Deforestation in Protected Areas in Madagascar and Correlation with Fire

Abstract
Madagascar's forests are both an integral part of the Malagasy economy and home to many endemic species. Ascertaining the main causes of deforestation can aid in understanding how conservation strategies can be deployed more effectively. This exploratory project focuses on deforestation caused by fire either natural or human caused, usually via slash-and-burn agricultural practices. To understand how much deforestation is caused by fire, we used remotely sensed data from 2000 to 2016, including deforestation data from University of Maryland’s Global Forest of Change database, and protected park outlines from the World Database on Protected Areas, and the US Geological Survey and NASA's LandSat program. 98 terrestrial parks were identified and separated by forest type (humid n=60, dry n=22, spiny n=8, and combinations humid/dry n=3, dry/spiny n=4, and spiny/humid n=1). On average, 5.66 square kilometers were lost to due to fire each year, accounting for less than 4% of the deforestation per year. However, some forest types experience more deforestation due to fire than others. Dry forests lost an average 12.6 square kilometers per year, similar to fire loss in humid forests, even though humid forests account for almost 3x more protected forested area. This preliminary analysis suggests that deforestation by fire may be more prevalent in certain forest types in Madagascar.
Collaborators: Mark Robinson, Melanie Polanco, Ashly Lopez
Skills
Madagascar's forests are both an integral part of the Malagasy economy and home to many endemic species. Ascertaining the main causes of deforestation can aid in understanding how conservation strategies can be deployed more effectively. This exploratory project focuses on deforestation caused by fire either natural or human caused, usually via slash-and-burn agricultural practices. To understand how much deforestation is caused by fire, we used remotely sensed data from 2000 to 2016, including deforestation data from University of Maryland’s Global Forest of Change database, and protected park outlines from the World Database on Protected Areas, and the US Geological Survey and NASA's LandSat program. 98 terrestrial parks were identified and separated by forest type (humid n=60, dry n=22, spiny n=8, and combinations humid/dry n=3, dry/spiny n=4, and spiny/humid n=1). On average, 5.66 square kilometers were lost to due to fire each year, accounting for less than 4% of the deforestation per year. However, some forest types experience more deforestation due to fire than others. Dry forests lost an average 12.6 square kilometers per year, similar to fire loss in humid forests, even though humid forests account for almost 3x more protected forested area. This preliminary analysis suggests that deforestation by fire may be more prevalent in certain forest types in Madagascar.
Collaborators: Mark Robinson, Melanie Polanco, Ashly Lopez
Skills
- GIS and raster manipulation
- Bioinformatics: R
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
Deforestation in Protected Areas in Madagascar

Abstract
Due to human activity, Madagascar is experiencing alarming rates of deforestation which endangers numerous species endemic to the island. To understand possible causes of deforestation, we used remotely sensed data from 2000 to 2019, including deforestation data from University of Maryland’s Global Forest of Change database and protected park outlines from the World Database on Protected Areas. ANOVA and linear models were run the different forest types (humid n=60, dry n=22, spiny n=8, and combinations humid/dry n=3, dry/spiny n=4, and spiny/humid n=1) against features of parks hypothesizes to correlate with increased deforestation, including length of roads within parks, populations within parks, age of parks, management type, and IUCN category. While deforestation rates did vary by forest type (humid 0.9%, dry 0.66%, spiny 0.32%, humid/dry 0.35%, dry/spiny 1.96%, and spiny/humid 0.23%) they did not differ significantly, nor did any of the hypothesized correlates of deforestation significantly explain deforestation rates. Interpretation of results is limited by the coarseness of data, such that management type had only two categories, and age of park was based on oldest portion with many parks being modified with area added more recently. Future research should focus on other possible correlates of deforestation, such as fire.
Collaborators: Galiba Anjum, Matthew Chu, Kayla Mora
Skills
Due to human activity, Madagascar is experiencing alarming rates of deforestation which endangers numerous species endemic to the island. To understand possible causes of deforestation, we used remotely sensed data from 2000 to 2019, including deforestation data from University of Maryland’s Global Forest of Change database and protected park outlines from the World Database on Protected Areas. ANOVA and linear models were run the different forest types (humid n=60, dry n=22, spiny n=8, and combinations humid/dry n=3, dry/spiny n=4, and spiny/humid n=1) against features of parks hypothesizes to correlate with increased deforestation, including length of roads within parks, populations within parks, age of parks, management type, and IUCN category. While deforestation rates did vary by forest type (humid 0.9%, dry 0.66%, spiny 0.32%, humid/dry 0.35%, dry/spiny 1.96%, and spiny/humid 0.23%) they did not differ significantly, nor did any of the hypothesized correlates of deforestation significantly explain deforestation rates. Interpretation of results is limited by the coarseness of data, such that management type had only two categories, and age of park was based on oldest portion with many parks being modified with area added more recently. Future research should focus on other possible correlates of deforestation, such as fire.
Collaborators: Galiba Anjum, Matthew Chu, Kayla Mora
Skills
- GIS and raster manipulation
- Bioinformatics: R
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
Comparison of Deforestation Rates in Protected and Unprotected Forests in Madagascar

Abstract
Most of Madagascar's forests are impacted by deforestation. As Madagascar is a biodiversity hotspot, deforestation rates disproportionately impact endemic species. Madagascar has many protected forested areas in the form of parks and nature reserves, but no studies thus far have examined the success of these protections in terms of reduction of deforestation rates. This project examined deforestation rates in Madagascar from 2000-2018, comparing deforestation rates of protected areas to those of unprotected ones. Protected areas were compared to similarly sized unprotected areas in the same forest type (i.e., dry, humid, or spiny). Data was gathered through satellite imagery and processed with R scripts. In general, there was no difference in deforestation rates between protected and unprotected regions when all forest types were combined (t-test = 0.84, n=76). However, this trend differed across forest types. Deforestation rates in dry forests (t-test = 0.37, n=20) did not differ between protected and unprotected areas. In humid forests, protected areas significantly reduced the deforestation rate (t-test = 0.04, n=34). Most surprisingly, in spiny forests, protected regions had HIGHER rates of deforestation compared to unprotected regions (t-test = 0.07, n=22). Differences in deforestation rates could be due to variation in management type or policies protecting the regions. Clearly more research is needed to understand the role of different attributes between protected areas to elucidate why some regions are better at lower deforestation rates than others.
Collaborators: Jason Jiang, Chenjie Gu
Skills
Most of Madagascar's forests are impacted by deforestation. As Madagascar is a biodiversity hotspot, deforestation rates disproportionately impact endemic species. Madagascar has many protected forested areas in the form of parks and nature reserves, but no studies thus far have examined the success of these protections in terms of reduction of deforestation rates. This project examined deforestation rates in Madagascar from 2000-2018, comparing deforestation rates of protected areas to those of unprotected ones. Protected areas were compared to similarly sized unprotected areas in the same forest type (i.e., dry, humid, or spiny). Data was gathered through satellite imagery and processed with R scripts. In general, there was no difference in deforestation rates between protected and unprotected regions when all forest types were combined (t-test = 0.84, n=76). However, this trend differed across forest types. Deforestation rates in dry forests (t-test = 0.37, n=20) did not differ between protected and unprotected areas. In humid forests, protected areas significantly reduced the deforestation rate (t-test = 0.04, n=34). Most surprisingly, in spiny forests, protected regions had HIGHER rates of deforestation compared to unprotected regions (t-test = 0.07, n=22). Differences in deforestation rates could be due to variation in management type or policies protecting the regions. Clearly more research is needed to understand the role of different attributes between protected areas to elucidate why some regions are better at lower deforestation rates than others.
Collaborators: Jason Jiang, Chenjie Gu
Skills
- GIS and raster manipulation
- Bioinformatics: R
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
Deforestation trends in protected areas in Madagascar

Abstract
Today, over 90% of Madagascar’s forests are being impacted by deforestation. With deforestation rates accelerating in a biodiversity hotspot region such as Madagascar, issues such as climate change, extinction, and habitat loss can worsen. Madagascar has many protected forested areas in the form of parks and nature reserves, but no studies thus far have examined the success of these protections in terms of reduction of deforestation rates. This project examined deforestation rates in Madagascar from 2000-2016, comparing deforestation rates of protected areas to those of unprotected ones. Protected areas were compared to similarly sized unprotected areas in the same forest type (i.e., dry, humid, or spiny). Data was gathered through satellite imagery and processed with R scripts. In general, protected regions exhibited lower rates of deforestation when compared to unprotected ones (t-test = 0.004). However, this trend differed across forest types. Deforestation rates in dry (t-test = 0.47) and spiny (0.84) forests did not differ between protected and unprotected areas. In humid forests, protected areas significantly reduced the deforestation rate (t-test = 0.001). Differences in deforestation rates could be due to variation in protection level or year protected status was enacted. Overall, protected areas ease deforestation rates somewhat, but more research is needed to understand the role of different attributes between protected areas.
Collaborators: Brandon Wong, April Wu
Skills
Today, over 90% of Madagascar’s forests are being impacted by deforestation. With deforestation rates accelerating in a biodiversity hotspot region such as Madagascar, issues such as climate change, extinction, and habitat loss can worsen. Madagascar has many protected forested areas in the form of parks and nature reserves, but no studies thus far have examined the success of these protections in terms of reduction of deforestation rates. This project examined deforestation rates in Madagascar from 2000-2016, comparing deforestation rates of protected areas to those of unprotected ones. Protected areas were compared to similarly sized unprotected areas in the same forest type (i.e., dry, humid, or spiny). Data was gathered through satellite imagery and processed with R scripts. In general, protected regions exhibited lower rates of deforestation when compared to unprotected ones (t-test = 0.004). However, this trend differed across forest types. Deforestation rates in dry (t-test = 0.47) and spiny (0.84) forests did not differ between protected and unprotected areas. In humid forests, protected areas significantly reduced the deforestation rate (t-test = 0.001). Differences in deforestation rates could be due to variation in protection level or year protected status was enacted. Overall, protected areas ease deforestation rates somewhat, but more research is needed to understand the role of different attributes between protected areas.
Collaborators: Brandon Wong, April Wu
Skills
- GIS and raster manipulation
- Bioinformatics: R
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
Relatedness in mouse lemur species

Abstract
Within-species genetic studies of primates have typically focused on parentage within groups or phylogeographic relationships among sparsely sampled populations. Most of these studies use a small number of microsatellites, which can only identify close relatives and may give a poor approximation of the overall relatedness in the population. This project compares relatedness within two sympatric species of mouse lemur (Microcebus berthae and M. murinus) in the Central Menabe region of western Madagascar. While M. murinus has a broad range, M. berthae is confined to a single dwindling forest. Samples were collected throughout the entire forest, with 45 samples from M. berthae and 120 samples from M. murinus. Using single polymorphism (SNP) data collected by ddRADseq, relatedness levels could be calculated to degrees unobtainable with microsatellite data. While both species had similar mean values of relatedness (~ 0.0), the patterns of background relatedness varied. Relatedness visualizations demonstrate a background of elevated relatedness or M. murinus, but not M. berthae. This pattern could be caused by the different distribution patterns each species displays within the forest, with M. murinus as more dispersed facilitating larger related networks.
Collaborators: Dr. Peter Kappeler, Dr. Livia Schäffler
Skills
Within-species genetic studies of primates have typically focused on parentage within groups or phylogeographic relationships among sparsely sampled populations. Most of these studies use a small number of microsatellites, which can only identify close relatives and may give a poor approximation of the overall relatedness in the population. This project compares relatedness within two sympatric species of mouse lemur (Microcebus berthae and M. murinus) in the Central Menabe region of western Madagascar. While M. murinus has a broad range, M. berthae is confined to a single dwindling forest. Samples were collected throughout the entire forest, with 45 samples from M. berthae and 120 samples from M. murinus. Using single polymorphism (SNP) data collected by ddRADseq, relatedness levels could be calculated to degrees unobtainable with microsatellite data. While both species had similar mean values of relatedness (~ 0.0), the patterns of background relatedness varied. Relatedness visualizations demonstrate a background of elevated relatedness or M. murinus, but not M. berthae. This pattern could be caused by the different distribution patterns each species displays within the forest, with M. murinus as more dispersed facilitating larger related networks.
Collaborators: Dr. Peter Kappeler, Dr. Livia Schäffler
Skills
- Next Generation Sequencing Library Preparation (ddRADseq method)
- Bioinformatics:
- Python scripts
- Stacks
- R
Deforestation in Madagascar's Forests

Abstract
Anthropogenic forest loss is severe, recent, and accelerating. In Madagascar alone, almost 40% of original forested areas have been lost since the 1950s. Deforestation negatively affects many already threatened, endemic species by destroying their habitats and lowering their populations. Deforestation of different forest types results in different climatic outcomes. For example, removal of humid forests leads to drier conditions at local scales. We used satellite imagery to quantify and compare deforestation trends across four different types of forests in Madagascar (humid, dry, spiny, mangrove). Deforestation in almost all forest types (except for mangroves) has been increasing through time. Between 2000 and 2016, approximately 3.7% of spiny forest, 2.3% of dry forest, 11.6% of humid forest, and 1.2% of mangroves have been lost. For a total loss of 38,042 square kilometers of forest over 16 years. Rates of deforestation varied, with the highest in humid forests (correlation coefficient = 0.8) and the lowest in spiny forests (correlation coefficient = 0.05). The slower rate of deforestation in Madagascar's spiny forests may be due to the lack of rosewood, an endangered and highly sought hardwood harvested illegally for international markets.
Collaborators: Catherine Zhou, Pablo De Castro Gomez
Skills
Anthropogenic forest loss is severe, recent, and accelerating. In Madagascar alone, almost 40% of original forested areas have been lost since the 1950s. Deforestation negatively affects many already threatened, endemic species by destroying their habitats and lowering their populations. Deforestation of different forest types results in different climatic outcomes. For example, removal of humid forests leads to drier conditions at local scales. We used satellite imagery to quantify and compare deforestation trends across four different types of forests in Madagascar (humid, dry, spiny, mangrove). Deforestation in almost all forest types (except for mangroves) has been increasing through time. Between 2000 and 2016, approximately 3.7% of spiny forest, 2.3% of dry forest, 11.6% of humid forest, and 1.2% of mangroves have been lost. For a total loss of 38,042 square kilometers of forest over 16 years. Rates of deforestation varied, with the highest in humid forests (correlation coefficient = 0.8) and the lowest in spiny forests (correlation coefficient = 0.05). The slower rate of deforestation in Madagascar's spiny forests may be due to the lack of rosewood, an endangered and highly sought hardwood harvested illegally for international markets.
Collaborators: Catherine Zhou, Pablo De Castro Gomez
Skills
- GIS and raster manipulation
- Bioinformatics: R
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
Genetic Structure and Gene Flow in Mouse Lemurs

Abstract
There is a critical need to ensure that the effects of current (and accelerating) habitat fragmentation are identified and differentiated from the genetic signatures of long-past events and environmental circumstances. The long term goals in this research area are to differentiate between the effects of recent habitat fragmentation and long-term evolutionary demographic history and population structure on primate species. Research results in this area will inform our understanding of ecological and evolutionary diversification in fragmented landscapes and offer guidance to conservation programs. The specific objective of this research proposal is to use genome-wide genetic data to compare and contrast genetic diversity, population structure, demographic history, and recent and historical gene flow in two sympatric mouse lemur species in western Madagascar: the endangered Mme. Berthe’s mouse lemur (Microcebus berthae) and the non-threatened gray mouse lemur (M. murinus).
Collaborators: Dr. Peter Kappeler, Dr. Livia Schäffler
Skills
There is a critical need to ensure that the effects of current (and accelerating) habitat fragmentation are identified and differentiated from the genetic signatures of long-past events and environmental circumstances. The long term goals in this research area are to differentiate between the effects of recent habitat fragmentation and long-term evolutionary demographic history and population structure on primate species. Research results in this area will inform our understanding of ecological and evolutionary diversification in fragmented landscapes and offer guidance to conservation programs. The specific objective of this research proposal is to use genome-wide genetic data to compare and contrast genetic diversity, population structure, demographic history, and recent and historical gene flow in two sympatric mouse lemur species in western Madagascar: the endangered Mme. Berthe’s mouse lemur (Microcebus berthae) and the non-threatened gray mouse lemur (M. murinus).
Collaborators: Dr. Peter Kappeler, Dr. Livia Schäffler
Skills
- Next Generation Sequencing Library Preparation (ddRADseq method)
- GIS and raster manipulation
- Species distribution modeling
- Landscape genetic analyses
- Bioinformatics:
- Python scripts
- Stacks
- R
Morphological Variation in Mouse Lemurs

Abstract
A major focus of evolutionary primatology is understanding how morphological variation is partitioned relative to ecological gradients and landscape barriers. Many studies of primates examine the distribution of morphological variation relative to neighboring and other closely related species in order to understand the relationship between morphology and environment. There is growing awareness of the diversity found within and between mouse lemurs at a genetic level. This newly described variation in mouse lemurs can also be studied using a morphological approach. I studied morphological variation in two species of mouse lemur (Microcebus murinus and Microcebus griseorufus) that are found in neighboring locations in Madagascar as well as being closely related sister species. Using skins collections from the American Museum of Natural History, I collected measurements on fur coloration differences. Pelage coloration differences between M. murinus and M. griseorufus are limited to facial and ventral locations, with dorsal coloration playing a minor role. Coloration is known to have several purposes, such as concealment, species/kin recognition, and sexual signals. Differences between species can be caused by several factors, such as genetic drift as well as selection by factors listed above. The causes for the differences in these mouse lemurs is unknown, but overall, M. griseorufus is lighter in coloration compared to M. murinus.
Collaborators: Joralyssa Conley, Aryanna Myles
Skills
A major focus of evolutionary primatology is understanding how morphological variation is partitioned relative to ecological gradients and landscape barriers. Many studies of primates examine the distribution of morphological variation relative to neighboring and other closely related species in order to understand the relationship between morphology and environment. There is growing awareness of the diversity found within and between mouse lemurs at a genetic level. This newly described variation in mouse lemurs can also be studied using a morphological approach. I studied morphological variation in two species of mouse lemur (Microcebus murinus and Microcebus griseorufus) that are found in neighboring locations in Madagascar as well as being closely related sister species. Using skins collections from the American Museum of Natural History, I collected measurements on fur coloration differences. Pelage coloration differences between M. murinus and M. griseorufus are limited to facial and ventral locations, with dorsal coloration playing a minor role. Coloration is known to have several purposes, such as concealment, species/kin recognition, and sexual signals. Differences between species can be caused by several factors, such as genetic drift as well as selection by factors listed above. The causes for the differences in these mouse lemurs is unknown, but overall, M. griseorufus is lighter in coloration compared to M. murinus.
Collaborators: Joralyssa Conley, Aryanna Myles
Skills
- Mentoring high school students as part of the Student Research Mentoring Program with the American Museum of Natural History
- Utilization of the Primate Collections at the American Museum of Natural History
Behavior and Dominance Effects in Blue Monkeys

Abstract
Many studies of cercopithecine monkeys implicate dominance rank as an important variable influencing fitness, but there is growing awareness of the paucity of data for many species, and variation in rank’s effects. We studied how dominance rank affected the exposure of wild adult female Cercopithecus mitis to predators. Dominance hierarchies in blue monkeys have been characterized as stable but shallow. Focusing on adult females in two study groups inhabiting the Kakamega Forest, Kenya, we monitored their location in the canopy and in individual tree crowns over a two month period. Although blue monkeys are generally arboreal, we found that high-ranking females were somewhat more likely than low-rankers to use the ground, which is likely risky. Otherwise, rank did not predict the location of individuals in the trees, and concomitant exposure to aerial predators. Our results confirm a previous report from a third group in our study population. They are also consistent with other findings suggesting that dominance rank has limited effects on behavior in this species, despite being readily detectable. Effects of dominance rank are clearly variable in cercopithecines.
Collaborators: Dr. Marina Cords, Colins Mutai
Skills
Many studies of cercopithecine monkeys implicate dominance rank as an important variable influencing fitness, but there is growing awareness of the paucity of data for many species, and variation in rank’s effects. We studied how dominance rank affected the exposure of wild adult female Cercopithecus mitis to predators. Dominance hierarchies in blue monkeys have been characterized as stable but shallow. Focusing on adult females in two study groups inhabiting the Kakamega Forest, Kenya, we monitored their location in the canopy and in individual tree crowns over a two month period. Although blue monkeys are generally arboreal, we found that high-ranking females were somewhat more likely than low-rankers to use the ground, which is likely risky. Otherwise, rank did not predict the location of individuals in the trees, and concomitant exposure to aerial predators. Our results confirm a previous report from a third group in our study population. They are also consistent with other findings suggesting that dominance rank has limited effects on behavior in this species, despite being readily detectable. Effects of dominance rank are clearly variable in cercopithecines.
Collaborators: Dr. Marina Cords, Colins Mutai
Skills
- 3 months field experience in Kakamega Forest, Kenya
- Collection of behavioral data (focal follows)
- Conservation activities with local communities