The potential of insect protein to reduce food-based carbon footprints in Europe : The case of broiler meat production

As part of the current climate change debate, concerns have been raised over the rapid expansion of soybean cultivation as the major protein source to feed animals for human consumption. At the same time, insect-based protein is rapidly emerging as an alternative protein source to animal-based protein, both as a potential substitute for soybean-based protein for the use in animal feed, and for direct human consumption. The assessment of climate impacts and market potentials of novel animal feed and food products related to such insect-based protein requires an assessment of their carbon footprints. This paper explores the extent to which insect protein could help to reduce the Global Warming Potential (GWP) associated with food consumption in Europe. The results from a quantitative review and scenario analyses provide two key insights. First, they support previous research suggesting that insect protein has the greatest potential to reduce the carbon footprints of European consumers, if insects are directly consumed as food. Second, they suggest not only that the use of insects as animal feed can substantially contribute to the sustainability of broiler production systems with regard to lowering GWP, but also that low-value side streams are key for improving this potential.


Introduction
In the face of increasing environmental, biodiversity and welfare concerns associated with traditional animal-based food production (Grunert et al., 2018;Bonnet et al., 2020;Bohnes and Laurent, 2021;Raven and Wagner, 2021), the search for new and alternative plant, insect and lab-based protein sources has accelerated (He et al., 2020;Santo et al., 2020;Kyriakopoulou et al., 2021). These protein sources have not only the potentials to replace traditional meat products designated for human consumption, but also to direct the use of feeds in animal production towards more sustainable practices (Hawkey et al., 2021;Williams, 2021), and contribute through better exploitation of side streams towards a more sustainable circular economy (Ghinoi et al., 2020;Ojha et al., 2020). To assess these potentials in terms of their climate impact, it is instrumental to understand the carbon footprint implications of switching from animal-based protein for human consumption to alternative protein sources, such as insect proteins (Sánchez-Muros et al., 2014). Carbon footprint-based analyses have thus emerged to complement taste-focused studies for the assessments of market potentials and consumer acceptance of novel food product attributes and protein sources (Burnier et al., 2021;Castro et al., 2020;Frigerio et al., 2020;Grebitus and SteinerVeeman, 2016;Llagostera et al., 2019;Steiner and PeschelGrebitus, 2017).
The production and consumption of animal-based food causes major environmental impacts (Poore and Nemecek, 2018), especially when ruminants are involved (Dyer et al., 2020;Röös et al., 2014). When considering the environmental impacts of different animal production systems, such as with regard to Global Warming Potential, the impact of broiler production is typically proportionally smaller when compared to other meat types (e.g., Nijdam et al., 2012;Dyer et al., 2020). We also observe and continue to expect that Western consumption patterns will continue to shift from red meat to white meat (González et al., 2020), in spite of the fact that animal-based protein intake of European consumers has gradually increased in the past, with a particular rise in the consumption of poultry (Westhoek et al., 2011). However, there is a growing interest and need to change European diets towards environmentally friendlier ones (e.g., da Silva et al., 2009;Kearney, 2010;Westhoek et al., 2014;Yang et al., 2021). In light of the above evidence, we chose broiler meat as the base scenario for the carbon footprint assessment in this paper, aiming to explore the extent to which insect protein could help to reduce the Global Warming Potential associated with food consumption in Europe.
A range of previous studies have assessed the potential for climate change mitigation of switching to alternative protein production and consumption options. Alternative options include the replacement of soybean-based feeds with other feed options (Ikiamba et al., 2021) and the replacement of animal-based proteins with alternative protein sources, including lab-grown and insect-based sources of protein (Alexander et al., 2017;Tuomisto, 2019;Bohnes and Laurent, 2021;Painter et al., 2020;Williams, 2021). Noting the land-use efficiency of insects and the fact that insects are typically nutritious with healthy high protein content, the use of insect-based protein promises to provide significant potentials for increasing food security and environmental sustainability of food (Bodenheimer, 1951; van Huis, 2013;Alexander et al., 2017). Insects do not use energy to maintain a high body temperature and are therefore very efficient in converting feed into edible meal protein (Nijdam et al., 2012; van Huis, 2013). When insects are mass produced, they can also contribute to the bioconversion of large quantities of food waste (Salomone et al., 2017;Fowles and Nansen, 2020;Guo et al., 2020;Ites et al., 2020). Further, insect protein production systems promise to have lower levels of environmental impacts due to a relatively lower carbon footprint (Blonk et al., 2008;Sánchez-Muros et al., 2014), and due to lower land and water requirements as compared to crop or animal protein production systems (Sánchez-Muros et al., 2014;Alexander et al., 2017;Salomone et al., 2017). Furthermore, the use of insect protein as a sustainable feeding-alternative in aquaculture is gaining significant attention in research and practice (Riddick, 2014;Llagostera et al., 2019;Lu et al., 2020;Toviho and Bársony, 2020;Poveda, 2021).
Despite a growing interest and research base on the potential of insect protein, evidence on the environmental sustainability of insects as an alternative protein source for the poultry industry in the European context is sparse (Netherlands: Blonk et al., 2008;Smetana et al., 2016Smetana et al., , 2019Germany: Ites et al., 2020), and previous studies focusing on insect-based food and feed production, in general, show large uncertainties (Salomone et al., 2017). Smetana et al. (2016Smetana et al. ( , 2019 provided a life cycle assessment (LCA) of insect-based food and feed production at the industrial level. Tallentire et al. (2018) studied the environmental impacts of incorporating novel ingredients such as insect protein into broiler diet formulations to replace soybean protein imported from outside Europe. Allegretti et al. (2017) evaluated the production and processing of Hermetia illucens (black soldier fly) larvae for insect meal as compared to the use of soybean meal in Brazilian poultry production systems. Abro et al. (2020) assessed the potential socio-economic benefits of using larvae meal for the Kenyan poultry sector. (Roffeis et al. 2018(Roffeis et al. , 2020 provide an ex-ante life cycle impact assessment of insect-based feed production and use a life cycle inventory analysis as a basis to analyze the economic performances of insect-based feed production systems for the West Africa context. Ites et al. (2020) assess the modularity of insect production and processing as a path to efficient and sustainable food waste treatment. They conduct an LCA assessment for Tenebrio molitor (mealworm) and H. illucens production, with specific consideration on the conditions of production in Germany.
The objective of this study is to contribute to the above debate by exploring the potential of insect protein to reduce the carbon footprint associated with European food consumption. We introduce three scenarios to compare alternative insect-based feed and food production systems with soybean-based feed and broiler meat production systems. In our scenarios, soybean meal-based feeds are replaced with insectbased feeds for broiler production, and broiler meat is replaced with food products made from insects for human consumption. Further, we compare industrially produced composite feeds based on insect protein production with alternative feeds produced from industrial side streams. Through these scenarios we address the following research question: how are European consumers' carbon footprints changing when they alter their food consumption patterns, by partially or completely replacing broiler meat with more insect-based protein food products? To answer this research question, we perform a quantitative review to assess the potential reduction in the carbon footprint associated with an average European consumer's diet.
The remainder of this paper is structured as follows. Section 2 introduces the study approach and methods, provides a selected review of relevant LCA studies applied to insect proteins, and describes the development of our data sets. This serves then as the basis for the selection of variables, calculation of functional unit and the scenario assessments. Section 3 discusses the scenario results, and the results of the sensitivity analysis. Section 4 outlines the study's conclusions, data validity, limitations, and indications for future work.

Materials and methods
First, we conduct a short and thematically focused literature review on the Global Warming Potential (GWP) values of broiler, insect, and soybean meal production from a variety of LCA studies (sections 2.1. and 2.2.), to generate a data set for further analysis. Based on the literature review, we assess the compatibility of insects as a protein source to replace soybean meal in broiler diets. The criteria for the studies included in our data set are presented in section 2.3. We used this data set to first assess the global warming impact associated with the consumption of conventionally farmed broiler meat, broiler meat farmed with insects or food products based on insects, and then outline three scenarios (Section 3). Fig. 1 illustrates the construction of items that we used in the analysis of the GWP of broiler and insect production.
We could in the following meta-analysis validate the robustness of the data, yet in the case of LCA, the quantitative analysis is challenging due to the heterogeneous nature of the studies. The production practices and processes of insect protein vary widely and contribute to this heterogeneity. For instance, the production of H. illucens as a part of the waste management processes for providing feed for the flies (Smetana , and Acheta domesticus (house cricket) production using industrial feed sources (Halloran et al., 2017) are fundamentally different protein production systems. In addition, there is a lack of peer-reviewed LCA studies on the production of specific insect species. Therefore, we attempted to improve the usefulness and validity of the results and data sets by conducting a sensitivity analysis, and by identifying the major factors affecting the results. In addition, it could be argued that because of the novelty of the insect industry, there are yet no best-practice examples and corresponding data available to inform insect production, to move towards lower GWP options. This is reflected in the fact that several of our source LCA studies use only one insect farm or a pilot scale farm as a case example, which we also indicated in the summary of characteristics of the insect-related LCA studies (Table 1). For these reasons, we examined scenarios by using only the best knowledge on production practices for each insect species.

Soybean meal as feed ingredient and its substitution with insects
Soybean meal is a widely used and major source of protein for broiler feed (van Gelder et al., 2008;Dei, 2011), yet it is also an ingredient that can contribute to negative environmental impacts in broiler feed production, for example due to its land use (e.g., Fearnside, 2001;Barona et al., 2010;Salomone et al., 2017), as well as through contributing negatively to social sustainability measures (Jia et al., 2020).
Soybean meal is a concentrated protein source from which most of the oil has been separated from soybeans. It is high on protein and contains a well-balanced composition of essential amino acids (Banszkiewicz, 2011). The average protein content of soybean meal is 46.0% (Florou-Paneri et al., 2014). Based on a literature review of the LCA studies on broiler production, the share of soybean meal in broiler feed ranges from 20.1% to 39.0%, and the value of the feed conversion rate (FCR) ranges from 1.6 to 2.8. However, typically the FCR in broiler farming is approximately 2.0, and average soybean meal content in broiler feed is 36.8% in EU ( van Gelder et al., 2008;Boggia et al., 2010;Thévenot et al., 2013;González-García et al., 2014;Cesari et al., 2017).
Insects are nutritionally rich in essential amino acids, and some insect species provide high good-quality fatty-acids Kouřimská and Adámková, 2016). It has been shown that insects can replace partly or all the soybean meal used in the feed without compromising the FCR value (Khusro et al., 2012;Makkar et al., 2014;Bovera et al., 2015;Józefiak et al., 2016;Lu et al., 2020). In our following analysis, we assume that replacing 1.0 kg soybean protein with the same amount of insect protein has no effect on the FCR value in broiler farming.

Insects as feed and food
The general EU legislation on insects for feed or food has distinguished insects as animals that are farmed to produce food or feed (Spiegel et al., 2013;Finke et al., 2015;Lähteenmäki-Uutela and Grmelová, 2016). The EU Commission defines Novel Food as food that had not been consumed to a significant degree by humans in the EU before May 15, 1997, when the first Regulation on novel food came into force. The Novel Food Regulation requires an authorisation before a novel food product can be placed on the EU market, which is then followed by a stringent scientific assessment from the European Food Safety Authority (EFSA). EFSA verifies, based on available scientific evidence, that the food does not pose a safety risk to human health. The regulation on novel foods has recently been revised, and has from June 1, 2021, authorized the placing on the market of dried yellow mealworm Table 1 Characteristics of the insect related LCA studies. A-LCA means attributional LCA. C-LCA means consequential LCA. Values accounting C-LCAs are given in brackets. as a novel food. 1 Several LCAs focus on insect production for human food (e.g., Oonincx and de Boer, 2012;Halloran et al., 2017), although the same insect species can also be produced for animal feed (e.g., Thévenot et al., 2018). We could argue that if the nutritional quality of insects produced beyond the farm gate is good enough for direct human consumption, the use of unprocessed insects should be acceptable also for feeding animals. For instance, SucklingDruckman et al. (2020) studied the production of A. domesticus for use in food and feed when the same production processes are applied while assuming that if insects are produced for food, they could also be used for feed production. However, we should also note that while most insect species are suitable for use in feed, the same species are not necessarily suitable for human consumption, keeping also EU hygiene regulations in mind (e.g., Regulation No 852/2004 and Regulation No 183/200).

LCA studies used to generate the data set
LCA is a standardized method which is internationally used to quantify the environmental impacts of a product throughout its entire life cycle (ISO 14040). The use of LCA has rapidly increased as a tool when studying environmental impacts of agricultural and food products (e.g. Ponstein et al., 2019;Roy et al., 2009;Thomas et al., 2020). The LCA studies that we selected for this study can be divided into attributional LCAs (A-LCA) and consequential LCAs (C-LCA). Broadly speaking, the A-LCAs focus on describing the impacts of existing production systems, and the C-LCAs focus on describing how the impacts could change based on decisions (Curran et al., 2005;Yang, 2016).
For those studies included in our analysis that encompass more than one insect species produced by the same production processes (e.g., Oonincx and de Boer, 2012;Halloran et al., 2017), we consider different insect species as a separate observation in the same studies. To ensure study validity and a robust, data-driven approach to the construction of the scenarios, we used only peer-reviewed studies ( Table 1) that are comparable by the use of variables and parameters.
For LCAs to be comparable, they should apply equivalent definitions of the system boundaries for the alternative production systems, and they should make similar delimitations within the systems (Tillman et al., 1994). To suit the context of our study, the LCAs should include similar system boundaries, a functional unit (FU) using mass unit or protein content, the GWP as an environmental indicator, and the results should be expressed in CO 2 -equivalents (CO 2 -eq.). We describe the requirements for similarities in greater detail in the following sub-sections.

System boundaries
Ideally, the LCA studies we used for our analysis should evaluate the various environmental impacts of broiler meat and insect products over their entire life cycle. However, in the literature (Tables 3-5), the impacts of broiler and insect farming are typically evaluated until the slaughter and/or package stage, and do not consider later stages, such as retail, use and disposal. For our study, we included (Fig. 2, below) LCAs which encompass also processing stages beyond the farm gate, such as drying, slaughter, mechanical separation and/or package (e.g., González-García et al., 2014;van Zanten et al., 2015;Cesari et al., 2017;Halloran et al., 2017;Thévenot et al., 2018), while keeping in mind that the LCAs on broiler and insect production consider different processing stages beyond the farm gate, yet as a commonality they consider that the actual product is ready for use.
The use of C-LCAs differ most strikingly in terms of system boundaries. For instance, farming of H. illucens for food or feed produces compost, which stems from the production and use of feed sources, and comes in the form of, for example, biowaste or manure. For this case, it is possible to include consideration of the avoided impacts, such resulting for example from differences in fertilizer use or waste treatment (e.g., Smetana et al., 2019), thus allowing for a subsequent data set to assess and highlight the systemic benefits of protein production from H. illucens as compared for example to protein production from broiler. However, the LCAs reporting the impact estimates of production of insect meal from H. illucens for substitution of other protein sources (e.g., van Zanten et al., 2015;Salomone et al., 2017) do not include consideration of the avoided impacts as this would create double counting. Which factors to include in C-LCA models is vital for the reliable evaluation of the impacts (e.g., Sillman et al., 2020). Therefore, we separately indicate which data was provided by the C-LCAs.

Functional unit (FU)
A well-defined functional unit (FU) allows to make adequate and informative evaluations across LCAs. The FU should be linked to one or more of the characteristics that are in common between the products. In the data sources employed, the LCA results are represented in terms of kg of CO 2 -eq. per kg of protein. Most of the studies report their results per kg or tonne of product at the processing or slaughterhouse gate. For our study, the nutrient-based FU was of primary importance, because we aim to present the results to inform recommendations for the consumption of protein-based food. Thus the FUs used in the literature are converted from kg of product to kg of protein based on the protein contents of the products, as applicable.

Global Warming Potential (GWP)
Protein production systems are sources of Greenhouse Gas (GHG) emissions, such as carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O). The impact of these emissions can be evaluated in different impact categories or by types of indicators (Hellweg and i Canals, 2014). The characterization factors used to convert GHG emissions other than CO2 into the common unit of the GWP impact category (CO2-eq) were derived from Myhre et al. (2013). In the data sources selection we used GWP 100 as the environmental indicator.

LCAs with different allocation methods
Some food and feed production systems may produce more than one single product. For example, H. illucens production systems may also include co-production in waste treatment and processing facilities. For such systems, the global warming impacts must be allocated to the various inputs and outputs of the production processes and divided over the products resulting from the system by using, for example, the economic, mass and energy allocation methods. Several problems are involved with the use of such allocation methods, because the use of the method, and the choices and assumptions made prior to the LCA analysis may all influence the robustness of the results (e.g., Ekvall and Finnveden, 2001;Cederberg and Stadig, 2003;Svanes et al., 2011). In other words, not considering these allocations could overestimate the actual impact of production in multiple-product systems. Therefore, for our data set, we used LCAs with diverse allocation approaches.

Selection of LCA studies on soybean meal, broiler and insect production
Most of the soybean meal used in the broiler industry is imported into the EU from Brazil and Argentina (van Gelder et al., 2008;Voora et al., 2020). To get the relevant data linked with appropriate analysis approaches, we selected LCAs focusing on the use of soybean meal from South America. Because the impact of land use or land use change on the total GHG emissions of soybean meal production is significant (Hörtenhuber et al., 2014), we considered only LCAs including land use or land use change impacts.
As the focus of this study is on European consumers, the LCA studies we selected have a corresponding focus on broiler production in the EU. Further criteria for selection was that soybean meal production must be included as a feed-ingredient variable in the broiler production function. This criteria could not be applied for the insect studies, because the European markets are yet to capitalize on the potential of insects for food and feed, and the corresponding value chains in emerging markets are highly globalized. Therefore, LCAs on insect production for regional feed and food systems need to take a global perspective.

Scenario I
Broiler production was selected as a baseline for scenario development and analysis. Broiler production systems can be categorised by farming programs into free-range broiler systems, where broilers have free access to pasture area, and conventional broiler systems, where broilers are grown indoors. In both systems, broilers are fed with industrially produced feed as their base diet (e.g., Lima and Nääs, 2005;Coletta et al., 2012). In this scenario, we assessed the carbon footprint of conventionally farmed broiler meat against the potential of reducing the food carbon footprint of protein consumption through an increase in insect consumption. We assumed that the protein intake from conventionally farmed broiler meat remains constant.

Scenario II
Emissions from feed production are typically the main source of environmental burden in broiler production. The protein sources used in feed and the feed efficiency rate have significant impacts on the GWP of broiler production systems (e.g., González-García et al., 2014;Leinonen and Kyriazakis, 2016;Cesari et al., 2017). Therefore, in this scenario, we assessed the GWP of insect meal production relative to soybean meal production, to examine the potential of insects as a more climate-friendly protein source for broiler feed. In this scenario, we examined how consumers' food carbon footprints change if they shift their protein consumption from conventionally farmed broiler meat, where soybean meal is typically used, to protein consumption from broiler meat farmed with insect-based feeds. To assess this scenario required that we first calculated the data sets for scenario I and scenario III.

Scenario III
Insects are generally assumed to be more environmentally sustainable compared to conventional animal-based protein sources (Alexander et al., 2017). This is largely due to relatively high FCRs and reproduction rates, because insects are cold-blooded ( van Huis, 2013;van Huis et al., 2013). To further assess the degree of their environmental sustainability, we evaluated LCA data on the GHG emissions from insect production. In this scenario, we investigated how consumers' food carbon footprints will change if they were to replace protein consumption from broiler meat with protein consumption from insect-based food products.

Calculations for scenario II
To be able to compare the GWP of the broiler protein production by using either soybean meal or insects as a feed ingredient, first, the impact of the feed ingredients must be known. Then, the amount of the consumed insect ingredients is to be calculated. We used the FCR value of 2 in the following calculations. For the mean value, we used the average of 36.8% as the share of soybean meal content in broiler feed. For the minimum and maximum (minmax) values, we used the shares of 20.1% and 39.0%, respectively (see section 2.1.). We used the GWP impact of total average (Tables 2-5) to calculate the mean value, and calculated the GWP of insect and soybean ingredients by using the following equation: where.
GWP feedingredient = kgCO 2 -eq . per kg of broiler protein GWP ingredient = kgCO 2 -eq . per kg of feed ingredient protein m broiler = kg of broiler FCR = kg of feed per kg of broiler protein n broilerprotein = kg of broiler per kg of broiler protein m − % ingredient = kg of feed ingredient per kg of feed n feedingredientprotein = kg of feed ingredient protein per kg of feed ingredient Once we calculated the GWP of the insect feed ingredients, we calculated the GWP of insect-fed broiler. We calculated this by using the minmax and mean values in the following equation: where.

Data sets
The scenarios I, II and III are developed from the data sets summarized and represented in Tables 2-5    were obtained are also given in the tables. Based on the Dutch Food Composition Database (NEVO), 2 which contains data on the composition of foods, Nijdam et al. (2012) estimated that the average protein content of poultry is 20.0% (without bones). We used this value in our study, where we compared the GWP values of broiler production in the different scenarios (Table 3).
In our analyses, we assumed that the insects are fed to broilers (Table 4) or consumed directly as food (Table 5). The production of insects was also divided into two feed categories: insects fed with industrially produced composite feed, and insects fed with industrial side streams. The specific insect species and their respective impacts are accounted for in Tables 4 and 5. Where the source LCA reported several protein content values for a given insect product, we used the average and minmax protein content values to calculate the impact values.

Results
The results of scenarios I to III are represented in Table 6. Scenario I uses the calculated data from Table 3. Scenario II uses the calculated data from Tables 2 and 4, which is the data from which the GWP of broiler production with insect feed was calculated using equations (1) and (2). Scenario III uses the data from Table 5.
By comparing the minimum (MIN.) GWP values for broiler production in baseline scenario I and scenario II, we can see that switching to feeding broilers with insects that are fed with industrially composite feeds can reduce the GWP impact of broiler production to a lesser extent than if broilers were fed with conventionally produced soybean meal.
When broilers are fed with insects produced with feeds from industrial side streams, the GWP impact is lower for both the minimum and mean value cases, 22.0% and 8.0%, respectively. However, scenario II also indicates that the maximum GWP value for insect-fed broiler production can be higher than that for conventional broiler production. In scenario III, where insects are used directly as a protein source for human consumption, the GWP values are lower in all considered categories, except for the maximum value of insects fed with industrially produced composite feeds. Further, when we consider in scenarios II and III also the systemic benefits by including in our scenario assessments the GWP values from the C-LCAs, the results show net-positive GWP impacts when H. illucens is used in waste processing for compost.

Sensitivities of different factors considering insect production
The results show a wide range of GWP values in scenarios II and III. Greatest uncertainty and impact on the sensitivities concern the location of production, insect species and sources of feed and energy.

Climatic conditions
The insect species G. bimaculatus (field cricket) shows the greatest range of GWP when industrially produced composite feeds are used, for both scenarios II and III. It also has the highest mean GWP value, being 38.0% higher than the total mean average (Table 4). If G. bimaculatus is not used to feed broilers in scenario II, the mean and maximum values decrease by 9.4% and 42.0%, respectively. We note that the GWP values for the species G. bimaculatus are obtained from two different LCA studies, which both use broiler feed as a feed source for crickets, and present scenarios to decrease the environmental impacts from G. bimaculatus production processes. The difference between both studies is related to the conditions of production, which is also the reason for the widely varying range of GWP values. Halloran et al. (2017) studied the production of G. bimaculatus in outdoor facilities and climatic conditions that are natural for species productivity, which means less energy is needed in the farming processes. SucklingDruckman et al. (2020) studied the case of UK production of G. bimaculatus in warm indoor facilities, where higher energy consumption is required to secure optimal levels of humidity and temperature. Consequently, the species of G. bimaculatus show lowest values in terms of GWP if the production systems are operated in optimal climatic conditions. The low GWP values associated with G. bimaculatus decrease the mean values in scenario II by 14.9%, and in scenario III by 30.6%. The other LCAs studied the production of other than cricket species, and the production systems under investigation were located in Europe (Table 1), which cannot be considered to explain the differences in GWP values.

Insect species
The use of different insect species may significantly influence the GWP of broiler production. If, for the example considering scenario II, industrial side streams would be used in the production of the fly species H. illucens and M. domestica as ingredients for broiler feed, the GWP values could be considerably lower. The mean values for H. illucens and M. domestica are 20.0% and 67.0% lower than the total average (Table 4). In other words, the mean value for the case of scenario II where industrial side streams are used to produce insects could decrease the overall GWP impacts in scenario II by 20.0%. The review of the GWP of different insect species for food use produces a similar wide range of GWP values. Only for the cricket species and if produced in the UK, the GWP of production is greater than that of broiler production (Suck-lingDruckman et al., 2020, Table 5).

Feed and energy
If climatic conditions are not considered, feed and energy consumption are the major contributors to the total GWP of insect protein production. Some LCA studies investigated the extent of influence of the feed and energy sources on the GWP impacts (e.g., Joensuu and Silvenius, 2017;Smetana et al., 2019). Specifically, Joensuu and Silvenius (2017) studied the GWP of Z. morio (mealworm) production processes and estimated the potential for GWP reduction, when feed or energy sources are changed to contain higher levels of renewable energy and better meet GHG reduction targets. The authors estimated that the GWP impact may change even by 22.0%, depending on the composition of the industrially produced composite feeds. When by-products are used as a feed and the share of renewables are increased, a GWP reduction of 67.0% can be achieved, when compared to reference production practices in this study. Smetana et al. (2019) studied the GWP of H. illucens production by using different shares of feed from side streams and renewables. Considering the best-case scenario with the highest amount of feed from side streams and the highest share of renewable energy, the Table 6 GWP values (MIN., MAX., and MEAN in kg of CO 2 -eq. per kg of protein) for each scenario and the number of observations. N means the number of observations of different production practices. The values accounting for the C-LCA results are given in brackets. GWP impact is reduced by 80.0% compared to the reference production practice in this study. Bava et al. (2019) studied the impacts of H. illucens production by using different feed sources. The GWP reduction can be 88.0% when insects are fed with feeds from side streams instead of industrially produced feeds. If avoided emissions are considered, e.g., by using flies to treat organic waste, the lowest GWP impacts are gained (Table 6). Table 7 summarizes the best practices for our three scenarios. In all casesexcept for the maximum and mean values in scenario II where broilers are fed with insects produced with industrially composite feedthe GWP is lower than for the base case of broiler production. If we exclude from our calculations the high GWP values for the H. illucens species produced with industrially produced composite feeds (Tables 4 and 5), then we obtain for all scenario cases that the GWP of production insects as a protein source for feed and food is lower than the GWP of conventional broiler production. Furthermore, when the GWP values from the C-LCAs are considered, the results are net-positive in all cases.

Discussion of scenario results and data validity
Most of the cases in scenarios II and III with insects fed with feed from industrial side streams suggest that the protein produced causes less GWP than protein production in scenario I. The values in scenario III suggest that insects are a good alternative protein source to reduce GHG emissions. The mean values in scenarios II and III are 6.3%, 50.0% and 67.8% lower than the mean value in scenario I, respectively. However, an even greater reduction potential exists, when using the best practices of insect production (Table 7) or when larvae are used to process organic waste (e.g., Smetana et al., 2019). By considering avoided impacts, the use of insects as animal feed can substantially contribute to the sustainability of the broiler production system with regard to lowering GWP. Our scenarios suggest that even net-positive impacts are possible.
However, although the data on insects used in this study is recent, it is relatively heterogeneous ( Table 1). The LCA studies on insects that were selected for scenario assessments differ in the use of functional units and allocation methods. Further, the studies apply different types of LCA (attributional LCAs (A-LCA) and consequential LCAs (C-LCA), see section 2.3.), which has an impact on the way the production processes around the system are investigated, while including or excluding considerations of the avoided impacts. Thus, our study should not be understood as a comparison between systems, but rather as a crossevaluation of the systems. Furthermore, the scenarios presented provide indicative estimates for the GWP of broiler and alternative insect protein consumption. Especially because the included peer-reviewed insect-related LCA studies are in many cases describing pilot scale, small scale, or even hypothetical insect production facilities, the results need to be interpreted with caution. On the other hand, state-of-the-art technologies for insect mass production are still in their early development stages (Cortes Ortiz et al., 2016), and therefore we could expect further considerable GWP improvements in the future.
The data sets we generated include non-ideal and ideal growing practices and processes, thus likely providing a realistic reflection of actual production practices. For instance, some of the high impacts are due to non-ideal growing conditions or due to the fact that the feed employed is not the most preferred one with respect to GHG emissions. Some studies like Bava et al. (2019) explore the production of flies by using industrial composite feed made for broiler production. In contrast, there were studies evaluating the impacts by using renewable energy sources and feed sources that are associated with low emissions (e.g., Joensuu and Silvenius, 2017;Smetana et al., 2019).
The data describing the GWP of soybean meal production is taken from studies focusing on soybean production in Southern America, which consider emissions from land use changes. By substituting nonsustainable practices with sustainable soybean production, the GWP reduction would be lower in scenarios II and III. For instance, if certified soybean that was not farmed in tropical areas was used instead of Brazilian soybean typically farmed in such areas, then the minimum value of soybean meal would be approximately 47.0% lower (e.g., Hörtenhuber et al., 2014). Beyond such obvious GWP reductions, our analysis including the comparison of insects-based protein and soybean meal from South America adds further insight with regard to emissions from land use, considering that the increasing soybean field expansion and associated deforestation as a result of incentives to produce more food and feed is likely not going to disappear any time soon (e.g., Fehlenberg et al., 2017), considering ongoing significant social sustainability issues (e.g., Brockhaus and Di Gregorio, 2014), and considering the fact that a major part of EU imports of soybean meal continues to come from South America (van Gelder et al., 2008;Karlsson et al., 2021).

Conclusions, limitations and further work
This study has explored the potential of insect-based protein to reduce the carbon footprint associated with food consumption in Europe through a quantitative review, focusing on Global Warming Potentials (GWP). Overall, it supports previous research suggesting that insect protein has the greatest potential to reduce the carbon footprints of European consumers, if insects are directly consumed as food. However, the results from our scenario analyses suggest that insect proteins considered may not necessarily perform better with regard to their GWP compared to broiler proteins. Depending on location of production and insect species used, the GWP of insect protein may be larger than that of broiler protein, even if insects are used for direct human consumption. In such a context, our study reveals that current practices of insect production for feed purposes are not yet efficient enough to significantly contribute toward a GWP reduction in European food consumption. As such, the study indicates that a precondition to achieve a GWP reduction of protein consumption is that side streams will need to be shown to be safe and thus allowed to be used as insect fodder. Even if that would turn out not to be the case, a careful selection of insect species for human consumption could unleash significant environmental sustainability benefits. Besides hygiene and legislation, also technological development choices can have an important impact on the potential to which insect-based protein may reduce food-based carbon footprints. Especially the energy sources affect potential climate benefits, and efficient upscaling is required for financial affordability to consumers.
Since insect-based feed has overall been found to have a lower potential to reduce the GWP compared to direct use of insect-based protein for human food, more research on product development, consumer acceptance and legal aspects of insect-based proteins will be decisive for the exploitation of the potentials of insect proteins in the future. In spite of the recently revised regulation on novel foods which has from June 1st, 2021 authorized the placing on the market of dried yellow mealworm as a novel food, in spite of other legal challenges that are currently in place for health reasons (e.g. EU Regulation No, 2016/429 on transmissible animal diseases), and in spite of undeveloped production practices and questions related to consumer willingness to use insects, there seems to be a clear market potential, especially if stakeholders cooperate (Cortes Ortiz et al., 2016;van Huis, 2020). Considering the potential benefits with which insect-based proteins could further contribute with respect to land use, GWP, and food security (Alexander et al., 2017;van Huis, 2013;Salomone et al., 2017), and considering these in the context of the EU's Circular Economy and Green Deal strategy (Domenech and Bahn-Walkowiak, 2019;Ghinoi et al., 2020;Johnson et al., 2021), it seems likely that insects will contribute more significantly with solutions for decreasing the burden of Europe's food systems on the environment. In particular, there seem large scale potentials for utilizing manure (Scarlat et al., 2018), and for utilizing biowaste from vegetable and animal sources (Lorenz et al., 2013). Both of these waste sources could be utilized as feed ingredient for insects, such as through flies, yet hygiene and safety issues remain to be resolved. If wastes were allowed to be used in insect production, insects could enhance food security with limited environmental impacts as part of novel circular economy strategies. As our results suggest, in best-case-scenarios, insects could even provide net-positive impacts, yet future research must demonstrate that side streams are safe to use.
On top of the above issues, there are also technological and thus efficiency aspects related to upscaling to be resolved, before mass production of insects becomes feasible (e.g., Cortes Ortiz et al., 2016;van Huis, 2020;Ites et al., 2020). To anticipate some of these technological potentials and to see what kind of potentials for reducing GWP there might be in the future with regard to insect production practices, the scenarios in this paper were based on only the best results of each insect species, which showed considerably lower GWP emissions in each scenario using insects (Tables 6 and 7). The results of the best-case scenarios suggest that there is potential in all of the insect species studied, yet the variation of the GWP values among different species is large.
In spite of the above insights, our paper suffers from a number of limitations. First, and while we acknowledge the need to include a multiindicator approach to environmental sustainability, our analysis is limited to the extent that it focuses on GWP as the single impact category. Further, the data used for this study is relatively heterogeneous, largely due to the diversity of available comparable studies underlying the analysis. Thus, the assessments presented in this study should be considered indicative (see e.g., Foster et al., 2006;Wiedemann et al., 2020;Röös et al., 2014), although our scenario analyses aimed to contribute to the robustness of the results. In light of the above data limitations, the results presented should not be understood as a comparison between production systems but rather as an approximate cross-evaluation of the systems. This is also the case because of the significant heterogeneity underlying broiler and insect-based food production. For example, differences in chicken breed, intensity of production, feeding programs, duration of chicken's life cycle, etc. will lead to differences in the carbon footprint of the final product (Castellini et al., 2012;Williams, 2021;Pelletier, 2008;Boggia et al., 2010). Similar heterogeneities apply to the insect-based food industry ( van Huis, 2016;Payne et al., 2016;Thévenot et al., 2018;Hunts et al., 2020;Pippinato et al., 2020). Furthermore, and despite their novelty, the scenarios constructed in this study need to be interpreted with caution, since they are indicative and relate to the carbon footprint of broiler and insect protein only. This is also the case because a sensitivity analysis in the form of Monte Carlo-based sensitivity analysis (e.g., Ponstein et al., 2019) could not be performed due to the nature of the underlying data. In spite of these limitations, we hope that this study's results provide a basis for an urgently needed clarification of the global warming impacts of broiler versus insect-based protein consumption and production (Ellingsen and Aanondsen, 2006;Weindl et al., 2020), for making a contribution to transitioning toward more sustainable food systems.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.