Forest Conservation with AI

A tale of the African Island of Madagascar

Deforestation in Madagascar

Madagascar is considered to be one of eight biodiversity hotspots across the globe due to its species richness and a loss of more than 70% of its original primary vegetation. It is estimated that 44% of Madagascar's natural forest has been lost since 1953, due to anthropogenic activity, which may be attributed to (i) the increasing population, (ii) practice of slash-and-burn agriculture, (iii) grazing, (iv) fuelwood gathering, (v) logging, (vi) economic development projects, (vii) cattle ranching, (viii) mining, etcetera.

The increasing deforestation results in an increasingly divided forest where at forest edges, the differing air temperature, moisture, and sunlight conditions impact both the abundance and distribution of species. Deforestation is a major threat to native species since 90% of species endemic to Madagascar are forest-dependent. The study area, Betampona Nature Reserve (BNR), provides a living laboratory for the studies of human–forest interactions, which has significant impact on our understanding of tropical forests and biodiversity at the global scale.

Betampona Nature Reserve

The Betampona Nature Reserve (BNR) is situated on the central-eastern coast of Madagascar

Once contiguous with the Zahamena rainforest, BNR is now isolated because of extensive deforestation which has resulted in it being one of few remaining tracts of primary rainforest on the eastern coast

BNR has an extremely high plant species diversity per hectare when compared to other rainforests. As a result of its isolated nature and small extent, fauna endemic to the BNR is believed to be at higher risk.

Additionally, non-native invasive plant or animal species also threaten the biodiversity within the BNR. Three habitat-altering invasive plant species seen prominently within the BNR are Molucca Raspberry (Rubus moluccanus ), Madagascar Cardamom (Afromomum angustifolium), and Strawberry Guava (Psidium cattleianum).

1

Molucca Raspberry

2

Madagascar Cardamom

Although native to Madagascar, its weedlike characteristics in degraded areas within and surrounding the BNR make it invasive in the region.

3

Guava

...

.

..

.

These are just a few locations within the BNR where these invasive plants are found.

Human Geography

A view from the study area

To effectively conserve biodiversity, an understanding of the human–environmental interactions is needed, since basic human needs have a direct impact on various conservation efforts through unsustainable agricultural practices, human encroachment, and bushmeat consumption.

As shown in the photos below, trees are cut down for cooking, then the cleared areas are used for tavy, a form of slash-and-burn agriculture to grow crops.

From left to right are photos of wood burning rock stove, newly cut areas to grow crops and tavy in the zone of protection. Image credit: MFG, 2018

Locals around the BNR face food insecurity because of the growing population, which is based on the increased residential land cover and decreasing agricultural yields. This food insecurity is further driving forest cover conversion into agricultural fields. Food insecurity leads to the consumption of livestock (chickens, ducks, etc.) and to the consumption of wildlife—particularly lemurs. Children around the BNR are reported to becoming weak, falling ill, and even performing poorly in school because of hunger. Health care services are costly and inaccessible to locals around the BNR. It was common for locals to use medicinal plants as a cheaper option.

Rice fields

Agriculture is the primary source of income for the locals around the BNR, with tavy being practiced by 80–85% of farmers. To ensure a source of income, locals around the BNR amplify and expand agricultural production. The poorest households around the BNR spend 2/3's of their income on food with many households experiencing chronic or temporary food insecurity, which is intensified by climate change-related events.

Remote Sensing

These human–environmental interactions are often studied by mapping the landscape with different land cover and land uses using Remote Sensing.

Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), and Deep Neural Networks (DNN) have been used for image classification. Further, a shift is seen toward the use of deep learning algorithms for the creation of classification maps with Convolutional Neural Networks (CNNs). CNNs are a type of Deep Neural Network model architecture that can extract various features automatically. To that end, new methods were developed, such as Fully Convolutional Networks (FCNNs) for image classification through pixel-wise semantic segmentation.

While considering the size of the BNR and more importantly, the rugged terrain, remotely sensed data is a top contender to map forest cover. The objectives were to develop end-to-end mapping of the tropical forest using fully convolution neural networks (FCNNs) with WorldView-3 (WV-3) imagery, to evaluate human impact on the environment using the BNR in Madagascar as the test site and to evaluate the impact of conservation efforts in the region by international organizations by analyzing land cover change from 2010 to 2019.

Land Cover Classification

The table below lists the classes and descriptions used in order to create the classification map. 11 classes were defined that relates to land cover (Open Water, Row Crops), land use (Residential, Fallow) and forest cover (Mixed Forest, Evergreen Forest, Molucca Raspberry, Shrubland, Madagascar Cardamom, Grassland and Guava)

Detailed class descriptions

Machine Learning algorithms were implemented to create a Land Cover Classification map using imagery collected in 2019.

Around 50% of the study area is covered by shrubland, 33% is covered by mixed and evergreen forests, and 4.7% is covered by invasive plant species. Another 8% of the study area is used for cultivation (Row Crops and Fallow). 

The presence of agricultural fields in the Zone of Protection (ZOP), which extends 100 m outside the boundary of the BNR, leads to an increase in human interaction in the buffer zone, which often has negative impacts on the biodiversity within the BNR.

Land Cover Change

Conservation efforts were quantified by investigating the land cover and land use change over time from 2010 to 2019. 

Land Cover Classifications (2010 vs 2019)

An increase is seen for the following classes: Evergreen Forest (1.5%), Mixed Forest (4.9%), Residential (44.3%), Shrubland (37.2%), Open Water (53.6%), and Madagascar Cardamom (47.5%).

A decrease is seen for the following classes: Molucca Raspberry (−55.7%), Row Crops (−56.1%), Fallow (−31.0%), Grassland (−63.4%), and Guava (−6.1%).

Within the BNR, results from the 2019 imagery show a 12.5% increase in regenerating Mixed Forest and a 0.7% increase in Evergreen Forest. The regeneration of Evergreen Forest is a lengthy process, and therefore, the success of the restoration work within BNR is only slightly apparent with a 0.7% increase in Evergreen Forest. However, given that the trend in 2010 was toward increasing deforestation and forest fragmentation, this is a worthwhile result.

Misclassification of Evergreen Forest and Shrubland may have been prevalent due to the small height of the trees. Additionally, the vegetation in these areas could be late inter-cropping periods when agricultural fields have been left fallow for several years, or it could be the early stages of a Mixed Forest or regenerating Evergreen Forest.

Knowing where and which LCLU types have been converted to Shrubland within the study area is very valuable to the MFG and their partners at the MNP. This knowledge targets awareness raising and conservation efforts in areas of particular concern for example, the deforestation of formerly evergreen forests that are mainly remnant “Classified Forests”. These “Classified Forests” have a legally protected status but little practical protection. 

Spatial Change

In order to visualize the trajectory of change spatially, a grid cell approach was used. This technique was used to combine pixel-level land cover and land use types. It is approximately the size of mature trees so that we can track the changes to the scale of individual trees. 

The image below shows the comparison between percent changes in Evergreen Forest and percent changes in Agriculture, highlighting the human impact (or lack thereof) on the BNR. Increases are seen in the Evergreen Forest within the BNR and surroundings and even the ZOP (outer edge of the gray polygon).

Change over time was calculated for forests

and for invasive plants

The success of Conservation Efforts by Madagascar Flora and Fauna Group (MFG), can also be quantified using the land cover classifications.

MFG, an association of zoos and botanical gardens, manages the BNR in collaboration with the Madagascar National Parks (MNP). Since 2007, the MFG has run a community-based native forest restoration project in the BNR’s ZOP, a 100 m buffer extending out from the BNR, in partnership with the Madagascar National Parks. Another initiative is the control of invasive Guava in the BNR as well as the control of local invasive plants Madagascar Cardamom and invasive Molucca Raspberry, both of which have contributed to the 0.7% increase in Evergreen Forest in the BNR.

Land use in the vicinity of protected areas is seen to negatively affect conservation efforts as human and ecological activities overlap in these areas. For the BNR, the size of this zone is taken as 100 m extending out from the BNR boundary. In the ZOP and 100 m within the BNR boundary, a decrease in anthropogenic land use is observed (residential areas, agricultural fields, and fallow land), alternatively, there is an increase in mixed and evergreen forests. These results are most likely a direct consequence of the MFG’s native forest restoration program and increased awareness-raising and lobbying by the MFG and Madagascar National Parks to discourage slash-and-burn agriculture in the ZOP. More recently, the MFG has initiated two efforts to reduce forest loss and promote more sustainable agriculture: the promotion of agroforestry and the distribution of fuel-efficient stoves.

Distribution of fuel-efficient stoves

One of the research objective was to evaluate the impact of using fuel-efficient stoves on forest conservation. To discourage cutting down trees for cooking, 2100 fuel-efficient stoves were distributed in September 2018, May 2019 and September 2019.

However, the effects of these stoves could not be quantified for the study period because 2-3 years were too short for satellite remote sensing.

Fuel-efficient stoves were distributed by foot.

Fuel-efficient stoves were distributed by canoe.

Fuel-efficient stove being used for cooking.

De-weeding Invasive plants

MFG has been involved in deweeding invasive plants from within BNR.

Planting Native Fauna

Native plant species were planted in the Zone of Protection, a 100 meter area surrounding BNR boundary

Study Area

Agroforestry

Within residential areas, the agroforestry has been promoted by the MFG and their agroecology specialists since 2005.

The increase in tree cover is indicative of native trees maturing over time and is also indicative of successful agroforestry efforts.

The local population has already been practicing agroforestry for many generations, so it is not possible to quantify the impact due to the MFG’s specific efforts. However, the provision of training, trees and equipment since 2005 seems to be making a positive difference to tree cover in residential areas despite the overall growth of many of the residential areas.

It should be noted that trees within the boundaries of residential areas, originally classified as Mixed Forests, were re-classified as Residential in this study to quantify the change in residential tree cover for agroforestry analysis. Based on the re-classified pixels, tree cover within residential areas was estimated for 2010 and 2019. A 32% increase in the tree cover in residential areas was observed in 2019, which may be associated with tree growth and increased fruit tree plantation within residential areas encouraged by conservation groups and local authorities to combat food insecurity.

For more information:

Please refer to the Remote Sensing Lab GitHub page for instructional labs.

Published paper can be accessed at the link below for more details on methodology and results.


Acknowledgements

Funding

The funding for this project is provided by Saint Louis University Geospatial Institute (GeoSLU), NASA #80NSSC20M0100, USGS AmericaView Grant (G18AP00077), and IUCN’s SOS Lemurs grant #2017A-093.

Data

The ground reference data were collected thanks to Madagascar Flora and Fauna Group’s in situ data collection team. A minimal cloud-free image was acquired thanks to multiple satellite tasking by Digital Globe.

Story Map

This story map was created by the members of the Remote Sensing Lab at Saint Louis University Gizelle Cota & Kevin Wells (grad students), mentored and edited by Dr. Sagan.

A view from the study area

Rice fields

Detailed class descriptions

Land Cover Classifications (2010 vs 2019)