“Having the support from the University of Florida Biodiversity Institute, my hope is that my research will inform efforts to conserve biodiversity in one of the most beleaguered ecosystems in the world, the Atlantic Forests of Brazil. I am evaluating a biodiversity conservation intervention in terms of its effectiveness (i.e., what did they actually conserve?), efficiency (i.e., how cost-effective was the intervention?), and equity (i.e., how were their welfare effects distributed?). The interventions are designed to support the Brazilian Forest Code, the main legal instrument used to protect and recover native vegetation on private land.
For the Atlantic Forest region, the Forest Code (FC) requires landowners to maintain ‘permanent protected areas’ (i.e. buffers zones around springs and rivers, sleep slopes, and hilltops), as well as 20% of their property as natural vegetation. Although the FC law could be a powerful conservation tool, low compliance reduces its impacts on forest conservation. To promote compliance, the FC uses both disincentives (e.g. law enforcement with fines) and incentives (e.g., payment for environment services). While economic incentives can promote conservation and forest restoration, law enforcement may be more effective and efficient under certain conditions. In recognition of Brazil’s size and diversity, I will use the Rio Claro municipality in Rio de Janeiro State as a case study. Rio Claro is especially suitable for my purposes because there are plans to extend its payment for environment services project to the entire state despite the absence of a rigorous evaluation. I plan to increase likelihood of my evaluation research helping to improve the design of its greatly extended version by collaborating with the responsible government agency.
My research will contribute to the biodiversity conservation literature by focusing on the restoration elements of the law. This aspect is normally neglected in consequence of the difficulty in using remote sensing to detect small and slow changes in vegetation structure in areas that are regenerating. For this purpose, I am using open sources software and freely available imagery to model vegetation changes. This approach has the potential to substantially reduce the cost of enforcement. Basically, I am mapping and monitoring forest regrowth by linking the temporally and spatially-extensive Landsat record with high resolution air photo image publicly available for the study area. I create models that predict forest regrowth with variables already available on-line as a first step to understand how socio-economic processes influence restoration and to explore how the ecological dynamics of forest regrowth relate to property boundaries. I will use my study site to: (1) validate the vegetation regrowth model; and (2) conduct interviews to understand compliance and voluntary participation in a payment for environment services project. The data I collect in the field will be combined with the remote sensing data to create control groups for evaluation of the impact of the payment for environment services. My findings should reveal whether the payment for environment services has a causal relationship with forest cover change. The results will allow me to explore which characteristics of the properties and property owners were associated with the projects’ outcomes and to understand where, when, why and for whom the PES intervention worked. I will address the effectiveness of the intervention using implementation cost data in combination with land-cover change outcomes. I will then compare these costs with a hypothetical high enforcement scenario.
The Forest Code is a potentially powerful tool to promote conservation of Atlantic Forest biodiversity but to capture these benefits and to avoid weakening of the regulations, policy-makers need to know what aspects of it works for whom and under what conditions.”
“I am investigating how the community of plants and animals in a Kenyan savanna is affected by an invasive ant species from Ethiopia. The impact of invasive species is often measured by the decline in native species diversity (i.e., how many species are found in a community, and how evenly those species are distributed). However, we do not clearly understand how species diversity is linked to important properties of the environment, such as the breakdown of dead material or the pollination of native flowers.
[My assistant and I using portable photosynthesis meter in the savanna at Mpala Research Centre.]
I’m trying to understand how species diversity influences environmental characteristics, to contribute to modern academic theories about those topics, and also to help Kenyan conservationists to understand the consequences of biological invasion. Over the next year, I will closely monitor changes to both species diversity and various environmental characteristics as the ant invasion spreads across two conservancies in Kenya for the next year. The tools that I use for my work are varied: portable electronics to measure the physiology of important savanna plant species; long-term field installations to measure termite activity; and various collection techniques to survey the native insect populations.
To learn more about my research, you can visit www.savannaecology.com, where I have posted multiple blogs, and also maintain a podcast series about research and conservation by myself and others in the region.”
“I am a Biology PhD student in the Soltis Lab at the Florida Museum of Natural History. My research will utilize big data to explore the relationship between polyploidy and niche adaptation in plants on a global scale. This project will test common assumptions about polyploidy in plants while generating new hypotheses and resources for future research.”
“My 2016-2017 UF Biodiversity Institute Graduate Student Fellowship has provided wonderful opportunites to build fundamentals in interdisciplinary field of remote sensing, image processing, and machine learning techniques to solve important knowledge gap regarding tropical forest biodiversity: In species-rich Tropical forest, diverse phenological strategies coexist, however not the pattern nor the cause of such variety is yet fully understood, due to limitation in monitoring methods.
To solve this gap, I use a year-long, bi-weekly high-resolution time series images that were acquired by UAVs (Unmanned Aerial Vehicles), over 50-ha plot in Barro Colorado Island (BCI), Panama. In estimating true phenology from the UAV dataset, expertise in overlapping fields of remote sensing, image processing, and machine learning is essential; e.g., in processing images, fundamental knowledge in theories and practice is required to extract relevant features from leaf, branch, and flower structures; in extracting crown-level information, proficiency in GIS software and programming languages is needed; in categorizing crowns to different levels of phenology phases, training and validating classification models is necessary.
During 2016-2017, I took the image processing and computer vision class (EEE 6521) from Electrical and Computer Engineering department, to develop a basis for image processing techniques. In addition, I developed a framework to fully utilize UF high performance computing (HPC) environment in generating and extracting textural features of individual trees in the UAV dataset, by employing relevant programming skills. The framework enables massive data processing to test multiple textural features of crown images, which could be used in estimating true phenology of tree crowns. I tested this framework on a subset of UAV datasets, whether extracted features correspond well with visual observations of leaf, branch, and flower cover of each crown (n=56); several features showed strong correlation to crown leaf cover (e.g. r=0.70, p<0.01).
The framework and techniques I have developed during UFBI fellowship will be combined with my previous work on automatically detecting leaf-on and leaf-off status of tree crowns (presented at Ecological Society of America, in 2016), to better estimate the individual-level phenology status of tree crowns. Accurate estimation of phenology status for thousands of individual trees for hundreds of species in the 50-ha plot in BCI will be a valuable information to study diverse phenological strategies in tropical forest.”