**4. Discussion**

Our findings demonstrate that forest projects that pursue ecological restoration and rehabilitation can deliver multiple benefits to local communities, which are grouped across the indicators listed in the analysis framework presented in Table 2. After ten-years of project implementation, local adaptive capacity has been enhanced, and mitigation has taken place. There have also been some tangible development benefits.

Perceptions of community and developers under both projects are aligned in terms of shortto medium-term e fforts that have built adaptive capacity (see Table 2). By training communities in alternative agriculture, building basic infrastructure and restoring natural capital (e.g., by increasing fish availability under Eden, and controlling forest use under Makira), forestry and tree planting projects improve availability of asset indicators (e.g., fishing equipment or access to credit) and flexibility indicators (e.g., changing cropping strategies) used to cope with climate change impacts. Increased community cooperation through the projects' networks results in improved social organization of adaptive capacity. Increased learning is evidenced by the learning of new knowledge on agriculture and environmental conservation under both Eden and Makira. Our findings show that such strengthened access to capital availability and resources fosters a range of development impacts, as evidenced by the indicators grouped across the five capital assets listed in Table 2 (i.e., human, social, natural, physical and financial). However, agency (or capacity and willingness to mobilize multiple resources to respond to pressing environmental change challenges in practice, as per Cinner et al., 2018 [10]) remains limited. Our findings indicate that increased capital, as evidenced by the indicators, does not automatically translate into enhanced adaptive capacity and development. For example, as recorded in the semi-structured interviews, while the delivery of agricultural training is reported to enhance knowledge and broaden adaptation and revenue generation opportunities, we note low adoption rates, excessive complexity and the lag time of benefits limit achievement of long-term impacts on the ground. The lack of long-term impact metrics limits capacity and willingness to progress towards more "transformative" adaptation and development [43]. This suggests a need to implement a more integrated, holistic approach to landscape managemen<sup>t</sup> and forest conservation.

The need for integrated climate-development approaches across time and space to sustain livelihoods is shown to be key in light of the high expectations of REDD+ and forest conservation projects, such as Makira, to foster mitigation-driven development [44]. We found that in instances where carbon verification and monitoring are in place, and enhanced carbon stocks from avoided deforestation are a formally recognized benefit, the capacity to generate carbon income and manage the revenues remains limited. We note that the carbon market is perceived as di fficult to access not only by 'non-carbon' projects such as Eden, but also by certified REDD+ actors such as Makira. If conservation projects like Eden were to invest in baselining and reporting of mitigation benefits, their capacity to access carbon funding streams would be augmented. This would help overcome major challenges reported across wider climate-development projects, which often lack the evidence base needed to attract continued investments [16]. Monitoring alone does not ensure carbon benefits will materialize, and this research stresses that institutional considerations are key in achieving impact. An adequate understanding of carbon cycles, combined with coordinated work between projects and public institutions across levels, is vital to ensure usability of carbon schemes [45]. For instance, if the capacity to e ffectively manage carbon funds amongs<sup>t</sup> WCS, Tany Meva (the carbon fund managing agency) and governmen<sup>t</sup> stakeholders was improved through better coordination, mitigation-development co-benefits of Makira would be enhanced and the carbon sequestered could result in increased

development benefits. This suggests that failure to engage diverse stakeholders poses major constraints to the effective delivery of the expected benefits [46]. The pursuit of multi-stakeholder cooperation should be considered as a key priority to enhance capacity of projects such as Makira to address complex climate-development challenges [47].

When looking at the reported community perceptions of climate and development needs and the benefits stemming from projects, there are instances where project benefits address the identified development needs, and instances where they do not. While socio-environmental benefits are appreciated in both projects, local capacity to make use of them is constrained by the need to solve urgen<sup>t</sup> challenges that are not related to forest conservation. Development is constrained by lack of access to basic healthcare, drinking water and education, combined with village isolation and limited access to credit. While Malagasy governmen<sup>t</sup> policies promote an environmental and socio-economic development discourse aligned with the international story of 'reforestation and REDD+ for development', the state is largely absent in the remote realities where projects operate. Non-state actors are expected to deliver where the governmen<sup>t</sup> cannot, due to lack of resources, capacity or corruption [48]. In pursuing a more holistic approach, adequate mapping of project benefits onto locally identified challenges must be ensured, rather than simply focusing on primary conservation goals. This indicates that for meaningful long-term engagemen<sup>t</sup> of multiple stakeholders in joint climate conservation and development to be achieved, it is essential that the adopted strategies are locally-relevant [49].

Uncertainty about how development benefits will be sustained raises concerns about the challenges reported more widely in international conservation and development programs [50]. There is a need to ensure that climate-development projects do not repeat the cycle of hype and disappointment of "conservation fads" [51] (p. 1). We argue that a system change that enables REDD+ and forestry projects to deliver successful adaptation and mitigation needs to include development. Findings here show that by setting up integrated operations across actors with varied operational foci can enhance the capacity to simultaneously tackle multiple challenges and deliver co-benefits. In thinking holistically, mixed operational and funding modalities should be identified at the outset of project development.

We observe that the monitoring and funding limitations reported in our cases derive from a lack of clear vision about the specific long term multi-benefits being sought. To define the missing holistic vision and achieve multiple wins, there is a need to better align community needs and objectives with project and donor commitments [52]. This is an area where climate-development approaches often fail to deliver practical advances, where stakeholder engagemen<sup>t</sup> is commonly pursued as a box-ticking exercise to meet demands of multiple donors, rather than addressing clearly identified community needs [53]. In strengthening community and project cooperation and alignment, knowledge co-production is recognized as a suitable approach to jointly address common priorities [54]. In our case studies we show that monitoring and knowledge co-production should be grounded in a theory of change that explicitly links project deliverables, outputs, outcomes and impacts over time, particularly envisioning the operational modalities, multi-stakeholder cooperation and continuity of funding needed to achieve climate-development wins [55].

By shaping a participatory theory of change, grounded in locally-identified priorities, the capacity of projects to generate a shared understanding and strengthen synergies among multiple stakeholders will be enhanced [46]. All these elements align with the definition of a holistic landscape approach, understood as a long-term collaborative process that brings together diverse stakeholders towards achieving multiple and sometimes conflicting objectives—in our case including adaptation, mitigation, and development [47,55]. These lessons are transferable to a range of internationally-led programs in Madagascar and across the globe, which integrate climate and development dimensions into conservation.
