sector of household emissions, primarily carbon emissions from energy consumption associated with heating, air conditioning, cooking, water heating, food consumption, clothing, housing, transportation, and communication. As a result, important signals in GHG emissions induced by food consumption from the elderly population were, to some extent, covered by other factors in household consumption. Limited investigations of the relationship between population aging and food consumption-induced GHG emissions in China and Japan also suggested that the increasing aged population was associated with growing GHG emissions or exerted a weak influence5,9,11.

The changes in diets and food consumption in the aged population are considered important factors influencing food demand and related GHG emissions. Older populations have different dietary needs and preferences than younger ones due to health considerations and the weaker digestive system12,13, though recent investigations suggested that older adults aged 65+ consumed more meat and dairy in some developed countries than the middle-aged group (age 45–49)3,14. This is, to a large extent, for protein and calcium requirements. As people age, their protein needs increase to help maintain muscle mass and prevent sarcopenia (muscle loss). Meat and dairy are rich sources of high-quality protein which can be particularly important for older adults. As a result, the aged population in these developed countries tends to contribute to more GHG emissions from a food consumption perspective. In China’s case, increasing food demand driven by growing population and personal income enhanced its agricultural GHG emissions because agricultural production is driven primarily by food demand. On the other hand, a rapidly growing aged population tends to reduce meat demand and consumption due to weaker digestive systems, economic factors (rapidly increasing meat price and low pension), and health considerations10,13. Therefore, the aged population will help reduce GHG emissions in China. The data from the China Health and Nutrition Survey (https://www.cpc.unc.edu/projects/china) reveal that, from 2004 to 2011, the meat (beef, pork, sheep, chicken) and cereal crops (wheat, rice, maize) consumption by the aged population of 60+ were 18.8% and 1.3% lower than the younger population of 60−. As a result, differing from the aforementioned studies3,5,9,11,14, lower meat consumption by older adults should help reduce food consumption-related non-CO2 GHG emissions.

Given its outsized impact on climate change and the relatively immediate benefits of reducing its atmospheric concentration, methane emission reduction is a critical and urgent element of global efforts to mitigate the most severe effects of climate change15,16,17,18,19. According to the statistics from the Food and Agriculture Organization (FAO), China accounted for ~27% and 30% of global meat and rice consumption in 2020 (FAOSTAT, http://www.fao.org/faostat/en/#data) and contributed significantly to global agricultural CH4 emission20, which is, to a large extent, attributed to its growing food consumption and demands associated with its vast population and rapidly increasing personal income21,22,23,24,25,26. Recently, China released its “Methane Emission Control Action Plan” (http://bbrmedia.cn/2023/1108/892117.html), in which emission mitigation from the agricultural sector is one of the major tasks in emission control because this sector emits about 40% of total CH4 emission in China. The EDGAR (Emission Database for Global Atmospheric Research, https://edgar.jrc.ec.europa.eu/dataset_ghg80#p1) has estimated the non-CO2 GHG emissions from China’s agriculture at 871.9 Tg CO2eq in 2020, accounting for about 39% of the country’s total, and by 2025, the agricultural emission is projected to maintain almost unchanged (about 40%), though considerable efforts have been made to mitigate emissions from agricultural activities27,28. Such high emissions pose great challenges to China’s non-CO2 GHG emission mitigation. On the other hand, to ensure the nation’s food security, the national government has also been encouraging agriculture expansion, suggesting that additional efforts need to be made to improve techniques and facilities for emission control. Since dietary shifts play a vital role in indirectly altering non-CO2 GHG emissions from agriculture and there are large knowledge gaps in this context, the present study shall focus on food consumption-induced non-CO2 GHG emissions associated with China’s rapidly aging population, which are directly released from enteric fermentation and manure management in livestock, as well as from staple food grain production (particularly rice)29. Considering significant connections between demographic and dietary transitions and non-CO2 GHG emissions, the change in population structure is likely an important driver contributing to the change in agriculture-related methane emissions, which is still poorly understood. The present study aims to fill this knowledge gap in the response of non-CO2 GHG emissions to population aging in China and to explore the extent of the population aging impact on agricultural non-CO2 GHG emissions.

Results

Drivers of non-CO2 GHG emissions from aging-related food consumption

By combining Chinese meat (beef, pork, sheep, goat, and chicken meat), dairy, and grain (rice, maize, and wheat) trade data with agriculture activities-driven non-CO2 GHG emission data from the FAOSTAT database, we quantified non-CO2 GHG emissions associated with food consumption in China. Figure 1a presents non-CO2 GHG emissions embodied in domestic food consumption in China during 2005–2020, showing a declining trend of food consumption-induced non-CO2 GHG emissions from 474.5 Tg CO2eq in 2005 to 447.5 Tg CO2eq in 2009 and growing after that to 513.4 Tg CO2eq by 2020, including 352.0 Tg CO2eq CH4 and 161.4 Tg CO2eq N2O. Enteric fermentation in ruminants, primarily from cattle, is the predominant source of CH4, leading to 117.8 Tg CO2eq CH4 induced by beef consumption alone in 2020. Beef consumption-based N2O emissions also reach 38.0 Tg CO2eq. Rice, on the other hand, is also an important source of CH4 and N2O emissions due to its significant consumption in China, releasing 126.0 Tg CO2eq CH4 and 39.9 Tg CO2eq N2O in 2020, respectively (Fig. S1). Figure 1b presents the three driving factors (Eq. (5) in “Methods”) and their respective contribution to changes in non-CO2 GHG emissions embodied in China’s food consumption every five years from 2005 to 2020 (“Methods”). Table S1 shows the respective contribution of three factors driving the changes in non-CO2 GHG emissions from food consumption.

Fig. 1: Non-CO2 GHG emissions (Tg CO2eq) embodied in China’s food consumption and emission drivers during 2005–2020.
figure 1

a For non-CO2 GHG emissions from food consumption in China, the red bar denotes CH4 emissions, and the blue bar represents N2O emissions. The solid yellow line scaled on the right Y axis denotes the total non-CO2 GHG emissions (CH4 + N2O); b three factors driving the changes in non-CO2 GHG emissions from food consumption estimated using LDMI decomposition analysis. Detailed model descriptions are presented in “Methods”.

Full size image

Among the three drivers, CBEI (Consumption-Based Emission Intensity) is the most prominent factor in reducing non-CO2 GHG emissions from food consumption during 2005–2010. The CBEI, defined as the food consumption-induced non-CO2 GHG emissions per unit of food consumption in China (including both imported and domestically produced food), is derived from the production-based emission intensity (PBEI) in food-supplying countries and the structure of food consumption sources. Benefiting from improving production techniques, China and other major food-supplying countries experienced a decline in PBEI from 2005 to 2010, meaning that the emissions released per unit of food production decreased. For example, CH4 PBEI and N2O PBEI for beef consumption in China decreased by 35.5% during this period (Fig. S2). Consequently, the CBEI of different food items in China declined significantly, leading to the overall non-CO2 GHG emission reduction of 71.5 Tg CO2eq from 2005 to 2010. Figure S2 compares CH4 and N2O CBEI of different food categories in China in 2005, 2010, 2015, and 2020. The highest CH4 CBEI is identified in beef consumption (Fig. S2). Compared to 2005, the CH4 CBEI and N2O CBEI for beef consumption dropped by 35.2% and 35.3% in 2010, respectively, leading to 51.1 Tg CO2eq CH4 and N2O emission reduction, which accounts for 71.5% of overall CBEI-induced non-CO2 GHG emission reduction (71.5 Tg CO2eq).

After 2010, the CBEI in China did not drop significantly, partly due to the significant change in the food (beef especially) supply structure. Figure S3 shows the total imported food and its proportion to total food consumption in China from 2005 to 2020. Imported beef from overseas only accounted for less than 1% of total beef consumption in China from 2005 to 2011. Beef imports from overseas have proliferated since 2012 under rapidly increasing beef demand and consumption. By 2020, imported beef has accounted for 24.0% of the total beef consumption in China. Since China’s major beef suppliers, such as Brazil, Uruguay, and Argentina, have higher PBEI than China (Fig. S2c, d), the CBEI of many food items in China did not decrease significantly but increased in beef and pork consumption during 2015–2020 (Fig. S2), despite the PBEIs in these countries declined, to some extent, due to their rising production efficiency (Fig. S2c, d). As a result, the non-CO2 GHG emission attributable to the changes in CBEI was reduced only by 21.5 Tg CO2eq from 2010 to 2015 and 5.7 Tg CO2eq from 2015 to 2020, respectively.

Among the three drivers, the CPAAP (food consumption per aging adjusted population) is the most critical driving factor in enhancing non-CO2 GHG emissions embodied in food consumption, contributing 57.6 Tg CO2eq, 42.5 Tg CO2eq, and 44.1 Tg CO2eq to growing non-CO2 GHG emissions during the three periods. The CPAAP, defined as the total food consumption of the population across all age groups divided by the non-elderly population (also referred to as aging adjusted population, AAP, estimated by total population minus aged population of 60 + ), provides an insight into the food consumption trends among the working-age and younger populations. Among different food items, increasing CPAAP for beef consumption is crucial in enhancing CPAAP-driven non-CO2 GHG emissions (Table S1). The CPAAP for different food items in China is presented in Fig. S4. The results illustrate a significant growth trend of beef CPAAP in recent years (especially after 2015), enhancing by 48.4% from 5.2 kg in 2015 to 7.7 kg in 2020 due to improvements in living conditions, changes in dietary structure, and remarkably declining AAP. Growing CPAAP from beef consumption caused 50.6 Tg CO2eq non-CO2 GHG emissions from 2015 to 2020 (Table S1). On the other hand, owing to the dietary structure change in China in recent years, the CPAAP for grain (rice, wheat, and maize) and pork tends to decrease (Fig. S4), which partly elevates the increase of CPAAP-driven non-CO2 GHG emissions from beef consumption from 2015 to 2020.

Interestingly, our result also yielded non-CO2 GHG emission reduction driven by the AAP at 0.4 Tg CO2eq from 2005 to 2010, 1.0 Tg CO2eq from 2010 to 2015, and 5.2 Tg CO2eq from 2015 to 2020, respectively, indicating that from 2015 onward, the AAP plays an almost equal role as the CBEI in the reduction of non-CO2 GHG emissions, together offsetting 24.7% of the CPAAP-induced non-CO2 GHG emission. Figure S5 presents annual variations of populations of the two age groups of 0–59 (AAP) and 60+ and their proportions to the total population from 2005 to 2020, clearly showing a declining proportion of the younger population of 0–59 and an increasing proportion of the aged population of 60+ during this period. The fraction of the aged population of 60+ in China’s total population increased from 11.03% in 2005 to 18.73% in 2020, growing by 69.79%. On the contrary, the proportion of the population aged 0–59 (AAP) to the total population decreased from 88.97% in 2005 to 81.27% in 2020. The population aged 0–59 declined from 1168.7 million in 2008 to 1147.7 million in 2020, suggesting that China has entered an aging society. Considering the differences in dietary structure between the aging and non-aging population, the food consumption (especially meat consumption) of the non-aging population tends to be higher than that of the aging population (Table S2 and Fig. S6); such rapid population aging in China is expected to cause a marked reduction in meat consumption. As a result, the population aging in China will likely become an important factor in inhibiting the increase of non-CO2 GHG emissions embodied in food consumption. As shown in Fig. 1b, the contribution of AAP to the food consumption-related non-CO2 GHG emission reduction enhanced from 0.4 Tg CO2eq from 2005 to 2010 to 5.2 Tg CO2eq Gg from 2015 to 2020, almost identical to CBEI’s contribution to emissions reduction, along with the growing aged population and decreasing AAP in the country. It is, therefore, worthwhile to further explore the associations between population aging and non-CO2 GHG emissions embodied in food consumption across China.

We estimated the difference in food consumption and related emissions from the Aging scenario (SR-A scenario) to the No-aging scenario (SCAS-A scenario, SI Text 4). The results indicate that the aging population significantly altered China’s food consumption structure, characterized by significantly declining meat (typically ruminant meat) and staple food grain (typically rice) consumption and increasing consumption of milk and maize. As shown in Fig. 2a, the absolute values of food demand difference are almost monotonically and linearly increased from 2005 to 2020, with the difference reaching 352,787 tons and 216,864 tons for beef and mutton in 2020, respectively. Given the high PBEI of ruminants (Fig. S2), the falling ruminant meat (beef and mutton) consumption induced by population aging led to an overall 9.0 Tg CO2eq non-CO2 GHG emissions reduction in 2020 (Fig. 2b).

Fig. 2: Impact of population aging on food consumption and related non-CO2 GHG emissions from 2005 to 2020 in China.
figure 2

a Changes in rice (dashed line, scaled on the left Y axis) and beef and mutton (solid lines, scaled on the right Y axis) consumption between Aging and No aging scenarios, estimated by Difffood = Foodaging – Foodnoaging, (kt). b Rice consumption-induced (dashed line, scaled on the left Y axis) and beef and mutton consumption-induced (solid lines, scaled on the right Y axis) non-CO2 GHG emission difference between Aging and No Aging scenario, estimated by Diffemi = EmiAging – EmiNoaging, where EmiAging stands for non-CO2 GHG emissions under Aging scenarios and EmiNoaging denotes non-CO2 GHG emission under No aging (constant age structure) scenarios, respectively (Tg CO2eq). c ACSI for rice (dashed line, scaled on the left Y axis) and beef and mutton (solid lines, scaled on the right Y axis), indicating changes in consumption (ton) per percentage change in aging rate (“Methods”). d AESI for rice (dashed line, scaled on the left Y axis) and beef and mutton (solid lines, scaled on the right Y axis), indicating changes in food consumption-induced non-CO2 GHG emission (Gg CO2eq) per percentage change in aging rate (“Methods”). The inner figures on the low-left corner show food consumption and emission differences between aging and no aging scenarios (a, b) and ACSI and AESI for other food items.

Full size image

On the other hand, grain consumption structure change driven by population aging contributes less to non-CO2 GHG emissions change under the SR-A scenario; the consumption of rice and wheat also decreased by 1,343,009 tons and 291,141 tons, leading to a total of 1.3 Tg CO2eq of non-CO2 GHG emission reductions in 2020. However, the consumption of maize, which has been often regarded as a health food (coarse grain) in China, slightly increased by 97,452 tons in the SR-A scenario in 2020 (Fig. 2a). Similarly, population aging also led to growing milk consumption at 1,071,888 tons, contributing to non-CO2 GHG emission increase at 0.9 Tg CO2eq in 2020 (Fig. 2b).

As a result, the food consumption-based non-CO2 GHG emissions driven by population aging reduced one order of magnitude from 1.0 Tg CO2eq in 2005 to 10.1 Tg CO2eq in 2020. This falling emission is higher than the total CH4 and N2O emission from all agri-food systems in some countries in 2020, such as Austria (9.0 Tg CO2eq) and Sweden (8.8 Tg CO2eq, https://www.fao.org/faostat/en/#data/GT).

We further introduced an aging-related food consumption sensitivity Index (ACSI) and an aging-related emissions sensitivity index (AESI) to measure the sensitivity of changes in food consumption and related non-CO2 GHG emission to varying aging rates, respectively (“Methods“), and to elucidate how consumption patterns and related emissions change in response to an aging demographic. These indices provide critical insights into the interplay between population aging trends and their implications for food consumption and environmental sustainability. The declining trend of the ACSI for meat and rice presented in Fig. 2c features the response of food consumption reduction to the increasing population aging from 2005 to 2020, during which population aging emerges as a national concern. ACSI for beef, in particular, shows a remarkable change from −2481.6 tons in 2005 to −7570.1 tons in 2020, indicating that a 1% increase in the aging rate under the SR-A scenario will lead to a decrease of 7570.1 tons in beef consumption in 2020. The decreasing trend indicates an increasing sensitivity of beef consumption to the aging rate over decades, revealing the growing impact of population aging on food consumption structures.

The beef consumption-induced non-CO2 GHG emission, demonstrating the highest sensitivity or AESI among those food items to the aging rate from 2005 to 2020, underscores the unique position of beef consumption and related non-CO2 GHG emission in response to demographic shifts (Fig. 2d). Despite a consistent negative sensitivity indicating that non-CO2 GHG emissions associated with beef consumption decrease as the population ages, this trend becomes markedly more pronounced after 2011, declining from −57.8 Gg CO2eq in 2011 to relatively high sensitivity of −133.4 Gg CO2eq in 2020. Compared to the ACSI, which declined from 2005 to 2020, the AESI increased from 2005 to 2011 and fell rapidly after that. This can be primarily attributed to a notable decrease in CBEI from 2005 to 2011 (Fig. S2). This reduction in CBEI initially masked the negative impact of population aging on beef consumption-induced non-CO2 GHG emissions, which became more evident as the CBEI stabilized and the aging sped up. Given the more rapid population aging in China in the coming years (https://population.un.org/wpp/), the effect of aging on food consumption-induced non-CO2 GHG emissions will become more evident, as will be elaborated below.

Figure 3 compares the differences in meat and grain consumption and related non-CO2 emissions between SR and SCAS scenarios with adjusted PCCR (per capita consumption ratio, “Methods”) from 2005 to 2020. Considering an intergenerational difference in meat consumption, the lower meat consumption by the elderly population in the past might not be unsubstantiated to be extrapolated to younger generations who will get old in the future. To account for the effect of changes in dietary behavior in different generations of the aged and non-aged population, we added and subtracted 5% and 10% to the baseline PCCR, defined as the ratio of per capita food consumption of the aging population to that of the non-aging population (referred to as adjusted SR-A scenarios, “Methods” and SI Texts 1 and 4) and recalculated food consumption and non-CO2 GHG emissions of the aged and non-aged population subject to these adjusted scenarios. Figure 3a illustrates the differences in the total meat consumption (scaled on the right Y axis) and related non-CO2 GHG emissions (scaled on the left Y axis) between SR-A and SCAS-A scenarios and between adjusted SR- and SCAS- scenarios, respectively. We observe, as expected, a downward trend in food consumption and related non-CO2 GHG emission difference after adding and subtracting 5% and 10% to PCCR, with increasing departures from the baseline case. Compared with the baseline case, a 10% reduction in the PCCR of each meat item (Table S2) enhances the impact of aging on meat consumption and related non-CO2 GHG emissions (Fig. 3a). The difference in total meat consumption between the aging and no aging scenario reaches 2,132,124 tons, leading to 12.7 Tg CO2eq non-CO2 GHG emissions reduction. The PCCR variation of ruminant meat (beef and mutton) contributed the most to these changes. This is because beef and mutton have the lowest PCCR at 57.9% and 56.8% in the baseline case, respectively (Table S2), implying the most significant difference in RMC patterns between aging and non-aging groups compared to the consumption of pork and chicken with higher PCCR between the two population groups (Table S2). Our results show that the 10% reduction of PCCR for beef and mutton consumption could decrease non-CO2 GHG emissions by 7.9 Tg CO2eq and 3.5 Tg CO2eq by 2020, respectively (Fig. 3a). In contrast, increasing 10% PCCR only reduces total meat consumption by 705,099 tons, which results in small decrease (6.9 Tg CO2eq) in non-CO2 emissions compared to the baseline scenario (Fig. 3a), indicating that more significant differences in consumption patterns between the aging and non-aging population tend to have more pronounced effects on both meat consumption and related non-CO2 GHG emissions. For food grains, the effect of PCCR switching of food grains on aging-induced food consumption-related non-CO2 emissions is less significant (Fig. 3b). In Table S2, we also present the PCCR from four surveys (CHNS, CNHS, CFCS, and UNIE) from 2004 to 2019 (SI Text 1). As seen, the PCCR does not alter markedly during this period, implying that the intergenerational difference in meat consumption is not significant.

Fig. 3: Difference in meat and grain consumption and related non-CO2 GHG emissions between SR and SCAS scenarios.
figure 3

a Differences of non-CO2 GHG emissions from meat consumption between adjusted SR- and SCAS- scenarios (bars scaled on the left Y axis). Emission difference is estimated by Diff_emi =EmiAging – EmiNoaging, where EmiAging and EmiNoaging stand for meat consumption-induced non-CO2 GHG emissions under aging (SR-) scenarios and no aging (SCAS-) scenarios, respectively. The left-to-right bars in each year represent the −10%, −5%, baseline, +5%, and +10% scenarios, respectively. The solid lines scaled on the right Y axis denote the difference in meat consumption between Aging and No aging scenarios, estimated by Diff_con = ConAging – ConNoaging, where ConAging and ConNoaging denote meat consumption under aging (SR-) and no aging (SCAS-) scenarios. Detailed scenario descriptions are presented in SI Text 4. b Difference of grain consumption (solid lines scaled on the right Y axis) and related non-CO2 GHG emissions (bars scaled on the left Y axis) between SR- and SCAS- scenarios.

Full size image

Projected non-CO2 GHG emissions associated with population aging from 2025 to 2050

To highlight the significance of population aging in altering food consumption and non-CO2 GHG emissions in the future, we projected food consumption and related non-CO2 GHG emissions from 2025 to 2050 at a 5-year interval under several scenarios (SI Text 5 and Figs. S7–S10) based on shared socioeconomic pathways30,31, referred to as SSP1-SSP5 scenarios. Food consumption-related non-CO2 GHG emissions from 2025 to 2050 tend to decline under all five SSP scenarios, with emissions in 2050 varying from 436.8 Tg CO2eq under the SSP1 scenario to 505.1 Tg CO2eq under the SSP5 scenario (Fig. S10).

To paint a general picture of the impact of the changes in the population aging on non-CO2 GHG emissions embodied in food consumption from 2025 to 2050, we set up constant aging structure scenarios (CAS or No aging scenario) corresponding to SSP1 to SSP5, referred to as SSP1_CAS to SSP5_CAS. The SSP_CAS scenarios assume a fixed aging rate (age structure) of 10% from 2025 to 2050. Other factors (total population, per capita food consumption for the elderly, and non-elderly population groups) under these scenarios vary with their respective SSP scenarios (SI Text 5).

Compared with food consumption under SSP Aging scenarios with variable age structure, higher food consumption (except for maize and dairy) can be identified in all five SSP_CAS scenarios, and the difference in food consumption between SSP and SSP_CAS scenarios grows from 2025 to 2050, indicating that the constant (fixed) age structure without future population aging would lead to more substantial food consumption (except for maize and dairy), and as a result, higher non-CO2 GHG emission. Taking beef consumption as an example, the SSP1 and SSP5 project the most rapid aging in China, with the proportion of the population of 60+ in the total population at 39.9% by 2050 (Fig. 4b). Under these two SSP scenarios, population aging leads to 1,920,605 tons and 2,011,719 tons reduction in total beef consumption in 2050, accounting for 15.1% and 15.2% of total beef consumption, respectively. Under the SSP3, with a relatively lower aging rate of 33.0% in 2050, aging will result in falling beef consumption by 1,360,282 tons (Fig. 4 and Fig. S11).

Fig. 4: Comparison of food consumption and related non-CO2 GHG emissions between the SSP (Aging) scenario and SSP_CAS (constant age structure or No Aging) scenario from 2025 to 2050.
figure 4

a Difference in total food consumption-induced non-CO2 GHG emissions (Tg CO2eq) between the SSP (Aging) scenario and SSP_CAS (No Aging) scenario from 2025 to 2050, estimated by Diff_emi = EmiSSP – EmiSSP_CAS, where EmiSSP stands for non-CO2 GHG emissions under SSP scenarios, and EmiSSP_CAS denotes non-CO2 GHG emission under SSP_CAS (constant age structure) scenarios, respectively. b Proportion of the population aged 60+ to the total population (%) for every 5 years in China under SSP1, SSP2, SSP3, SSP4, and SSP5 scenarios from 2025 to 2050; c difference of non-CO2 GHG emissions (Tg CO2eq) from RMC between the SSP5 scenario and SSP5_CAS scenario from 2025 to 2050 (color bars, scaled on the left Y axis). The solid line scaled on the right Y axis denotes the percentage changes in non-CO2 GHG emissions from RMC between SSP5 and SSP5_CAS scenario, estimated by Emifra = (Emi_rumiSSP - Emi_rumiSSP_CAS)/ Emi_rumiSSP × 100 (%), where Emi_rumiSSP stands for non-CO2 GHG emissions from RMC under SSP5 scenario, and Emi_rumiSSP_CAS denotes non-CO2 GHG emissions from RMC under SSP5_CAS scenario, respectively; d three factors driving the changes in non-CO2 GHG emissions from food consumption from 2025 to 2050 under SSP5 scenario, estimated by LDMI decomposition analysis.

Full size image

Figure 4 illustrates the differences in total non-CO2 GHG emissions from all food consumption (Fig. 4a), the proportion of the population aged 60+ to the total population (%, Fig. 4b), and the impact of population aging on non-CO2 GHG emissions under SSP5 scenario from 2025 to 2050 (Fig. 4c, d). The growing aged population of 60+ will play an increasingly important role in reducing non-CO2 GHG emissions from food consumption in the future. Again, the most notable reduction in non-CO2 GHG emissions occurs in the SSP5 scenario, where aging results in 34.5 Tg CO2eq reduction in total non-CO2 GHG emissions in 2050, accounting for 6.8% of total non-CO2 GHG emissions from China’s food consumption. This contribution should not be overlooked, considering China’s huge population and food consumption. The emission reduction subject to the other four SSP scenarios is 28.4 Tg CO2eq in SSP1, 28.1 Tg CO2eq in SSP4, 26.4 Tg CO2eq in SSP2, and 23.1 Tg CO2eq in SSP3, respectively (Fig. 4a). Notably, most of the total emission reduction can be attributed to the decreasing RMC induced by aging populations. For example, 86.8% of total emission reduction can be traced back to the reduced RMC, mainly due to beef’s high production-based emission intensity and decreasing RMC induced by population aging (Figs. S7 and S11). As shown in Fig. 4c, population aging leads to 29.9 Tg CO2eq reduction in the emissions from RMC in 2050 in the SSP5 scenario, accounting for 15.3% of total non-CO2 GHG emissions from China’s RMC.

To confirm the above results, we further apply LMDI (the Logarithmic Mean Divisia Index) decomposition analysis to explore the contribution of major driving factors to non-CO2 GHG emission under each SSP scenario (Fig. 4d and Fig. S12). The LMDI results show that the AAP overwhelms CBEI and CPAAP in the reduction of food consumption-induced non-CO2 GHG emission during 2045–2050 under all SSPs, of which the AAP yields the largest reduction of non-CO2 GHG emission by 47.4 Tg CO2eq under SSP5 scenario (Fig. 4d), followed SSP4 (43.4 Tg CO2eq) and SSP1 (41.6 Tg CO2eq), respectively. The contribution of AAP not only exceeds the effect of CBEI on non-CO2 GHG emission reduction (24.8 Tg CO2eq) but also cancels out the positive contribution of the CPAAP to non-CO2 GHG emission enhancement (43.5 Tg CO2eq under SSP5, for example). This is expected because the aged population in China is undergoing more rapid growth than the other driving factors, posing increasing impacts of aging on food consumption and related non-CO2 GHG emissions in the future.

Arguably, ppGDP is an important factor driving food consumption. Here, ppGDP was implicitly associated with CPAAP (“Methods”). To clarify, we take beef consumption-induced non-CO2 emissions under the SSP2 scenario to examine the contribution of major driving factors to non-CO2 GHG emission on a provincial level in China using a multiple regression model (MRM, SI Text 6). Instead of the AAP, the MRM considers an aging rate (the percentage of the population aged 60+ in the total population, AR), PBEI, the total population, and ppGDP across China from 2025 to 2050 (Fig. S13). The MRM result shows that the AR dominates temporally- and spatially altered non-CO2 GHG emission, contributing 59.4% to the provincial non-CO2 GHG emission, followed by ppGDP at 16.7% and the PBEI at 14.8%. The other factor is the provincial total population (POP), which accounts for about 9.2% of non-CO2 GHG emissions, considerably lower than population aging. This is expected because, as we mentioned in the Introduction, the AR in China has been undergoing more rapid change than the other driving factors. Table S3 presents regression coefficients and MRM statistics, including the model’s multicollinearity.

Impact of domestic migration

Considering the significant influences of population aging on food consumption and related non-CO2 GHG emissions, as well as domestic migration on age structure, we explore the consequences of changes in age structure induced by interprovincial migration from the perspectives of food consumption and related non-CO2 GHG emissions in China. We combined provincial population inflows (move-in) and outflows (move-out) and provincial population data to identify the changes in age structures with and without interprovincial migration in 31 provinces in China in 2020 (Fig. S14 and Table S4). Data were collected from the national census. The results show that the age structure changes induced by interprovincial migration significantly altered food consumption and related non-CO2 GHG emissions in most provinces. Non-CO2 GHG emissions (Fig. S15a) correspond well with the “move-in” immigrants (Fig. S15d) and the total population (Fig. S15c). Domestic immigrants, such as in the case of the Guangdong Province, also altered the latter. The evolution of non-CO2 GHG emission is not well associated with the provincial ppGDP (Fig. S15e). Since domestic migration altered the age structure significantly across China, those coastal provinces with high non-CO2 GHG emissions embodied in food, especially meat and rice consumption, correspond to a relatively low-aged population. Since the per capita food (except maize and milk) consumption of the population aged 60+ is lower than that of the population aged 60- (Table S2), an increase in the aging rate would decrease total food consumption and related non-CO2 GHG emissions. For instance, the proportion of the population aged 60+ to the total population increases from 16.5% under the “no migration” scenario to 18.1% under the “migration” scenario in Henan (Fig. S16), the province with the most significant net population outflow (Fig. S17), leading to −95.7 Gg CO2eq difference of non-CO2 GHG emissions between two scenarios. Likewise, similar situations occurred in Anhui, Sichuan, and Hunan. The increasing proportion of the population aged 60+ induced by interprovincial migration contributed 93.3 Gg CO2eq, 78.2 Gg CO2eq, and 68.1 Gg CO2eq to non-CO2 GHG emission reduction, respectively (Fig. 5).

Fig. 5: Provincial non-CO2 GHG emissions differences (Gg CO2eq) between the “migration” scenario and the “no migration” scenario in each province in 2020.
figure 5

a Non-CO2 GHG emission difference across China, estimated as non-CO2 GHG emissions under “migration” scenario minus non-CO2 GHG emissions under “no migration” scenario; b the percentage change in provincial non-CO2 GHG emission between “migration” scenario and “no migration” scenario. Positive differences are marked by the orange bar, and negative differences are marked by the blue bar. The full names of each province are presented in Table S4.

Full size image

On the contrary, the difference in non-CO2 GHG emission between the “migration” and “no migration” scenarios is positive in Beijing, Tianjin, northwestern, southern, and eastern China, particularly the eastern and southern seaboard areas of China in the Yangtze River Delta (YRD), including Shanghai, the largest megacity in China, and Zhejiang and Jiangsu provinces, and Pearl River Delta (PRD) or Guangdong province. These well-developed megacities and provinces have received the most young immigrants for decades. Interestingly, the highest percentage of the aged population of 60+ without considering migration across China occurred in Shanghai, reaching 35.3% (Fig. S16). However, the percentage of the population aged 60+ in Shanghai significantly fell to 23.4% under the “migration” scenario due to large young population inflows, leading to 334.2 Gg CO2eq of non-CO2 GHG emission increase (Fig. 5). Other details are discussed in SI Text 7.

Discussion

Our result reveals that the population aging-induced non-CO2 GHG emission reduction overwhelmed the effects of personal income and population. The rapid population aging in China could also alter the non-CO2 GHG emissions embodied in China’s food imports from overseas. Figure 6 shows the differences in beef consumption-induced non-CO2 GHG emissions between the Aging (SR-A) and No aging scenarios (SCAS-A) in major beef trade (exporting) countries. Negative values manifest the decline of non-CO2 GHG emissions embodied in beef consumption. Taking the beef imported from Brazil in 2020 as an example, our result shows that the population aging in China would reduce 1339.9 Gg CO2eq emissions in Brazil due to decreasing demand for imported beef products from Brazil at 33,847 tons. From this perspective, China’s population aging would benefit the mitigation of non-CO2 GHG emissions in those major beef-exporting countries as well. Our results suggest that population aging made an increasingly significant contribution to non-CO2 GHG emission reduction, which could offset the effects of population and personal income on non-CO2 GHG emissions from food consumption and demand. Such emission reduction associated with accelerated growth of the aged population would become more significant or even play a dominant role in the coming years, as shown in Fig. 4 and LMDI decomposition analysis (Fig. S12).

Fig. 6: Difference of non-CO2 GHG emissions (Gg CO2eq) embodied in beef consumption between the aging scenario and the no aging (CAS) scenario from 2005 to 2020.
figure 6

The dashed line shows the emission differences in China, scaled on the right Y axis, and solid color lines denote emission differences in five major beef-exporting countries to China and other countries with a small quantity of beef exported to China.

Full size image

Given the vast population and demand for domestic and global food products, the population aging in China might significantly affect CH4 and other GHG emissions from cattle farming and other agricultural activities. There is a need to integrate food security and climate policies to ensure that efforts to maintain food security for the aging population do not come at the expense of climate goals, including considering the carbon and methane footprints of different foods within food security strategies. Our results imply that to reduce global non-CO2 GHG emissions induced by food consumption in China, population aging should be considered in China’s long-term plans and strategies in the balanced and cost-effective mitigation of non-CO2 GHG emissions from its agriculture and food industry. The government could adjust agricultural policies to reflect the changing demand, such as reducing the emphasis on livestock production and redirecting subsidies and incentives towards agricultural practices and products that align with food security needs and non-CO2 GHG emission reduction targets. In addition, as food demand and related emissions decline due to population aging, agricultural land use policies could be adjusted to support carbon sequestration initiatives, such as afforestation or regenerative agriculture, further contributing to emission reductions. Integrating demographic shifts into long-term agricultural and climate planning would help ensure that emission reduction strategies align with expected changes in food consumption and production patterns, maximizing both environmental and economic benefits. To fulfill China’s carbon neutrality by 2060 (http://www.iea.org/reports/an-energy-sector-roadmap-to-carbon-neutrality-in-china) and to meet its food demand, expanding the cultivation area, enhancing agricultural output, and increasing the import of agricultural products from overseas have been China’s crucial agricultural strategies, which require substantial investments in agriculture resources. The present study suggests that the rapid population aging could help alleviate the pressure on China’s agricultural resources and ecological environment during the fulfillment of its carbon neutrality.

Methods

Non-CO2 GHG emissions embodied in food consumption

The total non-CO2 GHG emissions embodied in food consumption (EFC) are calculated by

$${\rm{EFC}}=\sum _{i}\sum _{j}({{PBEI}}_{{ij}-{CH}4}\times {{FC}}_{{ij}}+{{PBEI}}_{{ij}-N2O}\times {{FC}}_{{ij}})$$
(1)

where \({{FC}}_{{ij}}\) denotes the amount of food i supplied by a country j consumed in China (in tons of food), and \({{PBEI}}_{{ij}-{CH}4}\) and \({{PBEI}}_{{ij}-N2O}\) are CH4 and nitrous oxide (N2O) production-based emission intensity in food i supplying country j, respectively (in tons of emissions per ton of food production), defined as CH4 or N2O emissions per unit of a food product by country:

$${{PBEI}}_{{ij}-{CH}4}={{Emissions}}_{{ij}-{CH}4}/{P}_{{ij}}$$
(2)
$${PB}{{EI}}_{{ij}-N2O}={{Emissions}}_{{ij}-N2O}/{P}_{{ij}}$$
(3)

where \({{Emissions}}_{{ij}-{CH}4}\) and \({{Emissions}}_{{ij}-N2O}\) represent CH4 and N2O emissions emitted from food i production in the country j (in tons of emissions), respectively, and \({P}_{{ij}}\) is the amount of production of food i in the country j (in tons of food). CH4 and N2O emissions, food production, and food trade datasets in China are collected from the FAOSTAT database. CH4 and N2O are converted to CO2 equivalent (CO2-eq) according to their global warming potential with a time horizon of 100 years (GWP100-AR6, 27, and 273)32.

Drivers of non-CO2 GHG emissions

Following the widely used method in assessing carbon emissions and other resources, the Logarithmic Mean Divisia Index (LMDI) was employed to assess quantitatively and factorize the changes in food consumption-induced non-CO2 GHG emissions33,34,35,36,37,38,39,40,41,42,43,44. The total non-CO2 GHG emissions driven by food consumption (EFC) in China can be decomposed into three factors, defined as:

$${\rm{EFC}}=\sum _{i}\left(\frac{{{EFC}}_{i-{CH}4}}{{{TC}}_{i}}\times \frac{{{TC}}_{i}}{{AAP}}\times {AAP}+\frac{{{EFC}}_{i-N2O}}{{{TC}}_{i}}\times \frac{{{TC}}_{i}}{{AAP}}\times {AAP}\right)$$
(4)

where \({{\rm{EFC}}}_{{\rm{i}}-{\rm{CH}}4}\) and \({{\rm{EFC}}}_{{\rm{i}}-{\rm{N}}2{\rm{O}}}\) indicate CH4 and N2O emission embodied in food i consumption in China; \({{\rm{TC}}}_{{\rm{i}}}\) denotes the total food i consumption in China, AAP denotes the aging adjusted population, defined as the difference between the total population and the aged population of 60+ (total population minus population aged 60 + ).

We rewrite the three driving factors in Eq. (4) as

$${\rm{EFC}}=\sum _{i}\left({{\rm{CBEI}}}_{{\rm{i}}}\times {{\rm{CPAAP}}}_{{\rm{i}}}\times {\rm{AAP}}\right)$$
(5)

Equation (5) indicates that EFC is associated with (1) consumption-based emission intensity (CBEI) in China, defined as the amount of consumption-based non-CO2 GHG emissions (CH4 and N2O) per unit of food consumed (including both domestically produced food and food imported from other countries) in China, (2) food consumption per aging adjusted population (CPAAP), which is also associated with personal income or per person gross domestic product (ppGDP), and (3) the aging adjusted population (AAP) in China. A detailed LMDI analysis for quantifying the effects of these three drivers on non-CO2 GHG emissions is presented in SI Text 2. It is noted that since ppGDP also drives the market prices of a food item, the effect of food prices on food consumption and related GHG emissions is not investigated here. The model uncertainty and limitation are discussed in SI Text 3 and Fig. 1.

Scenario setup

To assess the impact of population aging in China on food consumption and related non-CO2 GHG emissions from 2005 to 2020, we first set up two scenarios with the annual change in aged population from 2005 to 2020 (referred to as SR-A) and fixed aging rate (the proportion of the population aged 60+ in the total population is fixed at 10%) for the same period (referred to as SCAS-A). Based on per capita consumption ratio (PCCR), which is defined as the proportion of per capita food consumption of the aging population to that of the non-aging population (Supporting Information (SI) Text 1 and Table S2), we estimated the consumption of each food item and related emissions in SR-A scenario and SCAS-A scenario to discern the signals of non-CO2 GHG emissions embodied in food consumption from the population aging.

We further introduced the Aging-related food Consumption Sensitivity Index (ACSI) and the Aging-related Emissions Sensitivity Index (AESI) for each food item to assess the sensitivity of changes in food consumption and related non-CO2 GHG emission to percentage changes in the aging rate:

$${ACSI}=\left({{Con}}_{R-A}-{{Con}}_{{CAS}-A}\right)/\left(\left({{AR}}_{R-A}-{{AR}}_{{CAS}-A}\right)/{{AR}}_{R-A}\right)$$
(6)
$${AESI}=\left({{Emi}}_{R-A}-{{Emi}}_{{CAS}-A}\right)/\left(\left({{AR}}_{R-A}-{{AR}}_{{CAS}-A}\right)/{{AR}}_{R-A}\right)$$
(7)

Where, \({{Con}}_{R-A}\) and \({{Con}}_{{CAS}-A}\) represent total food consumption under SR-A scenario and SCAS-A scenario, respectively, \({{Emi}}_{R-A}\) and \({{Emi}}_{{CAS}-A}\) are food consumption-induced non-CO2 GHG emissions under SR-A scenario and SCAS-A scenario, respectively, \({{AR}}_{R-A}\) and \({{AR}}_{{CAS}-A}\) denote aging rate (proportion of population aged 60+ to total population) under SR-A scenario and SCAS-A scenario, respectively.

Moreover, dietary patterns in different generations of the aged population might be subject to change; say, some older adults might switch their meat consumption to seafood or other animal products or become vegetarians in the future, which depends on complex factors and is hardly predicted. To collectively address the overall effect of the changes in dietary behavior, we adjusted the PCCR of each food item by adding or subtracting 5% and 10% to and from the baseline PCCR in SR-A. We recalculated the difference between SR-A and SCAS-A, thereby examining the sensitivity of food switching of older adults to non-CO2 emissions. Detailed descriptions of these scenarios are presented in SI Text S4.

We also set up five future scenarios to estimate the food demand and non-CO2 GHG emission between 2025 and 2050 associated with projected population aging in China on a 5-year interval subject to shared socioeconomic pathways (SSP1-5) under the Coupled Model Intercomparison Project (CMIP6) and the Intergovernmental Panel on Climate Change (IPCC)30,31. The scenarios range from sustainable development (SSP1) to high fossil fuel dependency (SSP5). We linked each driving factor with SSPs from 2025 to 2050. Detailed descriptions of each driving factor under five SSPs are provided in SI Text 5 and Figs. S7 and S8.

The contribution of interprovincial migration to population aging was considered in this study, though it is not included in the decomposition analysis. Over the past decades, China’s population has experienced significant domestic immigration among different provinces due to increasing gaps in economic and social conditions among provinces. Interprovincial immigration in China is featured by the young immigrants moving from less-developed southwestern, northeastern, and central China to well-developed eastern and southeastern China, particularly those coastal provinces, aiming to seek better education and living conditions45. As a result, such immigration would potentially lead to a growing young population and a reduction of the proportion of the aged population in the total population in those eastern seaboard areas of China, but it would also lead to growth in those less-developed regions. Other details are presented in SI Text 7.