A practical conservation tool to combine diverse types of evidence for transparent evidence-based decision-making

Making the reasoning and evidence behind conservation decisions clear and transparent is a key challenge for the conservation community. Similarly, combining evidence from diverse sources (e.g., scientific vs non-scientific information) into decision-making is also difficult. Our group of conservation researchers and practitioners has co-produced an intuitive tool and template (Evidence-to-Decision (E2D) tool: www.evidence2decisiontool.com) to guide practitioners through a structured process to transparently document and report the evidence and reasoning behind decisions. The tool has three major steps: 1. Define the Decision Context; 2. Gather Evidence; and 3. Make an Evidence-Based Decision. In each step, practitioners enter information (e.g., from the scientific literature, practitioner knowledge and experience, and costs) to inform their decision-making and document their reasoning. The tool packages this information into a customised downloadable report (or is documented if using the offline template), which we hope can stimulate the exchange of information on decisions within and between organisations. By enabling practitioners to revisit how and why past decisions were made, and integrate diverse forms of evidence, we believe our open-access tool’s template can help increase the transparency and quality of decision-making in conservation.


Introduction
Embedding the use of evidence in practice and policy to inform conservation decisions is increasingly recognised as best practice to achieve desired conservation outcomes and protect species, genetic diversity, and habitats (Addison et al., 2016;Rose et al., 2019;Gillson et al., 2019;Kadykalo et al., 2021b;Sutherland et al., 2013;O'Brien et al. in press).By learning from past successes and failures, we can understand how to do more of what works, and less of what does not work, reducing wasted resources previously spent on actions that are known to be ineffective, inefficient, or harmful (Sutherland et al., 2013(Sutherland et al., , 2020(Sutherland et al., , 2021)).Whilst avoiding wasted effort has always been important, it is perhaps even more relevant now when efficient, large-scale action is required to reverse current trends of unprecedented biodiversity loss (Leclère et al., 2020;Díaz et al., 2019).
A conservation practitioner planning to make a reasoned and informed decision using evidence needs to consider a variety of relevant information sources (Kadykalo et al., 2021b(Kadykalo et al., , 2021a;;Schwartz et al., 2018).Evidence in this decision-making context can be broadly defined as: 'Relevant information used to assess one or more hypotheses related to a question of interest' (Salafsky et al., 2019).Currently, there appear to be two main approaches to using evidence to inform conservation decisions (Sutherland et al. 2017): 1.) focusing on using 'local knowledge' derived from indigenous and local communities, practitioners, and stakeholders, typically without considering evidence drawn from wider contexts; and 2.) focusing on using scientific evidence (derived from peer-reviewed primary and secondary research in journals and syntheses, and increasingly from non-peer-reviewed reports and documents) to find generality in the effectiveness of conservation actions and provide wider, generic recommendations on best practice.
We believe that to truly inform evidence-based practice and policy in conservation, we need to combine these approaches to ensure relevant and reliable evidence from different sources of evidence are used to their full potential (Sutherland et al. 2017;Adams & Sandbrook, 2013).For example, in an ideal world, relevant evidence from the scientific literature on the effectiveness of conservation actions would be available across many different local contexts.However, there is often little to no available evidence from the scientific literature on the effectiveness of many conservation actions, and where evidence is available, actions have typically been tested on only a small subset of species and locations and may be of poor quality (Christie et al., 2020b(Christie et al., , 2020c(Christie et al., , 2020a;;Junker et al., 2020).The relevant information that can be taken from the generic recommendations of evidence collated at a global level (i.e., from syntheses of peer-reviewed research, such as systematic reviews and meta-analyses) is often perceived to be small, particularly by practitioners (Gutzat and Dormann, 2020;O'Connell and White, 2017;Walsh et al., 2019Walsh et al., , 2015)).This is often justified by the concern or perception that the effectiveness of conservation actions may vary considerably between different local contexts (e.g., habitats and species; Cook et al., 2013aCook et al., , 2013b;;Gutzat and Dormann, 2020;Levins, 1966;Shapin, 1998).
To determine, in practice, the likelihood that a conservation action will achieve its desired outcomes in a given local context requires complementing scientific studies and syntheses drawn from different contexts with evidence from more localised or contextualised sources that is relevant to the local context of interest (Cook et al., 2017;Sutherland et al., 2017;Adams & Sandbrook, 2013).These sources may include: i.) the non-peer-reviewed literature (also sometimes called the 'grey literature', which is sometimes partially included in scientific syntheses); ii.) decision-makers' own monitoring data, written experience (e.g., notebooks or logs), and research; and iii.) undocumented (or tacit) knowledge (e.g., Indigenous People and Local Community (IPLC) knowledge and experience).Just as for peerreviewed scientific evidence, the reliability and relevance of the evidence they provide must be carefully assessed (Kadykalo et al., 2021b(Kadykalo et al., , 2021a)).
In addition to evidence on the effectiveness of a conservation action, the costs and risks, feasibility, and acceptability to key stakeholders are also key factors to consider in decision-making (Kadykalo et al., 2021b).For example, consider the scenario of a conservation conflict involving the problem of controlling the numbers of geese to reduce crop damage in Scotland.In this scenario, using sport hunting (i.e., allowing private individuals to reduce the numbers of geese using permits) may be a more feasible action to take than government action to lift a ban on culling geese (Mason et al., 2018).
However, either type of action may be more or less acceptable to certain stakeholders (e.g., it may be politically challenging to lift a ban on culling) and have different levels of cost-effectiveness (e.g., sport hunting would bring in revenue at the same time; Mason et al., 2018).It is therefore very important that when combining diverse sources of information in conservation decision-making, that there is transparency and clarity over what evidence has informed decisions, as well as the thinking and reasoning used by the decision-maker (Schwartz et al., 2019).
There are a range of Decision Support Frameworks and processes to help conservationists make decisions using evidence (Bower et al., 2018;Schwartz et al., 2018;Wright et al., 2020), including: strategic foresight (e.g., horizon scanning and scenario planning; Sutherland et al., 2021); Structured Decision-Making (e.g., involving consequences tables and expert elicitation; Gregory et al., 2012bGregory et al., , 2012a)), Conservation Standards (e.g., results chains; CMP, 2020), theory of change (Rice et al., 2020), and argument maps (Keith et al., 2017); Bayesian Belief Networks (Newton et al., 2007), multi-criteria decision analysis (Adem Esmail and Geneletti, 2018;Knight et al., 2019), systematic conservation planning (Watson et al., 2011;Margules & Pressey, 2000), and cost-effectiveness or cost-benefit analysis (Cook et al., 2017).These frameworks can help collate diverse sources of information to improve decision-making, and often provide a stepwise, structured process for aiding those decisions, which may be particularly useful and warranted when decisions involve major investments of time and money, or where the negative consequences of failing to make the most optimal decision are high.
However, one major issue for practitioners in using these frameworks is that they may only be able to allocate hours or days to making many of their decisions (Sutherland et al., 2021), and thus perceive these tools as being too time-consuming to use.Other problems include the fact that some of these frameworks may not be widely known or understood by conservation practitioners, and may be perceived as too complex to use (particularly if they lack an intuitive user interface or reusable template).If more practitioners are to use evidence-based decision-making tools to inform their work, there is a need for more co-designed, user-centred tools (Rose et al., 2017;Sturm and Tscholl, 2019) that are more accessible, widely disseminated, and easier to use (Schwartz et al., 2018).
Transparency is another key issue that merits more attention in conservation decision-making.For example, whilst the frameworks mentioned earlier promote transparency in the structure and operation of the decision-making process, we believe that the conservation community would benefit from a tool that explicitly guides practitioners through transparently reporting the evidence and reasoning used to make decisions within a step-by-step process that integrates evidence from diverse sources.However, to the best of our knowledge, a freely available, co-designed, and interactive tool that achieves this has yet to be developed and we believe could play a key role in fostering more 'evidence bridges' between practitioners and researchers (Kadykalo et al., 2021a).
Here we adapt and apply the well-established Evidence-to-Decision framework used by the UK's National Institute for Health and Care Excellence (NICE) (Alonso-Coello et al., 2016) to create a versatile, co-designed decision support tool: the Evidence-to-Decision (E2D) tool.The goal of the Evidence-to-Decision tool is to make the evidence and reasoning behind conservation decisions more systematic and transparent to both internal and external stakeholders.The aim of this article is to describe the tool and explain its potential value to practitioners, providing a generic template for wider use and application in conservation practice.

Overview of tool and intended users
The Evidence-to-Decision tool (Fig. 1) draws upon the aspects of the Evidence-to-Decision framework described in clinical medicine (Alonso-Coello et al., 2016), as well as the Conservation Standards (CMP, 2020) and structured decision-making (Gregory et al., 2012b;Hamilton et al., 2020) frameworks from conservation.The tool guides users through three major steps to ultimately arrive at an evidencebased decision (Fig. 1).We refer to terms defined by the Conservation Standards (CMP, 2020) where possible.
The intended users of the tool are conservation practitioners who conduct interventions or actions to improve biodiversity in any field, sector (e.g., public or private), or location.Practitioners that contributed to the development and user testing of this tool typically worked in environmental Non-Governmental Organisations (NGOs), nature conservation-related governmental bodies, charities, or commercial agriculture companies.Their position in the organisation mostly ranged from lower to mid-level team leaders, area managers, and reserve managers, but also included some higher-level officers and managers in some relatively small NGOs.The tool was most suited to use by a single person, or in a team where a single person entered information into the tool with input from colleagues and stakeholders.The offline version of the tool can be used as a generic template to structure a transparent evidence-based decision-making process (see www.evidence2decisiontool.com and the Supplementary Information).
Figure 1 -The structure and implementation of the Evidence-to-Decision tool.Numbers, letters, roman numerals refer to the steps described in the main text.Figure 2 gives a worked example.N that Step 2 (B-G) is repeated for each action and that the size of each section is not meant to b guide -this will vary for each decision being considered and the evidence available.

Co-design process
The tool was created by a discussion of needs for a decision support tool with practitioners at various conservation organisations (Bat Conservation International, Berkshire Buckinghamshire and Oxfordshire Wildlife Trust, Froglife, Gloucestershire Wildlife Trust, Ingleby Farms, Kent Wildlife Trust, NatureScot, The Medway Valley Countryside Partnership, and The Woodland Trust).A team of practitioners and researchers worked collaboratively to co-develop and co-design the tool to iteratively improve its user-centred structure and functionality (Rose et al., 2017;Sturm and Tscholl, 2019).
We started by creating a prototype structure for the tool with some steps that are typically involved in making a decision as suggested by practitioners and drawing from the literature on Decision Support Frameworks (e.g., Schwartz et al., 2017, Bower et al. 2018).Practitioners on the author team and in the different organisations provided feedback on this structure and suggested additional steps, modifications to the order of these steps, and the types of important evidence that should be included (e.g., grey literature, local and indigenous knowledge, costs, and feasibility).We produced and usertested prototypes of the tool (using an online R Shiny (Chang et al., 2020) application) iteratively with different practitioners using real-life case studies to see how the tool could be used and improved (including scenarios with Master's students, typically from diverse practitioner-focused backgrounds, with at least three years professional experience in conservation).Many of the changes that were made included adding further guidance on how to assess scientific evidence and undocumented knowledge for biases, adding a dynamic summary table and text in the final step to aid weighing up the evidence, and explicitly integrating uncertainty into the tool through a scoring system and colour-coding of the summary table.

Structure and process of the Evidence-to-Decision (E2D) tool
In the following sections, we outline the different parts of the tool and explain how these workguidance on how to use the tool is provided in a downloadable guide on the tool website (www.evidence2decisiontool.com -where an offline template is also available) and in the Supporting Information.Questions are described in each section to act as prompts for the user to answer to guide their decision-making.
We illustrate the use of the tool and types of information a conservation practitioner using a simple hypothetical example where we are interested in reducing amphibian mortality along a road that runs through a nature reserve or protected area (Fig. 2).
Figure 2 -Diagram detailing a case study example of following the steps involved in using Evidence-to-Decision tool from Figure 1.This is implemented online via an R Shiny app (Chang e 2020): www.evidence2decisiontool.com.Note that Step 2 (B-G) is repeated for each action, with figure only showing the assessment of evidence for one action.The size of each section is not mean be a guide -this will vary for each decision being considered and the evidence available.For exam g the et al. th the ant to mple, in some cases far more evidence from undocumented knowledge may be available, and very little evidence from the scientific literature.

Define the Decision Context
What is the problem and desired outcomes?What is the relevant ecological, physical, and social context underlying the decision?
The tool prompts users to define and describe the decision context.Users are asked to give a brief description of the problem or direct threat, identify the desired focal target (i.e., species, group, or habitat), and state the ultimate goal of the conservation intervention or action (i.e., desired outcomes) (CMP, 2020).Users are encouraged to report important contextual knowledge, such as ecological (habitat types, species present), physical (e.g., location), and socioeconomic and cultural information (e.g., background on relevant stakeholders -people or groups with interests or concerns related to the context being considered; Franks et al., 2018).The tool asks users to identify constraints that may influence their decision such as regulatory structures/legislation, budget available, or personal/organisational values.

What does the available evidence suggest about the likely effectiveness, costs, acceptability, and feasibility of different potential actions?
This second step comprises several subsections, where the tool prompts users to consider different forms of evidence (Table 1) to assess the suitability of potential actions to achieve the focal targets and goals defined in the previous step.In 2.A.), users are asked to first identify a wide range of potential actions, and then in subsections 2.B-G.)they can assess whether the implementation of each action is supported by the available evidence.These subsections include the consideration of the desirable and undesirable effects on the focal target, and the costs, acceptability, and feasibility of each action.By considering the evidence, users can also identify whether certain modifications to each action are required.By the end of this step, users will have summarised and assessed the available evidence, providing the basis on which to make an evidence-based decision in the third and final step.For steps 2.B-2.F, users will provide simple summary scores for different sections (including the uncertainty associated with the evidence), which will be tabulated in the third and final step to help them make their decision.
It is important to note that if the user cannot access or find sufficient evidence to include in any of the subsequent sections, then the user should note this and make this added uncertainty in the evidence clear so this can be considered later when making a decision.

Which actions could be taken to address the problem?
Before considering the evidence for and against each action, users are asked to identify the potential actions they could take to address the focal targets and ultimate goal defined previously.This helps to ensure that users consider a wide range of potential actions and do not discount or miss out potentially useful actions that they may not have immediately considered (e.g., using techniques such as the 'vanishing options test', Red Teaming, Nominal Group Technique (Tanner et al., 2020; Table S4), Solution Scanning (Sutherland et al., 2014), and searching online databases such as Conservation Evidence (www.conservationevidence.com;Sutherland et al., 2019).We suggest that actions are defined broadly at first as later in the tool there will be time to consider if beneficial modifications can be made (2.F) based upon the evidence that has been considered (2.B-E).However, if users already have a prior understanding of possible modifications, these can be listed here separately as alternative actions and each considered separately (in this case, section 2.F may still be useful if there are minor, detailed modifications that the evidence suggests may be useful).

What do different types of evidence tell us about the desirable and undesirable effects of each action on the focal target? How certain are we of the credibility of this evidence?
Once potential actions have been identified, users are asked to summarise the available evidence on the effects of each action on the focal target and the uncertainty associated with that evidence.The tool prompts users to consider four forms of evidence from different sources: a.) peer-reviewed primary research; b.) evidence syntheses and summaries; c.) the non-peer-reviewed or 'grey' literature (a., b., and c. are collectively termed the 'scientific literature'); and d.) undocumented knowledge (see Table 1 for more detailed definitions and details of additional decision-making factors).Combining these diverse sources of evidence helps users to assess the relevance of evidence from the scientific literature (its local validity), avoids conflicts between different stakeholder needs, and importantly can give local and indigenous communities a sense of ownership over a project or conservation decision (O'Brien et al., 2021).
Table 1 -Differences between different forms of evidence for the purposes of this tool.When we use the term 'peer-reviewed', we refer to the formal process of peer-review in scientific journals, rather than organisational peer-review that is undertaken by some government bodies and non-governmental organisations.We provide a brief overview of the potential resources that users could access to retrieve and use evidence in Table 2 (which is also present in the tool guide -see Supplementary Information).Please note that this is not intended to be a comprehensive list, but does cover many of the major and readily accessible resources for evidence in conservation and natural resource management, as well as best practice advice on how to use evidence in decision-making.Templates containing basic qualitative benefit-cost information identified for all 175 NRCS Conservation Practices in the form of one-page documents.These are considered the first step towards an economic or financial analysis and designed so the user can easily review and discuss the benefits and costs of each conservation practice.

Type
Once these sources of evidence have been considered, users can then score the likely local effectiveness of the action and their certainty in their assessment.Users are given several choices, including: Harmful (the action is likely to have undesirable effects on the focal target); Ineffective (the action is unlikely to have desirable effects); Weakly effective (the action is likely to have weak desirable effects); Moderately effective (the action is likely to have moderate desirable effects); Highly effective (the action is likely to have strong desirable effects); Trade-off between benefits and harms (the action is likely to have both desirable and undesirable effects); Unsure (not confident enough to give an effectiveness score, for example, if there is no evidence available).The user will also give certainty scores to rate their confidence and certainty in their effectiveness score, which include the following categories of certainty: Very low (i.e., very low weight of evidence or no evidence available), Low, Moderate, High, and Unsure (not confident enough to rate certainty).

2.B.i. Scientific literature
How locally effective is this action likely to be based on evidence from the scientific literature?What is the overall certainty (reliability) of this evidence?
Users are asked to assess the available evidence on the effects of each action from peer-reviewed primary research, evidence syntheses and summaries, and the non-peer-reviewed ('grey') literature (see Table 1 for definitions).The tool asks users to consider: i.) the quality of methodological design (internal validity; i.e., how reliable is this evidence?);and ii.) the relevance of evidence to the question of interest (external validity; would the findings of this evidence apply/generalise to the local setting of interest?).
Critical appraisal of evidence is encouraged -whilst users are not expected to go back to the original sources of evidence syntheses and summaries, they are asked to think critically about how reliable a given study, synthesis, or summary may be with help from the tool's guidance document (including an evidence hierarchy and common biases to cautious of -see Table S1 and Fig. S1 adapted from Cooke et al., 2017 andMupepele et al., 2016).

How locally effective is this action likely to be based on your own monitoring data or notes?
What is the overall certainty (reliability) of this evidence?
As for the previous section, here users are asked to assess the reliability and relevance of any evidence they can provide from their own data, monitoring, or written experience (e.g., logbooks or notebooks) on the likely effectiveness of each action.This is separate to the grey or non-peer-reviewed literature as this evidence is usually internal (i.e., collected by the decision-maker or their organisation) rather than external, and is documented or recorded in the form of physical data or written observations -hence the distinction from undocumented knowledge (see Table 1).

2.B.iii. Undocumented knowledge
How locally effective is this action likely to be based on you and your stakeholders' knowledge?What is the overall certainty (reliability) of this knowledge?
Wheeler and Root-Bernstein (2020) suggest there is "not one unified definition for indigenous and local knowledge beyond it being the knowledge of indigenous and local people which often pertains to socialecological systems."We use the term 'undocumented knowledge' for the purposes of evidence-based decision-making to specify information that is not published or written down, which typically includes a knowledge holder's intuition, experience, wisdom, and values (also known as 'tacit' knowledge ;Tanner et al., 2020).For example, undocumented knowledge may include evidence that cannot be tied to a specific source or justified by a mechanism or explanation, but is simply 'known' by the knowledge holder.This may include forms of Indigenous and Local Knowledge (ILK; Kadykalo et al., 2021b;Reyes-García and Benyei, 2019;Wheeler and Root-Bernstein, 2020), such as indigenous storytelling (Fernández-Llamazares and Cabeza, 2018).Undocumented knowledge will play a particularly important role when there is little or no evidence available from the scientific literature and where there is concern that the effectiveness of the conservation action may not transfer well to the decisionmaker's local context (Christie et al., 2020b;Gutzat and Dormann, 2020).
In this section, users are asked to consider evidence from undocumented knowledge on the likelihood that each action would be effective in the user's local context and critically consider the uncertainty associated with such knowledge.In particular, users are advised to consider how the knowledge holder's experience, expertise, and skillset may affect uncertainty in the evidence they provideimportant biases to be aware of are described in the guidance document for the tool and Table S2 adapted from Tanner et al., 2020).

What information and evidence on the costs of this action are available and what does it tell us about its possible financial and wider non-financial risks and consequences?
Now that information has been gathered on the effectiveness of the action, users are asked to consider the likely cost-effectiveness of actions, and the wider effects that each action may have on non-target species, habitats, and stakeholders.

How much does the action cost financially and what are its resource requirements? What is the overall certainty (reliability) of these costs?
Resource requirements and financial costs can be broadly defined as the resources and finances required to implement a conservation action and form the core of assessing the cost-effectiveness of actions (e.g., Murdoch et al., 2007;Cook et al., 2017;Pienkowski et al., 2020).Users are asked to state the direct costs of implementation, which may include labour, time, consumables, overheads, and equipment/capital costs (see Iacona et al 2018 for a framework for recording direct implementation costs).If relevant, possible opportunity costs (e.g., loss of income), costs of future management and monitoring, and financial benefits are also important to include (e.g., avoided costs such as removing an invasive species and so not having to pay recurrent costs, or financial gains such as ecotourism value and Non-Timber Forest Products from implementing an action).Costs should be recorded alongside useful information (metadata) required to interpret cost data including date, currency, location, discount rates and time horizons if used.Iacona et al (2018) provide a process for reporting relevant metadata alongside costs.
Users are prompted by the tool to enter cost information from the scientific literature, guidance, and accounts, as well as from practitioners' experience and undocumented knowledge.Often there may be limited published data on the costs of interventions (e.g.White et al, in review; Pienkowski et al 2020) and in this situation users may need to rely on practitioner experience and other sources of evidence (e.g., IPLC knowledge) to estimate costs.Frameworks for thinking about types of cost and important input data may help with estimation (Iacona et al 2018).Users are encouraged to state the uncertainty associated with cost estimates (particularly if these are based on old data, taken from a different local context, or are anecdotal) and standardise costs on the same scale (e.g., cost per unit area or effort).
Users are asked to score the cost-effectiveness of each action (from very poor to high) and their certainty in this score (as for previous sections).

2.C.ii. Assess the non-financial costs, risks, and benefits for non-target species, habitats, and stakeholders
What are the wider non-financial costs, risks, and benefits of implementing this action?
Users are prompted to consider any potential undesirable and desirable effects of the action on species, habitats, and stakeholders that are not the focus of the action.For example, negative sociocultural or political outcomes from using pesticides, excluding access, or removing invasive species like reputational costs, loss of access, livelihood, or health costs.There may also be positive social or cultural benefits that the conservation action may provide to local communities or stakeholders that align with the strategic aims of the practitioner or organisation (if these were not the focal target of the action -e.g., farmers acting as stewards of their land; O' Brien et al., 2021), or if the action helps promote public engagement and/or citizen science projects -we suggest the US United States Department of Agriculture Conservation Practice Benefit-Cost Templates (see Table 2) may be useful to consider here (Hein et al., 2020).Costs, risks, and benefits on non-target species and habitats are also important to consider, such as whether types of grazing benefit the focal target (e.g., butterflies) but negatively impact other species (e.g., spiders).
Users will also be able to score the relative balance between these non-financial costs, risks, and benefits on non-targets from the following: costs/risks far greater, costs/risks slightly greater, trade-off between costs/risks and benefits, benefits slightly greater, benefits far greater, and unsure.As for previous sections, they will also be able to score their certainty from very low to high, with an option to select unsure.

2.D. Assess acceptability
Are the effects of implementing this action acceptable to all the key stakeholders?Are there sociocultural barriers to implementing this action?
We define acceptability as whether each action aligns to the values held by the practitioner and the key stakeholders (who were identified in Step 1: Define the decision context).Stakeholders will hold many human values (i.e., concepts or beliefs about desirable end states or behaviours that guide their choices and evaluation of outcomes; Schwartz and Bilsky, 1987) -see eight main types in Table S3.It is outside the scope of this tool to elicit these values directly from key stakeholders, so we suggest that the user and their organisation gathers this information using suitable methods already used by organisations (e.g., formal consultations, focus groups etc.) and summarise the key findings here.
We encourage users to provide background on the important concepts, beliefs, and motivations that stakeholders hold and how these relate to the decision context, and then consider how the potential effects of this action on targets (2.B.) and non-targets (2.C.ii.) align with these values.As for previous sections, users will be able to score their assessment of acceptability (from very low to high) and their certainty in this choice.

Can this action be successfully accomplished and properly implemented?
Users are asked to assess the feasibility of actions by considering both the costs (2.C) and acceptability (2.D) of the action to the user and key stakeholders.They are prompted to think about not only financial feasibility, but also social and operational feasibility; as in previous sections, users will then score their assessment of feasibility and their confidence in this choice.When assessing the feasibility of each action (and their certainty in this selection), users will be able to choose from the same options as in previous sections (very low to high).

How can the action be modified based on the previous evidence gathered?
After considering the previous evidence gathered, users are now asked to identify and consider modifications to improve each action's effectiveness, cost-effectiveness, feasibility, and acceptability.
Certain scientific studies may have trialled different methods of implementation, or the undocumented knowledge reported by the user may point to a better design or way of undertaking the action.For example, a structural action may also be too expensive to implement using certain materials, but using cheaper alternatives (e.g., suggested by undocumented knowledge) could increase its costeffectiveness and feasibility.In section 2.A, we suggested that actions could be defined broadly so that in this section beneficial modifications can be considered based upon the evidence gathered in previous sections.However, if users already detailed possible modifications as different potential actions in section 2.A, we suggest that this section could still be used to determine if any additional detailed modifications may be beneficial (i.e., depending on the level of detail specified in section 2.A).
Users will be able to score their assessment of the degree to which these modifications could improve this action, as well as their confidence in this choice in the same way as for previous sections.

How likely is this action to be locally effective based on all the evidence gathered? What is the overall level of uncertainty associated with these conclusions?
In this final stage of gathering evidence, the tool helps users summarise the likely local effectiveness of each action (whether modified or not), and the important cost, acceptability, and feasibility considerations.Users are prompted to reflect on the level of uncertainty associated with the evidence for and against the implementation of each action and whether it is sufficient evidence on which to base a decision.At this summarisation stage (and in Step 3: Making an Evidence-Based Decision), the tool guide also flags important biases to avoid that often affect organisational decision-making (such biases also affect undocumented knowledge; see Table S2), as well as approaches to counter these biases (see Table S4; adapted from Tanner et al., 2020).

3.A. Weigh up the evidence for and against different actions
Which action(s) are the best ones to implement to achieve the focal targets and goals defined at the beginning?Which action(s) should not be implemented?Justify these choices.
In this final step of the tool, the tool prompts users to carefully consider how each action tackles the original decision or problem being considered (in Step 1. Define the Decision Context).This involves weighing up how locally effective, cost-effective, acceptable, and feasible each action and whether its implementation is justified.There are many possible ways to do this, which are discussed below.A summary table is provided which collects the scores provided by users in previous steps (2.B-F) and displays them for each action side-by-side to aid comparisons -cells are coloured to allow users to visually assess uncertainty in their decision-making (see guidance below).
The tool encourages practitioners to note if actions are: 1.) retaining biodiversity and avoiding impacts, 2.) minimizing impacts, 3.) restoring or remediating impacts or 4.) compensating for impact or renewing biodiversity (corresponding to the Mitigation and Conservation Hierarchy; Arlidge et al., 2018;Milner-Gulland et al., 2021), and prioritise actions that avoid and minimize threats over restoration and compensatory measures (see Booth et al. 2019).A link to the diagram of this hierarchy is presented within the tool and tool guide.
A summary table is provided that automatically displays the scores (in the online tool) given by users in previous steps (2.B-G for each action) to allow a simple comparison across different decision-making factors.In addition, users are encouraged to first consider whether there is sufficient certainty in the evidence gathered to make a decision and the risks involved -the summary table helps assess uncertainty through colour-coding cells to represent certainty in the evidence gathered previously.For example, if insufficient evidence has been found earlier from the scientific literature or undocumented knowledge (e.g., fields left blank or with very limited information) then this added uncertainty can be taken into consideration here.We therefore encourage users to consider whether undertaking no action may be the optimal strategy, particularly if there is great uncertainty and/or great risk from undertaking any action.
If users believe there is sufficient evidence to support implementing some of the actions, we advise that they could start to determine the optimal actions to use by eliminating actions that are unlikely to be cost-effective (e.g., if they exceed budget limits), particularly where the evidence suggests other actions are either relatively less expensive (and equally effective) or more effective (and equally expensive).
Actions that are clearly unacceptable to the practitioner or key stakeholders, or are not feasible to implement could also be rejected relatively early on.Guidance and examples on possible trade-offs between different decision-making factors are provided in the guidance document.

What is the overall decision, what are the next steps, and why?
Once actions have been selected or rejected for implementation, the final step of the tool asks users to summarise and justify their decision.Users are prompted to set out the rationale and evidence behind their decision and to detail the next steps they will take.When deciding on the next steps, we recommend users consider drawing up a strategy to implement the actions they have selected.If there is too much uncertainty to make their decision, we also recommend investing in the use of a more detailed Decision Support Framework or tool to gather and assess the evidence more thoroughly (see Discussion for suggestions).Users may also wish to consider further research to test possible modifications to a particular action, better understand the risks of implementing an action, or consult more widely with stakeholders on ways to implement different actions to ensure they are cost-effective, acceptable, and feasible.
We recommend that implemented actions are then rigorously evaluated and reported to the wider community as part of the continual generation of evidence, regardless of the outcomes (i.e., positive, negative, or neutral).Many journals facilitate practitioners in publishing reports of tests of conservation actions (e.g., see the Conservation Evidence journal, Ecological Solutions and Evidence, Conservation Science and Practice, and material stored in the British Ecological Society's Applied Ecology Resources; Cadotte et al., 2020;Sutherland et al., 2020).

3.C. Document and report decision
Using the online tool, users can download a report that details the information they gathered and filled in throughout the entire process.We also provide an offline template version of the tool so that users can also create a documented report of their decision-making process (see Supporting Information).Documenting the evidence, logic, and reasoning behind the decision enables greater transparency in decision making and we suggest that these reports could be stored in 'decision libraries'.These could be used to disseminate and share information on how past decisions were made to internal and external practitioners, stakeholders, and organisations, enabling practitioners to revisit and reassess decisions based on new evidence or for new projects.This lends itself to the iterative concept of Adaptive Management and links this process in a complementary way to Evidence-Based Conservation (Gillson et al., 2019;Dubois et al. 2019).An illustrative example working through the Evidence-to-Decision tool is shown in Figure 2, whilst an example report can be retrieved from navigating to tab '3.
Make an Evidence-Based Decision' in the online tool (www.evidence2decisiontool.com).

Data sharing and security considerations
We have designed the online tool so that the data inputs and outputs users contribute or generate are not publicly available.The only time that data or text entered into the tool is stored is when the user bookmarks their work; when this occurs, the state of the tool is saved on a private shiny server maintained by Conservation Evidence, accessed via SSH for administration and uses HTTPS (i.e., uses an SSL certificate).This design was made to help keep any private or sensitive data entered by users away from the public domain (given that data security, particularly that of personal data, is both a legal and a moral duty) and this means that open, transparent, and public scrutiny of decisions made using the tool requires the user and their organisation to publish the downloaded reports.We believe, however, that this design will ultimately encourage greater uptake amongst practitioners and organisations that need to comply with data sharing and security legislation and rules when making decisions on sensitive issues, whilst enabling internal scrutiny of decision-making at the very least.
Clearly, we would encourage that reports be made public and open access for external scrutiny as soon as possible, potentially with redacted areas if data sharing and security continues is a concern.
The open sharing and publication of reports generated by the tool could be used as one desirable (but not essential) step towards accreditation or recognition for evidence use through schemes such as 'Evidence Champions' led by the Conservation Evidence project.

Exploration of strengths, limitations, and proposed use
The Evidence-to-Decision tool we have presented here has three key strengths.First, the tool enables users to make the rationale and process behind making decisions explicit and documentable; the online version of the tool allows users to fill in and produce a downloadable report that documents the users' decision-making process.We envisage that this tool could be used by organisations to make 'decision libraries' detailing how and when evidence-based decisions were reached, which can be documented and disseminated across the organisation, to stakeholders, and other organisations.This would allow others to see the logic and reasoning behind decisions made now and in the past.This is important because future staff or stakeholders can use these reports to look back to see why past decisions were made, and update or modify the reports to help them make decisions in the present or future.The process of documenting the decision-making process could also potentially help share and communicate common issues, successes, and failures, and promote greater sharing of knowledge on best practice in conservation (Schwartz et al., 2019).
Second, it presents a formalised approach to combining evidence from diverse sources on different aspects of a conservation management action's implementation to reach an evidence-based decision.
Previously, combining diverse forms of evidence has been a major, controversial challenge for evidence-based decision-making (Gillson et al., 2019;Gutzat and Dormann, 2020), where two different approaches have generally been pursued: 1.) focusing on making generalised recommendations from the scientific literature (which has been criticised for offering 'a view from nowhere'; Shapin, 1998); and 2.) focusing almost exclusively on what is perceived as 'locally relevant' evidence (Christie et al., 2020b;Sutherland and Wordley, 2017) that has been derived from the same or very similar specific context to the decision-maker's (such as Indigenous and Local Knowledge; Wheeler and Root-Bernstein, 2020).Neither of these approaches is reasonable or realistic in contemporary conservation (Sutherland et al. 2017;Adams & Sandbrook, 2013).If we place too much emphasis on generalised recommendations from scientific evidence, we risk alienating practitioners (reinforcing "a perception of the disconnected ivory tower of science"; Rose, 2018) and ignoring other important forms of locally relevant evidence that can guide decision-making (Wheeler and Root-Bernstein, 2020;Adams & Sandbrook, 2013).Alternatively, if we focus only on highly specific, locally relevant evidence from sources such as undocumented knowledge, we may ignore important scientific evidence, create conflict (e.g., see (Redpath et al., 2013), and limit our knowledge of effective actions to only those that have been conducted locally, potentially leading to misinformed decisions (Cook et al., 2010;Dicks et al., 2014;Persson et al., 2018).This tool's approach provides a way of harmoniously combining these two differing standpoints, particularly by drawing on the concept of local co-assessment of evidence (Sutherland et al., 2017).This method assesses the local relevance and applicability of scientific evidence to the local setting of interest through directly integrating undocumented knowledge as an equally valuable form of evidence.
Our tool further integrates the key factors of feasibility, acceptability, and costs that ultimately play a major role in practitioners' decision-making, facilitating the inclusion of evidence from a diverse group of stakeholders to contribute to an evidence-based decision (Kadykalo et al., 2021b(Kadykalo et al., , 2021a;;Wheeler and Root-Bernstein, 2020).The tool's integrative concept can also be seen as facilitating phronesis.
Aristotle talked of three intellectual virtues: episteme or scientific knowledge, techne or technical knowledge or 'know how', and phronesis characterised as prudence or wisdom in practice (Flyvbjerg, 2004).Therefore, by combining these diverse forms of evidence and real-world constraints within a transparent, structured decision-making process, we believe our tool provides a realistic, pragmatic way to facilitate more evidence-based decision-making by conservation practitioners.
The third strength of the tool is its versatility; we have deliberately designed as a generic template that can easily be modified and adapted through collaboration and co-design to produce customised versions (Rose et al., 2017).As has been demonstrated in medicine (Rosenbaum et al. 2018), this type of tool can be used in situations where there is a great deal of available evidence, and in situations where evidence is severely lacking or absent from any or all sources (e.g., scientific, undocumented knowledge, or otherwise).In either scenario, the abundance, sources, and quality of evidence used to inform decisions can be transparently documented and does not prevent use of the tool.Our collaborative team of researchers and conservation practitioners will continue to ensure that the tool is further refined, adapted, and embedded in the decision-making processes of more conservation organisations by promoting the tool with outreach activities, training, and guidance.We are particularly keen to expand the base of users to beyond those in the UK and USA to integrate more feedback from practitioners working in underrepresented parts of the world, particularly decision-makers who are (or work closely with) Indigenous Peoples and Local Communities (IPLCs).We believe that future work can improve upon the generic template we have developed and make aspects of the process more advanced in a modular fashion.The code used to develop and deploy the tool is open source and available from Zenodo (and linked GitHub repository -see Data Sharing and Accessibility).
To make the most of the tool, we strongly recommend that users first consider the Strategic Evidence Assessment Framework (Sutherland et al. 2021) to decide how much time they should invest in using the tool, collating evidence, and completing the various different sections.This framework suggests that the time invested in using evidence-based decision-making tools should be scaled based upon the uncertainty and the magnitude of risk associated with a decision (Sutherland et al. 2021).The versatility of the tool means that users could spend a great deal of time thoroughly completing each section, or could spend a smaller amount of time quickly reviewing a more constrained set of evidence without a detailed consideration of aspects such as costs for example.This is ultimately the responsibility of the user and their organisation to decide based upon weighing up the risks associated with the decision and the uncertainty associated with the likely effectiveness and consequences of any proposed actions.
For decisions carrying great risk and/or there is great uncertainty associated with the impacts of proposed actions, we would recommend that the tool is completed thoroughly alongside other Decision Support Frameworks that practitioners may already use (e.g., the Conservation Standards or structured decision-making tools).Some users suggested that, for decisions carrying high risk or high uncertainty, our tool could be as a starting point from which to implement more complex, comprehensive Decision Support Frameworks to thoroughly interrogate different aspects of the decision-making process (e.g., considering a wider theory of change using the Conservation Standards and Miradi).Equally, if practitioners are comfortable using a certain Decision Support Framework, they could potentially integrate some sections and concepts supplied by our tool's template into their existing processes to inform their decision-making.However, for decisions with low risk and low uncertainty, it may be deemed appropriate to undertake a rapid assessment of only the evidence that is rapidly and readily available using our tool.Either way, the advantage and principle of the tool is that whichever approach is adopted, the details of the evidence and reasoning used is documented and transparent so that it is clear whether the process was a shallow or deep dive into different sources of evidence (Rosenbaum et al. 2018).
Of course, like any decision-support tool, the Evidence-to-Decision tool has some limitations.Clearly, it can only act as a guide to users, and we can only encourage (but not enforce) the documentation, reporting, or sharing of their decision-making process.Therefore, the scope for internal and external individuals and organisations to review, quality check, and revisit decisions is the responsibility of the user (see Data Sharing and Security in Results).The tool also cannot stop users selectively picking the types of evidence they consider (e.g., ignoring scientific evidence) or falling foul of several decisionmaking biases (Tables 3 and 4, also highlighted in the tool), which has catalysed a movement towards more Evidence-Based Conservation (Cook et al., 2010;Sutherland et al., 2004).However, as has been found in medicine where the concept for this tool originated (Rosenbaum et al. 2018), this limitation should be counteracted by the transparent nature of the tool, enabling others to see exactly what evidence (if any), judgements, and reasoning informed the decision.

Current and future links to other decision-making tools and frameworks
The Evidence-to-Decision tool was designed to adapt a medical framework, which is typically used to assess whether to recommend the use of treatments and drugs to treat a specific disease, to the case of assessing whether certain conservation actions or interventions are likely to achieve a focal goal.This tool therefore sits within wider decision-making frameworks such as the Conservation Standards (CMP, 2020) and Miradi tool, which address the planning of conservation projects on a more holistic scale using results chains, for example.Our tool can therefore be used within this framework to more directly, and potentially in greater depth, examine the evidence for and against different alternative conservation actions or interventions to specifically address a single threat or goal.The outputs and decisions from using the Evidence-to-Decision tool can then be linked back into the Conservation Standards framework and Miradi tool to consider the wider strategy and how actions may interact.We are committed to future work to directly integrate our tool's process into frameworks such as these and particularly provide tools that explicitly assess the reliability and relevance of diverse sources of evidence (i.e., allowing the direct combination of scientific evidence with undocumented knowledge).
We believe that the tool in its current form works best for single project-based decisions with relatively short implementation or decision timescales (e.g., several hours or days to decide what to do) -but of course the tool is designed to be used over a longer timescale, whereby decisions are revisited and reconsidered based upon new evidence and insights, and actions adapted and modified to improve and refine outcomes (i.e., linking Adaptive Management and Evidence-Based Conservation in a complementary way; Gillson et al., 2019;Dubois et al. 2019).

Future Directions and Conclusion
Overall, we believe the power of the Evidence-to-Decision tool is to transparently show the evidence and reasoning that were used to make decisions for future reference and enquiry.This lays the foundation for greater internal and external scrutiny of how decisions were made and whether different sources of evidence were missed or ignored (whether intentionally, through ignorance, or issues of accessibility).We hope that we can encourage more practitioners to use the tool rigorously and integrate diverse sources of evidence into their decision-making (Rose et al., 2017;Sturm and Tscholl, 2019) by creating a community of practice around the tool, along with free accessible guidance on evidence assessment (e.g., in evidence synthesis -see https://synthesistraining.github.io/),and specific training on the tool's use (e.g., via online videos, outreach events, and guidance documents through repositories like Applied Ecology Resources; https://www.britishecologicalsociety.org/appliedecology-resources/search/; Cadotte et al., 2020).We are pleased to report that many of the practitioners that tested versions of this prototype tool are now working to embed the use of the Evidence-to-Decision tool in their organisations' decision-making processes, alongside the use of the Conservation Evidence database and website (www.conservationevidence.com),through the Evidence Champions scheme run by Conservation Evidence.We believe that further work to gather feedback and improvements for our tool and other Decision Support Frameworks (particularly from users from underrepresented backgrounds) is key; we can learn lessons from clinical medicine where inclusive and comprehensive user-testing forms a key component of the ongoing development of interactive Evidence-to-Decision frameworks (Rosenbaum et al. 2018).Such work will also help to promote the uptake of the tool amongst a wider, more diverse community of conservation practitioners.
Ultimately, we hope that the template provided by the Evidence-to-Decision tool offers a way forward to making transparent and evidence-based decision-making more routine and mainstream in conservation and related disciplines (e.g., natural resource management).We believe the tool could help in catalysing a cultural shift in conservation organisations and groups towards standardising and embedding evidence-based practice within or on top of existing decision-making processes.We encourage others to adapt, modify, apply, and operationalise the tool based on their specific needs so we can better equip, encourage, and help those working on the frontline of conservation to make more transparent decisions based on the best available relevant evidence.
Booth, H., Squires, D., Milner-Gulland, E.J., 2019.The mitigation hierarchy for sharks: A risk-based framework for reconciling trade-offs between shark conservation and fisheries objectives.Fish and Fisheries 21, 269-289.https://doi.org/https://doi.org/10.1111/faf.12429Bower, S.D., Brownscombe, J.W., Birnie-Gauvin, K., Ford, M.I., Moraga, A.D., Pusiak, R.J.P., Turenne, E.D., Zolderdo, A.J., Cooke, S.J., Bennett, J.R., 2018. Making , data, or books.Pre-prints, private reports, analyses, published reports, and data that are not peer-reviewed.See Applied Ecology Resources for a searchable database (https://www.britishecologicalsociety.org/ap plied-ecology-resources/search/).stakeholders such as that preserving traditions is important to the local community group.AcceptabilityInformation on how well the effects of an action align with the values held by stakeholders.It is judged to be unacceptable to implement an action that would limit access of local people to an area used for a local tradition.FeasibilityInformation, partly drawn from costs and acceptability, on whether the action can be implemented given the available resources, time, and conditionsIt is judged an action is not feasible based on the logistical difficulties in moving heavy equipment to the required location.

Table 2 -
A list of useful resources and guides that provide evidence and advice on evidence use in decision-making.This is not designed to be a complete comprehensive list of all resources available, but a starting guide as to what major resources are readily available.