Bioenergy crops, biodiversity and ecosystem services in temperate agricultural landscapes—A review of synergies and trade‐offs

The Paris agreement on climate change requires rapid reductions in greenhouse gas emissions. One important mitigation strategy, at least in the intermediate future, is the substitution of fossil fuels with bioenergy. However, using agriculture‐ and forest‐derived biomass for energy has sparked controversy regarding both the climate mitigation potential and conflicts with biodiversity conservation. The urgency of the climate crisis calls for using forests for carbon sequestration and storage rather than for bioenergy, making agricultural biomass an attractive alternative for fossil energy substitution. However, this calls for comprehensive assessments of its sustainability in terms of consequences for biodiversity and ecosystem services. In this review, we provide a first holistic overview of the impacts on ecosystems of land‐use changes from bioenergy crop production in temperate climates, by synthesizing results on both biodiversity and ecosystem service impacts. We found that bioenergy‐related land‐use changes can have both positive and negative effects on ecosystems, with original land use, bioenergy crop type and scale of bioenergy production being important moderators of impacts. Despite the risk of opportunity cost for food production, perennial crop cultivation on arable land had the lowest occurrence of negative impacts on biodiversity and ecosystem services. Growing biomass for bioenergy on surplus land has been suggested as a way to alleviate competition with food production and biodiversity conservation, but our results demonstrate that utilizing marginal or abandoned land for bioenergy crop production cannot fully resolve these trade‐offs. Furthermore, there is a lack of empirical studies of the biodiversity value of marginal and abandoned land, limiting our understanding of the sustainability implications of biomass cultivation on surplus land. We argue that future research and policies for bioenergy production must explicitly consider biodiversity and ecosystem services in combination to avoid potential trade‐offs between the two and to ensure sustainable bioenergy production.

The urgency of the climate crisis calls for using forests for carbon sequestration and storage rather than for bioenergy, making agricultural biomass an attractive alternative for fossil energy substitution. However, this calls for comprehensive assessments of its sustainability in terms of consequences for biodiversity and ecosystem services. In this review, we provide a first holistic overview of the impacts on ecosystems of land-use changes from bioenergy crop production in temperate climates, by synthesizing results on both biodiversity and ecosystem service impacts. We found that bioenergy-related land-use changes can have both positive and negative effects on ecosystems, with original land use, bioenergy crop type and scale of bioenergy production being important moderators of impacts. Despite the risk of opportunity cost for food production, perennial crop cultivation on arable land had the lowest occurrence of negative impacts on biodiversity and ecosystem services. Growing biomass for bioenergy on surplus land has been suggested as a way to alleviate competition with food production and biodiversity conservation, but our results demonstrate that utilizing marginal or abandoned land for bioenergy crop production cannot fully resolve these tradeoffs. Furthermore, there is a lack of empirical studies of the biodiversity value of marginal and abandoned land, limiting our understanding of the sustainability implications of biomass cultivation on surplus land. We argue that future research and policies for bioenergy production must explicitly consider biodiversity and ecosystem services in combination to avoid potential trade-offs between the two and to ensure sustainable bioenergy production.

| INTRODUCTION
Bioenergy is promoted as a substitution of non-renewable energy, to reduce greenhouse gas emissions and mitigate climate change (IEA, 2021). To meet the committed emission reductions under the Paris Agreement, the European Union (EU) has adopted a European Climate Law (Regulation 2021/1119; EU, 2021) with a legally binding target of climate neutrality by 2050, and a specific objective for the transport sector to achieve a 90% reduction in greenhouse gas emissions by 2050. Achieving these climate objectives can partly be done by a rapid expansion of renewable energy sectors. Bioenergy is currently the main source of renewable energy in the EU, and it is projected to constitute a significant share of renewable energy also in the future (EC, 2019;IPCC, 2018). According to the EU and IRENA (2018), as much as 55% of renewables in the EU could come from biomass in 2030. With the war on Ukraine affecting energy systems globally, bioenergy production could increase even further through the adoption of renewable alternatives to replace Russian natural gas in Europe (Osička & Černoch, 2022). At the same time, the war is also affecting the global food market, which creates incentives for increased domestic food production to ensure food security (Hellegers, 2022). In combination, this may exacerbate the competition for land for food and energy production. The sustainability of forest and agricultural biomass as a source of energy has, however, been the focus of a heated debate because of questionable climate mitigation contributions, accompanying land-use changes, negative effects on food security, as well as biodiversity and environmental impacts (cf. Dornburg et al., 2010;Kalt et al., 2020;Vera et al., 2022). In addition, the urgent climate crisis requires an increase in carbon stocks both short term and long term, why forests may have a larger role in carbon sequestration and storage, rather than fossil fuel substitution (Searchinger et al., 2022;Soimakallio et al., 2022). This suggests that agricultural biomass is crucial for fossil fuel substitution in the coming decades (Searchinger et al., 2022). To mitigate the negative consequences that could arise from agricultural bioenergy production, attention has been turned to the crop choices, management, land-use conversions and landscape dynamics involved in biomass cultivation (Bourke et al., 2014;Dale et al., 2011).
The first generation of bioenergy feedstock constitutes annual row crops, such as maize, wheat and oilseed rape, which are traditionally used for producing food and feed. The production of first-generation (1G) bioenergy crops has been criticized for driving up food prices and threatening food security, as well as for dislocating food production to other regions where the accompanying land-use changes (indirect land-use changes) could have a substantial impact on social and environmental values (Tilman et al., 2009). Focus has therefore shifted, both within research and the EU Renewable Energy Directive, Directive (EU) 2018/2001 (RED II), to second-generation (2G) feedstock that is dedicated crops for energy production unsuitable for human consumption, such as perennial lignocellulosic trees or grasses (Valentine et al., 2011). Yet, 2G energy crops account only for a fraction of the bioenergy crop production in the EU (Material Economics, 2021). A transition from 1G feedstock to 2G feedstock can partly solve the sustainability conflicts, not only by decreasing competition with food production but also as low-input perennial crops in general support more ecosystem services and biodiversity (Holland et al., 2015;Immerzeel et al., 2014). Furthermore, if these perennial crops are grown on surplus land, such as marginal or abandoned land, competition with food production could be reduced further (Dauber et al., 2012;Gopalakrishnan et al., 2011;Tilman et al., 2009). However, land can be defined as marginal based on management, biophysical characteristics or socio-economic context, which have resulted in a plethora of definitions, framings and applications of marginal land across spatial and temporal scales, regions and actors (Csikós & Tóth, 2023;Muscat et al., 2022). In an agricultural context, marginal land is fields in rural areas with unfavourable environmental conditions, where agricultural production is not cost-effective due to high production costs and limited productivity (Csikós & Tóth, 2023). Marginal land is often still used for agricultural production, or it can currently be non-managed and thus left idle, in fallow, or maintained as set-aside (Khanna et al., 2021). If there are no incentives for landowners to maintain any agricultural production, either due to socio-economic, environmental or biophysical factors, these fields will be abandoned (Dauber et al., 2012).
Using biomass as a source of energy places an increasing demand on the services provided in and by ecosystems . Large-scale bioenergy deployment can cause trade-offs between different ecosystem services, as well as with biodiversity (Humpenöder et al., 2018;IPCC, 2022;Smith et al., 2019), and with an increasing K E Y W O R D S biodiversity, bioenergy, ecosystem services, feedstock, land-use change, marginal land, sustainability demand for bioenergy, these aspects need to be highlighted and weighed against each other. Moreover, landuse change has been identified as one of the major drivers of global biodiversity loss (Purvis et al., 2019), and expansion of bioenergy crop production that displaces other land covers and uses could further exacerbate biodiversity declines (Hof et al., 2018;Humpenöder et al., 2018;Popp et al., 2017). The ecological impacts of land conversions extend beyond the scale of local fields, to also encompass negative externalities in the surrounding landscape, and unintended regional and global consequences because of teleconnected effects (Sun et al., 2017). Sustainable production of bioenergy and biofuel can only be ensured if there is a comprehensive understanding of how land-use changes to bioenergy crops affect the surrounding ecological systems and processes (Dale et al., 2010). The environmental and ecological impacts of bioenergy production systems on ecosystem services and biodiversity have been summarized by several previous studies (e.g. Donnison et al., 2021;Holland et al., 2015;Immerzeel et al., 2014;Tudge et al., 2021), but the combined impact on ecosystem services and biodiversity from bioenergy land-use changes is less studied. As biodiversity across spatial scales is critical for the delivery of ecosystem services, it is necessary to understand where and when synergies and trade-offs arise, and how these are connected to environmental change (Le Provost et al., 2023;Mace et al., 2012).
To generate a more comprehensive understanding of the sustainability of agriculture-derived bioenergy, we review how bioenergy-driven direct land-use changes in agricultural landscapes in temperate climates affect both ecosystem services and biodiversity. We focus on temperate climates because these are most relevant in an EU policy context. We limit our review to studies that consider originally non-forest land uses or land types that have been suggested within policy and/or research as potential production sites for agricultural bioenergy (arable land, grasslands, marginal land, fallows, abandoned land; cf. Carlsson et al., 2017;Dauber et al., 2012;Directive (EU), 2018;Janiszewska & Ossowska, 2022), and we include both 1G and 2G bioenergy crops. By comparing land with more or less intensive agricultural production (arable land, grasslands, marginal land) and non-managed land without any current production (fallows, abandoned land), as well as annual and perennial crops, we provide an overview of how ecosystem services and biodiversity are affected under multiple potential bioenergy futures. The aim of the review was to (i) synthesize and discuss the impact and potential trade-offs or synergies from land-use changes induced by bioenergy crop production on biodiversity and ecosystem services; (ii) analyze and discuss patterns in bioenergy-driven land-use change impacts and its importance for interpreting the results; and (iii) identify research gaps for bioenergy and ecosystem service impacts from introduction of bioenergy crops in the agricultural landscape.

| Search strategy
We conducted a systematic literature search in ISI Web of Science Core Collection during September 2020 with a search string (Supporting Information S1) constructed to access peer-reviewed research articles on bioenergyrelated land-use change in agricultural landscapes and its impact on biodiversity and ecosystem services. The search terms included keywords related to bioenergy and land-use change, in combination with biodiversity or ecosystem services. We screened all research fields and only included impact studies that had quantified or estimated impacts from converting a reference land use to a bioenergy crop. Hence, non-explicit bioenergy crops (such as row crops used for food or fodder) were not targeted but included when they were grown as 1G crops for energy purposes. The search generated a total of 356 hits that were screened for relevance based on a pre-defined decision tree (Figure 1). The initial screening was based on the article title and abstract and resulted in 143 potentially relevant studies. Out of these papers, all but one could be accessed, and these 142 studies were thoroughly read in a full-text review using the same decision tree as referred to above. From this literature, reviews were excluded to avoid double counting. Ultimately the procedure resulted in a total of 54 original research papers included in the review ( Figure 1).

| Review process
To structure the findings, we created a review matrix (Supporting Information S2) where each study was characterized based on taxa of biodiversity, examined ecosystem service(s), reference land use, bioenergy crop or feedstock introduced, data and methodology used, geographical information, production scale, as well as the general impact on biodiversity or ecosystem services. We subdivided studies into cases when they contained distinguishable types of land-use changes or measured impacts so that combinations of specific impacts from specific land-use changes were used as input for the analysis. The reference land was categorized into four groups: arable land, grassland, marginal land and non-managed land. These are all rather heterogeneous groups with both land uses and types of land within land uses, and for the sake of simplicity, they will all be referred to as land uses throughout this paper. This subdivision was based on the existing body of literature, but also on the reference land uses discussed as potential land for biomass production in the EU policy context (e.g. Janiszewska & Ossowska, 2022;Kluts et al., 2017). Despite their heterogeneity, these reference land uses provide an overview of the complexity of land-use change impacts from general policy recommendations. The categorization into reference land uses was largely based on the definitions made by the authors of the reviewed literature (for details on the categorization, see Supporting Information S3). The taxa studied as biodiversity indicators were divided into the categories of amphibians, birds, reptiles, mammals, invertebrates, plants, microorganisms and, in cases where no explicit organism group was studied, unspecified biodiversity. The category unspecified biodiversity included studies assessing impacts on farmland biodiversity conservation, mean species abundance (MSA) for total change in biodiversity and risk for high nature value areas. We categorized the studied ecosystem services according to the Millennium Ecosystem Assessment classification as provisioning, regulating, supporting or cultural ecosystem services (Millennium Ecosystem Assessment, 2005). However, we retained the specific ecosystem service in the matrix as input for the analysis.
We categorized data and methodologies used in each article by separating direct and indirect measurements, based on the type of data and analytical approach of the studies. Studies based on direct measurements included empirical studies such as surveys, transect walks, field experiments or sampling. Studies based on indirect measurements were generally not based on empirical collection of data but mainly consisted of modelling studies, simulations and life cycle analyses based on secondary data.
The scale of production and the scope of each study were categorized into local and regional scales, respectively. Local-scale studies were performed at one or several fields but conducted at a rather small spatial extent (plot-level or field-level measurements). Regional-scale studies assessed the impacts of a large-scale introduction of bioenergy crops for the extent of a landscape or region.
Following Holland et al. (2015), we categorized the general impact of the bioenergy-related land use for each case into a simplified impact assessment using the following categories: (i) positive impact; (ii) negative impact; and (iii) or neutral/mixed impact (if the case either concluded that the impacts were neutral or if the amount of positive and negative impacts reported were similar). After summarizing the cases, the total impact for specific ecosystem services and biodiversity taxa was then classified into five categories, following Immerzeel et al. (2014), as either a strong positive impact (if 75% or more of the cases reported a positive impact), moderate positive impact (if more than 50% but less than 75% of F I G U R E 1 The review procedure with the two-step screening process (title and abstract screening to the left, and full-text screening to the right) using a decision tree. The letter 'n' represents the number of studies. ES, ecosystem services; LUC, land-use change; WoS, Web of Science. The numbers in the text boxes represent the order of the review process steps.

No
Records identified through searching in WoS (n = 356) Has the study investigated impacts from agricultural bioenergy crop production on biodiversity or ES?
Has primary data been used to quantify/estimate impacts?  Identification Screening Eligibility Included the cases reported a positive impact), no or mixed impact (no impacts recorded or equally many positive and negative cases reported), strong negative impact (if 75% or more of the cases reported a negative impact) or moderate negative impact (if more than 50% but less than 75% of the cases reported a negative impact). To demonstrate the evidence basis for each land-use change assessment, the number of cases was indicated for each total impact. As the included studies have used different methods, and considered different taxonomy, land uses and spatial scales, a quantitative effect-size approach was not possible. Instead, we have qualitatively assessed and described in our results how likely it is that different factors have influenced results.

| Overview of the included research
In total 54 studies, published between 1994 and 2020, met the selection criteria ( Figure 1). From these articles, we retrieved 272 individual combinations (cases) of land-use changes and impact assessments (Supporting Information S4). Although the focus was on research conducted in all temperate regions, this review captured studies exclusively from Europe (25 studies) and North America (29 studies). A majority of the articles and 71% of the cases assessed impacts from the introduction of 2G bioenergy crops ( Figure 2). The most common 2G crops were different types of energy grasses, such as Miscanthus (Miscanthus × Giganteus), prairie grass (mixed species) or switchgrass (Panicum virgatum), followed by short rotation coppice (SRC) with willow (Salix spp.) or poplar (Populus spp). 1G energy crops comprised 28% of all the cases and the main 1G crop was maize (Zea mays). In European studies, there was a higher proportion of articles focussing only on 1G crops compared to North American studies ( Figure 2).
In the European studies, biodiversity assessments represented more than half of the studies (56%), whereas in North American studies, more than two thirds of the studies (69%) focussed on ecosystem services (Figure 2). Only four studies considered biodiversity and ecosystem services simultaneously. Overall, a large majority of the 272 cases covered ecosystem service impacts (76%). The most researched category was by far regulating ecosystem services, representing more than half of all ecosystem service cases (64%). Climate regulation stood out as the most investigated ecosystem service, with 35 cases (Figure 3a). Other well-investigated ecosystem services were the regulation of water quality (30 cases), provisioning of biomass and energy (26 cases), regulation of soil quality (23 cases) and erosion regulation (20 cases). The least studied category of ecosystem services was cultural ecosystem services, represented by only four cases. Among the biodiversity assessments, birds were the most studied taxonomic group. Out of the 25 articles that assessed biodiversity impacts, birds had been studied in 12 of these (corresponding to 28% of the 67 biodiversity cases). The least commonly studied taxa were microorganisms, amphibians and reptiles, each comprising only 3% of the biodiversity cases ( Figure 3b). The most used indicators for assessing the impact of bioenergy production on biodiversity were species richness and species abundance. Less common indicators were based on modelling or scenario assumptions for bioenergy, including quantitative indices such as changes in the species range and MSA, as well as qualitative biodiversity descriptors, such as intrinsic biodiversity value, and general outcome for biodiversity conservation. Studies assessing general outcomes for biodiversity conservation were for example based on species sensitivity scores, expert opinions and existing data.
The most common reference land use for both ecosystem service and biodiversity assessments (56% of all cases) was arable land, that is, land currently in use for production of annual crops (Figure 4). Regarding studies on ecosystem services, grassland was the second most common reference land use (23%), including ley fields, pastures and rangeland, native prairie, perennial energy grass fields or other types of natural or semi-natural grasslands. In contrast, the biodiversity impacts of planting energy crops on grasslands were only analysed in four cases. Conversion of marginal land was assessed in 49 cases. In these studies, marginal land was defined as (i) arable land used for production but characterized by low productivity; (ii) land with poor soil quality; (iii) land with high sensitivity to erosion or floods and (iv) land with other constraints for production. The remaining 12 cases assessed the impact of converting nonmanaged land, that is, set-asides, herbaceous buffers without management and abandoned farmland.
The type of methods used was, in all studies but one, directly linked to the spatial scale of feedstock introduction. Regional-scale studies were mostly based on indirect measurements, and it was by far the most common analytical approach and study scale in both ecosystem service and biodiversity assessments ( Figure 5). Notably, the proportion of negative impacts of introducing bioenergy crops on biodiversity was higher among the F I G U R E 3 Effects from land conversion to energy crop production on (a) ecosystem services, and (b) biodiversity, recorded from reviewed literature. The impacts are classified as positive, neutral/mixed or negative. NPP, net primary production. large-scale assessments based on indirect measurements (about 55% of cases), compared to local-scale studies based on direct measurements (less than 10% of cases). The same difference was not evident among studies on ecosystem services, where the proportion of negative impacts ranged from 25% to 30% independent of the spatial scale. Two of the three studies that included measurements on both biodiversity and ecosystem services concerned regional-scale studies, based on indirect measurements, whereas the third study was based on empirically collected data (direct measurement) in a field/ landscape-scale study.

| Impacts on biodiversity from bioenergy-driven land-use change
Compared to ecosystem services, the body of research on biodiversity impacts from bioenergy-driven land-use change in temperate climates is small, constraining the ability to generalize conclusions. In general, and as further detailed below, we found that an introduction or intensification of the cultivation of 1G energy crops in agricultural landscapes rarely has positive impacts on biodiversity, whereas cultivation of 2G crops has the potential to benefit several taxa when strategically placed.

| Arable land
Conversion of arable land to 2G crops had an overall positive effect on invertebrates, microorganisms, plants and

F I G U R E 4
Land-use change impacts on biodiversity and ecosystem services, categorized by reference land-use type.

F I G U R E 5
Land-use change impacts on biodiversity and ecosystem services, separated by the scale of production considered.
unspecified biodiversity (Figure 6). Amphibians and reptiles were only represented by one case each, both showing a negative response to arable land conversion ( Figure 6). Birds and mammals showed mixed responses to 2G crop introduction on arable fields, with equally many cases demonstrating positive and negative impacts. However, the positive impacts on birds and mammals were all associated with cultivation of energy grasses, while negative impacts were mainly linked to woody bioenergy crops. Cultivation of 1G crops or a mix of 1G/2G crops on arable land was covered by fewer cases than 2G crop introduction, with more mixed results ( Figure 6).

| Marginal land
The biodiversity impacts from converting agricultural marginal land to energy crop production were overall negative, in particular regarding 1G crop cultivation ( Figure 6). Birds and invertebrates (and unspecified biodiversity) were consistently negatively affected when 1G crops were introduced. A change from marginal land to 2G crops showed potential benefits for invertebrates, but no or mixed effects on birds and unspecified biodiversity. Amphibians, mammals and reptiles were only represented by one case, suggesting negative responses to the introduction of energy grasses. All cases of marginal land conversion were based on indirect measurements (such as modelling) and considered large-scale land conversions.

| Grassland
Conversion of grasslands to bioenergy crops, or changed management regimes on grasslands, was covered by very few cases ( Figure 6). The three cases studying unspecified biodiversity suggested a negative impact of converting grassland to 1G crops, but a mixed impact of 2G crops on grassland (one positive impact case and one mixed impact case). Similarly, one case suggested benefits to plant diversity and mixed impacts on invertebrates of biomass harvest (2G crops) on grassland. None of the investigated grasslands was identified as sites with high-nature value, and the cases that studied impacts on invertebrates and plants concerned the effect of harvesting previously unharvested grasslands.

| Non-managed land
The effects of converting non-managed land into bioenergy crop production were also only studied in a few cases. According to one case each, the conversion of setasides to 1G corn/rapeseed had a negative impact on birds, but the conversion of abandoned land into SRC of willow had positive effects ( Figure 6). Also based on only one case, plants decreased, and invertebrates showed mixed results from the conversion of set-asides into SRC of willow.
F I G U R E 6 Impact of land-use changes from reference conditions to firstgeneration (1G) and second-generation (2G) feedstock production on biodiversity, based on the number of cases. Impacts were assessed as positive, negative or neutral/mixed, and the bold numbers inside the cells (left) indicate the number of cases used in the impact assessment. The italic numbers inside the cell (right) show the division between direct measurements (top-right) and indirect measurements (bottom-right). Equally many positive and negative cases were considered as having mixed impact. Strong positive impact (≥75% of cases report a positive impact) Moderate positive impact (>50% and <75% of cases report a positive impact) No or mixed impact Strong negative impact (≥ 75% of cases report a negative impact) Moderate negative impact (>50% and <75% of cases report a negative impact) No data

Total number of cases Indirect measurements (modelling, LCA, scoring studies)
Direct measurements (empirical studies)

| Impacts on ecosystem services from bioenergy-driven land-use change
The impact on ecosystem services from the introduction of bioenergy crops in the agricultural landscape has been widely studied during the past decades, with much knowledge on certain specific ecosystem services and the conversion of active agricultural land of better quality. Less is known about introducing bioenergy crops on marginal land and non-managed land, but overall, our results show that cultivation of 2G crops on arable land or marginal land can benefit several ecosystem services.

| Arable land
There was strong evidence for an overall increased delivery of ecosystem services when perennial 2G crops are established on arable land previously used for production of annual crops (67 of 81 cases showed positive impacts). Both energy grasses and SRC/SRF of energy trees showed the same trend. Apart from increasing biomass and energy provisioning and contributing to climate regulation, the positive impacts included decreased erosion, decreased nutrient leaching, improved soil quality, increased habitat quality, higher water availability and increased pollination and disease and pest regulation (Figure 7). The main negative effects of converting arable land to feedstock production were reduced food and feed production and decreased net income for producers. Regarding the introduction of 1G crops on arable land, the impacts on ecosystem services were more mixed and had fewer assessments (26 cases) compared to 2G crops (87 cases). Yet, the results indicate that using arable systems to produce 1G crops increased primary production, climate regulation, biomass and energy provisioning, but decreased the potential for erosion regulation, and had negative effects on water quality and aesthetical values.

| Marginal land
Possibly because of the broad definition of marginal land, the results showed that some negative impacts on crop production remain if marginal land was converted to energy crop production ( Figure 7). Moreover, negative impacts on climate regulation could occur through increases in greenhouse gas fluxes and decreasing carbon storage. Nevertheless, introducing 2G crops on marginal land has the potential to increase the overall delivery of ecosystem services (Figure 4), not only by contributing with biomass for energy, but also by providing additional environmental benefits, such as reduced erosion, improved soil and water quality and improvements to recreational opportunities (Figure 7).

| Grassland
The conversion of grasslands to energy crops was associated with the highest fraction of negative impacts on ecosystem services (Figure 4). In particular, the conversion of grasslands to 1G crops was negative for all ecosystem services, except for income generation (Figure 7). For conversion to 2G crops or intensified use and harvest from existing grassland, the impacts were more mixed, as one perennial land use was replaced by another one, but often with the addition of fertilizers and pesticides. This increased the production and hence biomass and energy provisioning, primary production and soil quality ( Figure 7). However, increased external inputs were linked to negative impacts on water quality and water availability. Climate regulation, erosion regulation, pollination, food and feed production and flood regulation were not significantly affected by the land-use change or showed mixed impacts.

| Non-managed land
Finally, the category non-managed land, including setasides and abandoned farmland, was by far the least studied category, represented by only eight cases of ecosystem service impact assessments, all of which considered the introduction of 2G crops. Based on these few studied cases, converting non-managed land into 2G crops (mainly SRC of willow or poplar) could increase biomass and energy provisioning, and climate regulation, and contribute to increased habitat quality elsewhere. However, the same land conversion may decrease water quality, but have no or little effects on food and feed production, and soil quality (Figure 7).

| DISCUSSION
In this review, we provide a first systematic overview of existing research on the land-use change effects on both biodiversity and ecosystem services following bioenergy crop introductions in temperate agricultural landscapes. With 272 unique impact assessments of land-use changes to bioenergy crops, our results point to the complexity of impacts and the difficulty of making generalized recommendations regarding where, how and what bioenergy crops should be grown. For practical reasons, we categorized studies into four groups based on the studied land-use types, but this also implies broad and heterogeneous reference systems including a high diversity of land and crops. With these heterogenous categories come uncertainties in the interpretation of results, for example, whether a lack of impact could be attributed to the inherent diversity of the study systems or to the fact that the compared land types have low ecological contrast (cf. Kleijn et al., 2011). We confirm that original land use and crop choice are critical moderators of the consequences of conversion to grow bioenergy crops on both ecosystem services and biodiversity, but we highlight that the response of the two is not always synergistic. However, biodiversity and many ecosystem services are highly interlinked such that the supply of many vital ecosystem services is promoted by the conservation of biodiversity across spatial scales within the agricultural landscape (Le Provost et al., 2023).
Small-scale production of perennial bioenergy crops on arable land is, in relative terms, most likely to provide benefits for both ecosystem services and biodiversity. In contrast, the share of negative outcomes for biodiversity of large-scale introduction of 2G bioenergy crops was high. This can be explained by the major land-use changes that large-scale introductions of bioenergy crops require (Dauber et al., 2010), but possibly also because biodiversity proxies at this scale capture effects on rare species that are less likely to be observed in small-scale assessments (Jeliazkov et al., 2022). Importantly, and as further discussed below, despite their interconnectedness ecosystem services and biodiversity have very rarely been evaluated simultaneously, limiting our understanding of the potential of biomass production for bioenergy on farmland and the trade-offs or synergies between biodiversity and multiple ecosystem services that may arise. Inadvertently, this F I G U R E 7 Impact of land-use changes from reference conditions to firstgeneration (1G) and second-generation (2G) feedstock production on ecosystem services. Impacts were assessed as positive, negative or neutral/mixed, and the bold numbers inside the cells (left) indicate the number of cases used in the impact assessment. The italic numbers inside the cell (right) show the division between direct measurements (top-right) and indirect measurements (bottomright). Equally many positive and negative cases were considered as having mixed impact. Strong positive impact (≥75% of cases report a positive impact) Moderate positive impact (>50% and <75% of cases report a positive impact) No or mixed impact Strong negative impact (≥ 75% of cases report a negative impact) Moderate negative impact (>50% and <75% of cases report a negative impact) No data

Total number of cases
Indirect measurements (modelling, LCA, scoring studies)

Direct measurements (empirical studies)
knowledge gap could contribute to unsustainable recommendations for bioenergy policies. The European agricultural landscape and its provision of biomass for energy are strongly shaped by policies for energy, climate, biodiversity and agriculture, creating conflicting interests over what land should be used for (Material Economics, 2021). Our results show that cultivation of 2G energy crops on arable land likely has the highest occurrence of synergies with ecosystem service delivery and biodiversity conservation, by positively affecting nine of the 13 assessed ecosystem services and four of the eight taxonomic groups. This is in line with earlier studies, showing that perennial biomass plantations on temperate arable land can mitigate negative environmental impacts (cf. Englund et al., 2020), while also enhancing biodiversity (cf. Dauber et al., 2010;Werling et al., 2014). However, introducing 2G crops on arable land is controversial because of potential competition between food and bioenergy production, with associated displacement effects. Utilization of marginal land or abandoned farmland is therefore suggested and promoted as sustainable production sites for bioenergy (Campbell et al., 2008;Gelfand et al., 2013;Gopalakrishnan et al., 2011;Tilman et al., 2009;Vera et al., 2022). Yet, we only found a few studies assessing environmental impacts from energy crop cultivation on non-managed agricultural land, and they present none or very few benefits for ecosystem services and biodiversity. These results are in line with recent findings from Crawford et al. (2022) and Valujeva et al. (2022), suggesting that reintegrating abandoned land with intensive agricultural production rather will remove land that is essential for ecosystem services (such as carbon sequestration) and biodiversity locally and regionally. More studies on and an enhanced understanding of how we best utilize abandoned farmland and other non-managed land is needed, to avoid generating conflicts between energy, climate and biodiversity targets.
In contrast, planting marginal land with perennial energy crops has been frequently studied and these land conversions may offer several environmental benefits, including increased erosion control, water availability and soil quality. However, our results show little or no benefits to biodiversity from the conversion of marginal land to energy crop cultivation. The existing literature on the biodiversity of temperate marginal land that we have identified are all non-empirical studies, focussing on a large-scale introduction of bioenergy crops. Hence, little is known about local impacts on biodiversity from marginal land conversion, limiting our knowledge also of upscaled consequences. With the current knowledge basis, in addition to unspecified biodiversity, four of the seven taxonomic groups included in this review (amphibians, birds, mammals and reptiles) were at least partly negatively affected by a 2G crop cultivation on marginal land. This contrasts with recommendations from earlier studies (e.g. Núñez-Regueiro et al., 2019;Vera et al., 2022), where marginal lands are pointed out as the biodiversity-friendly option for energy crop production. We believe that this at least partly can be attributed to the wide and imprecise definition of marginal land (Csikós & Tóth, 2023;Khanna et al., 2021;Muscat et al., 2022), as well as whether or not marginal land is portrayed as a sustainable alternative to the conversion of natural ecosystems (Núñez-Regueiro et al., 2019). The fact that marginal land is often characterized by physical production limitations and low-input farming, could make it valuable for biodiversity (Strijker, 2005), and in general allow them to host more endangered species (Kleijn et al., 2011). More generally, an important note is that the wide definition of marginal land can also include non-agricultural lands, such as urban areas, former industrial sites and polluted land (Blanco-Canqui, 2016;Mellor et al., 2021), which however falls outside of the scope of this review. Based on our results, we show that placing energy crops in temperate agricultural landscapes comes with trade-offs, either for food production or for biodiversity conservation. Potentially such trade-offs can be at least partly alleviated, for example, by combining food and energy production in farming systems that integrate biogas production (Koppelmäki et al., 2021;Parajuli et al., 2018), or integrating small-scale production of perennial biomass for energy in the arable landscape and on agriculturally marginal land (Carlsson et al., 2017;Dauber & Miyake, 2016), but effects on biodiversity of these solutions have yet not been fully evaluated. Using marginal land and abandoned farmland for bioenergy crop production could be in direct conflict with biodiversity conservation, and more sustainable land management could rather be to improve their value for biodiversity (Navarro & Pereira, 2012;Plieninger & Gaertner, 2011). Moving forward, our results point to the same conclusions as Dauber and Miyake (2016) and Miyake et al. (2015), that identification of land suitable for biomass production needs to move beyond definitions solely based on productivity, physical characteristics and socio-economic settings, to also include ecological importance and ensure synergies between ecosystem service provision and biodiversity conservation in the agricultural landscape. Sustainable integration of biomass production in the agricultural landscape could not only provide direct environmental and ecological benefits, but it also has the potential to prevent arable land abandonment, and hence loss of farmland biodiversity, in regions with less favourable socio-economic conditions (Heinsoo et al., 2010;Kleijn et al., 2011).
Not only the definition of marginal land is wide and ambiguous, but so is also the definition of grassland (Bengtsson et al., 2019), with the reviewed studies including many types of managed grass-covered land. The impacts of bioenergy crop introduction will depend on the environmental and ecological values or conservation needs of the focal grasslands, which are globally recognized for delivering multiple ecosystem services and acting as biodiversity hotspots in agricultural production systems (Bengtsson et al., 2019). However, we did not capture more than three studies that assessed the impacts on biodiversity from bioenergy crop introductions on grassland. This can be compared with ecosystem service assessments that were analysed in 12 studies. There are studies that assess biodiversity responses to, for example, intensified production or introduction of harvesting on grassland for fodder or protein purposes (e.g. Allan et al., 2015;Klein et al., 2020), but because they fall outside of the bioenergy framing they were not captured in this review, even if the biodiversity impacts may be similar. In addition, grasslands tend to be avoided in bioenergy scenarios with the argument that they are known for their biodiversity importance and hence a conversion of this land should not be encouraged (Directive (EU) 2018/2001. Because of the fragmented literature, we argue that interpreting the effects of converting grassland systems into bioenergy production systems should be made with caution, paying particular attention to the ecological contrast that a conversion would likely cause (cf. Kleijn et al., 2011). Our results show that the conversion of all types of grasslands to annual 1G crop production had an overall negative impact on both biodiversity and ecosystem services. The negative impacts are also associated with a grassland conversion to 2G crops, where we show that water quality and availability can be reduced. In contrast, extracting biomass for bioenergy based on introducing new management regimes can have both positive and negative environmental impacts. An introduction of biomass harvesting on previously unharvested grasslands was associated with increasing soil and habitat quality, as well as plant diversity, with positive impacts for some invertebrates (in addition to contributing to the provisioning of energy and primary production). In addition, harvesting biomass from conservation grasslands for bioenergy production can increase interest in conservation as well as provide landowners with a new source of income (French, 2019). However, an intensified production and harvesting of grasslands could also create unintended negative environmental effects. Studies by Allan et al. (2015) and Schils et al. (2022) show that the multifunctionality of grasslands decreased with increased management intensity, causing negative impacts on biodiversity and several ecosystem services that are dependent on biodiversity for their delivery.
When it comes to energy crop choices, our results demonstrate that most synergies for biodiversity and ecosystem services are associated with 2G feedstock, particularly energy grasses. Perennial energy crops are recognized for harbouring higher biodiversity and ecosystem service delivery than 1G energy crops (e.g. Pugesgaard et al., 2014;Tudge et al., 2021;Vera et al., 2022), but with our results we also draw attention to the potential tradeoffs for some taxonomic groups and ecosystem services from 2G crop introduction. For example, amphibians, birds, mammals and reptiles, as well as the provisioning ecosystem services income and food/feed production, were at least partly negatively affected by 2G crop introduction on arable or marginal land. As also highlighted in previous sections, land conversions to bioenergy crops are complex and the surrounding landscape context (Werling et al., 2014), land-use history and local biophysical conditions (Vera et al., 2022) are as important as crop choice when it comes to the synergies and trade-offs for biodiversity and ecosystem services (Werling et al., 2014). Dauber et al. (2015) also highlight that the positive biodiversity effects of perennial energy grasses could be overestimated as the habitat suitability likely will diminish as the cropping system becomes more established and intensified over time. In addition, the results of converting just one or a few fields to energy crops will likely not yield the same effects as increasing the cultivation of energy crops across larger spatial scales. As shown in our results, an introduction of bioenergy crops, especially 2G feedstock, has been shown by several studies to benefit biodiversity if they are grown on a small scale and contribute to increased landscape heterogeneity (Baum et al., 2012;Stanley & Stout, 2013). In contrast, large-scale production of energy crops potentially results in large monocultures and/or aggregated high-intensity production around power plants or biorefineries that could deliver negative effects on biodiversity (Dauber et al., 2010;Holland et al., 2015).
Our review shows that agricultural bioenergy research strongly focusses on ecosystem services, rather than biodiversity. Within the two research strands, there are clear patterns of what ecosystem services or taxonomic groups the research focusses on. The majority of studies on ecosystem services focus on climate regulation, biomass and energy provisioning or water quality, whereas few studies considered cultural ecosystem services or biodiversity-mediated supporting services, such as pollination and biological pest control. Regarding biodiversity, birds and invertebrates made up half of all cases, and together with the group unspecified biodiversity almost 70% of the reviewed cases. As there is a generally poor correlation between taxonomic groups (Wolters et al., 2006), a thorough understanding of a few taxa does not necessarily allow for any conclusion on other species groups. Thus, there is a limited understanding of how less studied ecosystem services and taxonomic groups could be affected by increased feedstock production as the functionality of different ecosystem services and the diversity of different taxonomic groups are dependent on different ecological processes and can be expected to respond differently to land-use changes. Therefore, we view separate taxonomic groups and specific ecosystem services as important indicators of ecosystem state (Brandt et al., 2014). In addition, many studies used a coarse biodiversity measure, such as unspecified biodiversity, which risks failing to understand the underlying ecological mechanisms that drive biodiversity responses to land-use change (Duncan et al., 2015;Tscharntke et al., 2012). Finally, and importantly, we found only four studies evaluating both biodiversity and ecosystem services, three European studies based on indirect measurements (Gutzler et al., 2015;Schulze et al., 2016;van der Hilst et al., 2012) and one North American empirical study (Spiesman et al., 2019). As an example, Gutzler et al. (2015) investigated a bioenergy scenario for Brandenburg, Germany, with 1G maize cultivation on arable land, and its effect on landscape scenery, water quality, energy provisioning, climate regulation, soil erosion and farmland birds. Thus, as the first review that includes impacts on both ecosystem services and biodiversity from land-use changes to bioenergy crop production, we have identified a clear need for additional empirical studies that include multiple measures of biodiversity and ecosystem services to obtain a comprehensive understanding of the consequences of bioenergy crop introduction on farmland multifunctionality (cf. Birkhofer et al., 2018). Given the expected role of bioenergy in the European energy transition to climate neutrality, it is a rather disturbing fact that the assessment of consequences is based on a body of research that does not allow appropriate coverage of the combinations of transitions possible or the multitude of consequences that needs to be evaluated.
In conclusion, we show both synergies and trade-offs between land-use changes to bioenergy crops on the one hand, and conservation of biodiversity and the provisioning of ecosystem services on the other hand, but the research on ecosystem service impacts and biodiversity impacts are strongly segregated. While the ecosystem service impact from perennial crop introduction on arable fields has been widely studied, far less is known of alternative production sites and the response of species in the agricultural landscape. In the past decade, attention has been turned to marginal land and abandoned farmland for energy crop production to avoid conflicts with food production and nature conservation. However, the ecological contribution of marginal land and non-managed land has barely been researched. This is highly concerning as they can be important wildlife habitats, and conversion of these could be in direct conflict with biodiversity conservation. As several studies also have done before us, we show that introducing bioenergy crop production into the arable landscape results in yet another player in the competition for scarce land resources and multiple societal goals to meet. Any large-scale transformation of agricultural land will imply trade-offs between different ecosystem services or between ecosystem services and biodiversity conservation, where synergies are more likely to occur when bioenergy production is integrated into existing farming systems (Carlsson et al., 2017;Parajuli et al., 2018). Moving forward, joint recognition of multiple ecosystem services and biodiversity in land-use decisions is needed to avoid goal conflicts within climate change mitigation (cf. Sidemo-Holm et al., 2021). To be able to tackle the biodiversity and climate crises jointly, the integration of the two agendas is crucial (Pettorelli et al., 2021). Hence, we argue for a more holistic understanding of the complex environmental and ecological impacts of land-use changes caused by energy crop production, to guide future research and support policymaking.

AUTHOR CONTRIBUTIONS
Josefin Winberg, Henrik G. Smith and Johan Ekroos developed the concept and design of the methodology. Josefin Winberg collected and analysed the reviewed studies and their data. Josefin Winberg led the interpretation of results and writing of the manuscript with support from Johan Ekroos and Henrik G. Smith. All authors gave final approval for publication.

ACKNO WLE DGE MENTS
This study was supported by a grant to Yann Clough from the Swedish Research Council FORMAS (FORMAS project number 2018-01726) and is a contribution to the Strategic Research Area Biodiversity and Ecosystem Services in a Changing Climate (BECC). We thank the editor and the three anonymous reviewers for their valuable comments and suggestions, which helped us to improve the quality of the manuscript.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The data used in this review is publicly available in the SI.