Land-use-driven biodiversity impacts of diets—a comparison of two assessment methods in a Finnish case study

Agricultural land use and land-use change, especially from forests to agricultural land, to satisfy growing demand for food and feed, is a major cause of global biodiversity loss. International trade connects food consumption to land use across the world, leading consumption in one area to affect ecosystems elsewhere. However, methods for evaluating the effects of food consumption on biodiversity are still under development. Here, we evaluate two recent land-use-based biodiversity life cycle impact assessment methods, using the Finnish diet as a case study. Two different land-use-based biodiversity impact assessment methods (Chaudhary and Books 2018; Kuipers et al. 2021) were used to evaluate the biodiversity impacts of five dietary scenarios: the current Finnish diet and four alternative scenarios that involve a gradual reduction in the intake of foods of animal origin. The assessment was conducted using the previously developed FoodMin model. The model assesses the climate impact and nutritional quality of diets based on 90 food product groups. The countries of origin for each product group, as well as the land occupation and land-use change associated with the products, were determined using five-year averages from national import and international yield and land-use change statistics. The results showed that the biodiversity impacts vary depending on the assessment method used, with the difference in the dietary impacts being 60-fold in magnitude depending on the method. Most of the impacts were related to land-use change, which was linked especially to production of feeds, leading to the dietary impacts being reduced with intake of foods of animal origin. The occupation impacts were associated with different food groups than those of land-use change and were higher than the land-use change impacts for beverages, and sugars and sweets. Trade played a significant role in biodiversity impacts of diets, with over 85% of impacts being linked to imported foods and feeds. A comparison of the two assessment methods for the biodiversity impact of diet scenarios showed that results obtained with these methods should not be compared in absolute terms. Also, on a product-level basis, the methods lead to different relative results, and the choice of method can affect which products appear the most burdensome. The assessment also showed that land-use change had a greater impact than land occupation. Thus, ignoring the impacts of land-use change, particularly in feed production, might underestimate the contribution of animal-source foods. Hence, both factors should be considered in future LCAs. Further research is also needed to develop and unify the biodiversity impact assessment methods, including the clarification of units, improving the accuracy of land-use change methodology, and analyzing the impacts of various agricultural management practices.


Introduction
The rapid decline of biodiversity has become a major concern, with estimates suggesting that the global rate of species extinctions is tens to hundreds of times higher than has been recorded over the past 10 million years, and around one million species are at risk of extinction in the coming decades (IPBES 2019 Newbold et al. 2015), and thus, global as well as national goals concerning land use have been set to stem biodiversity loss. For example, the target to conserve at least 30% of terrestrial and aquatic areas was agreed upon in the Kunming-Montreal Global Biodiversity Framework at COP15 (CBD 2022). Agricultural expansion, in turn, is the most common cause of land-use change (IPBES 2019;Newbold et al. 2015), making agriculture the largest contributor to biodiversity loss (Bjelle et al. 2021;Wilting et al. 2021). Therefore, methods to assess the impacts on biodiversity of food production and consumption are needed to identify the potential measures to reduce biodiversity loss. Life Cycle Assessment (LCA) (ISO 2006) is a commonly used method to assess the environmental impacts of products and produce information for decision-making in various situations. For food products, most of the environmental impacts are associated with the agricultural stage of the life cycle (Poore and Nemecek 2018). However, in general, methods for biodiversity impact assessment are still under development and are incomplete in terms of coverage and level of detail (Marques et al. 2021;Verones et al. 2020).
Land-use-based methods have been among the most developed biodiversity impact assessment methods in the LCA context (Crenna et al. 2020). For example, the UNEP-SETAC Life Cycle Initiative (Jolliet et al. 2018) recommends using a land-use-driven method developed by Chaudhary et al. (2015) (updated by Chaudhary and Brooks 2018) to assess the land-use-driven global biodiversity impact of a product in LCA, prior to more advanced methods becoming available. A more recent assessment method for land-use-driven biodiversity impacts was developed by Kuipers et al. (2021).
The method of Chaudhary et al. (2015) and Chaudhary and Brooks (2018) provides country-specific characterization factors to assess biodiversity impact for different land uses, as well as land-use change, in terms of potentially disappeared fraction of species (PDF). The method of Chaudhary and Brooks (2018) includes three land-use intensities (minimal, light, and intensive) under every land-use class, and their impacts on mammals, birds, amphibians, reptiles, and plants. Following the UNEP (2017) recommendations to improve biodiversity assessment methods, Kuipers et al. (2021) included the fragmentation impact of habitats into the method. For that, they use the species-habitat relationship (SHR) model instead of the species-area relationship (SAR) model used by Chaudhary et al. (2015) and Chaudhary and Brooks (2018) to produce the characterization factors. However, the landuse classification in Kuipers et al. (2021) is broader and it includes fewer taxa; the method covers only the broad land-use types without intensity classes and their impact on vertebrates (amphibians, birds, non-flying mammals, and reptiles).
In both methods, the modeled biodiversity impact of agricultural production on species is directly dependent on the local species richness (Chaudhary and Brooks 2018;Marques et al. 2021). Therefore, when assessing the biodiversity impacts of food, the country of origin for each product needs to be systematically included. In addition to production site, also the amounts of products consumed are relevant when assessing food-related impacts. Diet-level impact assessments are able to consider both aspects, but previous diet-level biodiversity assessments (e.g., Crenna et al. 2019;Walker et al. 2018) have mostly neglected the location of production because they only apply general global characterization factors without spatial differentiation.
The need for dietary change has been recognized to improve both environmental and human health (Willett et al. 2019), but the biodiversity impacts of different dietary patterns have not been evaluated in previous diet-level assessments, apart from the study on average diets of a European sample population (Walker et al. 2018).
In this study, we assessed the land use and consequent biodiversity impacts of current and four alternative Finnish diet scenarios using two different impact assessment methods (Chaudhary and Brooks 2018;Kuipers et al. 2021). To our knowledge, these two methods have not yet been compared in any LCA case study. We evaluated how these assessment methods handle the variety of products originating from different locations and production systems, which include different land-use types and impacts arising from land use as well as land-use change. By comparing two recent methods, we aimed to assess the status and capability of land-use-driven biodiversity impact assessment methods in LCA and provide recommendations for future research.

Materials and methods
The land use and biodiversity impacts associated with the Finnish diets were done following the methodological procedure summarized below. The more detailed description of the procedure is presented in the following subsections: i) The dietary consumption data for 90 food groups in the assessed diet scenarios were acquired from the FoodMin dietary model (Sect. 2.1.) ii) The import quantities and countries of origin of each product group were traced (Sect. 2.2), and the land use and land use change related to production of each product group in each import country were assessed (Sects. 1 3 iv) The land use and land use change were characterized using the factors provided by Kuipers et al. (2021) and Chaudhary and Brooks (2018) (Sect. 2.2.), and the land use and biodiversity impacts of each product from each import country were aggregated through a weighted average calculation based on import quantities. v) The impacts of both imported and domestic products were integrated into the FoodMin dietary model, which calculates the final results based on consumption amounts and degrees of domesticity of each product group.

The diet scenarios
The current Finnish diet and four alternative diet scenarios were considered in the assessment, including "meat to half diet," "meat to a third diet," "a diet rich in fish and milk," and "a vegan diet." The dietary scenarios were derived from the FoodMin dietary model that has been developed in our previous work to simultaneously assess nutritional quality in relation to the Finnish nutritional recommendations (VNR 2014) and climate impacts of diet (Saarinen et al. 2019). In the diet scenarios, meat intake has been gradually reduced, and the rest of the diet has then been adjusted to meet the recommended intake of nutrients and food-based nutrition recommendations. Food consumption for the current adult diet is based on the FinDiet 2017 survey (Kaartinen et al. 2020;Valsta et al. 2018) and for other age groups data reported in earlier studies (Hoppu et al. 2008;Kyttälä et al. 2008;Montonen et al. 2008). The population-weighted average of the consumption of the 90 food groups is presented in the supplementary material.
The self-sufficiency rates of the products included in the diets were also taken from the FoodMin dietary model and are presented in the supplementary material. The degree of domestic origin of fish was higher in the diet rich in fish and milk, because increasing the consumption of domestic wild fish is encouraged also in the Finnish nutritional recommendations (VNR 2014), and the degree of domestic origin for legumes was decreased from 75 to 50% in the vegan diet.

Land use and biodiversity impacts
To assess the impact on biodiversity and land use of diets, biodiversity impact and land use for the 90 food subgroups were calculated and inserted into the FoodMin model (Saarinen et al. 2019). For biodiversity impact assessment, two different life cycle impact assessment (LCIA) methods were applied Chaudhary and Brooks 2018). Both methods provide country-specific taxa-aggregated characterization factors for land occupation impacts and land-use change impacts, which were used in this study. The global characterization factors were chosen over regional and average factors over marginal, following the Global Guidance for Life Cycle Impact Assessment Indicators (UNEP 2017).
The same degree of domestic origin of each product group in each diet scenario was used in this study as in the original FoodMin model (Saarinen et al. 2019, also presented in the supplementary material). For imported products, the countries of origin and imported quantities were obtained from foreign trade in agrifood product statistics, from which the average for years 2016-2020 was used (OSF 2022a, b, c, supplementary material).

Land use and land-use change data
The biodiversity impacts of imported plant products were assessed based on the land use of product (m 2 /kg). The land occupation (OCC) was calculated based on average yields in the years 2016-2020 derived from FAOSTAT statistics (FAO 2022a). The land transformation (land-use change) related to the production of each crop was assessed based on FAOSTAT Land use statistics (FAO 2022b). The land-use change was allocated over 20 years following the UNEP-SETAC (Koellner et al. 2013) andIPCC (2003) guidelines. The land-use type from which the conversion took place was not considered because the biodiversity impact assessment methods provide only one characterization factor for land transformation to cropland and one for land transformation to pastureland, assuming that the conversion takes place from natural state.

Imported products
The biodiversity impacts of imported plant products were assessed based on the land use for each imported product. For all plant products, the characterization factors for cropland from Kuipers et al. (2021) and factors for intensive cropland from Chaudhary and Brooks (2018) were used. The FAO statistics did not include yield data for all products from each reported country of origin in the import statistics (e.g., coffee from the Netherlands or Sweden). Since it is unlikely that these countries are the original producers, the missing production data was handled as errors in the import statistics, and therefore the import from these countries was excluded from the assessment.
The biodiversity impacts of imported animal products were assessed by applying the characterization factors from Kuipers et al. (2021) and Chaudhary and Brooks (2018) to the land use of animal production in different countries reported in the supplementary database of LCA studies compiled by Poore and Nemecek (2018). In the database by Poore and Nemecek (2018), the five most significant feeds used in animal production and their countries of origin are reported. For some feeds, the origin was reported as "Europe" or "Global." Based on FAO (2022c) statistics, the five largest feed exporters in Europe, and globally, were determined and a weighted average for these countries was used as the origin for those feeds.
In the Poore and Nemecek (2018) database, the land use for each product is reported separately for seed, arable, temporary pasture, fallow, permanent pasture, and loss. From Chaudhary and Brooks (2018), characterization factors for intensive cropland were used for seed and arable land, the factors for intensive pasture for temporary pasture, the factors for minimal pasture for permanent pasture, and the minimal cropland for fallow. From Kuipers et al. (2021), characterization factors for cropland were used for seed, arable and fallow land, and factors for pasture used for temporary and permanent pastures. For both methods, the land-use class "loss" was calculated based on loss percentage from each reported land use. The database by Poore and Nemecek (2018) does not include LCA studies from every country and for such missing countries the regional (Africa, Asia, Europe, NorthAmerica, Oceania, South and Central America and the Caribbean) values were created by defining the region for each LCA study and calculating the regional average.
The share of beef originating from beef cattle and dairy cattle was set to be the same as in the ecoinvent dataset (cattle for slaughtering, global) (Wernet et al. 2016). Farmed fish as a share of imported fish was assumed to be 56% (FAO 2022d). The biodiversity impact of different dairy products was assessed following the above-mentioned procedure to derive the impact for production as kg of milk, after which these impacts were multiplied by the volume of milk needed to produce one kg of each dairy product based on the default values given in PEFCR for dairy products (EDA 2018).
The factors used for standardization in Poore and Nemecek (2018) for processing yields of different products (e.g., kg of soy milk produced per kg of soybeans) are presented in the supplementary database (Table 4C food processing) and were applied in this study to assess the land use of both domestic and imported processed products.

Domestic products
The land use and biodiversity impacts of domestic plant products were assessed similarly as for imported products. The yield statistics covering 2016-2020 were used to calculate the land use per kg of product (OSF 2022a, b, c), and the characterization factors for cropland from Kuipers et al. (2021) and intensive cropland from Chaudhary and Brooks (2018) were used. For perennial plants, the characterization factors for light cropland were used, and for greenhouse production, the factors for (intensive) urban land use were used.
The domestic animal products were assessed based on data for the average feed compositions of Finnish broilers and pigs surveyed by Hietala et al. (2021) and Usva et al. (2023). The feeds included some co-products from food processing, such as soybean and rape meals. The land use for these products was derived from the ecoinvent database and characterized with the factors for import countries reported in the study by Hietala et al. (2021). The impacts per unit of live weight were converted to meat and edible offal by using the same factors as used in the Poore and Nemecek (2018) supplementary database. The land use and biodiversity impact of Finnish beef (weighted average of beef from dairy and beef cattle) has previously been assessed by Huuskonen et al. (2023) using the method by Chaudhary and Brooks (2018). These values were directly used in the FoodMin model, and the information on land use was utilized to calculate the biodiversity impact using the assessment method by Kuipers et al. (2021). The feed composition and feed usage regarding egg and fish production were obtained from Silvenius (personal communication), based on previous LCA studies (Silvenius et al. 2022). The share of farmed fish was set at 35% of domestic fish consumption based on consumption statistics (OSF 2022a, b, c). Due to a lack of data on Finnish lamb production, the impacts of Swedish production from Poore and Nemecek (2018) were used. The biodiversity impact of dairy products was assessed based on land use for raw milk calculated by Hietala et al. (2021). The same volume of raw milk as for imported products was assumed to be used per kg of processed dairy product. The shares of biodiversity impacts originating from imported feeds were calculated based on information on feeds and their countries of origin reported in the original publications by Hietala et al. (2021)

Results
The two biodiversity impact assessment methods used showed a similar trend for the diet scenarios, with only slight differences in product group contributions (Fig. 1). According to both methods, most of the biodiversity impacts were externalized-they related to imports (Fig. 1). However, the methods resulted in different scales of biodiversity impact, with the biodiversity impact results obtained with the method of Kuipers et al. (2021) being over 60 times lower than those obtained using the method of Chaudhary and Brooks (2018) (Fig. 1). Dietary biodiversity impact differed from land-use results, particularly in terms of product group contributions and share of imports (Figs. 1, 2 and 3). The contribution of dairy products, fats, and cereals was notably higher in total land use for diets than in dietary biodiversity impacts, and in contrast, the contribution of fish products and beverages was higher in terms of biodiversity impacts than in terms of land use (Figs. 1 and 2). The biodiversity impacts and land use for all products (domestic and imported separately) are presented in the supplementary material.
Biodiversity impacts and land use for diets decreased with reduced consumption of animal-derived foods, being highest for the current diet and clearly lowest for the vegan diet ( Figs. 1 and 2). The decrease in biodiversity impact was emphasized compared with land use-the impact of the vegan diet was only 30% of the biodiversity impacts of the current diet, while for land use it was about 50%. In the current diet, meats and dairy products made the greatest contribution to land use and dietary biodiversity impact regardless of the assessment method. However, for the vegan diet, beverages, legumes, and nuts made the greatest contribution to biodiversity impact, and cereals, legumes, and nuts to land use. The contribution of different product groups to total biodiversity impact differed slightly between the two assessment methods, with the Kuipers et al. (2021) method leading to a higher contribution for beverages, fish products, and meats, and the Chaudhary and Brooks (2018) method leading to a greater contribution of sugars, sweets, and fruits.
According to the results using both methods, over 85% of the biodiversity impacts related to consumed food were linked to imports of food and feed (Fig. 1), even though the share of imports accounted for only 38-50% of the total land use (Fig. 2). The share of externalized biodiversity impacts also varied among product categories, exceeding 98% for beverages, fish, meats, sugars, and sweets, whereas for cereals and vegetables it was less than 52% (Fig. 3). Externalization was much smaller for land use than for biodiversity impact (Figs. 1 and 2). It also targeted different product groups (Fig. 3).
The products making the greatest contribution to impacts for the current diet varied between biodiversity and landuse impact categories, especially in terms of different meats (Table 1, impacts of all products presented in the supplementary material). In the biodiversity impact for the current diet, poultry, fish, and pork made a greater contribution than beef, while the contribution of beef was greatest in terms of land use for the current diet, accounting for 19% of the dietary impact but for only 7 to 8% of the dietary biodiversity impact. Moreover, the contribution of milk was substantial in the land-use impact category but was not among the five products that contributed most in terms of biodiversity impact. Fig. 1 Biodiversity impact (PDF/person/day) assessed according to Kuipers et al. (2021) and Chaudhary and Brooks (2018) for Finnish diet scenarios, including the current and four alternative diets. The results include biodiversity impacts of land occupation and land-use change. The black asterisk represents the total biodiversity impacts originating from imported products (including also imported feeds used in domestic animal production) The biodiversity assessment methods divided the biodiversity impact only slightly differently into the impacts resulting from land occupation and land-use change. The share of impacts from land occupation was 26-75% using the method of Kuipers et al. (2021) and 28-73% when using the method of Chaudhary and Brooks (2018) (Figs. 4 and 5). In contrast, different product groups were related to biodiversity impacts resulting from land use and those resulting from land-use change. The biodiversity impacts of land-use change were reduced with a decrease in meat consumption using both methods. The biodiversity impact of land occupation was more even among the diet scenarios, with the occupation impact being associated especially with consumption of beverages, sugars, and sweets.

Discussion
We found notable differences between two LCIA methods in terms of the magnitude of biodiversity impacts, but similar relative differences between diet scenarios and impacts arising from land occupation and land-use change. An assessment considering only land-occupation impacts contributed substantially to product groups such as beverages and sugars, whereas inclusion of land-use change impacts led to the significant contribution of animal products. Moreover, our findings demonstrated the significant role of imported foods to the global biodiversity impacts of the Finnish diet, which is a typical western pattern diet (Päivärinta et al. 2020).

Impacts of diets
This study showed that when using a land-use-derived assessment method, the reduction in intake of animalsource foods can lead to a reduction in both biodiversity and land-use impacts. This was expected because animalsource foods are generally more land-use intensive than plant-based foods. Furthermore, most biodiversity impacts arise from land-use change, which is associated with global animal feed production, leading to appreciable impacts for products of animal origin and consequently similar impacts Fig. 2 Land use (m 2 /person/ day) for Finnish diet scenarios, including the current and four alternative diets. The results include biodiversity impacts of land occupation and land-use change. The black asterisk represents the total land use originating from imports (including also imported feeds used in domestic animal production) of the current Finnish diet. However, there are still uncertainties in the assessment of the biodiversity impacts of the diet scenarios, regarding possible changes in production systems caused by the diet change. In our study, the modeling was based on the life cycle assessments of current products and their related production chains, and therefore it does not consider the indirect impact of the dietary changes, e.g., changes in the use of by-products. Incorporating such dynamic systems-level changes into the assessment of diets requires further research.
There exist few studies on the biodiversity impacts of diets, and differences in the impact assessment methods used, which include factors such as land-use change, and countries of origin of products. This makes direct comparisons challenging. Our biodiversity impact result for the current Finnish diet (2.2E-12 PDF/cap/day assessed using the method of Chaudhary and Brooks 2018) was greater than the biodiversity impacts of average European food consumption in 2015 assessed by Crenna et al. (2019) (1.6E-14 PDF/cap/day), but lower than the biodiversity impact of self-reported diets in various European countries reported by Walker et al. (2018), with an average of 4.84E-12. However, the results are not directly comparable because the studies by Crenna et al. (2019) and Walker et al. (2018) applied global average factors (by Chaudhary et al. 2015;Chaudhary and Brooks 2018) and not spatially differentiated ones as applied here. Crenna et al. (2019) included only part of the total food consumption. Nonetheless, the relative reduction in dietary impacts when the consumption of animal products is reduced was similar to the results reported here. Fig. 3 The share of imported biodiversity and land-use impacts for current diet in each product category Furthermore, the contribution of various product groups to the total dietary impact reported by Moberg et al. (2020) for the Swedish diet was like that reported in our study for dietary land occupation impacts, with the highest contributions coming from beverages, sugars, sweets, and meats. They used the Chaudhary and Brooks (2018) method, but the results were converted to the unit of extinction rate (E/ MSY) and the land-use change was included differently. Their results cannot, therefore, be compared with the results of our study in absolute terms.
In Finland, poultry production relies more heavily on imported feeds than does the production of other meats, resulting in higher biodiversity impacts at diet and product level (product level results presented in the supplementary Fig. 4 Disaggregated biodiversity impacts (PDF/person/ day) arising from land occupation and land transformation, assessed using the method of Kuipers et al. (2021) for the current Finnish diet and four alternative diets Fig. 5 Disaggregated biodiversity impacts (PDF/person/ day) arising from land occupation and land transformation, assessed using the method of Chaudhary and Brooks (2018) for the current Finnish diet and four alternative diets 1 3 material). This is contradictory to Crenna et al. (2019), who established that pork and beef both contributed around 20% of the total biodiversity impact of European food consumption, which is considerably higher than the contribution for our study. In addition to products of animal origin, we identified coffee and chocolate as being significant sources of dietary biodiversity impact. The products with the highest impacts identified in this study were similar to those of Ahlgren et al. (2022).
Despite the wide coverage resulting from the dietary level assessment, a weakness of our study concerns blue foods, for which only the land-use impacts of feed production were covered. Therefore, wild fish and other captured blue foods have zero biodiversity impact, even though fishing has a sizeable impact on marine ecosystems (Halpern et al. 2008). This impact cannot be addressed using a biodiversity impact assessment method based on land use. Also, some plant products and mushrooms can be cultivated or collected directly from nature, but because of limited data, all products were assumed to be cultivated. In addition, this study only included the land use associated with agricultural production, excluding impacts related to production of inputs (e.g., fertilizers) and impacts of post-farm processing, which underestimates the total biodiversity impacts.
According to our results, most biodiversity impacts of Finnish diets are externalized, even though the degree of domesticity of Finnish food consumption is relatively high and most of the land use also occurs in Finland. The two biodiversity assessment methods used in this study led to notably different results, but still indicated that the majority (> 85%) of biodiversity impacts of Finnish diets take place outside of Finland. This is in accordance with biodiversity impacts for the Finnish food supply reported by Sandström et al. (2017), as well as biodiversity impacts of Swiss food consumption (Chaudhary et al. 2016). The high contribution of imported products is due to low characterization factors for Finland in the biodiversity impact assessment methods used. However, it should be noted that the assessment method for biodiversity impacts is developing and includes major uncertainties, which are discussed in the Sects. 4.3. and 4.4.
In general, importing food (to high-income countries such as Finland) is a controversial issue from the point of view of biodiversity impact. On the one hand, it has been estimated that international trade is an important driver of species loss in key global biodiversity areas, allowing high-income nations to externalize land use and the associated biodiversity loss to middle-and low-income countries (Sun et al. 2022). For example, the EU has restored forests by 12.6 Mha during the years 1990-2014, while 11.3 Mha was deforested to produce crops for consumption in the EU (Fuchs et al. 2020). On the other hand, international food trade can also reduce the resource use of food systems when trade flows from high resource use efficiency to areas of lower efficiency-it has been estimated that international trade has lowered global cropland demand by almost 90 Mha (Kastner et al. 2014). From a dietary perspective, the international food trade is also an essential provider of nutrient supply, especially for lower-income countries (Wood et al. 2018).

Comparison of the two biodiversity impact assessment methods
The two biodiversity impact assessment methods used in this study led to different results in terms of the scale and origin of impacts. The absolute results assessed using the method of Kuipers et al. (2021) were over 60 times lower than those obtained using the method of Chaudhary and Brooks (2018), which can be attributed to differences in the methods' backgrounds and the resulting characterization factors. Even though the relative dietary impacts assessed using the two methods were similar, the products contributing most to the total dietary impacts differed slightly between the two assessment methods. This indicates that the choice of method might lead to different relative results when comparing individual products. The most obvious differences are that the methods are built on different species data in terms of different taxa, species-area relationship (SAR) models, and land classification . Above all, however, the methods used to convert impacts on regional species richness into impacts on global species extinctions follow different principles, leading to different magnitudes for characterization factors. Chaudhary and Brooks (2018) apply a vulnerability score (0 < VS < 1) to each species group based on species' Red List status (IUCN 2017) and endemicity, whereas Kuipers et al. (2021) use global extinction probabilities, which are implemented such that globally they sum to 1 and thus result in notably lower global characterization factors (McLaren et al. 2021). Due to the difference in conversion of impacts on regional species richness into impacts on global species extinctions, the method of Kuipers et al. (2021) resulted in lower global impacts than the method of Chaudhary and Brooks (2018). However, considering the fragmentation effect, higher characterization factors at regional level were highlighted using the method of Kuipers et al. (2021). Also, the taxa aggregated characterization factors are slightly different using the two methods: Chaudhary and Brooks (2018) use the median of all ecoregion vulnerability score values per taxon as the weighting factor, whereas Kuipers et al. (2021) calculate taxon-aggregated characterization factors so that each taxon receives equal weight regardless of the number of species in each taxon.
The Chaudhary and Brooks (2018) method resulted in slightly higher impacts arising from land-use change than that of Kuipers et al. (2021), even though the land transformation factors are formatted following the same principles in both methods. Furthermore, neither method distinguished the land-use type from which the land-use change occurred but only offered one factor to be used for all types of land transformation. This may overestimate the impacts of land-use change because not all land-use change takes place from natural vegetation, and seminatural areas can maintain species that are absent in natural vegetation (Hyvärinen et al. 2019). Considering the previous land use as a reference state would also enable capturing potential improvements, such as when arable land is converted to forest.
Unlike the method of Kuipers et al. (2021), the method of Chaudhary and Brooks (2018) has three different intensity levels within each land use type, which enables consideration of the impact of land management on general level. However, in broad assessments (e.g., assessment of diets) that include a variety of products from several locations, it is challenging to define the production systems from which the products originate, which might limit the application of intensity levels in practice. Also, the differences in characterization factors of different intensity levels are relatively small and therefore do not necessarily distinguish actual production methods from each other . Because the production methods and crop types impact agricultural biodiversity, more detailed methods are needed to assess fully the impact of different agricultural practices (Toivonen et al. 2022), and the potential positive impacts of agriculture on biodiversity, such as sustaining rural biotopes and cultural habitats that support threatened species (Hyvärinen et al. 2019;Tiainen et al. 2020).

Global extinctions as a biodiversity measure
The two biodiversity impact assessment methods used in this study Chaudhary and Brooks 2018), both measure the biodiversity impact in units of PDF, which represents the fraction of species that may become extinct. In this study, we used the global characterization factors instead of regional ones, meaning that species loss is not considered in the results if the species still exists in other regions. Measuring global impacts on biodiversity is in line with other LCA impact categories-for example, the climate impact is always global (UNEP 2017)-but assessing global impacts ignores loss of a single species in a species-poor area having a different relative impact than losing one species in a species-rich area (Koellner et al. 2013). Because Finland belongs to an ecoregion that has relatively few endemic species, the assessed biodiversity impact for domestic products is low (Sandström et al. 2017). However, assessment methods that measure the impact on species are not fully able to consider the impacts at habitat level. For example, the potentially disappeared fraction of species assessed is very low in Finland, while 48% of habitat types in Finland have been classified as threatened in Red List assessments (Kontula and Raunio 2018; Hyvärinen et al. 2019). These national monitoring results for biodiversity indicators indicate that biodiversity loss is also a locally (or nationally) significant environmental problem that, for example, endangers the functioning of ecosystems. Information about it is very relevant for decision-making at many levels and thus is a relevant part of biodiversity assessment in LCA. There is clearly still a need for developing biodiversity assessment methods further, so that in addition to global biodiversity impacts, also the impacts to local biodiversity and habitat loss would be considered. One potential development could be the use of two different metrics, such as some adaptation of biodiversity intactness index (Scholes and Biggs 2005), alongside the extinction measures.
Apart from structural aspects, also the functional aspects of biodiversity should be considered. So far, methodology development has included the functional diversity into LCA, for example by Scherer et al. (2020), but the characterization factors generated by the method are still spatially limited, and therefore such approaches are not yet adopted for biodiversity impact assessment in LCA.

Land use and land-use change as biodiversity impact indicators
According to our results, dietary land use and biodiversity impact exhibited a similar trend for the diet scenarios. However, when looking at product contributions to dietary biodiversity impact in more detail, it is evident that biodiversity impact cannot be derived solely from the land area used for production. In the context of the Finnish diet, beef and dairy products contribute substantially to land use (in Finland), but far less to biodiversity impact. In contrast, poultry contributes much more to biodiversity impact than land use. This is due to the use of soy as broiler feed and the resulting land-use change.
In our study, most land-use-related dietary biodiversity impacts originate from land-use change, and the impacts from land occupation are lesser. It should be noted that the impacts from land occupation and land-use change are measured in different units, PDF/m 2 and PDF*year/m2, and the first metric has no time dimension while the latter does. Therefore, in principle, they would not be cumulative, and if still applied, in practice, the relationship between land occupation and land-use change impacts could change according to the period assessed. This issue should be clarified in future research.
In addition to habitat change due to land use and landuse change, climate change, nutrient pollution, overexploitation, and invasive species are identified as important drivers of biodiversity loss (IPBES 2019;MEA 2005). Crenna et al. (2019) showed that most biodiversity impacts are land-use-related, land use contributing to more than 50% of total biodiversity impacts, and the climate impact being the second highest contributor, generally accounting for more than 25% of the total impacts. However, the share of impacts caused by different drivers differs among products, and thus considering only land-use-related impacts might favor products with low land use but high impacts in other impact categories. Also, for the biodiversity drivers of overexploitation and invasive species, there is no established method to include them in LCA (Crenna et al. 2020), which is one of the aspects that further developments in methodology should consider.

Conclusions
The two assessment methods used in this study led to very different land-use-driven biodiversity impacts of diet scenarios in absolute terms, while the relative results were similar. Therefore, results obtained using the methods are not suitable for comparison in absolute terms. When comparing individual products, the methods can lead to different relative results. Furthermore, the study showed that the global biodiversity impacts of Finnish diets are highly externalized, despite the high degree of self-sufficiency in Finland. This highlights the global responsibility associated with high-income countries diets.
The majority of the biodiversity impacts arose from land-use change, and the impact of land occupation was smaller. The findings underscore the importance of considering land-use change impacts in food LCAs, which has generally been overlooked in previous biodiversity impact assessments. Because the impacts of land-use change are linked especially with feed production, ignoring the impacts could underestimate the relative contribution of animal-source foods.
The comparison of two different biodiversity impact assessment methods in this study showed that methodology development and unification of assessment methods for biodiversity impact, even regarding methods derived solely from land use and land-use change, are still needed. Future developments regarding land-use-driven methods should consider the clarification and unification of units of land use and land-use change, as well as the accuracy of methodology concerning land-use change impact assessment. From a wider perspective, the biodiversity impacts within agricultural systems should be more profoundly considered (i.e., the impact of different production practices), and the inclusion of different aspects of biodiversity alongside the global impact on species should be covered.
Funding Open access funding provided by Natural Resources Institute Finland (LUKE).

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on request.

Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.