Environmental Footprint and Nutritional Intake of 1,000 Cantonese and Sichuan Dishes in China

Abstract

Transforming diets toward sustainability requires more than broad recommendations, which begin with a detailed understanding of what people eat. In many cultures, especially in Asia, food is consumed in the form of complex, culturally specific dishes rather than isolated ingredients. Yet, most existing assessments of dietary sustainability remain disconnected from how meals are experienced in daily life. Here, we develop a comprehensive dataset capturing both the nutritional composition and environmental footprints of 1,000 recipes from Sichuan and Cantonese cuisines, two of the most influential and widely consumed culinary traditions in China. We combine ingredient-level nutritional analysis with life cycle assessments that account for emissions, land use, and water consumption across production, transportation, and cooking. Our data reveal large intra-cuisine variation in health and sustainability metrics, shaped by ingredient combinations and cooking techniques. By aligning real-world culinary practices with quantitative sustainability metrics, this dataset lays the groundwork for culturally grounded dietary transitions and supports a new generation of food policy, public guidance, and behavioral research rooted in dietary habits.

Background & Summary

Food systems have the dual capacity to enhance human well-being and mitigate environmental degradation, yet they currently exacerbate both^1,2. With the expanding global population and shifts in dietary preferences, food production and consumption patterns have become major drivers of environmental stress, intensifying land-use change, freshwater consumption, biodiversity loss, and greenhouse gas emissions (GHGs)³. Among these impact, land use and water footprint rank among the largest biophysical pressures exerted by modern food systems, with agricultural land conversion occupying nearly 40% of the ice-free terrestrial surface⁴ and food production accounting for roughly 70% of global freshwater withdrawals^4,5.

Concurrently, poor diets represent the leading global risk factor for morbidity and mortality, fueling the rise of chronic diseases such as diabetes, hypertension, and metabolic disorders^6,7,8. To address these global challenges, the EAT-Lancet Commission proposed the “Planetary Health Diet” framework in 2019, advocating for dietary transitions that align human nutritional needs with planetary boundaries¹. While this framework has inspired policy initiatives, particularly in high-income nations, the widespread adoption of sustainable dietary patterns remains constrained by deep-seated dietary habits, cultural traditions, and regional disparities in food availability^9,10. Achieving both human and environmental health, therefore, necessitates not only the broad implementation of evidence-based dietary guidelines but also context-specific adaptations that account for local food systems and sociocultural realities¹¹.

As previous research has pointed out, assessing the nutritional intake and environmental footprints of dishes enables a more precise understanding of how daily food choices impact both health and sustainability^12,13,14. Unlike aggregated food-group analyses, dishes represent the final form of food consumption, balancing the nutritional composition of multiple ingredients while also reflecting cultural preferences¹⁵. Estimating resource utilization and emissions at the dish level is particularly important for developing effective demand-side strategies to reduce food system emissions. While some recent studies have begun to evaluate diets at the dish level, most have not jointly estimated nutrient intake and environmental impacts in a culturally specific context^16,17,18,19. In practice, sustainable food consumption should consider cultural acceptability, nutritional adequacy, and environmental sustainability²⁰. Advancing dish-level assessments of nutrition and environmental footprints across diverse dietary cultures will enhance the applicability of sustainable diets, making them more accessible and actionable for individuals and communities.

As one of the world’s most populous nations and a major food consumer, China possesses a highly diverse food culture shaped by distinct regional differences. The country’s long-standing adaptation to its natural environment and the strategic utilization of local resources have fostered the development of eight major regional cuisines, namely Shandong, Sichuan, Cantonese, Jiangsu, Fujian, Zhejiang, Anhui, and Hunan cuisines, each reflecting deep-rooted culinary traditions^21,22. Among them, Sichuan and Cantonese cuisines stand out due to their wide popularity and rich culinary diversity. These two cuisines are among the most representative and frequently consumed in China, covering both inland and coastal food systems^23,24,25. For instance, Sichuan cuisine skillfully employs spicy, aromatic seasonings to balance the humid, hot climate, using traditional fermentation techniques to extend ingredient shelf life and create a unique eco-adaptation system²⁶. In contrast, Cantonese cuisine draws on the advantages of the Pearl River Delta and abundant coastal produce, using short supply chains to highlight pure ingredient flavors while modestly lower logistical energy demands²⁷.

While these regional food cultures represent a valuable heritage, some of their culinary traditions have raised concerns regarding public health and environmental sustainability. For instance, the heavy seasoning characteristic of Sichuan cuisine often results in high sodium intake²⁸, while the frequent use of rich meats and slow-cooking techniques in Cantonese cuisine might lead to elevated levels of purines and fats^28,29. So far, extensive epidemiological evidence has established strong associations between high-fat, high-sodium diets and conditions such as hypertension, while excessive consumption of purine-rich foods is linked to increased risks of cardiovascular diseases, diabetes, and gout^30,31. Beyond their health implications, the diversity in cooking techniques, ingredient choices, and dietary preferences across these regional cuisines also results in notable variability in their environmental footprints^32,33. Given that food is rarely consumed in isolation but rather in the form of prepared dishes, analyzing food sustainability at the dish level is crucial for both scientific accuracy and public engagement. By focusing on these two representative Chinese cuisines, insights can be obtained into how culturally embedded food choices shape both human and planetary health.

To bridge the gap between global sustainability goals and culturally grounded dietary realities, our dataset provides an integration of nutritional and environmental indicators at the dish level. Through diversified data collection methods, including traditional publications, the internet, and market surveys, 1,000 recipes for Sichuan and Cantonese cuisines were gathered to construct an ingredient-level recipe database. Our dataset quantifies the emissions associated with regional-specific cooking methods, which is a relatively underexplored focus in existing research³⁴. By integrating data from standardized documents and markets, our dataset includes the emissions produced by common kitchen appliances such as gas stoves, induction cookers, ovens, rice cookers, pressure cookers, and refrigerators. In addition, due to the unique characteristics of production processes, energy structures, and combustion efficiency, which vary in different countries^35,36, we used localized environmental impact data to accurately reflect the environmental footprints of Chinese cuisines. By revealing the connections between food choices and dietary culture, this dataset provides insights for designing sustainable and culturally appropriate dietary practices.

Methodology

Recipe collection and ingredient data standardization

Before accounting for the nutritional and environmental trade-offs for each recipe, a systematic data collection approach was employed to compile the recipe database of Sichuan and Cantonese cuisines, covering traditional publications^{37,38,39,40,41} and online platforms^42,43,44. A rigorous screening was applied to ensure research relevance and quality standards, with a data cutoff date of December 31, 2024. A total of 1,000 representative recipes (500 from Sichuan cuisine and 500 from Cantonese cuisine) were chosen, covering categories such as cold dishes, soups, meats, seafood, poultry, tofu products, vegetables and fungi, shark fin and abalone, staple foods, and desserts. These recipes collectively involved approximately 730 unique standardized ingredients.

To enhance the data consistency and accuracy, vague descriptions (e.g., “appropriate amount”, “a little”, etc.) in the collected recipes were standardized by assigning precise numeric values. The standardization process relied on large language models and practical culinary experience, supplemented by culinary articles published on online platforms⁴⁵. Additionally, the ingredient names in the original data were standardized to ensure consistency. Duplicates were removed, errors were corrected, and formatting was unified. The ingredients were then mapped to relevant food categories in the Chinese Nutrition Society’s Food Composition Database⁴⁶ and the Life Cycle Assessment (LCA) database from the National Institute of Advanced Industrial Science and Technology⁴⁷ for obtaining nutrition and environmental impact data. This standardization process provided a robust foundation for subsequent quantitative analysis, environmental impact assessments, and health-related research.

Nutritional components estimation

To estimate the nutritional content of each dish, this study utilized standardized food composition tables along with the ingredient weights specified in each recipe. The primary data sources included not only the Food Composition Database but also the Eshian Food Database⁴⁸, with nutritional equivalents substituted where necessary for less common ingredients. A total of 28 nutritional indicators were estimated, covering energy, protein, fat, carbohydrates, cholesterol, total dietary fiber, vitamins, and minerals, as presented in Table 1.

Table 1 Type and unit of nutrients.

The nutrient content calculation is based on the edible portion of each ingredient, where non-edible portion of the ingredient (e.g., skin, bones, etc.) is excluded from the total weight W_edible (g), as Eq. (1):

$${W}_{{edible}}=W\times \frac{E}{100}$$

(1)

where W represents the total weight (g) of the ingredient, and E is the proportion of the edible part, sourced from the Food Composition Database. The actual contribution of each nutrient was calculated as Eq. (2):

$${N}_{{actual}}=\frac{{N}_{{standard}}\times {W}_{{edible}}}{100}$$

(2)

in which N_standard denotes the nutrient content per 100 grams of ingredients in the standard food ingredient list. To unify the units, convert the energy value from kilojoules to kilocalorie.

Finally, the total nutritional value of the dish was calculated by Eq. (3) by summing the contributions from all ingredients:

$${N}_{{dish}}=\sum _{i}{N}_{{actual}}$$

(3)

This approach ensured accurate estimation of the energy, protein, fat, vitamins, minerals, and other nutritional components of each dish, providing robust data and scientifically grounded assessment for subsequent sustainability analyses.

Carbon footprint accounting

Carbon footprint accounting is one of the core aspects of this study, aiming to quantify the GHG emissions associated with the production, transportation, and cooking processes of ingredients.

Carbon footprint embodied in food ingredients

The embodied carbon footprint of each ingredient from collected recipes was assessed using LCA, including production, manufacturing, and transportation stages. The data source for the carbon emission factors is the LCA database from the National Institute of Advanced Industrial Science and Technology⁴⁷, which provides global warming potential (GWP) values converted to CO₂ equivalents based on the IPCC 6th Assessment Report⁴⁹. This database ensures region-specific accuracy, reflecting China’s unique energy structure, production processes, and combustion efficiency⁵⁰.

The carbon footprint of each ingredient is calculated by matching it to the corresponding food category in the LCA database and using the emission factor for that category. The carbon footprint ${E}_{g}$ (kg CO₂eq) of each ingredient is calculated as Eq. (4):

$${E}_{g}=\sum _{i}{Q}_{i}\times {I}_{g,i}$$

(4)

where Q_i is the weight of ingredient i (kg) and I_g,i is the emission intensity of ingredient i (kg CO₂eq/kg).

Carbon emissions from cooking process

This study also quantifies the carbon emissions from various cooking methods. Emission factors were sourced from the latest national and regional carbon emission data provided by the Ministry of Ecology and Environment of the People’s Republic of China, as well as publicly available energy emission coefficients^51,52. The national average CO₂ emission factor for electricity is 0.6205 kgCO₂/kWh, with carbon emission coefficients for natural gas and coal gas being 2.8603 kg/m³ and 0.8162 kg/m³, respectively. Below, we quantify the carbon footprint of five of the most common cooking styles in Chinese households.

Gas stove

As one of the most widely used kitchen appliances, a gas stove’s carbon emissions depend on fuel type, thermal efficiency, and load conditions. Given its significant market share, natural gas stoves were selected as the reference for calculations. The thermal efficiency ${E}_{{gs}}$ of the gas stove is assumed to be 59%, based on the GB 30720-2014 standard from the National Standards of the People’s Republic of China (GB) for household gas appliances. The rated thermal load ranges from 3.5 kW to 5.23 kW, and a value of 4.365 kW was used. The gas consumption, based on the lower heating value of natural gas ${H}_{{ng}}$ (37.62 MJ/m³), is calculated as Eq. (5):

$${Q}_{{gas}}=\frac{{Q}_{{rated}}}{{E}_{{gs}}\times {H}_{{ng}}}$$

(5)

here, Q_rated is the rated heat load converted to the energy consumed per minute (m³/min). On this basis, combined with the carbon emission factor of natural gas 2.8603 kg/m³, the carbon emissions of the gas stove are further estimated under both high-flame and low-flame modes.

Although the above analysis is based on the average performance of gas stoves with a level 2 energy efficiency rating, actual carbon emissions may vary due to differences in appliance models, fluctuations in thermal efficiency, and specific usage scenarios. Factors such as cooking duration, ambient temperature, and the cooking requirements of different ingredients can influence the actual gas consumption and carbon emissions.

Induction cooker

With the increasing electrification of kitchen appliances, induction cookers have gradually gained popularity. This study further examines the carbon emission characteristics of induction cookers. According to Household Induction Cooker Energy Efficiency Limits and Levels (GB 21456-2014) standard, the rated power range of induction cookers is 700 W to 3500 W. Generally, the cookers are divided into two categories, determinized by greater than 1200 W and less than or equal to 1200 W. Based on market data for typical home induction cookers, an average power of 1200 W was used. With this power, the electricity consumption per minute is:

$${Q}_{{elec}}=\frac{{P}_{{avg}}}{T}$$

(6)

where T denotes the cooking time when using the induction cooker.

The carbon emissions are calculated based on the electricity consumption and the electricity carbon emission factor (F = 0.6205 kgCO₂/kWh) as Eq. (7):

$$E={Q}_{{elec}}\times {\rm{F}}\times {10}^{3}$$

(7)

Thus, the carbon emissions per minute for full power (100%) and low power (25%) are estimated separately as: E_high = 0.02 × 0.6205 kgCO₂/kWh × 10³ = 12.41 gCO₂/min and E_low = 0.02 × 0.25 × 0.6205 kgCO₂/kWh × 10³ = 3.1gCO₂/min.

Oven

Ovens are commonly used in daily cooking, and their emissions depend on power, usage time, and operation mode. Although China does not yet have mandatory energy efficiency standards for ovens, the industry standards (QB) QB/T 5625-2021 Energy Efficiency Limits and Levels for Electric Ovens provides guidelines. Ovens typically have a power range of 800 W to 2500 W, with fluctuations in power usage between 50% and 90% of the rated power. This study used a market survey to estimate an average value of 1650 W as the reference power. The electricity consumption per minute is:

$${Q}_{{elec}}=\frac{{P}_{{avg}}\times 70 \% }{60\min }=0.01925{\rm{kWh}}/\min $$

(8)

Using the electricity carbon emission factor, the carbon emissions per minute are calculated as:

$${E}_{{oven}}={Q}_{{elec}}\times F\times {10}^{3}$$

(9)

Rice cooker/Pressure cooker

Rice cookers and pressure cookers, due to their versatile functions and high frequency of use in households, are key contributors to kitchen appliance carbon emissions analysis. According to the Energy Efficiency Limits and Levels for Rice Cookers (GB 12021.6-2017), the rated power of rice cookers does not exceed 2000 W. Based on market surveys, the power range for typical kitchen appliances is generally between 300 W and 1000 W. An average value of 650 W was used for the calculation. The energy consumption per minute is:

$${Q}_{{elec}}=\frac{{P}_{{avg}}}{60}=0.010833{\rm{kWh}}/\min $$

(10)

The corresponding carbon emissions, using the electricity carbon emission factor E_cooker can also be calculated similar as Eq. (9).

Compared to other high-energy-consuming appliances such as ovens, rice cookers and pressure cookers generate lower carbon emissions per minute. Additionally, these devices typically achieve high energy efficiency during use. For example, pressure cookers utilize increased internal pressure to elevate temperature, enabling faster cooking with reduced energy consumption. Similarly, modern rice cookers use smart temperature control and multiple cooking modes to further optimize energy usage. These features not only reduce carbon emissions but also enhance cooking efficiency, making them ideal choices for low-carbon kitchens.

Refrigerator

Although refrigerators are not directly used in cooking, their cooling function is essential before and after cooking. Refrigerators are typically classified by volume into small, medium, and large sizes. Small refrigerators generally have a power rating not exceeding 100 W, medium refrigerators range from 100 W to 200 W, and large refrigerators range from 200 W to 400 W. The power of typical household refrigerators ranges from 90 W to 200 W, rarely exceeding 300 W.

According to the national mandatory standard Energy Consumption Limits and Energy Efficiency Levels for Household Electric Refrigerators (GB 12021.2-2015), the energy consumption of refrigerators is divided into baseline consumption, load-related consumption, and overall consumption. For simplification, the reference value was set to 150 W, which is the middle range for a medium-sized refrigerator. The energy consumption per minute is as Eq. (10), and the corresponding carbon emissions follows Eq. (9).

It should be noted that the energy consumption of refrigerators is not solely for cooking-related cooling. For instance, refrigerators may store various ingredients beyond those required for cooking. Household refrigerators typically have both refrigeration and freezing compartments, with the power for the freezing section approximately 1.5 times that of the refrigeration section. Since freezing is not involved in the cooking process of the recipes, the cooling emission factor E_c for the refrigeration function is set at 0.4, with a usage rate R of 5% based on volume. Thus, the carbon emissions related to cooling for cooking purposes are calculated as:

$${E}_{{cooling}}={E}_{{fridge}}\times R\times {E}_{c}$$

(11)

Water footprint accounting

To assess resource use efficiency and ecological pressure in environmental impact analysis, this study utilizes water footprint factors from the LCA database to analyze water use for each recipe. The quantification of water footprint factors involves three categories: withdrawal water, consumptive use, and non-consumptive use during the production and processing of ingredients. Withdrawal water refers to the total amount of water extracted from natural resources, while consumptive use includes water contained in the product, evaporation, and plant transpiration during agricultural production. Non-consumptive use refers to the portion of water that is not consumed and can be reused.

The water footprint of a dish is calculated using Eq. (12):

$${E}_{w}=\sum _{i}{Q}_{i}\times {I}_{w,i}$$

(12)

where E_w represents the water footprint of the dish (m³), Q_i is the amount of ingredient i (kg), and ${I}_{w,i}$ is the water footprint factor for ingredient $i$ (m³/kg). This factor accounts for the flow of water used in the production process, including withdrawal, consumptive, and non-consumptive uses. Consumptive uses (e.g., evaporation and plant transpiration) are treated as major environmental stressors, while non-consumptive use (e.g., return flow) is considered less impactful.

The water resources model assumes a balance, where the total withdrawal water equals the sum of consumptive and non-consumptive water use:

$${W}_{{withdrawal}}={W}_{{\rm{consumptive}}}+{W}_{{\rm{non}}-{\rm{consumptive}}}$$

(13)

Each ingredient’s water footprint factor is calculated by considering all relevant unit processes. For example, in agriculture, irrigation water is primarily sourced from surface water and groundwater, with rainwater contributing directly to crop growth. The model allocates total water usage to consumptive and non-consumptive categories based on agricultural irrigation efficiency and crop evapotranspiration data.

The calculation formula for the water footprint factor is:

$${I}_{w,i}=\frac{{W}_{{withdrawal},i}-{W}_{{\rm{non}}-{\rm{consumptive}},i}}{{P}_{i}}$$

(14)

where ${W}_{{withdrawal},i}$ is the total water required for producing ingredient $i$, including surface water, groundwater, rainwater, and seawater; ${W}_{{non}-{consumptive},i}$ is the non-consumptive water that can be reused; and ${P}_{i}$ is the production yield of ingredient $i$ (kg). For example, producing 1 kg of rice requires 7.5 m³ of water, of which 6.3 m³ is non-consumptive (e.g., irrigation water return) and 1.2 m³ is consumptive (e.g., evaporation and plant absorption). Therefore, the water footprint factor ${I}_{w,{rice}}$ for rice is: ${I}_{w,{rice}}=\frac{7.5{{\rm{m}}}^{3}-6.3{{\rm{m}}}^{3}}{1{\rm{kg}}}=1.2\,{{\rm{m}}}^{3}/{\rm{kg}}$.

For other ingredients, the model uses specific industry data to refine calculations. For instance, the ratio of surface water to groundwater for tap water usage might be 77% to 23%, with each portion allocated between consumptive and non-consumptive use. The model further clarifies the specific use of water services, such as for potable or industrial water, to define the pathways for water resource utilization.

Land footprint accounting

As agriculture is a primary driver of land transformation, particularly in crop production, variations in planting methods and land use efficiency influence the extent of environmental impact. The land footprint reflects the human occupation of biological resources through land use. Evaluating the land footprint associated with a dish can help quantify the long-term environmental impact of food production, providing scientific evidence for optimizing dietary structures and promoting sustainable development.

Specifically, the land footprint of a dish is evaluated based on the amount of raw materials required and the corresponding land transformation factor, as shown in Eq. (15):

$${E}_{l}=\sum _{i}{Q}_{i}\times {F}_{l,i}$$

(15)

where ${E}_{l}$ represents the land footprint of the dish (m²), ${Q}_{i}$ is the amount of ingredient $i$ (kg), and ${F}_{l,i}$ is the land transformation factor of ingredient $i$ (m²/kg).

The land transformation factor ${F}_{l,i}$ reflects the land resource demand per kilogram of raw material, typically based on permanent or long-term land use changes that occur during production. For example, this includes the conversion of forests into farmland or wastelands into pasture. The calculation process is shown in Eq. (16):

$${F}_{l,i}=\frac{{A}_{l,i}}{{Y}_{i}\times {T}_{i}}$$

(16)

where ${A}_{l,i}$ is the land area (m²) used in the production of raw material $i$, ${Y}_{i}$ is the annual yield of raw material $i$ (kg/year), and ${T}_{i}$ is the time period (years) over which land transformation occurs for the production of material $i$. For example, the land transformation factor for rice production is determined by the cultivated area, the rice growth cycle, and the rice yield per unit area. Due to the longer growth cycle and limited yield per unit area, the land transformation factor for rice cultivation is typically high. In contrast, vegetable cultivation generally has a lower land transformation factor, reflecting higher land use efficiency.

The land use database used in this study incorporates official data sources such as the 2015 Crop Statistics and 2015 Livestock Production Cost Statistics published by Japan’s Ministry of Agriculture, Forestry, and Fisheries^53,54. These datasets were utilized to refine land transformation factor calculations across various agricultural activities.

Uncertainty analysis

To quantify the impact of parameter variability on our dish-level nutrient and environmental estimates, we performed a Monte Carlo simulation. Five groups of parameters, including ingredient nutrient concentrations, environmental-impact coefficients, edible-portion fractions, cooking times, and appliance emission factors, were independently perturbed via normal distributions (σ = 10% for nutrient and impact data; σ=5% for edible-fraction, time, and emission-rate data). For each of the 1000 recipes, 1000 iterations were executed in Python, yielding distributions of dish-level outcomes. From these simulations, we extracted means, 95% confidence intervals, and kernel density estimates for all 28 nutrient and three environmental indicators (see FigShare⁵⁵ for the simulation code and results).

Data Records

The database is stored and published, available on FigShare⁵⁵ as a folder with several spreadsheets. The files named ‘Sichuan cuisines.xlsx’ and ‘Cantonese cuisines.xlsx’ contain information on 500 recipes from Sichuan and Cantonese cuisines, respectively, including nutritional content and environmental impact data. The recorded cuisines are categorized across a wide range of dishes, including beef, pork meat, livestock meat, soy products, cold dishes, soups, seafood, poultry, vegetables, cubilose and abalone, staple foods, and desserts. In total, approximately 730 unique standardized ingredients were used in the recipes, with ingredient names standardized and mapped to standard food categories in ‘Portions and Weights.xlsx’. The data collection process was completed by December 31, 2024, ensuring that all recipes reflect up-to-date culinary practices. Nutritional components, including energy, protein, fat, carbohydrates, vitamins, and minerals, were calculated for each dish based on the edible portions of ingredients. Environmental impact data was gathered, including the carbon footprint, water footprint, and land footprint for each recipe, with the carbon footprint covering the entire lifecycle of ingredients, including production, transportation, and cooking processes. This dataset is structured in Excel format and is available for academic research and further sustainability analyses.

Technical Validation

To validate the reliability and consistency of our dataset, we presented the nutritional profiles (Fig. 1) and compared the per-serving environmental footprints of Cantonese and Sichuan cuisines (Fig. 2) with data from previously published studies and databases (Table 2). To quantify uncertainty in our dietary and footprint estimates, we performed 1000-run Monte Carlo simulations for each dish (see Methodology). Inputs, including ingredient nutrient contents, environmental coefficients, edible-portion fractions, cooking times, and appliance emission factors, were perturbed by ±10% or ±5% (normal distribution). Figures S1–S4 present the resulting 95% confidence intervals and density estimates, confirming that our core findings are robust to plausible variations in parameters. Specifically, the results of other nutrients are presented in Figures S5, S6.

Fig. 1

Per-Serving Nutrition Intake in Cantonese and Sichuan Cuisines.

Fig. 2

Per-serving environmental footprints in Cantonese and Sichuan cuisines.

Table 2 Carbon footprints for dishes from previous studies.

The nutrition intake validation is conducted by referring to previous research^56,57, which indicates that per-serving energy content in main dishes ranges from 137 to 261 kcal, protein from 8.8 to 118.4 g, fat from 6.1 to 17.3 g, and carbohydrates from 0.6 to 7.3 g. This comparison confirms that the protein, vitamin, and carbohydrate profiles of Cantonese and Sichuan dishes broadly align with the general nutrient ranges reported in multi-country analyses. In addition, we normalized every recipe’s nutrient content to a common baseline of 2,000 kcal so that all micronutrients could be benchmarked against the Chinese Dietary Reference Intakes (DRI)⁴⁶ regardless of the original portion size, as shown in Figure S7. The resulting density profiles (e.g., manganese, selenium, zinc, calcium, vitamin C, and Vitamin A) fall well within the recommended 80-120% DRI window for both cuisines, further confirming the plausibility of our nutrient accounting.

As shown in Fig. 2, the carbon footprint values for the sampled Cantonese and Sichuan dishes in our study fall predominantly within the range of 400-1,500 gCO₂eq per serving, with the average carbon footprint for all dishes at 2,692.54 gCO₂eq. We then quantitatively compared the per-100g environmental footprint values of Cantonese and Sichuan cuisines with those from previous studies. By reporting an average carbon footprint of 368.97 gCO₂eq/100 g across 1,000 sampled dishes, ranging from 36.84 to 1,924.45 gCO₂eq per 100 g.

To validate our findings, we compiled carbon footprint data from previous studies alongside the AGRIBALYSE v3.1 database^{17,58,59,60,61,62,63}, standardizing all values to g CO₂eq per 100 g. Table 2 presents the minimum, maximum, and mean footprints reported by previous research. As shown in Fig. 3, the distribution of our 1,000 Cantonese and Sichuan dishes (mean 369 g/100 g, shaded red) falls squarely within the envelope defined by these earlier works (overlaid colored bands and lines). This close alignment across multiple international datasets and methodological frameworks reinforces the robustness and generalizability of our carbon-footprint estimates.

Fig. 3

Comparative distributions of carbon footprints per 100 g across dishes in this study and other literature sources.

In addition, our dataset incorporates water footprint estimates, a parameter less frequently detailed in comparable works. Our dataset’s water footprint range (8.43-108.96 L per 100 g) is generally consistent with the findings of Frederic et al.⁶⁴, which reported a wider range of 12.88-2,066.42 L per 100 g for Chinese takeaway dishes. For water footprint in AGRIBALYSE v3.1, they report a median of approximately 112 L per 100 g (using scarcity-weighted m³ world-eq), whereas our Cantonese and Sichuan dishes range from 8.4 to 108.9 L per 100 g. The higher AGRIBALYSE values reflect its scarcity-weighted methodology, but both datasets remain within the same order of magnitude (10¹-10³ L vs. 10¹-10² L), indicating reasonable consistency.

The discrepancies might be attributed to that our dataset includes a variety of smaller, simpler dishes, including cold appetizers and side dishes, which generally have a lower water footprint per 100 g. While our dataset includes estimates for carbon and water footprints, we note the absence of comparable land use data in previous studies. Unlike carbon and water footprints, land use footprint estimates for cuisines remain scarce, limiting direct cross-validation.

Overall, these cross-references to prior research demonstrate that our estimates of environmental footprints and nutritional content fall within the broad spectrum of published values. Minor discrepancies likely reflect methodological differences, such as system boundaries, functional units, or underlying life cycle data, as well as region-specific factors related to Chinese culinary practices. Nonetheless, the observed convergence of our results with existing literature supports the robustness and applicability of the dataset in analyzing the environmental and nutritional implications of traditional Chinese cuisines.