Incentive-based conservation in Peru: Assessing the state of six ongoing PES and REDD+ initiatives

Incentive-based conservation has gained ample notoriety over recent decades, particularly across Latin America where targeted incentives feature prominently in environmental services initiatives, such as for carbon storage or watershed regulation. Here we first develop an analytical framework for assessing the Peruvian initiatives of conservation incentives. We then identify six ongoing interventions that have introduced incentives conditional upon compliance with voluntary environmental commitments. We collected information from secondary sources and conducted semi-structured interviews with thirty nationaland local-level stakeholders. We scrutinized the extent to which such initiatives featured impact-oriented design and implementation elements, as typically recommended in the state-of-the-art literature on Payment for Environmental Services (PES) and Reducing Emissions from Deforestation and forest Degradation (REDD+). We found only limited adoption of such recommendations, including spatial targeting, payment differentiation, enforced conditionality, and customized measures nurturing locally perceived equity and transparency. We argue, supported by a still incipient rigorous evidence from impact evaluations, that suboptimal design and implementation choices probably have influenced outcomes towards limiting the sought-for environmental and welfare impacts. We discuss three critical aspects for upscaling: overcoming financial and legal constraints, strategic involvement of non-government stakeholders, and more impact-oriented design of the interventions.


Introduction
Command-and-control regulatory approaches have traditionally dominated conservation policies, often implying negative social consequences for local populations, such as the eviction of communities due to the establishment of protected areas (Brockington and Igoe, 2006). To boost effectiveness and better reconcile environmental conservation with development goals, local participation, incentives, and benefit sharing have been increasingly seen as essential features of desirable conservation practices over recent decades, thus deviating from an exclusionary 'fencing' approach . Ensuring local buy-in and benefits from sustainable resource management interventions is thus recognized as a significant priority for conservation initiatives (Wali et al., 2017).
Providing incentives is viewed as a promising way to deal with social issues arising from the uneven distribution of costs from the implementation of conservation initiatives (TEEB, 2010). These costs generally affect vulnerable populations in economically marginal yet environmentally sensitive areas. In this sense, a first generation of incentive-based initiatives, labeled as Integrated Conservation and Development Projects (ICDPs), emerged especially in protected area management since late 1980 ′ s, and were implemented "to stabilize land use outside protected boundaries and to increase local incomes, in order to reduce the pressure for further exploitation of natural resources in the protected area" (Wells and Brandon, 1992, p. 3). However, the first generation of project experiences revealed problems affecting their implementation, including difficulties in addressing fundamental tradeoffs between conservation and development objectives Wells and McShane, 2004).
A second generation of conservation incentive-based initiatives emerged in response to the problems encountered by ICDPs, involving conditional direct payments to compensate for foregone incomes and added costs incurred by land managers that provide environmental / ecosystem services (ES) (Ferraro and Kiss, 2002). The key concept here has been Payment for Environmental Services (PES). Theoretically, the greatest innovation differentiating this generation of conservation initiatives from the previous one is that payments or rewards are to happen only conditionally, i.e. as a quid pro quo for the provision of ES (Wunder, 2015). On the international climate agenda, this groundbreaking conditionality approach to incentivize ES provision was also reflected in the global initiative known as Reducing Emissions from Deforestation and forest Degradation, conservation of existing carbon stocks, sustainable forest management and enhancement of forest carbon stocks (REDD+). REDD+ was originally conceived as a result-based payment to countries for reducing their emissions of greenhouse gases (GHG) from deforestation (Angelsen, 2017). It was envisaged that these incentives from carbon markets to countries should be passed on to the land managers on the ground, in a type of multilevel PES scheme (Wertz--Kanounnikoff and Angelsen, 2009). However, including much smaller than anticipated REDD+ financing flows, coming so far mostly from bilateral donors, REDD+ has since evolved more towards a multi-objective set of policies addressing many connected concerns including rural poverty, biodiversity conservation, climate change adaptation, indigenous rights, and natural resource governance (Angelsen, 2017;Duchelle et al., 2018;Sunderlin and Sills, 2012;Wunder et al., 2020b).
Latin America has been one of the most fertile regions for the adoption of incentive-based conservation strategies such as PES and REDD+, given the typically large and resource-rich yet threatened nature of its forests, and a tradition for numerous development-oriented policies promoting natural resources exploitation (Balvanera et al., 2012;Ezzine de Blas et al., 2017). In addition to multiple local initiatives throughout the region, national PES programs in Costa Rica and Mexico are among the oldest worldwide, and have paved the way for other similar initiatives recently implemented elsewhere in the region, such as in Ecuador, Brazil, and Peru (Rosa da Conceição et al., 2015). As part of its climatic compromises within the frame of the Paris Agreement, Peru recently committed to unconditionally cut its GHG emissions by 30% until 2030, and an additional 10% cut conditioned by the availability of international support (Perú, 2020). Alongside the public National Forest Conservation Program (NFCP) and local REDD+ projects , other parallel initiatives aligned with REDD+ goals have focused on implementing actions inside indigenous territories, covering jointly around 15% of the Peruvian Amazon (Espinoza Llanos and Feather, 2012). Furthermore, various local watershed-based PES-like incentive schemes (Tristán Febres, 2018) have also emerged over the last fifteen years.
Despite increasing adoption of incentive-based conservation initiatives during the last decades, only few country-level assessments exist in the literature (e.g. Suich et al. (2017); Pagiola et al. (2013)). Nevertheless, we know that contexts, including national-level finance, legislations and upscaling often matter a lot for appropriate incentive development (Börner et al., 2017). Hence, we aim to look for commonalities and differences among six ongoing Peruvian initiatives, scrutinizing to what extent their design and implementation has incorporated typical recommendations from the state-of-the-art conservation incentive literature (Engel, 2016;Wunder et al., 2018Wunder et al., , 2020a. In Peru, an institutional milestone for replicating these initiatives was the enactment of Law No. 30215, "Law of Rewards for Ecosystems Services Mechanisms" in 2014, which formally introduced "voluntary agreements that establish actions of conservation, recovery and sustainable use to ensure permanence of ecosystems" as responsibility of the Peruvian State. We also discuss the role of the Peruvian State in the emergence and design choices of the analyzed initiatives. Our analysis may thus have relevance for other countries where nationwide legal frameworks for promotion of incentive-based conservation are either being designed (e.g. Brazil) or implemented (e.g. Colombia, Vietnam) (Jackson, 2018;To et al., 2012). We structured this paper as follows: in Section 2 we present the analytical framework. In Section 3 we describe the research methods, while the findings are presented in Section 4. Section 5 discusses the findings and presents concluding remarks.

The analytical framework
For the purpose of this paper, we define incentives as inducements aimed to make nature conservation (and ES provision associated with it) more valuable for land managers, as compared to degrading practices (Börner and Vosti, 2013;Emerton, 1999). The policy instruments that this definition covers include both positive financial and/or non-financial inducements (e.g. in-kind rewards consisting in training, equipment and agricultural inputs). Conversely, penalties and sanctions count as disincentives aimed to discourage nature degradation. These and other instruments may operate simultaneously in the form of 'policy mixes' customized to specific contexts (Bouma et al., 2019).
Given its wide adoption worldwide (Salzman et al., 2018), we are especially interested in the second generation of conservation incentives as described above, that is those featuring some level of conditional delivery of performance-based rewards. In the Peruvian context, this element is present in three types of interventions: Hence, throughout this paper we will use the term conditional conservation incentives (CCI) to generically refer to these initiatives. Typically, the performance conditionality that is contractually fixed and rewarded refers thus to outcomes (e.g. forest area conserved), rather than the longer-term ES impacts properly (forest carbon retained, stable water flows, etc.). Our analytical framework focuses on three aspects: (1) preconditions and enabling factors for CCI emergence; (2) design and implementation features potentially influencing their effectiveness; and (3) environmental and socioeconomic outcomes observed so far.

Preconditions and enabling factors for CCI emergence
The emergence of CCI is favored by the presence of certain economic preconditions, and institutional enablers (Wunder et al., 2020a). First, economic preconditions need to ensure that ES users (individual or collectively) are willing to pay enough to compensate land managers for mitigating their potentially ES damaging actions. Theoretically, this means that ES users' maximum willingness to pay (WTP) must meet at least ES providers' minimum willingness to accept (WTA) to make feasible the emergence of voluntary PES contracts. In practice, frequently this alleged sine qua non for PES is altered by both ES users' WTP and providers' WTA being also influenced by non-monetary values, agents' risk perceptions, expectations for volatile commodities prices, or existent land rights (Kosoy et al., 2007;Wunder, 2013). Although some studies using techniques derived from the economic theory aim to capture such parameters, real-world complexities often disconnect results from policy (Ferraro et al., 2011). Furthermore, upfront transaction costs needed for establishing CCI mechanisms may be unexpectedly high (Nantongo and Vatn, 2019;Rendón Thompson et al., 2013). In practice, ES users need to assume total or at least part of the costs to warrant ES provision, and ES providers need to be voluntarily enrolled in order to enable them to manifest their actual WTA.
Secondly, emergence of CCI also depends on the existence of adequate institutional structures. For instance, collective action is often required among both ES users and providers, so they need to be able to self-organize to act upon their aforementioned economic interests. Third actors (e.g. government agencies and local NGOs) typically play important mediating roles between ES recipients and providers, typically for fund management, monitoring and administrative functions. Frequently, these intermediaries also help assuming considerable CCI transaction costs, as it has been observed for many local REDD+ initiatives (Luttrell et al., 2017). Land tenure clarity is also an important institutional enabler, since ES providers must be able to exert exclusion rights into their enrolled areas vis-à-vis external actors (Wunder, 2013). Provision of more secure land tenure rights has sometimes been offered to ES providers as part of the reward packages (Montoya-Zumaeta et al., 2019;Suyanto, 2007).

CCI design and implementation features
Numerous contextual, design and implementation factors likely codetermine to what degree CCI produce successful outcomes (Börner et al., 2017;2020). In line with the recent conservation incentive literature (Engel, 2016;Ezzine-de-Blas et al., 2016;Wunder et al., 2018), we focus on the identified key design and implementation choiceswithout dismissing the co-determinant role of other contextual, less manageable factors (e.g. climatic change, leakage displacement, etc.) for the attained outcomes. Accordingly, the recent CCI literature highlights the four critical design elements: • Spatial targeting, which refers to prioritizing interventions on high-ES density areas (Wünscher et al., 2008) and/or those under major threat (Alix-Garcia et al., 2008), and doing so at different scales (national, project-or plot-level), when scarce funding makes enrollment of all willing landholders unrealistic. • Payment differentiation, which implies landholders are paid variable rates, e.g. according to their levels and costs of ES provision. Varying payments is key for CCI efficiency when ES providers constitute a heterogeneous group (Engel, 2016). • Conditionality, which as a key distinctive PES feature requires effective monitoring of compliance and sanctioning of non-compliance, the latter often being neglected (Sommerville et al., 2009;Tacconi, 2012;Wunder et al., 2018). • Customization of payment modalities, e.g. cash vs. in-kind, collective community-based vs. individual landowner contracts, timing (exante vs. ex-post) and duration of the contract (short vs. long) have all been claimed to influence environmental performance and socioeconomic outcomesincluding through feedback loops vis-à-vis perceived CCI equity and transparency (Engel, 2016;Pascual et al., 2014;Tacconi, 2012). Customization implies per definition that we cannot a priori say which design option is the most promising.

Outcomes and impacts
We also consider relevant to analyze the outcomes (and where possible, impacts) of CCI interventions, especially vis-à-vis conservation and wellbeing objectives. The key role that spatial targeting, price differentiation, and enforced conditionality have on environmental additionality (that is, the CCI bottom-line contribution to targeted ES provision, or at least forest-cover proxies) was empirically confirmed by meta-analyses of global PES datasets (Ezzine- de-Blas et al., 2016;Wunder et al., 2018). Furthermore, low perceived levels of equity and transparency in this type of initiatives can potentially erode participants' wellbeing (Hejnowicz et al., 2014;Maldonado et al., 2019).
The measurement of conservation impacts has arguably lagged behind that of policy interventions of other sectors such as education and health (Ferraro and Pattanayak, 2006), but has rapidly gained ground (Börner et al., 2017 Many rigorous PES impact evaluation studies using experimental or quasi-experimental methods show small but statistically significant levels of additional benefits for both conservation and wellbeing objectives (Börner et al., 2017;Duchelle et al., 2018;Snilsveit et al., 2019). Nevertheless, such studies have been much concentrated on few countries, principally Brazil, China, Costa Rica, and Mexico (Snilsveit et al., 2019).

Information sources
We rely on information from both primary and secondary sources. Thirty semi-structured interviews were conducted by the first author between February and March 2019 with CCI-related representatives at national and local levels ( Table 1). We complement data from interviews with in-field information previously gathered by the first two authors during study missions.
Primary information was collected according to the following sequence: first, information from the Ministry of Environment (MINAM) and the National Superintendence of Sanitation Services (SUNASS) helped us preselect the case studies that are presented below. Then, local informants from each selected initiative were initially contacted via email or telephone to explain the study context. Appointments were made with the understanding that interviews were voluntary, and information provided would be treated confidentiality. Face-to-face semi-structured interviews were conducted after formal prior informed consent was received following the Human Ethics Protocol 2018/435 of the Australian National University. Interview questions tackled aspects concerning CCI history, implementation, key actors, legal norms, rules, outcomes, and difficulties, among others. Simple descriptive statistics were estimated using MS-Excel to analyze interview responses. Also, we collected secondary information including journal articles, publicly available grey literature, and internal reports, including documents referred to us by our informants.

Selection of case studies
To select case studies for our analysis, firstly we elaborated an initial long list containing more than sixty CCI initiatives across the country in diverse implementation stages (Supplementary Data), based mainly on information provided by informants from MINAM and SUNASS. Currently, these two government organizations are the most prominent stakeholders promoting this type of initiatives nationwide. We complemented such information with available previous nationwide inventories of CCI, such as Entenmann (2012), Quintero and Pareja (2015), MINAM (2016), Tristán Febres (2018), and Simonet et al. (2018). From the long list, we picked initiatives meeting the following two requirements: (1) to-date evidence of continuous activities on the ground (at least three consecutive years of operations); and (2) sufficient availability of reliable secondary information about the local context in which the initiative emerged, as well as its design and implementation features. As result of this screening, the following six initiatives were Table 1 Interviews conducted for this study.

CCI overview, emergence preconditions and enablers
Firstly, we provide some general information about the six CCI initiatives analyzed here. Three of these are subnational REDD+ projects aiming to reduce forest threats arisen from multiple drivers, including construction and improvement of roads in Madre de Dios (Chávez et al., 2014;Naughton-Treves, 2004), San Martín (MINAM, 2010), and Ucayali (Ugarte-Guerra, 2009). Another two initiatives, QWF and MRHESM, use PES mechanisms to enhance the provision of hydrological services to farmers and potable water users by conserving targeted ecosystems in the northern departments of Piura (Andean páramos) and San Martín (cloudy rainforest), respectively (Quintero and Pareja, 2015). The NFCP, conceived as a upscaled PES program, aims to make contributions in reducing deforestation inside indigenous communities territories located across the Peruvian Amazon (Giudice et al., 2019). As of December 2017, 188 indigenous communities in nine departments (Cerro de Pasco, Loreto, Junín, Cusco, Amazonas, Ucayali, San Martín, Madre de Dios, and Huánuco) managing more than 1.8 millions of forest hectares were participating in the initiative (PNCBMCC, 2018).
Intervention areas of selected REDD+ projects in Madre de Dios, San Martin, and Ucayali cover 308,000, 182,000 and 128,000 ha, respectively. QWF and MRHESM initially identified ecosystems with high potential to provide watershed regulation services, and sought to enroll only prioritized upstream zones; thus their intervention areas are both smaller by design (see Section 4.2.3).
The alleged economic preconditions for CCI emergence require ES recipients to be willing to pay (WTP) at least as much as ES providers are willing to accept (WTA) in compensation (see Section 2). Some empirical evidence shows moderate but still significant WTP from especially watershed ES users across the country to conserve the ecosystems providing these (Carbajal and Lucich, 2014;Lucich and Gonzales, 2015;Montoya-Zumaeta and Nolazco Cama, 2015). ES recipients in our six Peruvian cases include local water users, international carbon markets buyers, national taxpayers, private funders, and foreign governments (such as Germany and Norway) who catalyze development funding to the implementation of CCI. On the supply side, Börner et al. (2016) found in a study supporting PNCB implementation at the scale of the Peruvian Amazon that opportunity costs, an important proxy for ES providers' WTA, are quite heterogeneous, and influenced by multiple factors. Among our study cases we found that factors strongly reducing WTA are legal restrictions (MRHESM, AMCI, and BN-REDD), internal collective restrictions (NFCP, and REDD-U), and distance to markets (QWF). However, our interviews seemed to confirm that ES providers were enrolled voluntarily in all the six analyzed cases, either through individual agreements (as in the case of AMCI, MRHESM, and BN-RDD), or by obtaining collective acceptance from communities according their internal rules (in the case of NFCP, REDD-U, and QWF).
Meanwhile, implementers in the analyzed cases included a national private non-profit organization with expertize in carbon finance (REDD-U), an international non-government organization (AMCI), multi-actor platforms (QWF and MRHESM), a regional producer organization (BN-REDD), and a decentralized national government office within the  With regard to land tenure security as an institutional precondition, in four out of our six study cases ES providers have legal recognition of their rights, as the State has granted collective land rights to peasant and indigenous communities participating in QWF, NFCP, and REDD-U; as well as individual rights awarded to Brazil nut concession managers participating in BN-REDD. This recognition allows them to obtain economic benefits from forest ecosystem services as established by the Law 29763 "Forestry and Wildlife Law" enacted in 2011. Given that MRHESM and AMCI intervention sites overlap with state-owned conservation areas and their buffer zones, migrant farmers participating as ES providers in these are only recognized as de facto land users, with considerably limited usufruct land rights. Therefore, allegedly only land uses compatible with conservation objectives are allowed by the State in areas where both initiatives are being implemented (Kowler et al., 2016). The clarification of tenure aspects in the MRESHM which included the establishment of a conservation area substantially increased its transaction costs (Montoya-Zumaeta et al., 2019). Although some encroachments inside intervention areas as well as overlaps with conflicting land uses (logging, mining, oil and gas extraction) were reported during interviews, these are considered relatively infrequent, and are typically solved quickly with help from CCI implementers and facilitators. The aspects described in this section are presented in detail in Table 2.

Rewards vs. payments
In-kind rewards have been preferred to cash in compensating ES providers within the selected CCI cases, although the two modalities were in principle combined in both REDD-U and BN-REDD. The mostly discontinuously provided rewards have included financial and technical assistance for sustainable livelihoods (e.g. beekeeping, eco-tourism, sustainable forest management, cocoa and coffee agroforestry), sanitation infrastructure, seeds and other reforestation inputs. Yet, some rewards were still delivered annually according to scheduled contracts (MRHESM and AMCI), or through collective investments plans (QWF and NFCP). Interestingly, the preference for in-kind rewards lies strongly on the implementer side, while our small sample of consulted ES providers would prefer either cash or an in-kind with cash combination ( Fig. 2A). BN-REDD and REDD-U contracts did include fixed profit shares from carbon credits sales that would be distributed to enrolled participants. Nevertheless, according to our informants in none of these projects cash payments have taken place so far, due to difficulties to obtain funds from voluntary carbon markets. In all the analyzed initiatives, implementers' preference for in-kind rewards respond also to uncertainties to establish a continuous flow of funding, and to a lesser extent their doubts about the capacity of cash payments to trigger sustainable local welfare gains (Fig. 2B). The existence of legal restrictions on the use of public funds for direct cash payments has influenced the in-kind choice in some initiatives, including the two mostly funded with public investments, NFCP and MRHESM (Quintero and Pareja, 2015).

Enforced conditionality
Contracts in four out of the six examined CCI cases (FNCP, MRHESM, AMCI and QWF) explicitly stated reward delivery to be conditional upon environmental compliance. Yet, another question is to what extent conditionality by design was also enforced in practice. We decomposed enforced conditionality into two elements: periodic compliance monitoring, and sanctioning incompliance. Remote sensing technologies are used to monitor land-cover changes in NFCP and the three REDD+ projects, given the relative simplicity to infer carbon storage dynamics from spatial analyses (Ferraro et al., 2015). For both watershed initiatives (QWF and MRHESM), land-cover changes were monitored, but focus here has mainly been on detecting farm-level land clearances, discharge of contaminated water, etc. More recently (and infrequently), direct measurements of water quality parameters have also been adopted by both initiatives.
In relation to sanctioning, Montoya-Zumaeta et al. (2019) report for MRHESM that at least one noncompliant participant was not only not rewarded, but also sentenced to a prison term due to clearance of forest enrolled into the initiative for coffee expansion, and existence of legal restrictions to his plot that went beyond the CCI conditionalities. Conversely, Giudice et al. (2019) report occasional non-sanctioned infractions against contracted environmental obligations, committed by communities enrolled into the NFCP pilot phase, although a graduated sanctioning system (which includes successive written warnings, partial reductions in rewards, and retirement from the initiative as a last resort) has been implemented more recently, according to our informants. A similar graduated sanctioning system is being applied in QWF and  In-kind rewards: technical assistance, reforestation, sanitation infrastructure, shaded coffee systems, support for benign activities: beekeeping, sustainable tourism, handcrafts.
In-kind rewards: technical assistance, reforestation, support for sustainable activities (avocado and coffee production).
In-kind rewards: technical support, forest enrichment, training. Being considered: community cash payment of half of surplus above minimum carbon price (~US$1.5 per tCO 2 ) from carbon offsets in voluntary markets.
In-kind rewards: technical assistance for coffee, postharvesting equipment, sanitation (ecological kitchen and toilets), small livestock (e.g. guinea pigs), and incomes from sustainable activities: birdwatching, communal tourism, and cultivation products (pitahaya, cupuassu). Hydrological priority areas were pre-identified using Soil and Water Assessment Tool (SWAT). But then action included non-priority areas (e.g. Almendra watershed).
Hydrological priority areas were pre-identified using Soil and Water Assessment Tool (SWAT), but some identified areas are common-pool, and one nearby community could not be enrolled. Rewards are negotiated with each household; agreements compiled in annual conservation contracts.
Participation and (potential) benefits were discussed with each concessionaire. Carbon sales and disbursement are still pending.
(continued on next page) AMCI, although our informants also report that occasional non-sanctioned infractions have occurred in these.

Spatial targeting
Spatial targeting efforts usually focus on some combination of high ES levels, high threat/leverage, and low costs (Engel, 2016;Wunder et al., 2018). Most of our CCI cases are located in departments that over the last two decades have ranked high in term of their deforested areas, such as Madre de Dios, San Martin and Ucayali (MINAM, 2017). In Madre de Dios, the BN-REDD deliberately targeted threatened forests containing large volumes of carbon, located alongside the Southern Trans-Oceanic Highway (Naughton-Treves, 2004), yet its intervention area also entails no-road access forests. Likewise, in San Martin both AMCI and MRHESM intervention areas overlaps with vulnerable forests located near to the Fernando Belaunde Road, yet some areas with substantially less access were also preventively enrolled. Furthermore in MRHESM, the Soil and Water Assessment Tool have enabled implementers to identify which parts of the targeted watershed are most sensitive towards land-use changes and degradation (Quintero et al., 2009), nonetheless the Almendra micro-watershed that was not identified as a major potential contributor of sedimentsthe main ES focusbut being part of the local conservation area, it was included into the CCI initiative. In Ucayali, in the REDD-U most of the areas enrolled into the initiative are located in relatively low-threatened remote indigenous territories with quite limited road access.
Given its national scope, the NFCP have managed to enroll communities from departments containing more recent deforestation hotspots such as Huánuco and Loreto in addition to those mentioned above (MINAM, 2017), however its strategy was only imperfectly threat-targeted: several forest areas with very low deforestation risk were consciously enrolled in the program for the 2011-15 period, due to participant communities' own preferences and implementers' acceptance (Giudice et al., 2019). On the other hand, the intervention area in Piura where the QWF operates faces other type of threats, such as overgrazing and mining pollution (Albán Contreras, 2017). Like in MRHESM, remote sensing-based tools were also applied in QWF to identify areas with higher potential for provision of hydrological services during the CCI design phase (TWINLATIN, 2008). Nevertheless, a strategic upstream community of highest hydrological ES priority could not be enrolled into the initiative. In sum, some targeting in area enrollment is certainly applied across the six initiatives, but variably and imperfectly so as either the uneven spatial risk distribution is only lightly considered, or low-threatened zones are included even consciously.

Reward differentiation
A differentiation of payments/rewards is in conservation incentive schemes recommended so as to give priority to the enrollment of high-ES/high-additionality land areas, even when some of those are associated with higher landowner provision (incl. opportunity) costs (Engel, 2016;Wunder et al., 2018). Some implementers state to have customized in-kind rewards considering participant preferences (e.g. in investment plans of NFCP and QWF), and/or their socio-economic status (AMCI, MRHESM), seeking to achieve social improvements. However, customizations have been done using simple criteria such as the size of enrolled area such as in NFCP (Börner et al., 2016), or participants' preferred sustainable economic activities (in QWF), not the variable cost or quality of ES provision, as CCI theory would alleged (see above). Hence, within the six analyzed CCI efforts to differentiate prospective payments or rewards were not motivated by environmental additionality or cost efficiency. Based on informants' testimonies, we infer that the limited effort to differentiate payments for efficiency responds to two main reasons. Firstly, implementers may prioritize equitable rewards using fairness criteria (e.g. egalitarian, pro-poor) that outweigh environmental targets (Börner et al., 2016;Pascual et al., 2010). Secondly, implementers' underlying assumption is that poverty is a primary driver of ecosystem degradation. Hence, by addressing participants' needs, allegedly they will support forest conservation.

Equity
In this subsection we analyze two elements closely related to perceived equity within CCI (McDermott et al., 2013): (1) participation of providers in decision-making processes, and (2) distribution of benefits. As to the former, top-down decision-making structures are predominant in our selected cases, sometimes generating strong discomfort to ES providers. For instance, non-consulted forthcoming changes in MRHESM implementation resulted in strongly negative household-stated perceived wellbeing (Montoya-Zumaeta et al., 2019).
Secondly, although in some cases (QWF, NFCP, MRHESM, and AMCI) benefits had been negotiated before participants' enrollment (or their communal representatives, for collective agreements), it is not clear to what extent they actually came to benefit. For instance, for MRHESM we identified spillover of some benefits toward non-enrolled participants like training and agricultural inputs that were consciously delivered beyond the group of conservation agreements signatories, as a political strategy to obtain greater acceptance in the intervention area, allegedly motivating broader participation. Likewise, in the BN-REDD project some CCI-contracted ES providers claim they received no benefits (such as technical support and training), while non-enrolled Brazil nut concessionaires participated instead.

Transparency
We can define transparency for our purposes as the timely and reliable provision of information to all relevant stakeholders (Kolstad and Wiig, 2009) involved with CCI initiatives. Transparency contributes to the sustainability of these initiatives; the inclusion of transparent strategies can avoid perceptions of corruption and make stakeholders' responsibilities publicly more accountable (Tacconi, 2012).
In this respect, all CCI initiatives have established communication channels between implementers and ES providers (Table 3). In NFCP and QWF, both consulted implementers and enrolled participants concur that fluent communication exists within such initiatives. However, in MRHESM and all the three REDD+ initiatives seemingly information about fund management and benefit distribution do not trickle easily through to the multiple stakeholders, including the proper ES providers. Here, interviewees' responses differ: while implementers consider that information is fluid, other consulted informants perceive that communication with implementers and facilitators is claimed to be infrequent and unclear, therefore causing ES providers' dissatisfaction towards the formers' role. Particularly for BN-REDD, this perception has seemingly raised some suspicions about mismanagement of carbon sales revenues.

Environmental additionality
Environmental additionality is the incremental contribution that an initiative achieves in terms of ES provision, compared with a counterfactual business-as-usual scenario (Tacconi, 2012). The assessment of counterfactualsmeaning "what would have happened without the intervention?" -is central to any quantitative environmental impact evaluation (Ferraro, 2009;Ferraro and Hanauer, 2014). Our primary sources here are quasi-experimental impact evaluations, such as was recently performed for NFCP and MRHESM. Giudice et al. (2019) reported small positive conservation effects from NFCP, surprisingly on non-contracted rather than contracted community forests, inside enrolled indigenous communities' territories (557 ha of forests conserved during 2011-15). The effects on contracted forest were negligible. Meanwhile, Montoya-Zumaeta et al. (2019) found that MRHESM conserved between 7.91 and 22.25 ha of primary forest into the ZOCRE Rumialba for the 2010-16 period, with at least one third of it being attributable solely to the CCI component. However, beyond of impacts on the land-use proxy of forest conserved, its actual contribution to the provision of targeted hydrological services per se has still not been rigorously assessed. A recent impact evaluation performed for BN-REDD did not find statistically significant effects of the CCI initiative on avoided deforestation and forest degradation for the 2012-2018 period (Montoya-Zumaeta, 2021).
For the remaining initiatives, only less rigorous information about their impacts is available, such as analysis funded by the same implementers. In relation to AMCI, an implementer-commissioned study found that the initiative had reduced deforestation by 24%, thus avoiding the emission of 5.6 millions of CO 2e tonnes during the 2012-16 period (CI, 2017). Similarly, a report prepared by the implementer of the REDD-U initiative claims that it contributed to the avoided deforestation of 4856 ha, equivalent to a reduction of 2.8 million tonnes of CO 2e in the 2013-2018 period (AIDER, 2019) -about one third of what the intervention originally planned to mitigate. Nevertheless, these estimates tend to overestimate actual CCI contributions since the effects of other policies and confounding factors (e.g. lower agriculture returns) are not properly controlled for (West et al., 2020). In QWF, 80 ha were reforested with native species between 2014 and 2016 (Albán Contreras, 2017). Although relatively small, this contribution would arguably not have happened without the initiative, as no other reforestation program or related policy is taking place in the same site.

Socioeconomic and wellbeing impacts
All our Peruvian CCI case studies featured participants' adoption of environmentally friendly economic activities. These included agroforestry coffee and cocoa production (NFCP, MRHESM, AMCI, REDD-U), community tourism (NFCP, MRHESM), reforestation with native species (QWF), and sustainable forest management (REDD-U, BN-REDD). Implementers made investments to promote adoption of such activities (see Section 5). Rigorous counterfactual based evidence about hoped-for positive effects on participants' incomes and livelihoods in our selected CCI cases is still inconclusive: indications exist for only three out of the six (Table 4). Montoya-Zumaeta et al. (2019) find significant positive effects of MRHESM in-kind rewards on households' incomes and assets, though recent self-reported wellbeing was strongly undermined by participants' unmet expectations. Also using a quasi-experimental approach, Solis-Chavez (2017) did not find statistically significant evidence that participation in REDD-U and BN-REDD had increased participating households' environmental incomes or assets. No welfare-oriented impact assessment exists for the NFCP, but Börner et al. (2016) using simulations find that the current design leads to an inequitable concentration of benefits in relatively few, land-abundant communities. Giudice et al. (2019) suggest ample adoption of alternative economic activities (cocoa and coffee production, ecotourism, etc.), though we do not know to what extent these have improved incomes. Likewise, similar interventions were introduced in QWF (Albán Contreras, 2017) and AMCI (CI, 2017) to mitigate land-cover threats in intervention areas; however, robust estimates for socioeconomic outcomes have not been reported so far.

Discussion and conclusion
As elsewhere in Latin America, in Peru the introduction of CCI initiatives to promote the sustainable provision of ES has attracted much attention by conservation practitioners. Yet, most of the 60+ Peruvian initiatives adopting CCI we list in Appendix 1 are still in very early implementation stages. According to our informants from MINAM and SUNASS, their limited progress is explained by some level of distrust from local actors, e.g. municipalities and drinking water companies: they still view performance-based payments with a certain skepticism. In that sense, the role of third parties (mainly NGOs) has been very important on contributing firstly to consolidate economic pre-conditions in some of these initiatives by, for example, gathering and disseminating information to ES users about the importance of targeted ecosystems (such as in both watershed-based CCIs) with the hope that these environmental consciousness campaigns could stimulate their WTP. Similarly, in some cases intermediary NGOs have also put some downward pressure on providers' WTA in MRHESM and AMCI, by enhancing enforcement of pre-existent legal land-use restrictions: a more credible local regulatory threat has reduced the business-as-usual profit potentials of land users, and thus curbed their conservation opportunity costs, making them ceteris paribus more willing to accept payment offers. Therefore, involvement of NGOs either as implementers or facilitators during early adoption of CCI trials seems to be key for their progress, including for shaping realistic expectations among the local actors. In all our case studies, NGOs typically backed by external donor funding assumed considerable CCI startup costs, and remained actively involved in QWF, AMCI and REDD-U, providing funds and legal-technical assistance in matters such as land-tenure issues (cf. Section 4).
Legal constraints and funding shortages seem to have a determinant role in some of implementers' design choices, such as vis-à-vis the type of rewards. For instance, legal restrictions determined that in-kind rewards (e.g. public investments) were provided, rather than continuous cash payments (NFCP and MRHESM). QFW, mostly funded privately, made temporary investments in communities aiming to generate lastingly positive environmental outcomes. Given the slow development of voluntary carbon markets (Ecosystem-Marketplace, 2019) with considerable declines in traded carbon volumes (dropping from 131.4 millions of CO 2e tonnes to 98.4 millions for the 2010-2018 period) and prices (peaking at US$6.20 per CO 2e tonne in 2011, then dropping to on average US$3.01 in 2018, though substantially lower prices were negotiated by some projects, as in the case of BN-REDD); alleged cheaper in-kind rewards in the form of technical assistance, reforestation and agroforestry inputs, promotion of sustainable economic activities have been also used as incentives in the analyzed REDD+ initiatives.
Hence, strong reliance on limited and discontinuous external funding mechanisms for CCI implementation (e.g. temporary public investment projects and/or donor funding) rather than actual ES recipients' payments jointly with implementers' doubts about cash payments have seemed to be crucial for design preferences in the analyzed initiatives.
Where the often small offered per-hectare payments thus enough to cover local WTA? This depends on several contextual factors. If the enrolled areas were de facto little threatened, then opportunity costs would also be close to zero; hence even low payments are attractive. Furthermore, when large forest areas were enrolled simultaneously (e.g. for some indigenous lands in the PNCB), then even small per-hectare rates could add up to significant amounts, but creating both unequal and environmentally inefficient outcomes as "communities with large forest reserves and low population densities will receive the largest transfers" (Börner et al., 2016, p. 413). The same study makes the case that spatially well-targeted and differentiated payments mimicking local opportunity costs could 'buy out' much current deforestation in the Peruvian Amazon at moderate costs. This also constitutes an overall fairly optimistic outlook for CCI implementation, if only design and implementation could be improved.
Although in-kind rewards have been said to be preferable in situations with elite capture and local market constraints (Asquith et al., 2008), aspects regarding their divisibility, repeatability, and delivery costs can drag into the opposite direction (Engel, 2016). Moreover, in all our cases implementers noted in interviews that they preferred temporary over permanent rewards, due to difficulties to ensure a continued flow of funding (Fig. 2). This scenario would make investments in sustainable livelihoods alternatives a logical choice, if economic activities were clearly profitable and resolved 'permanently' the environmental externality problem (Engel, 2016). Nevertheless, further effort is required to generate rigorous evidence to support or challenge such strategies.
Furthermore, we think that in the analyzed cases environmental additionality could be boosted by incorporating spatial targeting and payment differentiation more decidedly in subsequent stages of analyzed CCI implementation, as our findings reveal that little application of these have taken place so far. These design features seem to be under-utilized due to administrative ease, coupled with challenges to implement differentiated payments considering implementers' equity perspectives. Adopting smaller but continuous payments (either in cash or in-kind), preferably coming from actual ES recipients rather than from temporary investment projects with multiples objectives, could create more realistic expectations. This should be possible especially in relation to watershed payments, which in opinion of our MINAM and SUNASS informants currently represent the best economic opportunity in Peru to apply CCI to natural resource management: if there is a 'business case' for adapting upstream water management, the challenge would be to get water users to accept the payment of a premium to ES providers.
In addition, periodic compliance monitoring and sanctioning systems should also be adopted or refined. These two elements are critical for CCIs implementation, as their absence could imply a more systemic problem of moral hazard: that enrolled participants are rewarded yet defecting on their environmental commitments, thus undermining the credibility of the CCI mechanism at its foundation . As demands to use increasingly scarce conservation funds in more efficient ways have grown over the past decade (Ferraro and Pattanayak, 2006), this aspect requires more attention. Eventually political pressures may set the bar higher in terms of CCI initiatives' effectiveness, as findings from impact evaluations are revealing many small or even negligible impacts as consequence of recurrent implementers' design choices, guided by practicalities rather than goal-oriented streamlining .
Making decision processes more transparent is also an essential aspect that should be incorporated by implementers, donors, and facilitators, including so as to strengthen perceptions of procedural equity. As explained above, this matter was raised particularly for BN-REDD and MRHESM, where widespread distrust among actors relate respectively to low financial transparency and the introduction of new land-right arrangements. In line with Tacconi (2012), we argue that transparency is a highly relevant aspect to achieve and maintain sustainable CCI impacts, particularly those related with improved socioeconomic conditions. Specific strategies can include launching of participative spaces to inform stakeholders on management aspects of common interest (e.g. through periodic village workshops), and broadening communication beyond websites (e.g. periodic informative brochures, radio or/and television). These strategies have been used successfully in agricultural development programs (Basics Oliveira, 1993), and thus might be adapted for contexts where CCI are being implemented.
To conclude, the design of adequate incentives in conservation initiatives is complex, and some choices apparently prioritizing practicality and simplicity can eventually bring about important tradeoffs in attaining conservation and social objectives, thus potentially undermining environmental additionality and wellbeing impacts alike, as our analysis of Peruvian CCI cases has revealed. To some extent, these initiatives can also become broader politically motivated instruments, pushing environmental effectiveness and efficiency concerns somewhat into the background (Rosa da Conceição et al., 2015Conceição et al., , 2018. In the above, we have flagged priority issues that, from our perspective, deserve consideration by stakeholders involved with the implementation and promotion of CCI initiatives. These could guide the implementation of subsequent phases of the selected cases, or during design stages of nascent similar initiatives in Peru and beyond.

CRediT authorship contribution statement
Javier Gustavo Montoya-Zumaeta and Sven Wunder have conceived the study. Javier Gustavo Montoya-Zumaeta acted as lead writer, Sven Wunder and Luca Tacconi supplemented.