This dataset is part of the Ecosystem Services and Poverty Alleviation (ESPA) programme. These data represent the central quantitative datasets from a mixed method, quasi-experimental study of the effects of an environmental intervention (national implementation of Wildlife Management Areas - WMAs) on Tanzania's rural population. The study focused on changes in wealth, livelihoods and wellbeing 2007-2014 for households in villages that are members of a WMA, compared to households in matched control villages that are not in WMAs, and causal attribution of those changes. The study covered 6 of the then 18 WMAs, (3 in northern rangelands and 3 in southern Miombo woodlands), and surveyed a total of 47 villages (both 'inside' WMAs and matched 'outside' non-WMA villages). In all, 13,578 households in these villages were wealth ranked on locally derived, village-specific criteria for both 2014/5 and (by recall) for 2007. From this sample frame we surveyed a stratified random sample of 1924 household heads (including 187 female heads of household) and 945 wives of household heads. Questions to household heads addressed household composition, land and livestock assets, resource use, income generating activities and income portfolios, participation in decision-making in natural resources management, and perceived benefits and costs of conservation, at the time of the 2014/5 survey and also by recall for 2007. Related questions addressed women’s perceptions of changes since 2007 in access to land and natural resources, production, income-generating activities, human-wildlife conflict, participation in WMA management; and overall costs and benefits of WMAs. Though there is also considerable qualitative data, much of this is politically sensitive and therefore not deposited here: interested researchers may contact the PI for partial access. Environmental data not already in the public domain are being deposited in the NERC Environment Information Data Service.Rural people across the global south are caught between competing land demands for large-scale cultivation, global conservation, and local needs. These can in theory be integrated locally through community-based natural resource management (CBNRM) and payments for ecosystem services (PES): where communities can decide on and benefit directly from natural resources, they may invest in and manage those resources in ways that are more socially and environmentally sustainable. CBNRM/PES initiatives are being rolled out across the global south, but there are conflicting views as to how well they work, for whom and under what circumstances. This is partly due to the complexity and multidimensionality of the ecosystem services (ES) and poverty alleviation (PA) outcomes involved, and the inevitable tradeoffs, but also to the hitherto limited use of either qualitatively or quantitatively rigorous impact evaluation approaches that are independent, control for confounding factors and ensure the voices of the most marginalized are heard. As well as being limited by generally weak research design, studies to date have often failed to account for the ways political sensitivities around changing access to and use of ecosystem services may compromise data quality and mask differentiated impacts. PIMA seizes a unique policy moment, with Tanzania's poverty reduction strategy Mkukuta driving nationwide implementation of CBNRM/PES-based Wildlife Management Areas (WMAs), and other countries in the region considering comparable initiatives. The WMAs comprise different ecosystems (rangeland, miombo), socio-political structures (long-established/ethnically uniform vs recent, heterogeneous constituent villages), and a broad range of ecosystem services (water-regulating and -supplying, provision of forest products, grazing, livestock, crop and wildlife production, cultural services both local and global (from locally significant social and ritual spaces, to heritage and tourism). Quasi-experimental comparison of social and ecological outcomes for established WMAs with statistically matched non-WMA areas (within the same ecosystems) offers an ideal opportunity for rigorous impact evaluation. PIMA combines analysis of remotely-sensed, public-domain MODIS and NDVI data, with cutting edge study of governance, and new data from qualitatively and quantitatively rigorous, differentiated survey of livelihoods and resource use histories, structured within a quasi-experimental research design. PIMA brings together a powerful international research team to work with strongly-rooted civil society organizations to ensure research excellence and development impact. Building on ongoing stakeholder engagement, with input sought from users, practitioners and policymakers at all stages pre- to post-project, PIMA ensures findings will be of direct use locally, nationally and internationally. PIMA 's framework and approach create channels for grassroots users to make experienced change in ecosystem services quality and quantity, and in poverty and wellbeing, more clearly heard by policymakers and practitioners, as well as highlighting tradeoffs and best practice lessons. Establishing what works, why and for whom will be of use not only to the one million rural people directly affected by WMAs, but will deliver insights and best practice lessons generalizable to the many millions more whose livelihoods and wellbeing are to be shaped by comparable CBNRM/PES initiatives. The findings delivered, and the mechanisms piloted, will give local users and national and international policymakers and practitioners the insights and tools to improve interventions through creating better upward and downward accountability. PIMA findings will be of use locally to rural people making collective and individual resource use decisions, through national levels, to international donors deciding how to invest scarce resources for ecosystem services and poverty alleviation.

We carried out a quasi-experimental impact evaluation of the social impacts of WMAs, adopting a Before/After, Control/Impact (BACI) design, collecting data from 24 villages participating in 6 different WMAs across two geographical regions (northern Tanzania semi-arid rangelands; southern Tanzania miombo woodlands), and 18 statistically matched control villages. Across these 42 villages, we collected participatory wealth ranking data for 13,578 households. Using this as sampling frame, we selected within each village 10 ‘very poor’ households (according to wealth ranking as of 2007); 10 households with household members in ‘leadership’ positions (as of 2014/15); and 20 households from the rest of the village register (the ‘other’ stratum). Alongside the main household survey, we also randomly sampled 20 wives of male heads of sampled households, with the same relative proportions within the three strata as for households in the main survey. This gave quantitative surveys with a stratified random sample of 1,924 household heads and 945 household heads’ wives. All data were collected in 2014/15, with questions on recall of conditions in 2007 (based on participatory, site-specific event calendars) providing a baseline. Questions addressed household demographics, land and livestock assets, resource use, income-generating activities and income portfolios, participation in natural resource management decision-making, perceived benefits and costs of conservation. Datasets permit research on livelihood and wealth trajectories, and social impacts, costs and benefits of conservation interventions in the context of community-based natural resource management. Detailed qualitative work (not deposited here, but analysed and written up in some depth and published) provides explanatory context. A plain language statement was gone through with each potential respondent, explaining the research collaboration and its aims and methods, and the interviewees’ right to decline prior to or drop out at any time during the survey. Respondents were then asked for their informed consent to participate in the survey. Participation was voluntary and no cash payment was offered. Small inexpensive but locally appreciated courtesy gifts such as tea, sugar or phone credit vouchers were offered as a thank you after the survey.