Theses Doctoral

Navigating Challenges and Opportunities in India’s Afforestation Initiatives: Balancing forest restoration, biodiversity conservation, and livelihood benefits through effective evaluation of private and public afforestation programs

Vanamamalai, Anubhav

Global restoration and conservation agendas increasingly depend on converting policy commitments and financial investment into lasting ecological recovery. Yet evidence of implementation is often mistaken for evidence of success. Funds spent, land enrolled, saplings planted, and communities engaged show that action has occurred, but not whether ecosystems recover, planted trees survive, or people and wildlife coexist more safely. This dissertation examines this gap in India across three linked scales, namely, a national compensatory afforestation policy, a regional voluntary agroforestry program on private farmland, and the local social-ecological pressures facing protected-area-adjacent communities expected to sustain these efforts. It combines satellite remote sensing of vegetation change, household-linked sapling survival modelling, and occupancy modelling of human-elephant conflict to evaluate whether conservation actions produce measurable outcomes once ecological context, household decision-making, and imperfect observations are accounted for.

At the national scale, the dissertation assesses India’s US $6.7 billion compensatory afforestation system, administered through CAMPA, as a public mechanism for offsetting forest diverted to non-forest use. Using three decades of Landsat-derived Enhanced Vegetation Index time series (1995-2024) across 25,000 compensatory afforestation sites, it examines vegetation trajectories before, during, and after program implementation. The analysis combines structural breakpoint detection with parametric and non-parametric trend approaches to identify abrupt vegetation changes, long-term greening or decline, and post-intervention recovery. The results show a marked shift from widespread decline in the pre-CAMPA period, when only 16.1% of pixels in compensatory afforestation sites showed an increase in vegetation between 1995 and 2009, to substantial greening between 2010 and 2024, when 83.2% of pixels exhibited increases. A change in vegetation was typically detected about 2.5 years after project initiation. However, gains were frequently not sustained beyond formal project completion, varied sharply across states, biomes, and ecoregions, and in naturally open ecosystems often indicated woody-cover gain rather than ecologically appropriate restoration.

The dissertation then moves to a voluntary smallholder agroforestry program on private farmlands around Bandipur and Nagarahole National Parks in Karnataka. Pairing 523 household surveys with field-verified sapling survival counts, the analysis applies a zero-inflated beta binomial model to assess how farmland resources, management practices, and farmer perceptions on agroforestry shape household-level and species-level survival. The study found that over 60% of households experienced complete sapling failure within the first year, with overall survival at 4.45% across the program. Baseline irrigation access was the primary factor preventing total household-level sapling failure, while farmer perceptions explained the largest share of conditional survival, exceeding the contributions of material resources and management practices. Households that were uncertain about the intended use of agroforestry products showed remarkably lower conditional survival rates (2.8%) than those intending to sell (16.5%). Within farms, the most-preferred species survived at nearly twice the rate of the least-preferred species, suggesting that farmer priorities shaped post-planting care.

Finally, the dissertation examines local constraints that shape whether agroforestry adoption is viable in the same landscape, where communities manage agricultural livelihoods amid frequent human-elephant interactions. Analyzing over 20,000 verified farmer-reported incidents of conflict between 2015 and 2023, the study employs a Bayesian spatiotemporal occupancy model to separate and map occurrence and detection of human-elephant conflict incidents. Cropland extent and seasonal nighttime light radiance, an indicator of human activity and presence, most strongly predicted conflict occurrence, with mean occupancy around 16%. Occurrence peaked during the post-monsoon season, while mean detection probabilities of around 31.7% indicated substantial underreporting of incidents even under systematic monitoring.

Together, the chapters show that restoration and conservation-linked land use change cannot be evaluated at the point of policy authorization, fund disbursement, sapling distribution, or incident reporting. The dissertation’s central conclusion is that conservation investments produce sustained ecological results only when programs are designed and implemented around local ecological and social realities, evaluated through verified outcomes rather than administrative proxies, and monitored long enough to distinguish implementation from recovery.

Geographic Areas

Files

  • thumbnail for gsas-dissertations-000900.pdf gsas-dissertations-000900.pdf application/pdf 4.45 MB Download File

More About This Work

Academic Units
Ecology, Evolution, and Environmental Biology
Thesis Advisors
DeFries, Ruth S.
Degree
Ph.D., Columbia University
Published Here
October 28, 2026

Notes

Wildlife Conservation, Afforestation, Agroforestry, Farm Management, Ecology