Towards evidence‐based biodiversity assessment tools for agroforestry systems

Ecol Solut Evid. 2025;6:e70133.   | 1 of 13 https://doi.org/10.1002/2688-8319.70133 wileyonlinelibrary.com/journal/eso3 Received:23January2025 | Accepted:13September2025 DOI: 10.1002/2688-8319.70133 RESEARCH ARTICLE Towards evidencebased biodiversity assessment tools for agroforestry systems Jari Vandendriessche1 | Maxime Eeraerts1 | Pieter De Frenne1 | Paul J. Burgess2 | Laura CumplidoMarìn2 | Sonja Kay3 | Margherita Tranchina4,5 | Paul Pardon6 | Kris Verheyen1 1Department of Environment, Forest & Nature Lab, Ghent University, MelleGontrode, OostVlaanderen, Belgium; 2SchoolofWater,Energyand Environment, Cranfield University, BedfordCranfield, Central Bedfordshire, UK; 3AgriculturalLandscapesandBiodiversity,Agroscope,Zürich,Switzerland; 4ScuolaSuperioreSant'Anna,Pisa,Toscana,Italy;5ScuolaUniversitariaSuperiorePavia,Pavia,Lombardia,Italyand6Flanders Research Institute for Agriculture,FisheriesandFood,Merelbeke,Oost-Vlaanderen,Belgium ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttribution License, which permits use, distribution and reproduction in any medium, providedtheoriginalworkisproperlycited. ©2025TheAuthor(s).Ecological Solutions and EvidencepublishedbyJohnWiley&SonsLtdonbehalfofBritishEcologicalSociety. Correspondence Jari Vandendriessche Email: jari.vandendr[email protected] Funding information HORIZONEUROPEFood,Bioeconomy, NaturalResources,Agricultureand Environment,Grant/AwardNumber: 101059794 Handling Editor:FlorentNoulekoun Abstract 1. Agroforestrycanhelptoconservebiodiversityandenhancemultipleecosystem services such as carbon sequestration, microclimate regulation and nutrient cycling. However, in land planning and biodiversity certification schemes it remains difficult to quantify the effect of agroforestry on biodiversity across time and space. 2. Here we combine insights from a second-order meta-analysis, a stakeholder questionnaire, and a review of biodiversity assessment tools to establish a route towards more accurate estimates of agroforestry effects on biodiversity. Via a synthesis of crosstaxa metaanalyses, we evaluated the impact of agroforestry and landscape structure on biodiversity. Complementing the literature evidence, we performed a stakeholder questionnaire to determine the perceptions and preferencesofstakeholderswithregardstobiodiversity. 3. Themeta-analysessynthesisindicatespredominantlypositiveornoeffectsof agroforestry practices on biodiversity, albeit with contextual nuances such as landscapestructureandsystemdesign.Thequestionnairerevealedstakeholders' recognitionofbiodiversity'spivotalroleinagroecosystemsandawillingnessto supportmethodstoassesstheeffectsofagroforestryonbiodiversity.Therewas a preference for userfriendly, webbased tools that integrated mapping features andcheckliststailoredtodiverseagroforestrytypes.Finally,weevaluated73existing biodiversity tools in terms of their capability of incorporating agroforestry components.Thetools'reviewrevealedlimitationsintermsoftheirspecificity, accessibility or capacity to encompass multifaceted agroforestry designs. 4. Practical implication. Our threefaceted approach provided comprehensive insights about the building blocks required to develop an evidence-based and 2 of 13 | VANDENDRIESSCHE et al. 1 | INTRODUCTION Biodiversity is vital for wellfunctioning ecosystems, in part because of its role in supporting ecosystem services such as pest control, pollination andnutrient cycling(Daineseet al., 2019; LindemannMatthies et al., 2010; Millennium Ecosystem Assessment, 2005). Despiteitsimportance,biodiversityisgloballyunderthreat(Butchart et al., 2010;Sachsetal.,2009).Theproblemisincreasinglyrecognizedbypolicymakers(EuropeanCommission,2011)whoaretaking measures to preserve and restore biodiversity. In Europe, this is now embodied by, among others, the Green Deal and the Biodiversity Strategyto2030(EuropeanCommission&Directorate-Generalfor Environment, 2021). During recent decades, the expansion and intensification of agriculturehavebeenkeydriversofbiodiversitydecline,mainly through driving the loss of natural and seminatural habitats and small landscape elements (Balmford et al., 2012; Tilman et al., 2001; Tscharntke et al., 2012). In many European areas, agricultural landscapes generally consist of a matrix embedding fragmented patches of remnant natural vegetation, often referred to as seminatural habitats, where nondomesticated biodiversity persists (Goulson, 2021). Maintaining and increasing these seminatural habitats is critical to halt biodiversity loss in agriculturallandscapes(Eeraerts,2023; EstradaCarmona et al., 2022; Perfecto&Vandermeer,2010). One way of increasing biodiversity in intensive croplands and grasslands is to introducetrees and shrubs(Leakey,1996). These woodyelementsmayco-produce,forinstance,nuts,fruit,corkor wood. Examples of such agroforestry practices include silvoarable (i.e. tree-crop associations) and silvopastoral systems (i.e. tree- livestockassociations)andmorespecificallyalleycropping,hedgerows and windbreaks, food forests, forest grazing (FG) practices, scatteredsolitarytrees,etc.(Dmuchowskietal.,2024; MosqueraLosada et al., 2009). Agroforestry can increase biodiversity- mediatedecosystemservices(Udawattaetal.,2019)byproviding habitat, food, shelter and the provision of more diverse resources to multiple species (Jose, 2009; McAdam et al., 2009) and might enrichthestructureofalandscape.However,Statonetal.(2019) argued that, especially in temperate regions, both the practical knowledgeandscientificunderstandingoftheeffectsofagroforestryonbiodiversityarestilllimited.Publishedstudiesontheeffects of agroforestry on biodiversity in comparison to monoculture croplands and intensively managed grasslands represent mixed results(Imbertetal.,2020;Pardonetal.,2019;Plieningeretal.,2015; Varah et al., 2020).Theneteffectofagroforestryonbiodiversity islikelytobedependentonthefocalcrop,theageofthesystem, the tree species, their management and the surrounding landscape (Klettyetal.,2023).Structurallycomplexlandscapesarereportedto have generally high levels of biodiversity as they offer a variety of habitatsbyformingacomplexpatchworkofsemi-naturalhabitats (Concepciónetal.,2008; Eeraerts, 2023;Tscharntkeetal.,2012). Both conceptual and empirical research demonstrate a mediating effect of landscape structure on the net biodiversity effect of agroecologicalmeasures(Lichtenbergetal.,2017;Scheperetal.,2013; Siramietal.,2019).Itisofexpectancethatthismediatingeffectof landscape structure on biodiversity is also present in agroforestry systems. Given the current biodiversity crisis, it is important to properly account for possible biodiversity gains that could be attained byagroforestry.Additionally,biodiversityisacomplex,multi-taxa conceptdependentonspatialandtemporalscales,whichmakesit difficulttoscoreitproperly.Asaresult,farmersinterestedinbiodiversity are uncertain about their options to enhance biodiversity (Birreretal.,2014; Dwyer et al., 2023). Atpresent,weareunawareofaspecificpolicyormanagement supporting tool available to assess the biodiversity benefits of agroforestry inEurope. Such a toolmight (1)enablefarmers to assessandunderstandthebiodiversityontheirfarmsand(2)provide a clear interpretable metric that can be used to report downstreamalongthevaluechain.Thismightenabletheappropriate labelingofagroforestryproducts(e.g.forcertificationschemes), which is important to both farmers and value chain actors, includingtheendconsumer.Hence,weaimtoidentifythekeybuilding blocksofanevidence-basedanduser-friendlytoolthatpredicts the biodiversity benefits of agroforestry implementation in agroecosystems.Toachievethis,weimplementedamultifacetedapproach to enable us to gain comprehensive insights about the requiredbuildingblocks. userfriendly tool for predicting the effects of agroforestry on biodiversity. Specifically, our interdisciplinary synthesis underscores the potential of agroforestryinpromotingbiodiversitywhileemphasizingtheneedforanevidence- based, usercentric tool to effectively assess biodiversity within agroforestry systems,accountingforlandscapecontextandsystemdesign.Suchatoolshould be constructed with input data for different agroforestry types, across taxa and updateablewhenknowledgegapsarefilled. KEYWORDS agrienvironmental measures, biodiversity conservation, decision aid system, ecosystems services,landuse,questionnaire,stakeholderperception,sustainableagriculture | 3 of 13 VANDENDRIESSCHE et al. 2 | MATERIALS AND METHODS Our approach was threefaceted. Firstly, we compiled scientific evidence, based on published metaanalyses, about agroforestry systems in the broad sense and their effect on multitaxa biodiversityinagroecosystems.Asampleevidenceoflandscapestructure influencing biodiversity effects generated by agroecological measures(Batáryetal.,2011)ispresent,oursearchalsoincludedmeta- analyses on landscape structure in agroecosystems. Additionally, implementation of agroforestry implicates the landscape context, as structuralelementsareimplemented.Thus,somelandscapemetrics couldbeusedasaproxyforagroforestryimplementation(applied onabiggerscale).Thecollecteddatawillserveasthescientificproof forbiodiversityeffectsprovokedbyeitheragroforestryorlandscape structure(i.e.landscapecomplexity,configurationandcomposition). Secondly,aquestionnairewasdistributedtodifferentstakeholders in the agroforestry value chain within Europe to explore the qualitative and quantitative requirements and preferences concerning a biodiversitytoolforagroforestrysystems.Thisaspectcanserveas a wish list potential users have for a biodiversitytargeted agroforestrytool.Thirdly,wecompiledadatabaseofavailablebiodiversity toolsthatareusedtopredictbiodiversityinagroecosystems.With this third aspect, we evaluated the collected tools upon the criteria established by the metaanalyses and questionnaire. 2.1 | Metaanalyses: Literature search and processing Existing metaanalyses are valuable in this sense as they provide a quantitative summary of published scientific research on a specific topic. In the metaanalyses, the estimated effects of case studies are convertedtocomparableeffectsizes,whichcanbecontrastedand usedtosummarizetheeffectsacrossalargerangeofcontexts,taxa and scales. Over time, the number of metaanalyses has increased, even within one particular topic; thus, several efforts have already led to the combining of multiple metaanalyses into comprehensive articlesfocusingonagriculturalpractices(Beillouinetal.,2019; Bonfanti et al., 2023;Dmuchowskietal.,2024;Makowskietal.,2021). Weperformedasystematicliteraturesearchtoidentifysuitable metaanalyses for our objectives. Our search was performed on 29 February 2024, using the following search terms: (biodiversityORagrobiodiversityORarthropodsORcontributions)AND(agricultureORagroecosystemORagroecosystemsOR farmlandORsilvopastoralORlandscapeORforest)AND(agroforestry OR silvopasture OR hedgerows OR hedges OR "scattered trees"ORwoodyOR"forestgrazing"OR"livestockdisturbances" OR "food forest" OR "landscape complexity" OR "landscape structure")AND(meta-analysisOR"metaanalysis"ORmeta-analysesOR "metaanalyses"OR"systematicreview"OR"meta-synthesis"). Weoptedforasinglequerydesignedwithfourcompartmentsto screentheliteraturewewanted,implyingthefollowing:(i)afilteron biodiversityasatopic,(ii)afilterontheecosystemsastudyshould becarriedoutin,(iii)afilteronthespecificcases(i.e.agroforestry andlandscapestructure)and(iv)thestudybeingameta-analysis. InsertingthisqueryinISIWebofScienceCoreCollectionand inScopusresultedin217and128articles,respectively,ofwhich 244uniquerecords.Titleandabstractswerescreenedtoseeifthey wererelevanttothisstudy'skeyobjectives:(1)thestudyquantitativelysynthesizedexistingliterature(i.e.ameta-analysis),(2)the study includes either agroforestry elements or landscape structure effects(i.e.landscapecomplexity,landscapeconfigurationandlandscapecomposition)and(3)thestudyusedoneormorebiodiversity metricsasresponsevariables.Studiesfocusingongeneticbiodiversitywereexcluded(e.g.intra-specificgeneticsorgeneticbiodiversity).Thisapproachresultedin53suitablemeta-analyses. Hereafter we assessed the full texts and excluded metaanalyses (1)inwhichlessthan50%oftheunderlyingstudieswereperformed inecosystemsrelevanttotheEuropeanscope(i.e.temperate,borealandMediterranean),(2)didnotcompiledataforagroforestry elements or landscape structure meta-analytically (e.g. narrative reviews),(3)usednon-agroecosystemcomparators(e.g.forests)or (4)fellbeyondourparticularscopeofagroforestrysystemsorlandscapecontexts(e.g.interactioneffectsofagri-environmentalmeasureslikeorganicfarmingorimplementingwildflowerstripsacross landscape gradients, effects of mowing). More information about these decisions is available in Text S1,alongwiththePRISMAdiagramofourapproach.Thisresultedinafinalsetof12meta-analyses. Wealsoidentifiedsevenmeta-analysesassessinginteractioneffects betweenlandscapestructureandagri-environmentalmeasures(e.g. flowerbelts,hedgerows,grassstrips,cropdiversification).Notone identified metaanalysis specified an interaction between landscape structure and agroforestry implementation; thus, the available evidencerepresentstheclosestlinkpossible. From the studies that were identified as suitable metaanalyses forourstudy,weextractedtheoveralleffectsizesasdetermined byeachmeta-analysis.Effectsizesofinterestweretheeffectof either agroforestry or landscape structure on biodiversity (i.e. richness,abundance).Whenextractingtheeffectsizes,thehighest level of detail was retained regarding biodiversity functional groups(i.e.overallbiodiversity,plants,soilfaunaandmicrobiota, pollinators,naturalenemiesandpestspecies),typesofagroforestry(e.g.agroforestryinthebroadsense,silvoarable,silvopastoral,hedgerows,scatteredtrees)andlandscapecharacteristics(i.e. composition,1 configuration2 and complexity3).Thedegreeofreplicatesandstandarderrorperoveralleffectsizewasextractedto calculatecomparablestudy-leveleffectsizes.Detailsaregivenin Text S1.Werepresentedthesevaluesqualitativelywithindividual effectsizes. 1Thatis,theratioofdifferentbuildingblocksinalandscape(e.g.reductionofintensive agriculture,higherpercentagesofseminaturalhabitat). 2Thatis,thespatialarrangementoflandscapebuildingblocks(e.g.smallerplotarea, betterconnectivityofseminaturalhabitats). 3Encompassingavarietyoflandscapestructurefactors(includingcompositionaland configurationalfactors). 4 of 13 | VANDENDRIESSCHE et al. Next, we carried out a secondorder metaanalysis for the agroforestry part as our main interest. Most of the considered studies used either log-response ratio effect sizes (agroforestry meta-analyses)orFisher'sZeffectsizes(landscapestructuremeta- analyses).Therefore,werecalculatedallothereffectsizestothese, ifpossible(adetailedapproachisavailableinText S1).Wecalculated the mean response ratio across all studies according to Hedges etal.(1999): All related formulas are available in Text S1, k represents the number of studies and wi* and θi represent respectively the random-effectsweightsandtheeffectsizeforstudyi.Wevisualizedpooledeffectsizesalongwithcorresponding95%confidence intervals. 2.2 | Questionnaire design and collection Second,weundertookaquestionnaireinwhichwesoughttodisentangle the stakeholders' needs and wishes for a biodiversity tool designed to aid biodiversity estimation in European agroforestry systems. We specifically targeted four actors' groups: (1) Policymakers and administrations concerned with applying agroforestryrelated regulations at regional, national and European levels,whosetthescenefortheadoption(ornot)ofagroforestry; (2)farmers,landownersandbyextension,farmadvisers(whowere categorizedseparately)playinganactiveroleindesigningandmanaging agroforestry and whose choices determine the agronomic, economic, environmental and social performance at farm level and beyond; (3) stakeholders in the value chain including wholesalers, retailers, organizations trading the carbon sequestration and biodiversitybenefitsofagroforestry,andfinalconsumersseeking verification of the benefits of agroforestry in clear and accessible terms;and(4)researcherswhohelptoassesstheperformanceof agroecosystems and communicate their results. The questionnairewasdistributedwithinsevenstakeholdergroupslinkedtothe DigitAFproject(https:// digit af. eu/ ).TheseweresituatedinFinland, Czechia,theNetherlands,Germany,theUnitedKingdom,Italyand Belgium(Tranchinaetal.,2024).Thebiodiversity-relatedquestionnaireconsistedoffoursegments,encompassingquestionsupon(1) theimportanceofbiodiversityanditsassessment,(2)therequired specifications and relevance of an agroforestrybased biodiversity tool,(3) thedesignof sucha tooland(4)interestsconcerningits validation(seeText S2forthefullquestionnaire).Summaryresults of the questionnaire are presented. The questionnaire was completed by 82 stakeholders across thesevenregionsbetweenAprilandSeptember2023.Respondent characteristicsandrespondents'opinionsontheinclusionofagroforestry types can be retrieved in Figures S1 and S2. 2.3 | Biodiversity tools: Search and processing We performed a broad search of biodiversity tools between November 2022 and December 2023 to build a comprehensive database of tools used to assess biodiversity in agroecosystems. Adetailed description of this process,anoverviewtableand the correspondingPRISMAdiagramareavailableinText S3. From this database, totaling 73 tools, we selected 37 tools that are able to assessbiodiversityinagroforestrysystems(processalsodescribedin Text S3).Forthe37selectedtools,wecategorizedtheinput–output relations in different categories, distinguishing between the input variables required by the tools and the outputs they provide. As such,therearetoolsusinghabitatcharacterization(structurequality),bioticindicators(samplingofspeciesgroup)orbothasinputs. Additionally,wehighlightedthetypologyconcerningdifferentagroforestry structures incorporated in the tools, referring to their applicability in specific agroforestry types. Furthermore, the tools were evaluated on their open source availability. Text S3 provides any additionally required clarifications. Wealsoestablishedapointofview(POV)ofthestakeholders to compare the tools efficiently, which we assessed by means of the questionnaire outputs, assigning to each category the percentage of stakeholders valuing that particular category. To best display this dataset on tooltype, scale, applicability, input and output, we performedNonmetricMultidimensionalScaling(NMDS)usingthe functionmetaMDSwithBray–Curtisdistanceinthepackagevegan (Oksanenetal.,2022).Weexploredandplottedboththetoolsspace (sitesofmetaMDS-output)andtheunderlyingvariables(speciesof metaMDS-output).Allpackages andprogramsusedfordatahandling are available in Text S4. 3 | RESULTS 3.1 | Metaanalyses Weidentifiedsixmeta-analyses(totaling19effectsizes;3ofthese werenotconvertibletolog-responseratios)thatassessedtheeffects of agroforestry on biodiversity in agroecosystems, relevant to aEuropeancontext(Figure 1a).Onlyalimitednumberofagroforestry features were assessed through metaanalyses for different taxa.Weretrievedstudieswhereinagroforestrywaspresentedas agroforestry in general, silvoarable, silvopastoral, hedgerows and scattered trees were represented. However, the low number of metaanalyses and the highly variable nature of agroforestry lead to highly variable results in the quantitative secondorder recalculationtechniques.Assuch,weobtainedinsignificantresultsforagroforestryinthebroadsense(overall:0.30 ± 0.54,plants:0.83 ± 0.91, natural enemies: 0.22 ± 0.55, pest species: −0.33 ± 0.67). On the other hand, panels consisting of only one study tended to have smaller confidence intervals (e.g. silvoarable + natural enemies: 0.22 ± 0.17, hedgerows + overall: 0.41 ± 0.05). Therefore, we additionallyvisualizedtheeffectsqualitativelyperstudy(Figure 1b). Pooled effect size: 𝜃∗= k ∑ i=1 w∗ i𝜃i k ∑ i=1 w∗ i | 5 of 13 VANDENDRIESSCHE et al. Qualitative results overall supported the premise of mostly positive orneutraleffects:onlyoneeffectsizereportednegativeresultson biodiversity(pestspecies).Thus,theidentifiedmeta-analysessupport the hypothesis that agroforestry generally enhances or maintains biodiversity. We also identified another six meta-analyses that assessed the effect of landscape structure on biodiversity in agroecosystems (i.e. not necessarily implying agroforestry), accounting for 35 effect sizes. Landscape structure effects on biodiversity were mostly positive, meaning that an increase in landscape structure is beneficial for biodiversity, and more consistent when compared to the results of the agroforestry part(Figure S3).Followingthelimitednumberofmeta-analysesforboth agroforestry context and landscape structure, we could also not detect any metaanalysis detecting the interaction between both. However, seven meta-analyses in our review tackled interactions between agro-ecological measures (of which agroforestry FIGURE 1 Effectsofdifferentagroforestrytypesonbiodiversitylevelsofdifferenttaxonomicgroups.(a)Thesecond-ordereffectsizes (log-responseratios)with95%confidenceintervalsascalculatedwith16effectsizes(3effectsizesrepresentedbyhedge'sg values were deletedinthisanalysis,seeSection2).(b)19effectsizesconcerningagroforestrytypeandfunctionalgroupasreportedorrecalculated fromsubindicesinthemeta-analyses,representingpositive,negativeornon-significantresults.Theunderlyingeffectsizesarebasedon abundance,diversityandrichnessmetricsandarecomparedtotheirbaselines(agroecosystemswithoutagroforestry).Thegreyshadedrow representsasummaryviewonagroforestryingeneral.Thecoloursofthedotsandsilhouettesappointataxa'sfunctionalityinagricultural context:Green,plants;black,soilfaunaandmicrobiota;orange,pollinators;blue,naturalenemiesandred,pestspecies.e,GarcíadeLeón etal.(2021);f,KoellnerandScholz(2008);h,Mupepeleetal.(2021);i,Prevedelloetal.(2018);k,Statonetal.(2019);l,Torralbaetal.(2016). Taxasymbolswereretrievedfromphylo pic. org. 6 of 13 | VANDENDRIESSCHE et al. implementation might be considered an example) and landscape structure, and these all agreed that landscape structure mediates theeffectofthesemeasures onbiodiversity (Batáryetal., 2011; Lichtenberg et al., 2017; Marja et al., 2019, 2024; Pérez-Sánchez et al., 2023; Sánchez et al., 2022; Tuck et al., 2014). Landscape structure here mainly influenced the magnitude of the effects of agroecological measures on biodiversity, rather than the level of significance.Thegeneralobservationisthateffectsaregreatestin simple landscapes compared to cleared and complex landscapes, as in the latter the baseline biodiversity is already too low or very high, respectively, for measures to be effective. We detect mostly positive effects of increasing landscape structurebyinfluencinglandscapecomposition(exceptfornoeffectinpestspecies)andlandscapeconfiguration(exceptforpest speciesandnaturalenemies).Similarly,inmostcases,landscape complexity has a positive effect on biodiversity. This evidence supports the general finding that landscape structure influences local biodiversity, with higher biodiversity levels in more diverse landscapes. 3.2 | Questionnaire Mostofthe82respondentsrecognizedtheimportanceofbiodiversityinagroecosystems(Figure 2a),with79.3%valuingbiodiversityas ‘veryimportant’andanadditional13.4%valuingbiodiversityas‘important’.About60%ofthestakeholdersindicatedthatabiodiversity assessmenttoolconcerningagroforestrywasimportant(veryimportant = 34.2%andimportant = 26.8%;Figure 2c).However,despitethe acknowledged importance of biodiversity, most of the respondents did not have experience with existing biodiversity tools: only nine respondents(11.0%)claimedtouseortobefamiliarwithexistingtools. FIGURE 2 Therelevanceofbiodiversityandatoolassessingbiodiversityaccordingtostakeholders,expressedinpercentageof stakeholders.(a)Importanceofagrobiodiversityaccordingto82stakeholders.(b)Perceivedimportanceofatooldesignedtoassess biodiversityinagroforestrysystems,accordingtothestakeholders.(c)Desiredinterfacesthestakeholderswantforabiodiversitytool.CP, computerprogram;Excel,spreadsheettool;QGIS,geographicinformationsystemtool;R,programming-wisetool;WT,websitetool.(d) Geographicscaleforwhichthestakeholderswantabiodiversitytooltoworkwith.(e)Practicesthatmightbeincorporatedinabiodiversity tool(addressedtofarmersandadvisorsonly).BioChar,checklistforpresenceorabsenceofseveralcharismaticandeasy-to-recordkey species;FarmChar,checklistformanagementpracticesandhabitatson/closebythefarmorfield;Woody,theamount/characterizationof woodyspeciesatthefarm/field.(f)Needsofstakeholdersinrelationtoahelpfuloutputofabiodiversitytool(EffAF,effectsofdifferent agroforestrytypes;MetricBio,totalbiodiversity;MetricSpGr,metricsfortaxa;PropAct,proposedactionstowardsimprovingbiodiversity). | 7 of 13 VANDENDRIESSCHE et al. Concerning the functioning of a tool, most stakeholders expressed preferences towards tools accessible through websites orspecificallydesignedcomputerprograms(Figure 2b).Themost preferred scale for the operation of a tool was at the farm scale, but without a lot of distinguishment as approximately half of the stakeholderswereinterestedinatoolthatcouldoperateatafieldor landscapescale(Figure 2d).Stakeholdersexpressedadesireforthe tooltointegratemappingfeatures(79.3%)andchecklistsforfarm characteristics(89.0%)asinputs(Figure S4). About70%ofstakeholdersidentifiedthebiodiversityeffectsof differentagroforestrytypes(EffAF)andproposedactionstowards improving biodiversity (PropAct)as useful outputs for a biodiversitytool.62.2%and58.5%ofrespondentsidentified,respectively, metricsfortaxa(MetricSpGr)andtotalbiodiversity(MetricBio)as desirable(Figure 2f). Aseparatepartofthequestionnairewasdirectedtowardsfarmers and advisors, the likely assessors of farm-scale biodiversity, where we tried to disentangle the effort that they were willing to investtoassessbiodiversity.Thestakeholdersreportedthatthey werepreparedtoallocatetime(13.8 h/yearonaverage)fortheassessmentofbiodiversity,and89.0%expressedaninterestinusinga checklistforon-farmhabitats,structuresandmanagementpractices in combination with specifications of the agroforestry elements. However, more than half of the respondents were still interested inassessingbasicbiodiversitymetrics(e.g.withachecklistofcharismatic species to encounter on a farm; Figure 2e).Inafree-form section of the questionnaire, some respondents suggested options to enable farmers to assess farmland biodiversity by using either a checklistofseveralspeciesorenablingphotographidentification. Other detailed suggestions indicated interests towards biodiversity differences in agroforestry designs. 3.3 | Tools Weidentified37toolsthatwereconsideredusableinagroecosystemsandintegratedatleastoneformofwoodycharacterization, or were solely built upon onsite biotic metrics, meaning they could beappliedinagroforestrysystems(Text S3).However,mostofthe toolsareunabletoevaluatediverseagroforestrydesigns.Withinthe 37 tools in our database, only 17 were able to accommodate more than one distinct agroforestry type, thereby allowing comparison of differentagroforestryoptions.Particularlynoteworthytoolswere thecapacityoftheBiodiversityMetric,SALCA-BDandEcological FocusAreacalculator(EFA)toaccountforsixandtwotimesfive agroforestrytypes,respectively(ofseventypesdefinedinText S3). Animportantobservationisthatmostofthesetoolsonlyuseapresence/absence criterion for agroforestry; more detailed agroforestry characterizationsareasofyetnotpresentinbiodiversityestimation toolkits. These selected tools used different approaches to assess biodiversity:someofthekeyapproachesbeingusedandpotential pitfalls are illustrated below with examples of tools in the database.TheBiodiversityMetric(NaturalEngland)differentiates over 100 habitats and requires a detailed mapping of the area that willbeassessed(Panksetal.,2022).However,thiscomplicated distinction of habitats could impede accessibility for nonexperts. Towardsagroforestryapplication,thisbroadscope(i.e.goingfar beyondagroecosystems)andimpededaccessibilityaremajorlimitations. The Credit Point System (IP-SUISSE & Schweizerische Vogelwarte)implementsanentirelydifferentapproach.Itmakes useofanexhaustiveanduser-friendlyquestionnaireformat(Birrer et al., 2014).However,thisapproachfallsshortinprovidingquantitative values for separate species groups, and its narrow agroforestry designrelated implementation options limit its applicability fromanagroforestryperspective.Thesefirsttwotoolsprovide deterministic,evidence-basedoutputs.Conversely,GLOBIO(PBL Netherlands Environmental Assessment agency) adopts a data- driven output for predicting different habitat types based on its access to a global biodiversity database, offering an empirical estimation for habitat assessment. Its global approach, however, imposes limitations on the number of assessable habitats, rendering it incapable of accommodating various agroforestry types and lacking distinctions among ecosystem regions (Alkemade et al., 2009).Anotherapproachtoavoidtheuseof deterministic inputs or the need for yet established empirical databases is theuseofactual,in-situbiodiversitydata(Freymanetal.,2016), butthedatagatheringforsuchtoolsmightrequireexpertknowledge. Using data from open-source biodiversity databases like theimplementationoftheGreenspaceBirdCalculator(Australian Museum Research Institute; Callaghan et al., 2020) can solve this.AtheoreticalsetuplikethiswasalsodiscussedbyBimonte et al. (2021). The incorporated processes, however, disconnect from insitu habitat specifications, troubling the comparison of agroforestryimplementationwithbaselinesystems.Toenhance this connection, highresolution data have to be available to enable parcelwise biodiversity estimates. InourNMDS-analysisofbiodiversitytools,weobserveddistinct patterns of tool relationships (Figure 3). Cool Farm Tool, Ecosystem Services Assessment Tool for Agroforestry (ESAT-A) andHabitatandBiodiversityAssessmentToolwereidentifiedas having distinct characteristics, suggesting substantial differences intheapproachesusedbythesethreetools.Thisdistinctionin tool relationships was primarily influenced by programming tool (R),maps,andEffAFtypes.FG,appandGISwerealsoimportant variables, but only had presence in the StakePOV variable. We refer to Text S3 for the interpretation of the variables mentioned. Thepresenceorabsenceofthesevariablessubstantiallyimpacts the positioning of tools along the first axes of our analysis (Figure S5).TheNMDSanalysisalsohighlightsthestakeholder's POV (stakePOV). Whereas Biodiversity Metric, Nature Smart Cities4andBPIcoincidedmostwithstakePOV,thethreetools showingtheleastcoincidencewereESAT-A,EIIandBirdscalculator. 4NatureSmartcitiesincorporatestwoapproaches,forfurtherinformation,wekindly refer to Text S3. 8 of 13 | VANDENDRIESSCHE et al. All of these latter tools are almost solely built on biotic characterization. 4 | DISCUSSION Inthisstudy,weidentifiedthekeybuildingblocksofanevidence- based tool that can be used to predict the biodiversity benefits of agroforestry implementation in agroecosystems. From the questionnaire,stakeholderswhohaveaninterestinagroforestry reported thatthey wouldlikeatool(1)to provideguidance on measuresthatcanpromotebiodiversity,(2)toassesstheeffects ofdifferenttypesofagroforestryand(3)toaddressarangeof target species groups. Additionally, stakeholders articulated a desire for tools that integrate mapping data and checklists for farmcharacteristics,mainlyfocusedonwoodycharacterizations to define agroforestry and stressed the need for enriching functionalities within agricultural tools and the need for adapting to uncertaintyanddynamicfactors.Accordingtoourquestionnaire, stakeholders are also willing to spend time to assess biodiversity, preferring comprehensive tools adapted to their intended endusers, ideally available through websites or computer programsinsimplifiedgraphicalinterfaces.Accessibilityandusabilityareanoften-occurringshortcominginagriculturaltools(Zhai et al., 2020).Stakeholdersalsoexpressedtheutilityofintegratingbothchecklistsforfarmcharacteristicsandmappingfeatures. Farm characteristics could encompass specific habitat features and management strategies, mainly tailored towards agroforestry. Unfortunately, a tool encapsulating all these features is presentlyunavailable(NMDSoutputinFigure 3;seealsoStewartetal. (2022)).Thetoolsmostcloselyalignedwiththesepreferences(e.g. theBiodiversityMetric,NatureSmartCities,BPI,EFAandHabitat Hectares)arerelativelychallengingintermsofusercomplexity and—most importantly—are not specifically tailored to agroforestry.Thisadditionallymeansthattheyarenotproperlycapable of accounting for effect differences between different agroforestrytypes,oneofthemainconcernsforstakeholders.Therefore, there appears to be an identifiable need for either a new or an adapted biodiversity assessment tool for agroforestry systems that is both evidencebased, but still sufficiently userfriendly to facilitateimplementationandapplication.Thiskindoftoolshould (1)informfarmersaboutpotentialbiodiversityimplementations, (2)reportthesepotentialeffectstopolicyactors,urgingtherewarding of agroforestry implementation. Ourstudyhighlightsthatagroforestrystakeholdersacknowledge the importance and value of biodiversity (García de Jalón et al., 2018;Herzogetal.,2012).Wedoacknowledgethefactthat thepoolofstakeholdershere(i.e.highlyeducatedandinvolved in an agroforestryproject)probably causes bias (however,they are probably also the target public to facilitate the use of a future biodiversitytool).Previousresearchhasshownthatbiodiversity perception changed especially with farmers that first had indepth interactions with biodiversity researchers about the role of biodiversityandwildlife(Gabeletal.,2018; Noe et al., 2005),underpinningthatacitizenscienceapproachcouldbebeneficialtowards creatingfarmer–researcherinteractions(e.g.FrameWORK'sfarmingclusters;Banksetal.,2023; Hager et al., 2022).Thesescientist–practitionerinteractionsareimportantto‘buildbridges’.This also confirms the findings that farmers and advisors consider the outputofproposedmeasurestobeveryvaluable.Also,rewarding farmers for the potential biodiversity outcome of naturefriendly management practices (e.g. agroforestry implementation) might give more flexibility in management, hence promoting farmers' engagementandautonomy,butalsoemphasizestheneedforan FIGURE 3 NonmetricMultidimensionalScaling(stress = 0.196)ofthedifferentcategorizationsofthebiodiversitytools.StakePOV(in blue)representsthestakeholders'pointofviewbasedonthequestionnaire. | 9 of 13 VANDENDRIESSCHE et al. assessment tool enabling selfassessment of these potential outcomes(Tasseretal.,2019). Any future tool should start from a baseline context where agroforestry implementation is being considered and should determinegeneratedbiodiversityeffects.Thestakeholder-appointed criteria require wellsuited input data for different agroforestry types, compared with these baselines, across taxa. However, such detailed scientific information is not yet available in existing metaanalyses(Figure 1).Thesedoacknowledgethepositivebiodiversity value of agroforestry interventions as in line with results from thetropics(DeBeenhouweretal.,2013;Schrothetal.,2004)and the work of the European Joint Research Center (Makowski et al., 2021;Schievanoetal.,2025)forlandscapefeatures(including agroforestry)5 supporting the premise that structurally and functionally more complex landuse systems result in greater biodiversity. However, two major challenges were identified with our approach. Firstly, gaps on the metaanalysis level occur concerning different agroforestry types and in distinguishing different functional groups. No pollinatorfocused metaanalysis was available (however, there are certain studies addressing the topic; e.g. Varah et al., 2020)andalackofdataonseveralotherfunctional groups (e.g. soilfauna, natural enemies, pest species)occurred. Wecouldonlyretrievemeta-analysesforagroforestryingeneral, silvoarable,silvopastoral,hedgerows,scatteredtreesandFG(Li& Jiang, 2021).Otherpractices(e.g.foodforestapproaches)were not covered. Most metaanalyses combined comparisons among agroforestrytypesoramongagroecosystemsand(semi-)natural habitats(e.g.forest).Thisvarietyinbaselinesmakesitimpractical to estimate the effect of agroforestry compared to baseline agroecosystems as the net effect can depend on the baseline land use selectedforthecomparison(Boinotetal.,2022).Themoreuseful approach for the question addressed here would be to compare an agroforestry treatment with a pure control. Second,thelownumberofavailablemeta-analyseslimitsthe power of the performed secondorder analyses as variation within theresultsisconsiderable.Thisisanindicationofcontextdependency in the magnitude of the effects, perhaps influenced by the agroforestry design or management factors (Jose, 2009; Kletty et al., 2023) and the limited data available. Kletty et al. (2023) reported, as such, unequivocal, but most often positive, biodiversity effects of silvoarable agroforestry and identified aspects influencingbiodiversityoutcomes.Theydefinedaffectingaspects such as the taxa, diversity metrics, management, age, site location, environment, climate, landscape structure, comparison type and samplingmethod.Agroforestryinthebroadsensecanencompass manydifferentsystemtypes(Dmuchowskietal.,2024; MosqueraLosada et al., 2009), which will generate additional variation. Theseeffectsmightbepartlyextractablewithdataavailableona casestudy level, but are as of now not available on a metaanalysis level,eitherduetovariabilityinbaselines(e.g.forests,agroecosystems)and/oradifferenceinscope(i.e.theprimarytargetisto express the overall added value of agroforestry instead of comparingittobaselinesonlyorrepresentingsystemdifferences)(Boinot et al., 2022; Mupepele & Dormann, 2022).Additionalcasestudies followed by complete metaanalyses, starting from primary studies, are thus needed to establish a database better suited towards cross-taxaEffAFtypesandlandscapecontexts.Thisisparticularly relevant given that surrounding landscape structure generally has aninfluenceonbiodiversity.Thisobservationisrelevant,asthe implementation of agroforestry systems inherently contributes to increased landscape quality. Notably, several metaanalyses have identified interactions between landscape structure and agrienvironmental measures with respect to biodiversity outcomes (Batáryetal.,2011;Scheperetal.,2013).However,itisimportant to acknowledge that meta-analytic data specifically addressing these interactions within the context of agroforestry remain unavailabletodate.Also,veryfewcasestudiesaccountforthislandscapeinteractioninanagroforestrycontext(Klettyetal.,2023), highlighting the need for additional research and compiling efforts. If we translate these observations from our secondorder metaanalysis to applicability for a tool input, we argue that some characteristics(e.g.agroforestrytype,age,treeandcropspecies) might need to be assessed empirically. However, current empirical evidence would not meet the data requirements needed to account for highly variable moderators (e.g. detailed agroforestrydesignsandmanagementstrategies).Themainreason for this lies in the wide range of agroforestry application modalities incombinationwiththerelativelylownumberofstudies(Kletty et al., 2023).Onewaytoincludethesefactorsistouseexpert opinion. Another way to achieve this is by composing dynamic models,butatpresent,theirfunctioningisstilllimited(Rahman et al., 2023).Additionalinclusionofmappingfeaturescouldenable remotehabitat(quality)andlandscapeassessment.Furthermore, this option enables landscape structure inclusion to account for mitigating effects between agroecological measures and landscape structure as literature suggests. This way,targetedinputs can be straightforward andtailored to agroforestry systems; therefore, enhancing easy application for agroforestrystakeholders,ifatleastthesestakeholdersareawareof theexistenceofatool.Thequestionnairehighlightedthatonlyalow numberofstakeholdersknowanybiodiversitytool.Existingliteratureconfirmstheimportanceofhavingtoolstoaiddecision-making alongside video tutorials and technical guides for communication purposes(Blissetal.,2019)anddeterminingthepotentialneedfora tool,alongwithrequired/desiredcriteria(Gravesetal.,2005).Also, insightsinfarmers'andadvisors'informationbehaviourarecrucial to distribute knowledge. Peer-to-peer communication still seems to be the most important facet, but website information, images, printedmaterialsandvideotutorialsarevaluedhighlyaswell(Kiraly et al., 2023).Thus,synergizingtoolreleaseswithworkshops,guidelines and tutorials might ensure successful utilization (De Vetter et al., 2022). 5h t t p s : / / d a t a m . j r c . e c . e u r o p a . e u / d a t a m / m a s h u p / J R C _ F P _ E V I D E N C E _ L I B R A R Y / i n d e x .  html.