The recent call to halt biodiversity loss by protecting half the planet has been hotly contested because of the extent to which people might be excluded from these landscapes. It is clear that incorporating landscapes that implicitly work for indigenous people is vital to achieving any sustainable targets.We examine an attempt to balance the trade-offs between conservation and development in Enonkishu Conservancy in the Maasai Mara, using a working landscape approach. Mobile livestock production strategies are theoretically consistent with wildlife-based activities and can present a win-win solution for both conservation and development. We explore the success and failings of Enonkishu’s evolving attempts to achieve this: addressing the criticism of the conservation sector that it fails to learn from its mistakes. We found that Enonkishu has had considerable positive conservation outcomes, preventing the continued encroachment of farmland, maintaining and improving rangeland health relative to the surrounding area, while maintaining diverse and large populations of wildlife and livestock. The learning from certain ventures, that failed, particularly on livestock, has created institutions and governance that, while still evolving, are more robust and relevant for conservancy members, by being fluid and inclusive. Diverse revenue streams (beyond tourism, including a residential estate, livestock venture and philanthropy) enabled Enonkishu to withstand the pressures of Covid-19. Livestock is crucial for defining the vision of the conservancy, and the institutions and governance that underpin it.

Rangeland health: Fourteen 25 metre transects across Enonkishu were sampled four times per year since 2018 using five equally spaced (5-metre) quadrats. Four transects are within control areas that are not included in Enonkishu’s grazing plan and ten transects are located within areas included in the grazing plan. Eighteen parameters are examined in each quadrat to describe cover, soil surface description (capping), litter (the amount of dead plant debris covering the soil surface), and plant species (grass, tree, shrub, forbs or sedge). The corresponding ratings are re-calibrated such that a rating of “5” indicates the best possible score, with “0” indicating the worst possible score. As an example, the parameter of Plant Density rates 5 if there is 100% plant cover, with a score of 0 indicating no plant cover .

Wildlife Monitoring:  Data to estimate numbers of large mammals in Enonkishu conservancy were collected twice per month since June 2016 via two 2km strip transects. All mammals located within 100m (as verified by rangefinder) from the transect are recorded. Population estimates are then calculated for the conservancy using Jolly II methodologies (Jolly, 1969). In Naretoi wildlife data was collected twice per month since 2019. The small size of Naretoi and the constant communication between the teams counting ensures that these counts are precise.

Livestock: Livestock in the conservancy were recorded at the end of each month and births, deaths, and treatments among the conservancy herd are recorded in real time.  Each quarter, the conservancy herd is valued by those familiar with current market cattle sale prices.

Socio economic survey:  The socio-economic survey data were from a larger stratified random sample of conservancies in the Northern Mara. The total sample frame of 414 landowners (defined as the head of the household registered as a lease payee) was based on landowners in each conservancy. From this, a random sample of 140 households was chosen, stratified by conservancy. In Enonkishu, 41% (n = 11) of landowners were sampled. Each survey began with an explanation of the purpose of the survey, how data would be used, confidentiality measures, the participant’s rights, and sought their consent before proceeding. The survey was conducted using the ODKcollect application in English, Swahili or Maa (all translated and independently back translated to ensure precision) depending on the preference of the respondent. Responses from surveys were reviewed daily once surveys were uploaded. A random selection of 7% of respondents were called to validate and confirm the data collection process.

Remote sensing of changes of land under cultivation: We downloaded Google earth timelapse (https://earthengine.google.com/timelapse/) imagery for each year from 1984, at a 30m resolution and obtained from the Landsat series of satellites. All cultivation areas were digitised in QGIS.  In R using sf the total cultivation area and proportion of cultivation area in each conservancy was calculated over time.

Three irrigation pivots on 1000 acres of Olerai Farm were transformed into a safari property called ‘Naretoi’.