Tropical forests contain more than half of the world’s biological diversity and play key roles in sustaining human life. In many parts of tropical Latin America, forests continue to be degraded and destroyed despite many efforts to protect them. Determining the extent to which conservation efforts work under different contexts can potentially lead to future improvements of forest cover outcomes at lower financial and social costs. However, rigorous impact evaluations of conservation interventions are still rare. This dissertation reports on estimated impacts of conservation and development interventions on tropical forest cover in Guatemala and Colombia and seeks to explain the quantitative results through qualitative data from surveys and interviews. The primary aim is to provide critical information to improve the efficacy of conservation policies.
In Guatemala, the conservation intervention in question is a land tenure clarification project. In Colombia, I compare two conservation interventions (land acquisitions and payments for ecosystem services) for cost-effectiveness. I collected and consolidated spatial data on land acquisitions to create a novel dataset of these interventions. I use quasi-experimental methods with novel remote sensing data to estimate impact on forest cover. Field interviews and surveys inform the quasi-experimental analysis and explain results. This research therefore works across scales and uses multiple methods to tackle questions of project or program effectiveness. In the Colombian Andes, public land acquisitions (PLA) as investments in watershed services covered an area substantial enough to be considered within the broader PA network. However, PLA mostly occurred in lower cost (and relatedly, less-threatened) ecosystems, with some exceptions, such as the high Andean Páramo hills and mountains and Medium dense dry forest on hills and mountains. The percent of area distributions covered by PLA for a subset of species showed mostly marginal increases in protection coverage, with some exceptions, including Hapalopsittaca fuertesi, and endangered parrot, and Spizaetus isidori, an endangered eagle. Half of the area of mapped land acquisitions overlapped with legally protected areas. PLA as investments in watershed services are not guided by ecosystem endangeredness, but rather are moderated by important social and economic factors, including governmental income, land costs, and land tenure informality.
In the southeastern subregion of Antioquia, where there is rich empirical data for both PLA and PES (payments for environmental services), we found that there are large differences in the level of implementation of PLA and PES, as well as the geographical characteristics of where they are implemented, with PES covering more area in places with more land tenure informality, lower incomes, and ecosystem threat. Despite PES being more broadly implemented, the impacts of PLA on percent increases in forest cover were consistently higher than those of PES. The overall relative area impacts were dependent on the choice of land cover dataset, but if we believe our ensemble model dataset that we considered most reliable, PLA had a higher area impact than PES. We did not find compelling evidence that either intervention was more cost-effective than the other. We identified unintended consequences of each intervention that should be considered when managers are making decisions about how to implement either of these interventions, including unequal distributions of benefits, perceived undue burden on the participants in either intervention, especially for those who were within PAs.
The degree of forest cover gain (or of avoided loss) could be improved through more efficient targeting of PES and PLA. Subgroup analyses suggest that effects on forest cover are higher outside of protected areas PAs, in lower elevation areas, and (for PES) in areas with higher deforestation pressure. In Guatemala’s PAs, we were only able to conduct counterfactual impact evaluations with one type of PA. These were PAs where multiple activities were permitted (with internal zoning), moderate previous levels of deforestation, low drug trade organization (DTO) control, and low to medium land tenure security prior to treatment. We find some effects of the demarcation and cadaster treatments within these PAs – specifically that they prevented approximately 200 ha of forest post-treatment. The other PA types, including those with already-secure land tenure prior to treatment as well as strict protection and nearly no deforestation prior to treatment, strict PAs under DTO control, and buffer zones do not show positive forest outcome post-treatment through non-robust impact estimation methods, but interviews and the literature suggest that there is little reason to expect positive impacts in these areas. We do not have interviews or other data to explain patterns in strict PAs with low deforestation pressure prior to treatment and low levels of insecure land tenure. Common themes amongst intervention studies in this dissertation suggest that developing the right indicators for implementation progress, the efficient targeting of interventions, inter-institutional cooperation and data management could improve conservation outcomes.
| Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/49359 |
| Date | 01 October 2024 |
| Creators | Reboredo Segovia, Ana Laura |
| Contributors | Nolte, Christoph |
| Source Sets | Boston University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
| Rights | Attribution 4.0 International, http://creativecommons.org/licenses/by/4.0/ |
Page generated in 0.0017 seconds