Benecial land-use change in Europe: deployment scenarios for multifunctional riparian buffers and windbreaks

The land sector needs to increase biomass production to meet multiple demands while reducing negative land use impacts and transitioning from being a source to being a sink of carbon. The new Common Agricultural Policy of the EU (CAP) steers towards a more needs-based, targeted approach to addressing multiple environmental and climatic objectives, in coherence with other EU policies. In relation to this, new schemes are developed to offer farmers direct payments to adapt practices beneficial for climate, water, soil, air and biodiversity. Multifunctional biomass production systems have potential to reduce environmental impacts from agriculture while maintaining or increasing biomass production for the bioeconomy across Europe. Here, we present the first attempt to model the deployment of two such systems, riparian buffers and windbreaks, across >81.000 landscapes in Europe (EU27 + UK), aiming to quantify the resulting ecosystem services and environmental benefits, considering three deployment scenarios with different incentives for implementation. We found that these multifunctional biomass production systems can reduce N emissions to water and soil loss by wind erosion, respectively, down to a “low” impact level all over Europe, while simultaneously providing substantial environmental co-benefits, using less than 1% of the area under annual crops


Introduction
Climate change, population growth and changes in per capita consumption will add further pressure on managed as well as natural and semi-natural ecosystems (IPCC, 2019). At the same time, many global scenarios that meet the targets set by the Paris Agreement include a rapid increase in the use of biomass for bioenergy, which is often envisaged in conjunction with carbon capture and storage (IPCC, 2018). Significant mitigation is also expected through changes in land use practices to reduce greenhouse gas (GHG) emissions and provide carbon storage in soils and vegetation (IPCC, 2019). In addition, more sustainable land-use strategies are anticipated regarding other environmental impacts, related to, e.g., water, soil and biodiversity, driven by various environmental policies (DeBoe, 2020). In short, the land sector needs to increase biomass production to meet multiple demands while reducing negative land use impacts and transitioning from being a source to being a sink of carbon.
Multiple policies are being developed in the EU to take on this double challenge. For example: (i) the European Green Deal (EC, 2019) has a vision of achieving multiple sustainability and climate neutrality goals by 2050. To decarbonize the energy system, the EU will, e.g., prioritize solutions based on sustainable bioenergy, that needs to comply with strengthened sustainability criteria (EC, 2018a); (ii) the Biodiversity Strategy for 2030 (EC, 2020a) includes general goals for the protection and restoration of nature and biodiversity, requiring high-diversity landscape features on at least 10% of the agricultural area; and (iii) the Farm to Fork Strategy (EC, 2020b) aims to promote the transition to sustainable food production, to protect the environment and preserve biodiversity, and to tackle climate change. Connected to all these policies is the new European Union (EU) Common Agricultural Policy (CAP) for the period 2021-2027, that aims to drive the sustainability transition in agriculture, and enable farmers to play a key role in tackling climate change, protecting the environment, and preserving landscapes and biodiversity (EC, 2018b).
Three of the CAP´s nine specific objectives will concern the climate and environment, namely (i) contribute to climate change mitigation and adaption, as well as sustainable energy, (ii) foster sustainable development and efficient management of natural resources such as water, soil and air, and (iii) contribute to the protection of biodiversity, enhance ecosystem services and preserve habitats and landscapes (EU, 2019). The new CAP also introduces "Eco-schemes" to incentivize farmers to adopt measures for contributing to the environmental and climate objectives. The schemes offer the possibility to grant direct payments for implementing practices beneficial for climate, water, soil, air, and biodiversity (EU, 2019). Member States will be free to define such schemes, but they are voluntary for farmers.
Multifunctional biomass production systems are designed, located and managed to reduce environmental impacts from agriculture while maintaining or increasing biomass production for the bioeconomy (Christen and Dalgaard, 2013;Styles et al., 2016;Ferrarini et al, 2017a;Englund et al, 2020a;Englund et al, 2020b). Such systems could therefore contribute to the objectives of the emerging EU policies and be eligible for compensation within the EU Eco-scheme framework. Recently, an assessment of the potential for multifunctional biomass production systems in Europe indicated a substantial potential for effective mitigation regarding soil loss by wind and water erosion, nitrogen emission to water, losses of soil organic carbon (SOC), and recurring floods (Englund et al. 2020a). This assessment did not, however, provide any quantitative estimates concerning large-scale deployment of specific systems, such as impact mitigation and biomass output.
Two examples of systems to explore further in this context are riparian buffers and windbreaks. Riparian buffers can consist of woody and/or herbaceous crops, located along watercourses with the primary purpose of retaining nutrients (Christen and Dalgaard, 2013;Ferrarini et al., 2017a;Styles et al., 2016). Buffer strips can deliver additional multiple benefits associated with the characteristics of the perennial crops, most notably flood mitigation and reduced streambank erosion. They are also likely to enhance SOC, if established on land historically used for annual crop production, enhance conditions for biodiversity by, e.g., improving landscape connectivity, and protect agricultural fields from wind and water erosion (Englund et al., 2020a). Windbreaks are strips of woody crops, such as poplar or willow, located within or between fields to protect agricultural land from wind erosion (Osorio et al. 2019). Similar to riparian buffers, multifunctional windbreaks may lead to co-benefits such as enhanced SOC and reduced nutrient leaching (Englund et al., 2020a). Both riparian buffers and windbreaks can be designed using high-yielding species and harvested for biomass, thus increasing land-use efficiency by maintaining agricultural productivity in the landscape while providing environmental benefits. Buffer strips are also recognised as acceptable options by the EU to help protect soil fertility and increase soil organic matter, reduce nutrient losses and soil erosion and enhance carbon sequestration. Buffer strips also contribute to ensure connectivity among habitats, thus avoiding habitat fragmentation through the implementation of the Birds and Habitats Directive (European Commission, 1979;European Commission, 1992).
Here, we model the implementation of multifunctional riparian buffers and windbreaks in >81.000 landscapes in Europe (EU27 + UK), aiming to quantify the resulting ecosystem services and environmental benefits, considering three deployment scenarios with different incentives for implementation. We also estimate GHG emissions savings from utilizing the biomass produced in the multifunctional systems to replace fossil materials, combined with increases in soil organic carbon. Finally, we discuss implications associated with the land-use change, such as for current agricultural production, and relevant land-use policies that may incentivize deployment .   3   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19   20  21  22  23  24  25  26 2 Results

Riparian buffers
A total of 7574 landscapes were identified as suitable for riparian buffers in the "Biomass" scenario, and 5705 landscapes in the "Low-impact" and "Food-first" scenarios. The difference is due to higher requirements of N emissions to water to incentivize buffers in the latter two scenarios. These landscapes are predominantly located in north-western Europe (Fig. 1). In most landscapes, SRC was identified as the highest yielding buffer option, with willow as the most suitable SRC species, in terms of productivity. In some countries, however (most notably France, Belgium, and Italy) grass was identified as the highest yielding buffer option.
The degree of N emissions to water in suitable landscapes, and the degree of impact mitigation using different buffer designs, varies substantially, both within and between countries (Fig 1; see also Englund et al. (2020a)). While the impact is already at a "low" level in 1870 of the 7574 landscapes in the "Biomass" scenario, it cannot even be reduced to a "low" level in 1031 other landscapes using only narrow buffers. Figure 1: Establishment of riparian buffers in the three deployment scenarios, and the share of maximum narrow buffer area needed to achieve "low" N emissions to water in each landscape.
In the "Biomass" scenario, 1.4 million hectares (Mha) of double-wide SRC buffers are established, corresponding to 4.6% of the area under annual crops in the affected landscapes and 1.3% of the total area under annual crops in EU28. These buffers result in about 900 kt of avoided N emissions to water, while delivering over 16 Mt DM biomass annually. (Table 2; Table S1) In the "Low-impact" scenario, there is a large spread in total buffer area, since farmers can freely decide which buffer option to implement. It ranges from about 69 kha (if only narrow buffers are implemented) to 431 kha (if only double-wide buffers are implemented), corresponding to 0.3-2.1% of the area under annual crops in the affected landscapes and 0.1-0.4% of the total area under annual crops in EU28. These buffers result in 371 kt of avoided N emissions to water, while delivering between 0.8-4.9 Mt DM of SRC biomass or 0.8-2.1 Mt DM of grass biomass. (Table 2; Table S1) In the "Food-first" scenario, a total of 101 kha of narrow (49 kha) and wide (52 kha) buffers are established, corresponding to 0.5% of the area under annual crops in the affected landscapes and 0.1% of the total area under annual crops in EU28. As in the "Low impact" scenario, these buffers result in 371 kt of avoided N emissions to water, delivering about 1.2 Mt DM SRC or grass biomass annually. (Table 2; Table S1)

Enhanced SOC
The potential of riparian buffers to enhance SOC in specific landscapes and in different scenarios depends on existing accumulated SOC losses and thus the potential to increase SOC by establishing perennials, combined with the total buffer area. Given the large variation of both factors, there is also a large variation in the degree to which riparian buffers can contribute to enhancing SOC (Fig. 2). In the "Biomass" scenario, given the large areas used for buffers, relative to the other scenarios, SOC increases are the greatest. By 2050, the SOC increase relative BAU amounts to almost 33 Mt C, corresponding to avoided GHG emissions of 4 Mt CO2-eq annually. In the other scenarios the corresponding numbers are 1.1-6.3 Mt C and 0.1-0.8 kt Mt CO2-eq/y for "Low impact" and 2.

Avoided soil loss by water erosion and sediment retention
In the "Biomass" scenario, riparian buffers directly avoid about 3.3 Mt of soil loss by water erosion, corresponding to 6.3% of soil loss by water erosion on cropland in landscapes where buffers are established, and 1.2% of total soil loss by water erosion on cropland in EU28. The median landscape contributes with 26% of the reductions in soil loss by water erosion necessary to achieve a "low" impact, at the landscape scale. They can also retain an additional 49 Mt of soil eroded on cropland outside the buffers. At EU28 level, 19% of all soil loss by water on cropland is retained in the respective landscapes. (Table 2; Table S1) In the "Low-impact" scenario, total avoided soil loss by water erosion ranges from 192 kt (only narrow buffers) to 1150 kt (only double-wide buffers), corresponding to 0.5-3.1% of all soil loss by water erosion on cropland in landscapes where buffers are established, and 0.1-0.4% of total soil loss by water erosion on cropland in EU28. The median landscape contributes with 2-11% of the reductions in soil loss by water erosion necessary to achieve a "low" impact, at the landscape scale. They also retain an additional 13-16 Mt of soil eroded on cropland outside the buffers, totalling 38-50% direct and indirect avoided soil loss by water erosion within the buffer landscapes At EU28 level, 5-6% of all soil loss by water on cropland is retained in the respective landscapes. (Table 2; Table S1) In the "Food-first" scenario, total avoided water erosion is 291 kt from narrow (126 kt) and wide (165 kt) buffers combined, corresponding to 0.8% of soil loss by water erosion on cropland in landscapes where buffers are established, and 0.1% of total soil loss by water erosion on cropland in EU28. The median landscape contributes with 2.2% of the reductions in soil loss by water erosion necessary to achieve a "low" impact, at the landscape scale. They also retain an additional 16 Mt of soil eroded on cropland outside the buffers, totalling 42.6% direct and indirect avoided soil loss by water erosion within the buffer landscapes. At the EU28 level, 5.6% of all soil loss by water on cropland is retained in the respective landscapes. (Table 2; Table S1) Although there is a notable variation between different landscapes, countries, and regions ( Fig. 3), riparian buffers are considered as having limited potential for reducing soil loss by water erosion at the European scale. However, this does not consider that the potential for avoiding streambank erosion, which has not been modelled, could be substantial. The potential for retaining eroded soil in buffers and thus avoiding sedimentation in watercourses appears substantial, especially within the buffer landscapes but also at the EU28 level. It should, however, be noted that sediment retention does not mitigate the negative effects of water erosion on eroded cropland.

Windbreaks
A total of 7483 landscapes are classified as having a "medium" or higher effectiveness concerning wind erosion mitigation. However, in 6315 landscapes, the impact is already at a "low" level. Given the assumptions for wind erosion mitigation potential, i.e., that windbreaks cannot reduce wind erosion beyond the threshold for a "low" impact, these landscapes are not subject to windbreak implementation in any of the deployment scenarios. Windbreaks are therefore implemented in 1168 landscapes, in all deployment scenarios. The largest modelled windbreak area is in Denmark, followed by the UK, the Netherlands, and Spain. As noted for riparian buffers, willow is typically higher yielding in northern Europe, while poplar is typically higher yielding in southern Europe (Fig. 4).
As for riparian buffers, the degree of soil loss by wind erosion varies substantially, both within and between countries (Fig 4; see also Englund et al. (2020a)). The median implementation level required to achieve a "low" impact level is about 16%, but with large variations; the implementation level is <1% and >100% in about 4% of all landscapes, respectively. In the "Biomass scenario, it is always 100%. In the "Low-impact" and "Food-first" scenarios, it is determined by the implementation necessary to reduce wind erosion to a "low" level but not beyond. impact" scenario, the highest yielding options are always used, in the "Food-first" scenario, the option is also affected by the ambition to minimize total windbreak area.
In the "Biomass" scenario, total windbreak area ranges between about 1,7-2.3 Mha. This corresponds to 1.6-2.1% of the current area under annual crops in EU28 and 30-33% of the current area under annual crops in the landscapes where windbreaks are established. Estimated wind erosion mitigation is about 5-13 Mt avoided soil loss annually, corresponding to 9-23% of total soil loss by wind erosion in EU28. Biomass production from windbreaks sums up to 3-24 Mt DM/y. (Table 3; Table S2) In the "Low-impact" scenario, the total windbreak area is significantly smaller, 185-555 kha, corresponding to 0.2-0.5% of the current area under annual crops in EU28, and 2.7-8.2% of the area under annual crops in the landscapes where they are established. Wind erosion mitigation is about 13 Mt avoided soil loss, or 23% of total soil loss by wind erosion in EU28, regardless of how the windbreak options are combined. Total windbreak biomass production is about 2-6 Mt DM/y. (Table 3; Table S2) In the "Food-first" scenario, the total windbreak area is 312 kha, of which 190 kha with SRC windbreaks and 212 kha with SRF windbreaks. This corresponds to 0.3% of the current area under annual crops in EU28, and 4.6% of the area under annual crops in the landscapes where windbreaks are established. As for the "Low-impact" scenario, wind erosion mitigation is about 13 Mt avoided soil loss, or 23% of total soil loss by wind erosion in EU28. Total windbreak biomass production is about 3 Mt DM/y. (Table 3; Table S2) As for riparian buffers, there is a notable difference between the "Biomass" scenario and the other two scenarios. However, in this case, the spatial deployment is the same in all scenarios. Here, the difference is instead solely explained by the difference in implementation level necessary to achieve the defined impact mitigation. In the "Biomass" scenario, buffers are implemented as defined in all landscapes, while in the other scenarios, windbreaks are only implemented to the extent where the impact is reduced to a "low" level at the landscape scale. In most landscapes, this means implementing windbreaks to a lesser extent than in the "Biomass" scenario, although in some landscapes to a considerably greater extent, as seen in Fig 4.

Co-benefits
Enhanced SOC As explained for riparian buffers, effects on SOC depend largely on the windbreak area. In the Biomass scenario, the total SOC increase is therefore the greatest, 38-45 Mt C by 2050, corresponding to 139-165 kt CO2-eq of annual GHG emissions. In the other scenarios, total SOC increases are 4-11 Mt C for "Low-impact" and 6 Mt C for "Food-first". Corresponding annual GHG emission savings are 15-40 and 22 kt CO"-eq/y, respectively. (Table 3; Table S2) Unlike for most of the assessed co-benefits in this study, windbreaks could potentially play a significant role in restoring SOC in landscapes where they are established (Fig. 5). This is particularly the case in the "Biomass" scenario, where a substantial share of current cropland is used for windbreaks. In the other scenarios, where the implementation level is, in general, more limited, it can still play a significant role in restoring SOC where wind erosion is severe and the implementation level high. In other landscapes, it can contribute to varying degrees to restoring SOC, depending on the implementation level. The contribution to restoring SOC could be further increased if the location of the windbreaks is rotated during replanting, since the positive effects on SOC decrease over time. At the European level, however, the positive effect on SOC is small, given that most agricultural landscapes are not subject to windbreak implementation in these deployment scenarios.

Avoided soil loss by water erosion
In the "Biomass" scenario, windbreaks avoid between 1.9 Mt (if only SRF windbreaks) and 3.2 Mt (if only SRC windbreaks) soil loss by water erosion. This corresponds to about 1% of total water erosion in EU28 but a more significant 20-33% of total water erosion in the landscapes where windbreaks are established (Fig 6). The median contribution of windbreaks towards reducing water erosion down to a "low" level is 95-112%. (Table 3; Table S2) In the "Low-impact" scenario, windbreaks avoid between 0.2 Mt (only SRF) and 0,6 Mt (only SRC) of soil loss by water erosion, corresponding to 0.1-0.2% of total water erosion in EU28 and 2-7% of total water erosion in the landscapes where windbreaks are established. The median contribution towards reducing water erosion to a "low" level is 4-12%. (Table 3; Table S2) In the "Food-first" scenario, windbreaks avoid 0.3 Mt of soil loss by water erosion, corresponding to 0.1% of total water erosion in EU28 and 10% of total water erosion in the landscapes where windbreaks are established. The median contribution towards reducing water erosion to a "low" level is 10%. (Table 3; Table S2) This indicates that, in principle, no further measures to reduce water erosion are necessary in landscapes with windbreaks, given the level of windbreak implementation in the "Biomass" scenario.
In the other scenarios, where the implementation level is generally lower, the role of windbreaks in reducing water erosion is less, albeit still, significant (Fig. 6). It should be noted that some landscapes have a greater reduction in water erosion in the "Low-impact" and "Food-first" scenarios, than in the "Biomass" scenario, as these scenarios allow for an implementation level >100% (Fig. 4). As for the other co-benefits, the contribution to reduced soil loss by water erosion is marginal on the European scale.
To estimate windbreak sediment retention, it is necessary to know the orientation of windbreaks relative slope, as this strongly influences the sediment trapping efficiency. While this is technically possible, windbreak sediment retention has not been assessed here.

Avoided N emissions to water
In many landscapes where windbreaks are established, their effect on N emissions to water is substantial (Fig. 7). This is particularly the case in the "Biomass" scenario, in which they suffice to reduce the impact to a "low" level in most landscapes, totalling 20-22 kt avoided N emissions to water.
In the other scenarios, the variation is large. In some landscapes, i.e., where wind erosion is severe and the implementation level high, N emissions are reduced to a "low" level or beyond, while in other landscapes, the contribution is only marginal (Fig. 7). This is especially seen in the "Food-first" scenario, where the windbreak area is optimized and thus the lowest. Total avoided N emission to water in the "Low-impact" and "Food-first" scenarios are 1,6-4,8 and 2,9 kt, respectively. (Table 3, Table S2 In many landscapes where windbreaks are implemented, N emissions to water is also high, indicating that windbreaks can be equally effective as riparian buffers in this respect. This exemplifies how one measure could suffice to simultaneously resolve two major environmental issues. This reduces the cropland area needed for impact mitigation and thus increasing land-use efficiency. It also highlights the need to focus on multiple objectives simultaneously and to adopt a landscape perspective.

Potential additional co-benefits and trade-offs
As discussed, several additional co-benefits of riparian buffers and windbreaks are possible, although difficult to quantify without taking more landscape-specific characteristics into consideration. To exemplify this, the likeliness of flood mitigation (by riparian buffers and windbreaks, respectively) and wind erosion (by riparian buffers), is indicated in Fig. 8. The likeliness of mitigating recurring floods is high or very high in 1/5-1/7 of all landscapes where riparian buffers are established and in 1/6 of the landscapes where windbreaks are established. As wind erosion is severe in much fewer landscapes than recurring floods, the likeliness of wind erosion mitigation by riparian buffers is generally lower; being high or very high in only 2-4% of the affected landscapes.

Figure 8: Likeliness of flood mitigation and wind erosion mitigation in riparian buffers and windbreaks.
These results illustrate that there may be several important additional co-benefits than what was possible to quantify on this large scale. To provide a more comprehensive, and precise, understanding of the benefits of strategic perennialization, it is therefore necessary to study a broad range of environmental aspects on a smaller scale. This also applies for possible negative effects of the otherwise "beneficial" LUC, as discussed below.

Discussion
Multifunctional biomass production systems in the form of energy crop cultivations, designed and utilised as riparian buffers and windbreaks, can lead to significant environmental benefits with limited negative effects on current agricultural production. These systems can reduce N emissions to water and soil loss by wind erosion in Europe down to a "low" impact level, while simultaneously providing substantial environmental co-benefits, utilizing less than 1% of the area under annual crops in the EU. The biomass produced in these multifunctional systems can at the same time be used to displace fossil materials and energy carriers in the emerging bioeconomy. The maximum total biomass production of SRC in riparian buffers and windbreaks amount to some 300 PJ and 450 PJ per year, respectively, in the Biomass scenario (see Tables 2, 3, S1, and S2), in this case utilizing 3.4% of the area under annual crops in the EU. The GHG emissions savings obtained from this biomass depend on what biobased product is produced (and how) and which other product is displaced. Assuming that the biomass is converted to either (i) Fischer-Tropsch diesel, methanol, and dimethylether (DME) based on farmed wood, displacing fossil diesel and gasoline, or (ii) electricity and heat, displacing coal and oil, respectively, the corresponding GHG emissions savings are 29-44 Mt CO2-eq/y (see SI for more information). In the same scenario, an additional 9,5 Mt CO2-eq/y is stored in soils until 2050 (annual average, see Tables 2, 3, S1, and S2). The total GHG emissions savings of riparian buffers and windbreaks, in this deployment scenario, consequently reach about 38,5-53.5 Mt CO2-eq/y, or 1-1.4% of total EU-28 GHG emissions in 2018 (Eurostat, 2020). However, it must be stressed that the results vary substantially between the different deployment scenarios: when the implementation is limited to achieve a certain mitigation level, but not beyond, the corresponding areas, biomass production, environmental benefits, and GHG emissions savings are significantly lower.
While based on high-quality and high-resolution data, and assumptions founded in scientific literature, the model is subject to several critical uncertainties and limitations, e.g., deployment scenario assumptions, thresholds used for classifying environmental impacts and impact mitigation effectiveness, and the assumption that co-benefits are linearly proportional to the share of annual crops used for the different systems. The results presented here should therefore be interpreted as indicative for large-scale deployment of riparian buffers and windbreaks, given the set of assumptions in the different scenarios. For a thorough description of background and input data, and classification of environmental impacts and mitigation effectiveness, see Englund et al. (2020a). Additional uncertainties and limitations specific to this study are presented in the Supplementary Information.
The specific outcomes of large-scale deployment of riparian buffers and windbreaks are very difficult to predict. However, our findings clearly demonstrate that it is possible to produce biomass feedstock together with multiple and significant environmental benefits, using a limited amount of land. The assessment and implications of these biomass plantation systems are, by no means, exhaustive. As the establishment of perennial crop plantation stripes on agricultural land generate variations in the landscape, they have direct benefits for biodiversity (Weih and Dimitriou 2012, Baum et al, 2012, Vanbeveren and Ceulemans, 2019, and can play important roles concerning the preservation of sensitive species such as semiaquatic amphibians (Ficetola et al, 2009 Furthermore, as these perennial energy crop plantations modify the moisture regime, micro-climate, vegetative structure, and productivity, depending on the management and location, they can act as barriers against fire (Petit and Naimen, 2007). However, at the same time, these stripes can accumulate dry biomass fuel and thereby become corridors for fire movement, which suggests additional considerations concerning management and species selection in fire sensitive areas. The effects on water availability in dry areas should also be considered, as fast-growing grass and SRC species have high water demands (Fischer et al. 2018;Maleski et al. 2019). It is thus important to consider not only benefits but also trade-offs in landscape-specific studies, to support well-informed decisions concerning the deployment of different perennialization strategies.
Utilizing agricultural land for buffers and windbreaks implies that current agricultural production is impacted. The consequences of this are difficult to quantify, as there are multiple landscape-specific aspects that determine how the introduction of buffers or windbreaks affect current agricultural production. For example, riparian buffers could be established entirely on cropland or to a varying degree on existing, unmanaged, buffers, thus limiting the reduction in cropland area. In the former case, agricultural production will initially decrease, but if yield levels increase on the cropland areas that benefit from reduced erosion or flooding mitigation (Li et al., 2016;Osorio et al., 2019), this could more or less outweigh the negative effect of reduced cropland area. For windbreaks, significant parts of the agricultural landscape are converted from annual crops (1/3 for SRC windbreaks and 1/9 for SRF windbreaks at 100% implementation) in the different scenarios. This can, however, fully or partly (depending on the implementation level) be compensated for by yield increases on cropland sheltered from wind erosion (Osorio et al. 2019). Finally, the different co-benefits of buffer and windbreaks (e.g. SOC increase, see Lal, 2020) can also have positive long-term effects on yield levels.
In case yield improvements cannot fully compensate for losses in cropland area, the potential connection between establishment of biomass production systems on cropland and indirect land use changes, causing, e.g., biodiversity impacts and land carbon losses, remains an additional concern (Sumfleth et al., 2020;Khanna et al., 2017;Takaes Santos, 2020;Berndes et al., 2012). The same concern exists when lower-yielding alternative farming practices are introduced to reduce environmental impacts, as exemplified by studies associating organic food with additional emissions (Smith et al., 2019), or foregone carbon sequestration (Searchinger et al., 2018), due to the need for more land to compensate for lower yields. Examples of such carbon leakage in other sectors include those associated with electrification of vehicle fleets, which may result in significant up-front carbon emissions due to large demand for batteries with significant embedded emissions, and low net reduction on CO2 emissions from displacing petrol and diesel use in ICE vehicles due to battery charging with carbon-intensive electricity (Hill et al., 2019;Xiong et al., 2020). Indirect land use change and carbon leakage need to be taken into account and addressed in appropriate ways. We argue that decision-making in the context of societal sustainability transitions needs to reflect a holistic view on both supply-side and demand-side mitigation, and be based on data and insights from many scientific disciplines and complementary methodologies covering a multitude of sustainability indicators. Taking the example of organic meat, consumers who are motivated to buy organic meat for environmental and ethical reasons may also buy fewer animal-based products in the first place (Baudry et al., 2017;Treu et al., 2017). Thus, the larger land demand associated with one individual food An important barrier towards wider implementation of multifunctional systems has been described as the lack of markets, or policies, compensating producers for enhanced ecosystem services and other environmental benefits (Englund et al., 2020b). Realizing the deployment scenarios presented here thus require significant policy efforts to generate sufficient incentives. Such efforts are now becoming visible in the EU. One potential opening may be the introduction of Eco-schemes in the new CAP. Ecoschemes will offer farmers the possibility to grant direct payments to adapt practices beneficial for climate, water, soil, air and biodiversity (EU, 2019). As illustrated in this assessment, such practices could include the establishment of multifunctional biomass production systems. This is a practical example of how to combine climate and agriculture policy goals, by producing long-term sustainable biomass feedstock which could replace fossil fuels and improve agricultural production in general. A critical aspect is therefore to significantly increase the knowledge of the opportunities with multifunctional biomass production systems among all EU member states before designing and introducing country specific Eco-scheme options in the new CAP.
While this study provides novel insights into large-scale strategic perennialization for beneficial LUC, it covers only two options, that are relevant only in parts of the EU. There are multiple additional systems that could be implemented to address other primary impacts (Englund et al. 2020a;Englund et al. 2020b). Such systems can be modelled similarly to what has been done here, to show a more complete picture of the potential for beneficial LUC through widespread deployment of multifunctional biomass production. Furthermore, intra-landscape modelling, considering a broad range of possible benefits and trade-offs, is necessary to fully understand the effects of introducing these kinds of systems at the landscape scale. Such information can also be valuable to support local or regional spatial planning, where different goals and objectives, from different stakeholders, need to be considered (Busch, 2017

Material and methods
Two main types of multifunctional biomass production systems were modelled: riparian buffers and windbreaks. For each system, we modelled the implementation in individual landscapes, under different deployment scenarios where strategic perennialization is incentivized, and quantify the corresponding (i) areas; (ii) amounts of biomass produced and the corresponding energy content; (iii) primary benefit; and (iv) co-benefits.
All cases of strategic perennialization are modelled to achieve a primary objective, i.e., to mitigate a specific environmental impact, driven by specific incentives. The mitigation effect is designated "primary benefit". In many cases, additional co-benefits are likely, i.e., the perennialization aimed at the primary objective will have a beneficial influence on other objectives, thereby increasing the total benefits for the environment and society. Co-benefits are often subject to uncertainty and depend on local circumstances and implementation practices, among other factors (Smith et al., 2020;Jia et al., 2019). Here we differentiate between "co-benefits" that can be generally expected, although not being explicitly targeted, and "potential co-benefits" that may or may not occur, depending on landscapespecific characteristics.
Unless otherwise specified, the term landscape refers to polygons in a dataset with over 81.000 subwatersheds across the EU27+UK; thus used synonymously to "sub-watershed" or "sub-catchment" (see Englund et al. 2020a for full justification and further details on the conceptual approach). The spatial analysis and operations, including all the database aggregation queries, were performed in GRASS GIS (GRASS Development Team, 2019) with projection EPSG:3035, whereas the cartography and some specific GIS operations were done in QGIS (QGIS.org, 2020).

Degree of environmental impact and effectiveness of strategic perennialization
Current N emissions to water, soil loss by water erosion, soil loss by wind erosion, recurring floods, and accumulated losses of SOC were estimated by Englund et al. (2020a) for the same landscape dataset as used for the modelling presented here. The impacts are described both in absolute terms, based on spatial indicators, and in relative terms on a five-step scale from "very low" to "very high". Englund et al. (2020a) also estimated the effectiveness of strategic perennialization, for each impact and in each landscape, by combining the degree of impact with the density of annual crops. As for the impacts, the effectiveness for the different impacts is described on a five-step scale from "very low" to "very high". Further methodological details are available in Englund et al. (2020a).

Deployment scenarios for strategic perennialization
To illustrate different outcomes, three deployment scenarios were designed. In all scenarios, it is assumed that certain environmental impacts are addressed at the EU-level via incentives for strategic introduction of perennials into intensively managed agricultural landscapes, to mitigate specific impacts of concern. It is also assumed that there is a high demand for biomass produced in the multifunctional system within the bioeconomy. The scenarios are conceptualized as follows, and further described under the respective multifunctional systems below: In scenario 1 (Biomass), there are incentives for implementing MBPSs in all landscapes where the effectiveness in reducing the impact of concern is classified as above "low" (defined below). The degree of the impact, and the corresponding impact mitigation potential, in individual landscapes are not considered. Incentives are such that farmers select the multifunctional systems that are expected to result in the highest mitigation, by default, while maximizing biomass output.
In scenario 2 (Low impact), incentives for mitigating impacts of concern are generally stronger than in scenario 1, but there are no incentives when impacts are below a defined "low" level at the landscape scale. Local and regional circumstances will determine whether farmers will prefer to favour biomass production in the multifunctional system or to minimize effects on current agricultural production. Different options for strategic perennialization may therefore be implemented in different landscapes.
In scenario 3 (Food first), incentives are the same as in scenario 2, with the addition that impacts on food production are disincentivized. Farmers therefore seek to minimize the area used for the multifunctional system, while achieving mitigation where impacts are above the "low" level at the landscape scale.

Modelling of riparian buffers
The primary benefit of riparian buffers is considered to be reduced N emissions to water. In this case, this refers to diffuse N emissions that reach watercourses by runoff or shallow groundwater (Grizetti et al., 2012). Co-benefits include (i) enhanced SOC, (ii) avoided water erosion and (iii) retention of sediment. Potential co-benefits include (i) flood mitigation, and (ii) reduced wind erosion.

Design options and deployment scenarios
Five different riparian buffer systems have been defined (Table 1), based on three different widths. First, a 5 m "narrow" option represents mandatory requirements in Italy and is based on empirical experiments by Ferrarini et al. (2017a). Second, a 21 m "wide" option represents a buffer with 100% buffer strip efficiency (BSE), based on a regression analysis by Ferrarini et al. (2017a). Third, a 50 m "double-wide" option represents a system where agricultural practices and economic aspects are taken more into account (Styles et al., 2016). In this system, only half of the buffer area is harvested each season, thus reducing temporal variations as a result of temporarily decreased BSE after harvest when all biomass is removed at once. The first two systems were modelled with either Short Rotation Coppice (SRC), i.e., fast growing tree species cultivated as "coppice" with multiple shoots from the stump, or with grass. The third was modelled only with SRC. BSE for SRC and grass was determined based on "buffer width necessary to obtain a given value of BSE" for "bioenergy crops", i.e., miscanthus and willow, as reported by Ferrarini et al. (2017a  "Biomass" scenario Farmers are assumed only to implement double-wide SRC buffers, as this provides economical and practical advantages due to larger cropping sites, maximum biomass production from the SRC buffers, and maximum impact mitigation. In all landscapes, the highest yielding SRC species is used, to maximize land-use efficiency.
Spatial deployment: Buffers are established in landscapes where the effectiveness of strategic perennialization for mitigation N emissions to water has been classified as at least "medium".
Buffer design options: Only double-wide buffers are implemented, as defined in Table 1.

"Low-impact" scenario
Farmers can implement any of the five defined buffer systems, depending on what is most favourable in their respective landscapes. The extent of the implementation in each landscape is, however, limited by the degree of impact mitigation necessary to reduce the impact down to a "low" level, at the landscape scale. In all landscapes, the highest yielding SRC and grass species, respectively, are used.
Spatial deployment: Landscapes suitable for implementation include those where both (i) N emissions to water and (ii) the effectiveness of mitigating N emissions to water, have been classified by Englund et al. (2020a) as at least "medium".
Buffer design options: All buffer options defined in Table 1 are possible.

"Food-first" scenario
Farmers establish narrow buffers to the greatest extent possible to minimize effects on food production. Where narrow buffers do not suffice to reduce the environmental impact down to a "low" level, at the landscape scale, wide buffers are used as a complement. Double-wide buffers are not implemented as they do not result in a higher primary impact mitigation than wide buffers. In all landscapes, the highest yielding SRC and grass species, respectively, is used, to ensure maximum landuse efficiency.
Spatial deployment: As for "low-impact". Buffer design options: Narrow SRC or grass buffers where they suffice to reduce the impact down to a "low" impact level. Wide SRC or grass buffers elsewhere.

Primary impact mitigation potential
The BSE:s for the respective buffer designs are defined in Table 1. The impact mitigation potential depends on the current primary impact and the BSE of the different buffer designs. The following was calculated for all buffers designs in the different deployment scenarios, for each landscape:  Maximum impact mitigation (kg N/ha/y), estimated as the product of current N emissions to water and the respective BSE.
 Maximum amount of avoided N emissions per year, estimated as the product of maximum impact mitigation and total landscape area.
 Necessary impact mitigation to achieve a "low" impact level, estimated as the difference between current N emissions to water and the upper threshold for the impact class "low".
 Share of impact mitigation necessary to achieve a "low" impact level, estimated as the quotient between necessary impact mitigation to achieve a "low" impact level and maximum impact mitigation.

Buffer areas
For each buffer option in the three deployment scenarios, the buffer area in each individual landscape was estimated as the product of buffer width (times two, assuming that it is established on both sides of the watercourse) and the total length of primary and secondary drains in each landscape. The latter was calculated based on a river dataset from the European Catchments and Rivers Network System (ECRINS) project (European Environment Agency, 2012). This dataset was developed in parallel with the sub-catchment dataset that is the basis for the landscape dataset used here. To calculate total river length, all river polylines were first cut by the landscape polygons and the sum of the length of each polyline in each landscape was then calculated and added as a new attribute to the landscape dataset using the QGIS tool "sum line length".
The above approach resulted in the maximum buffer area for the three buffer designs, considering only buffer widths and lengths of primary and secondary drains in each landscape. The maximum area was thus used in the "Biomass" scenario, in which deployment is driven by incentives for maximized multifunctional biomass production and maximum impact mitigation, not limited by a certain degree of impact mitigation or implications for food production.
In the "Low-impact" and "Food-first" scenarios, however, deployment is driven by incentives to reduce N emissions to water to a "low" level, but not beyond. This means that the buffer area in many cases will be smaller than the maximum area, since the maximum area results in a greater impact mitigation than what is necessary to fulfil the objective. To estimate the buffer area needed to achieve a "low" impact in each landscape, it was assumed that BSE, at the landscape scale, is proportional to the share of maximum buffer area. For example (cf. Table 1) if a BSE of 30% is necessary to reduce N emissions down to a "low" impact level, 50% of the maximum area for narrow buffers (having a BSE of 60%) is needed, and 30% of the maximum area for wide or double-wide buffers (having a BSE of 100% The buffer area for the different buffer design options in the "Low-impact" scenario was therefore estimated as the product of share of impact mitigation necessary to achieve a "low" impact level (see previous section) with maximum buffer area.
Finally, in the "Food-first" scenario, deployment is driven by incentives to reduce N emissions to water to a "low" level, but not beyond, but also by incentives to minimize impacts on food production. Since farmers would seek to optimize (in terms of buffer area) the impact mitigation, they utilize narrow buffers to the extent that they suffice to reduce the impact down to a "low" impact level, at the landscape scale, and wide buffers elsewhere. We therefore first assessed in which landscapes that narrow buffers suffice, by comparing impact mitigation necessary to achieve a "low" impact level with maximum impact mitigation for narrow buffers specifically. Where the former exceeds the latter, wide buffers are implemented. In other landscapes, only narrow buffers are used. The corresponding buffer areas were then calculated as for the "Low impact" scenario.

Biomass from buffers
Biomass production from riparian buffers was estimated for each suitable landscape and for each buffer option in the three deployment scenarios, as the product of buffer areas, estimated as described above, and the yield of the highest yielding SRC and grass option, respectively. The latter was identified using pan-European yield simulations at NUTS3 level (Dees et al., 2017). Simulated yields for SRC willow, SRC poplar, other SRC, miscanthus, switchgrass, and reed canary grass, using a "medium" yield-input management level, were identified for each landscape by first spatially joining landscapes to NUTS-3 regions, and then joining the database tables. Yields are expressed as t DM ha/y. The energy output is calculated as the product of biomass production and energy content of the harvested biomass; 18.7 MJ/kg DM, for both SRC (Nordborg et al., 2018a, 2018b and grass (Baxter et al., 2014).

Enhanced SOC
The effects on SOC from establishing riparian buffers were based on SOC simulations of permanent grasslands in relation to a business-as-usual (BAU) SOC scenario (Lugato et al., 2014), available for download at the Joint Research Centre European Soil Data Centre (ESDAC; https://esdac.jrc.ec.europa.eu/).
The SOC simulations are spatially explicit and provide BAU SOC estimates (t C/ha) for 2010, 2020, 2050, 2080, and 2100, assuming a continued rotation with the four most dominant crops in each area. They also provide SOC values in relation to these BAU values for multiple management options, including a permanent grassland system, in which the BAU rotation is replaced by permanent grassland. It is here assumed that SOC effects of establishing permanent grassland on cropland is representative of SOC changes in riparian buffers, as these are permanent perennial systems with documented positive effects on SOC (Ferrarini et al. 2017b).
SOC values relative BAU for permanent grassland were rasterized to 100 m and new SOC values were added to the landscape dataset by identifying the median value within each landscape (GRASS: v.rast.stats). BAU values are referred to below as "SOCbau_[year]" and SOC increases relative BAU from implementation of permanent grassland are referred to as "SOCinc_ [year]. SOCinc values in the dataset are expressed in relation to 2010. They were therefore re-estimated with 2020 as base year, to be able to represent SOC changes from current levels while maintaining 2050 and 2080 as points in time for assessment. SOCbau did not require re-estimation as it represents a continuation of BAU landuse. SOCbau_2020 was thus considered representative for current SOC.
To reflect that SOC tends to increase more rapidly early after the introduction of a new land-use system (Lugato et al., 2014), SOCinc_2020 was assumed to represent the change in SOC during the first ten years, i.e., between 2020 and 2030 ("SOCinc_first10"). SOC changes during the remaining period (i.e., 20 and 50 years, for 2050 and 2080, respectively) was calculated by subtracting SOCinc_first10 from SOCinc_2050/2080, representing SOC changes in 30/60 years following the first 10 years ("SOCinc_last30/60). Since we require SOC changes in 20/50 years, these values were downscaled by 20/30 and 50/60, respectively ("SOCinc_last20/50"). Finally, SOC increases by 2050/2080 relative BAU could be calculated as the sum of SOCinc_first10 and SOCinc_last20/50.
At this point, SOC changes per hectare of riparian buffers by 2050/2080 relative BAU, with base year 2020, have been estimated. SOC changes per hectare of cropland were then calculated as the product of SOC changes per hectare of riparian buffers and share of area under annual crops used for riparian buffers in each deployment scenario, for all individual landscapes.

Avoided water erosion
Soil erosion within SRC systems can be considered marginal (Ferrarini et al. 2017a). Fully replacing annual crop production with SRC is therefore assumed to reduce soil erosion nearly completely on that specific land. Consequently, the share of land under annual crop production used for riparian buffers indicates the share of reduced soil loss by water erosion, at the landscape scale.
First, avoided soil loss by water erosion per hectare and year was calculated in each landscape, for each riparian buffer option, as the product of share of area under annual crops used for riparian buffers and current soil erosion by water on land used for annual crop production. The total amount of avoided soil loss by water erosion per year in each landscape was calculated as the product of avoided water erosion per hectare and total area under annual crop production. The share of avoided water erosion relative total water erosion was then calculated as the quotient of amount of avoided soil loss and total soil loss in each landscape. Finally, the degree to which riparian buffers could contribute to reducing soil loss by water erosion down to a "low" impact level, at the landscape scale, was estimated as the quotient of avoided soil loss by water erosion per hectare and year and soil loss by water erosion above the threshold for "low" impact.

Sediment retention
Sediment retention was quantified based on the assumption that all soil loss by water erosion on cropland at the sub catchment scale is destined to end up in watercourses within the catchment. Total sediment loads in each landscape are thus equivalent to soil loss by water erosion, as estimated by Englund et al. (2020a), reduced by avoided water erosion as estimated above. Empirical studies have shown that buffer widths of 3 m, 6 m, and 7 m, can have sediment trapping efficiencies (STE) of 66%, 77%, and 95%, respectively (Ferrarini et al., 2017b). Based on this, trapping efficiencies of 75% for the narrow buffer and 100% for the wide and double-wide buffers, were assumed.
In the "Biomass" scenario, sediment retention at the landscape scale was estimated as the product of STE and total sediment load, as estimated above. In the other scenarios, where buffers are only implemented to the extent where N emissions to water is decreased to a "low" level, sediment retention is calculated as the product of sediment retention in the "Biomass" scenario and the share of total buffer area that is needed in each landscape to achieve "low" N emissions to water, for each buffer design.

Modelling of windbreaks
The primary benefit of windbreaks is mitigation of soil loss by wind erosion. Co-benefits include enhanced SOC and reduced N emissions to water. Potential co-benefits include flood mitigation.

Design options and deployment scenarios
Two design options for windbreaks have been defined. First, "SRC windbreaks" refer to the establishment of SRC willow or poplar, with a rotation period of 3-4 years and a height of 5 m (Tahvanainen, 1996). Second, "SRF windbreaks" refer to the establishment of SRF poplar, with a rotation period of 15 years and a height of 20 m (Hjelm et al., 2015).
In both the SRC and the SRF design, a windbreak width of 50 m is assumed, for practical and economic reasons, since larger cropping sites reduce management costs, following the same reasoning as for double-wide riparian buffers. The distance between windbreaks is assumed to be 20H, i.e., 20 times the windbreak height (Osorio et al., 2019).
In both designs, half of the windbreak is harvested at a time, allowing for constant windbreak functionality.

"Biomass" scenario
Farmers can implement either SRC willow/poplar or SRF poplar, depending on which is most favourable in their respective landscapes. The design option is not affected by incentives to maximize biomass production, as biomass production per hectare is the same in both options. The design option is also unaffected by incentives to maximize mitigation of soil loss by wind erosion, as this is assumed to be identical for both options.
Spatial deployment: Windbreaks are established where the effectiveness of strategic perennialization for mitigating soil loss by wind erosion is classified as at least "medium". However, as it is assumed that windbreaks can only achieve mitigation down to a "low" impact level (see below), landscapes also need to have a current impact level of at least "medium" for soil loss by wind erosion.
Buffer design options: Where willow is higher yielding than poplar: SRC willow windbreaks. Where poplar is higher yielding: either SRC or SRF poplar windbreaks.
"Low-impact" scenario Farmers can implement either SRC willow/poplar or SRF poplar windbreaks, depending on what is considered most favourable in their respective landscapes, but only to an extent where the environmental impact is mitigated down to a "low" level, at the landscape scale. considered more favourable for other reasons. In all landscapes, implementation is limited to what is necessary to achieve a "low" impact, at the landscape scale.
"Food-first" scenario Since farmers are incentivized to minimize impacts on food production, they seek to limit the share of cropland used for windbreaks to less than or equal to the expected resulting yield increases. Farmers therefore seek to achieve impact mitigation down to a "low" level, in each landscape, while limiting the area used for windbreaks to a maximum of 10% of the area under annual crops (Osorio et al. 2019).
Spatial deployment: As for "Biomass" and "Low-impact" Buffer design options: Farmers implement SRC willow or poplar windbreaks, depending on what is highest yielding, where the area needed for SRC windbreaks to reduce impact down to a "low" level does not exceed 10% of the cropland area in the landscape. Where a larger share of the existing cropland area is needed, SRF poplar is used.

Primary impact mitigation potential
Soil loss by wind erosion is classified as "low" (and rarely "very low") in a vast majority of agricultural landscapes across Europe (Englund et al., 2020a). Windbreaks are therefore assumed to be able to reduce soil loss by wind erosion to a "low" level, i.e., less than 5 t/ha/y (Englund et al., 2020a), but not to a "very low" level. It is further assumed that a windbreak distance of 20H may not suffice to achieve this level of impact mitigation (i.e., down to a "low" level) in all landscapes (Osorio et al., 2019;Brandle et al. 2004). An assumption was therefore made that the defined windbreak options, if fully implemented in all agricultural fields currently under annual crops, suffice to reduce wind erosion from a "high" to a "low" impact level, at the landscape scale. Landscapes with a higher current impact than "high" therefore need shorter windbreak distances to achieve the desired impact mitigation. It is further assumed that landscapes with a current impact lower than "high", and thus in need of lesser impact mitigation to achieve a "low" level of wind erosion, at the landscape scale, require lesser mitigation efforts. In such cases, a greater distance between windbreaks, or windbreaks implemented only in selected parts of the landscape, may suffice to achieve the desired impact mitigation.
Based on these assumptions, a windbreak implementation level was estimated for each landscape, based on the need for impact mitigation to reach "low" soil loss by wind erosion. The implementation level was assumed to decrease linearly from 100%, in landscapes with wind erosion of 10 t soil loss /ha/y (the upper threshold for "high" impact), to 0% for landscapes already having a "low", or lower, impact (<= 5 t soil loss / ha / y). Landscapes exceeding the threshold for "very high" receive an implementation level greater than 1 to reflect that the distance between windbreaks need to be shorter (minimum 11H for the landscape having the highest wind erosion, 16.8). This implementation level was calculated for all landscapes using Formula 1. Note that the implementation level in the "Biomass" scenario is always 100%. Having established the implementation level in each landscape, the following can be calculated: Necessary impact mitigation to achieve a "low" impact level (t soil loss/ha/y) estimated as the difference between the upper threshold for the impact class "low" and current soil loss by wind erosion.
Maximum impact mitigation (t soil loss/ha/y) for all windbreak options, estimated as equal to necessary impact mitigation to achieve a "low" impact level. In the "Biomass" scenario, however, "simpler" incentives are assumed, in which the landscape-specific potential for impact mitigation is not considered. The implementation level in the "Biomass" scenario is therefore always 100%. In landscapes where an implementation level greater than 100% is necessary to achieve this degree of impact mitigation, necessary impact mitigation to achieve a "low" impact level will exceed maximum impact mitigation. In the "Biomass" scenario, maximum impact mitigation was therefore set manually to 8, i.e., the difference between the upper thresholds of "high" and "low" impact, in landscapes where the implementation level exceeds 100%.
Maximum amount of avoided wind erosion (t soil loss/y) estimated as the product of maximum impact mitigation and total area under annual crops.
Share of impact mitigation necessary to achieve a "low" impact level, estimated as the quotient of necessary impact mitigation to achieve a "low" impact level and maximum impact mitigation.

Windbreak deployment areas
For each landscape, and for the two windbreak designs, total windbreak area needed to reduce wind erosion to a "low" level was calculated as the product of (i) implementation level, as calculated above, (ii) the share of land under annual crops needed for establishing windbreaks at 100% implementation, and (iii) the total area under annual crops.
The total share of agricultural land under annual crop production needed for establishing windbreaks, at 100% implementation, is calculated as the quotient of windbreak width and windbreak distance, i.e., 1/3 for SRC windbreaks and 1/9 for SRF windbreaks.

Biomass production in windbreaks
Biomass production from windbreaks was estimated as for riparian buffers, but considering only SRC willow and SRC poplar. The yield for SRF poplar was assumed to be identical to SRC poplar (Ayllot et al., 2008;

Co-benefits
Avoided N emissions to water N emissions to water from SRC or SRF systems can be considered marginal Ferrarini et al. 2017). Following the same assumptions as for water erosion, avoided N emissions to water (per hectare and year) was estimated in each landscape as the product of share of area under annual crops used for windbreaks in the two different windbreak designs, and current N emissions to water. The total amount of avoided N emission to water in each landscape was then calculated as the product of avoided N emissions to water (per hectare and year) and total area of the landscape. Finally, the degree to which riparian buffers could contribute to reducing N emissions to water down to a "low" impact level, at the landscape scale, was estimated as the quotient of avoided N emissions to water (per hectare and year) and current N emissions to water above the threshold for "low" impact.

Enhanced SOC
Effects on SOC estimated as for riparian buffers.

Avoided water erosion
Avoided soil loss by water erosion estimated as for riparian buffers.

Potential additional co-benefits 4.5.1 Flood mitigation
As detailed above, riparian buffers and windbreaks are likely to have positive effects in reducing flooding events. However, no empirical data for quantifying such effects has been found. The extent to which this benefit may occur is therefore considered to differ between different landscapes, depending on landscape-specific characteristics. For example, if the predominant wind direction is parallel to watercourses in a landscape, windbreaks will be established predominantly perpendicular to contour lines and thus have limited effect on regulating water flows. Furthermore, for riparian buffers, the cause of flooding in a landscape may be predominantly caused by the land-use upstream of the main catchment drain, i.e., in another landscape.
No attempts were therefore made to quantify this co-benefit. Instead, we attempt to indicate the likeliness of mitigated or avoided flooding events as a result of the establishment of riparian buffers or windbreaks. This was done by assuming that the likeliness is directly correlated with the effectiveness of strategic perennialization in mitigating recurring floods (Englund et al., 2020a). An effectiveness of medium corresponds here to a likeliness of medium, etc. The effectiveness of strategic perennialization in mitigating recurring floods was therefore identified for each landscape where riparian buffers and windbreaks are introduced, respectively, in the different deployment scenarios.

Wind erosion mitigation
Estimated in the same way as for "flood mitigation". Establishment of riparian buffers in the three deployment scenarios, and the share of maximum narrow buffer area needed to achieve "low" N emissions to water in each landscape.

Figure 2
Average increase in SOC on cropland by 2050, relative BAU, in each landscape, in different deployment scenarios and with different riparian buffer options.

Figure 3
Degree to which establishment of riparian buffers contribute towards reducing soil loss by water erosion down to a "low" level, in the three deployment scenarios for different buffers options Top row: the implementation level used in the different scenarios. In the "Biomass scenario, it is always 100%. In the "Low-impact" and "Food-rst" scenarios, it is determined by the implementation necessary to reduce wind erosion to a "low" level but not beyond. Bottom row: the different windbreak options that are implemented. In the Biomass and "Low impact" scenario, the highest yielding options are always used, in the "Food-rst" scenario, the option is also affected by the ambition to minimize total windbreak area.

Figure 5
Average increase in SOC on cropland by 2050, relative BAU, in each landscape, in different deployment scenarios and with different windbreak options.

Figure 6
Degree to which establishment of windbreaks contribute towards reducing soil loss by water erosion down to a "low" level, in the three deployment scenarios for different windbreak options Degree to which establishment of windbreaks contribute towards reducing N emissions to water down to a "low" level, in the three deployment scenarios for different windbreak options Likeliness of ood mitigation and wind erosion mitigation in riparian buffers and windbreaks.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. strategicperennializationSI.pdf