Severe decline in large agroforestry trees in India over the past decade

doi:10.21203/rs.3.rs-3340055/v1

Download PDF

Biological Sciences - Article

Severe decline in large agroforestry trees in India over the past decade

https://doi.org/10.21203/rs.3.rs-3340055/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 15 May, 2024

Read the published version in Nature Sustainability →

Version 1

posted

You are reading this latest preprint version

Agroforestry trees represent a vital part of India's landscapes, however, despite their significance, a current lack of robust monitoring mechanisms has contributed to an insufficient grasp of their distribution in relation to management practices, as well as their vulnerability to climate change and diseases. Here we map all agroforestry trees, excluding block plantations, in India for each year 2010/2011 and 2018-2022. We reveal a count of about 0.6 billion trees, and track each tree crown over the past decade. Around 11±2% of the large trees (about 96 m2 crown size) mapped in 2010/2011 were disappeared by 2018. Moreover, during the period 2018-2022, more than 5 million large farmland trees (about 67 m2 crown size) have vanished, partly due to altered cultivation practices, where trees within fields are perceived as detrimental to crop yields. These observations are particularly unsettling given the current emphasis on agroforestry as a pivotal natural climate solution, playing a crucial role in both climate change adaptation and mitigation strategies.

Biological sciences/Ecology/Tropical ecology

Earth and environmental sciences/Ecology/Ecosystem services

Wildfires, fungi, insects and droughts cause considerable tree mortality in forests and forested ecosystems¹. Climate change, human appropriation and mismanagement are further aggravating tree die-offs, which have reached alarming levels in recent years^2,3. These events are well documented and the public awareness is high, because operational monitoring systems based on satellites are able to locate and quantify forest diebacks and deforestation¹. However, only little is known on potential widespread diebacks and human appropriation of trees outside forests, such as in drylands or agroforestry systems. Agroforestry trees have drawn a lot of interest in relation to tree planting and restoration as an underexplored pathway towards climate mitigation, but little attention is paid to the protection of existing on-farm trees in well-established agroforestry systems, and how they are being affected by the above-mentioned threats.

According to a recent state-of-the-art land cover map⁴, 56% of India is covered by farmland, and only 20% is covered by forest. While the separation of forests and perennial tree plantations is not always clear, it is unequivocal that these classifications miss a large part of India’s trees, which are scattered within croplands, in urban areas or grow as hedgerows^5,6. These multipurpose trees, such as Prosopis cineraria, Azadirachta indica, Madhuca longifolia, Acacia nilotica, Dalbergia sissoo, Syzygium cumini, Sesbania grandiflora, Albizia procera, Artocarpus heterophyllus and Cocos nucifera, provide a variety of ecosystem services related to environmental conservation (for example, soil fertilization, shade), and products used for consumption (for example, fruits, fuelwood, fibre, mulch, medicine, fodder) or the generation of income (for example, wood, sap, medicine, fibre)^6-9. India is well known as a globally leading nation regarding the large-scale implementation of agroforestry, that is the integration of multipurpose trees in farmland, and even urban systems⁶. However, the systematic quantification of tree resources within these systems is challenging at national scale¹⁰, and current mapping approaches mainly capture block plantations but less individual trees on farms¹⁰. India has the largest agricultural area cover in the world, which makes tree cover and count in India uncertain, and possibly makes it one of the most overlooked woody elements globally.

Agroforestry trees experience a certain dynamic in their populations, as they grow in highly managed ecosystems. The monitoring of these trees has become fundamental over the past decade, not only to quantify available resources but also to identify possible climate related diebacks, as observed in forests¹. A prominent example of a tree grown on farms is the Neem tree (Azadirachta indica), which can grow tall with canopies of up to 20 m in diameter, and live up to several hundred years¹¹. Recent newspaper reports indicate that thousands of Neem trees in India have been infected by fungi over the past years, possibly leading to diebacks, which had already been identified as a potential problem several decades ago^12,13. Fungi, such as Phomopsis azadirachtae, attack old trees in particular, while younger trees are stronger and survive the infections. Due to the large number of trees at national scale, it is a challenge to inventory the number of trees at a single or repeated basis, and the extent of mortality from diseases, climate or human disturbances remains unknown.

Previously it has not been possible to study changes in non-forest trees over time. Individual trees that grow scattered in farmlands can be mapped from sub-meter resolution satellite imagery, but these are not available as time series over larger areas, such as the country of India, and thus cannot be used for quantifying dynamics at the level of single trees in agricultural fields¹⁴. The RapidEye archive is an often overlooked resource of satellite imagery, providing global coverage from micro-satellites back to 2009. Together with PlanetScope (starting 2017), these satellite constellations provide a unique opportunity of repeated global coverage with images at a spatial resolution of 3-5 m, which allows the identification of individual trees^15,16. Here, we developed a heatmap-based tree detection approach^17,18 and deployed the RapidEye and PlanetScope satellite constellations to study changes in agroforestry trees at single tree level in India from 2010-2022.

We define agroforestry trees as adult trees that can be identified as individuals within croplands, here delineated by a previously published state-of-the-art land cover map (10-m resolution)⁴. This includes trees in fallow fields, hedgerows and trees along rural roads and rivers, but excludes recently planted trees, which are not detectable in our images (see Methods). Groups of trees and perennial plantations that form larger closed canopies, such as palm and rubber trees, are typically classified as “treecover” in the baseline land cover map and masked out in our study. Dense tree plantations can be part of agroforestry systems but are highly managed and include a high temporal dynamic from harvest and replanting, which makes this type of land use less relevant for this study, which is aiming at longer term changes in old agroforestry trees related to disturbances rather than recurrent management practices.

We formed country-wide custom mosaics of high-quality satellite scenes for each year 2010-2011 (RapidEye), and 2018-2022 (PlanetScope), and trained two deep learning tree detection models: one for RapidEye and one for PlanetScope, respectively. We merged the results from 2010 and 2011, and also from 2018-2022 to ensure robustness in particular towards variations in image quality (see Methods). The tree crown centers of individual trees mapped in 2010 or 2011 were tracked and equivalents were identified during 2018-2022 (Fig. 1a,b). Here we used a buffer area of 20 m around the tree center to account for potential shifts in satellite image pairs. If a tree center mapped in 2010 or 2011 was not found in one of the annual images between 2018-2022, it was counted as disappeared (Fig. 1b). When developing the models, we aimed for a very low false positive rate from the 2010/2011 detections (3%), meaning that only very few misclassifications exist. This method is tailored towards accurately detecting the disappearance of large trees and consequently is not designed for a wall-to-wall mapping of all single trees in 2010/2011 as it may miss small trees. Therefore, we did not attempt to quantify the gain of small trees and net changes over the full period of analysis.

For the 2010-2018 change, we focused on trees that have been detected with a high confidence (>0.8) in the RapidEye period (see Methods), as these are typically large trees that are clearly identifiable, such as the Neem tree. This means that the instances of detection and changes reported are reliable but conservative, and implies that the real numbers are likely to be higher. As a result, we refrain here from reporting absolute numerical values of change. Results show that 11±2% of the high confidence trees mapped in 2010 or 2011 were not detected in the 2018-2022 images (Fig. 1). The 2% uncertainty was quantified by a manual checking of 1000 random samples. We found that trees detected by RapidEye with a confidence >0.8 have an average crown size of 96 m² (see Methods), which implies that the trees covered by this analysis were generally mature trees that have reached a later stage of development, and such a high loss rate of mature trees over less than a decade is unexpected. The disappearance of mature farmland trees was observed in many areas, but numbers rarely exceed 5-10%, except for areas in central India, in particular in the states Telangana and Maharashtra, where we document massive losses of large trees (Fig. 1). Here, several hot-spot areas have lost up to 50% of their large farmland trees, with up to 22 trees per km² disappearing (Figure 1c-e). Also smaller hot spot areas of loss are observed, such as in eastern Madhya Pradesh around Indore (Extended Data Fig. 1).

We then focused on the period 2018-2022, where higher quality PlanetScope imagery were further used for annual wall-to-wall tree detections (Fig. 2), although the detection reliability for small trees with a crown area below 10-20 m² is relatively low¹⁵ (Extended Data Table 1; Methods). We mapped about 0.6 billion trees within the landcover classes of cropland, urban, and bare. The average number of trees per hectare is 0.6 (SD 1.6), and the highest densities are found in the northwest (Rajasthan) and southeast of India (Chhattisgarh) with up to 22 trees per hectare. We then tracked individual trees during 2018-2022. Here, the condition was that a tree that was either detected either in 2018 or 2019 with a high confidence and was not detected in the three consecutive years 2020-2022 was classified as disappeared, which ensures that the tree had actually disappeared and did not simply fail to be detected. We developed a metric termed ‘change confidence’, which quantifies the certainty to which a given tree actually disappeared (see Methods). We focus exclusively on reporting tree losses accompanied by a high change confidence (>0.7). This approach acknowledges that the real figures may surpass those presented here, while intentionally omitting fluctuations of smaller tree and losses of trees characterized by considerable uncertainty in their reported numbers. We find that 5.3 million trees (2.7 trees per km²) observed in 2018/2019 were not detected in 2020-2022. Trees with a change confidence above this threshold on average have a crown size of 67 m². The uncertainty in the number of disappeared trees was 21%, quantified by manually checking a random sample of 1000 trees classified as disappeared. This number of disappeared trees, considered a conservative estimate due to the method applied, was still high considering that a majority of the losses must have occurred between 2018 and 2020. Several hot-spot areas stand out: The example shown in Figure 3b-d is not a local exception from the rule, as we observed similar situations all over Indian croplands, reflecting a considerable national scale thinning of India’s large agroforestry trees over such a short time period. Some regions have lost more than 50 trees per km².

A potential driver of tree losses is climate change, with rapid increases in temperature in central and southern India¹⁹, and unfavorable trends in rainfall and drought conditions over the past decade (Extended Data Fig. 3a-c). However, compared to the long term mean precipitation, the past decade was above average for most areas²⁰ (Extended Data Fig. 2d), and there is little evidence supporting climatic changes as the direct and main reason for the observed losses. We thus conducted qualitative interviews with villagers from the Telangana, Haryana, Kerala, Maharastra, Andrah Pradesh, Uttar Pradesh, Kashmir, and Jammu provinces, all affected by the disappearance of trees in our maps (Fig. 1). All participants in the interviews verified the considerable reduction in the populations of mature trees within and along fields. These trees were predominantly removed owing to alterations in the cultivation practices. The establishment of new boreholes, which resulted in an augmented water supply, facilitated the expansion of paddy rice fields with the objective of boosting crop yields. The decision to remove trees within fields is often driven by the perception that their benefits are relatively low, and concerns that their shading effect, in particular for trees with a very large crown such as Neem trees, may adversely affect crop yields. While all interviewees reported that native trees within fields have become rare, several report that block plantations have increased. Fungis and climate reasons on the other hand, were not perceived to have an impact on tree populations by the interviewees.

The outcome of this study is twofold: First, we provide a tool that enables the monitoring of every individual tree every year at sub-continental scale. The imagery employed are not free of charge, but the cost is order of magnitudes lower than conventional sub-meter imagery¹⁵. We developed a deep learning-based method that can be trained with simple point labels, which can be rapidly optimized for regional use^17,18,21-23. The database does not only provide a complete inventory of mature agroforestry trees, but also offers the opportunity to locate and quantify trees that have disappeared. The database includes detection confidence values for each year 2018-2022. If a tree that has been detected with a high confidence (above a given threshold) is not detected anymore over several consecutive years, there is a high probability that the tree has disappeared. By combining the confidence values of each year, a change confidence can be derived, quantifying the uncertainty of wether a tree was has disappeared or remained. In this study we refrained from making statements on low or medium confidence changes in order to take a conservative approach to the reporting of losses, but the dataset will be open for various applications. The database can be updated at an annual basis and scaled to global scale, potentially being a tool to support scientists, agroforestry practitioners, land managers, and policymakers¹⁰. Potential applications beyond inventorying are the monitoring of illegal logging of trees and the monitoring of survival rates of restoration areas.

Second, our study reveals a concerning trajectory, documenting a considerable depletion of India's large agroforestry trees over the past decade. This finding is particularly unsettling given the current emphasis on agroforestry as an essential natural climate solution, playing a crucial role in both climate change adaptation and mitigation strategies, as well as for livelihoods and biodiversity^24-28. A certain loss rate is natural, and the cutting of trees is also part of agroforestry management systems, and also not every lost tree is related to climatic disturbances or human appropriation, just as not every mapped tree contributes to livelihoods of local communities. Also, there are large areas in India where observed losses do not exceed expected dynamics. Nonetheless, an observable trend is emerging where established agroforestry systems are replaced with paddy rice fields, which are being expanded and intensified, a development facilitated by the availability of newly established water supplies. Large and mature trees within these fields are removed, and trees are now being cultivated within separate block plantations typically with lower ecological value.

At a first glance, this may contradict official reports and recent studies stating that tree cover in India has increased considerably in recent years^29-30. It is however important to note that we only report gross losses and did not consider tree gains as a separate class. We also masked out block plantations. We thus cannot provide information on net tree changes, and our results do not contradict reports that there has been a net increase in planted trees outside forests, as tree planting is encouraged and actively carried out in India. However, newly planted trees do not always contribute to biodiversity, and it can take a long time until they efficiently contribute to livelihoods. It is also unclear how newly planted trees survive in times of climate change. In light of the vast scale of global forest mortality caused by recent wildfires and droughts, the figures presented here may appear relatively modest, yet the substantial loss of particularly large agroforestry trees is disconcerting, as virtually every large tree that has been lost may impact on the ecosystem, and in the long-term the well-being and sustenance of communities.

This study generates country-wide custom mosaics for RapidEye for 2010 and 2011 (5-m resolution), and PlanetScope (3-m resolution) for 2018, 2019, 2020, 2021 and 2022. We trained deep learning models to detect individual non-forest trees for each year, and studied changes between years by tracking the tree crown center over the years. We calculated a change confidence for each tree, which is composed of a multi-year detection confidence to be able to quantify the uncertainty of the detection and of the change. The 2010/2011 maps detected mostly larger trees, but differences in image quality impeded exhaustive wall-to-wall mapping. The newer maps available after 2018 provide a more spatially consistent mapping of all agroforestry trees, however the mapping was designed to exclude dense plantations and small/young trees. We did not set a strict height or crown size threshold for the Planet data analysis, but a previous study has shown that trees below 4 m height and with a crown size below 10 m² are likely not to be detected¹⁵, and that trees with crown sizes above 20 m² can be mapped with a relatively high confidence. Hence, comparing the 2010 maps with the 2018-2022 maps gives an overview of the disappearance of large trees, while the tracking of trees during 2018-2022 includes all trees that have reached a certain crown size, excluding newly planted trees.

Monitoring trees at the level of individuals over time requires consistency in high quality images and reliable tree detection methods. While this would be easier to address by using sub-meter imagery, these data are not available as time series over large areas. Imagery from nano-satellites are available at 3-5 m resolution at a high temporal frequency, but the quality of the images captured from several hundred different satellites varies substantially. This concerns in particular the image sharpness, and smaller trees will not be visible in images that are blurred: a type of information which is not obvious from the scene metadata. Imagery at this spatial resolution is not suitable for pixel-wise stacking as image composites due to variations in the viewing angle. What is most important for the detection of trees is sharpness, and most other aspects, such as the number of spectral bands, can be neglected. Our solution to this issue was to filter each downloaded scene for sharpness using a blur kernel²³, and if a certain threshold was not reached, the image was blacklisted and the area was filled with different images until the threshold was passed.

The method used for the tree detection in this study diverges from previous deep learning based approaches^14,15. Studying changes at tree level at continental scale requires a highly accurate and robust detection method that works over a variety of satellite images and landscapes. Here, we developed a heatmap based detection method^17,18 that can be quickly trained with a large amount of point labels, and is also being able to separate adjacent tree crowns reliably.

Image preparation

The RapidEye constellation was operational from 2009 to 2019 and consisted of 5 nano-satellites. The spatial resolution is 5 m and the spectral bands are red, green, blue, red edge, and near infra-red. The PlanetScope constellation consists of 130+ nano-satellites with a resolution of 3-4 m and red, green, blue, and near infra-red bands. We formed custom 1x1 degree mosaic tiles, each consisting of about 30 scenes for RapidEye, and about 120 scenes (Dove Classic) or 60 scenes (Super Dove) for PlanetScope¹⁵. A histogram matching algorithm with Landsat and Sentinel-2 scenes from the same time period was applied for each tile to adjust differences in colors and form a homogeneous mosaic. The images were acquired following a phenological window, derived from the MODIS phenology product (MOD09): For areas with deciduous trees, determined by the WorldCover map⁴, images were acquired from the period between “senescence” and “mid-greendown”. This particular period was selected as it represents the period after herbaceous vegetation has passed its productivity peak, whereas trees still have green leaves, faciliating the detection of tree crowns. For areas with evergreen trees, images were acquired between the “mid-greendown” and the “dormancy”, which is a period were herbaceous vegetation are not productive, but trees have full green leaves. If too few images were available meeting this criteria, we progressively extended the length of the time window until a maximum of 60 days was reached. If too few images were still to be found, we loosened the filtering criteria, allowing progressively lower “visible-confidence-percent” values down to 60 to be included. We applied strict filtering criteria to use only images that are entirely free of clouds and have a low sun elevation (below 50), which facilitates the identification of trees by their shadow. The ground sample distance (GSD) of PlanetScope images varies between 3.1 and 4.7 m, and we only retained images that have a GSD close to 3.1 m, and never used imagery with a GSD above 4 m.

The identification of small trees is only possible in sharp images, and since there is a large variation in image sharpness between scenes which cannot be seen in the meta-data, we applied a blur kernel to estimate the sharpness of the scene directly after download. The blur kernel is described in ref.²³, and we followed their recommendation to set a threshold of 0.23. We disregarded all scenes below this threshold, and redownloaded the areas until the scene was classified as sharp. We only applied the blur kernel in images where the forest, shrubland, water, bare soil and wetland classes accounted for less than 50% of the scene coverage, as the calculation of the blurriness score was found to be unreliable over these land cover classes.

Since there is no reliable metadata on cloud cover in RapidEye images, we calculated the standard deviation for each downloaded scene, and disregarded all scenes where the blue band had a very high standard deviation, which is typically the case in cloudy images. This did not guarantee fully cloud-free images, but, together with the blur kernel, removed most of the contaminated scenes.

The entire framework was automated and the filling of a 1x1 degree tile required about 2-3 hours using a university connection, but can be performed in a parallelized way, so the downloading, filtering and processing of one year requires less than one week for all India. All processing was done locally, and the raw data for one year is about 2.5 terrabyte. As a final step, we applied a Contrast Limited Adaptive Histogram Equalization (CLAHE) to each mosaic and normalized the bands to values between 0-255.

Training data

We manually labeled trees with point labels at the tree crown center, using separate models for PlanetScope (about 130,000 labels) and RapidEye (about 100,000 labels), including labels from all years. For each labeled tree, we used high resolution images in Google Earth and Bing to verify if the trees existed. A large number of sub-meter images around 2010 were available from Bing maps and were used to verify labels from RapidEye, while Google Earth images were from recent years and used to verify labels from PlanetScope images. The labels were generated as in iterative process. In the first round, labels were generated over a variety of landscapes and image conditions (that is different viewing angles, sharpness levels, etc), being spatially well distributed across India. Then a first model was trained and trees were predicted over the study area. Subsequently, labels were added in areas where the performance was not satisfying, according to a visual inspection, and this process was repeated until the result was visually satisfying.

Trees are more challenging to map in RapidEye images at 5 m (6.5 m ground sample distance) as compared to PlanetScope data. They become clearly visible if they cover 4 pixels, which would translate to a crown size of >100 m². To reduce the number of false positives in our classififcation, we did not aim at a wall-to-wall mapping of trees in RapidEye images in 2010, but instead opted for a very low misclassification rate. Consequently, we only labeled clear examples, resulting in some heterogeneity in the tree maps, related to image quality, remaining cloud cover, sun elevation, and visibility of the trees, generating a sample set of trees from 2010 with a certain randomness. By adding a second year, 2011, the combined results became more spatially consistent and homogeneous.

Tree detection

To produce the most reliable assessment of tree densities, we adapted a detection approach previously used to count dense objects, such as persons in a crowd. Our method is inspired by ref.¹⁷ and ref.¹⁸ and uses Convolutional Neural Networks (CNNs) to produce a confidence map indicating the location of individual trees. The peak of the confidence map (“heatmap”) is assumed to be the center of a tree crown (Extended Data Fig. 2). The advantage of this method over previous methods^14,15 is that it can be trained by point labels, which enables a rapid generalization and adjustment over large areas and different scene quality levels.

Heatmap-based detections typically transform discrete point labels into continuous Gaussian kernels, more suited to smooth optimization. It irequires deciding on fixed kernel scale, which is not straightforward here given the high level of variability in tree sizes, image quality, and potential offsets between labeled and actual tree centers. Ref. ²¹ proposed a solution with adaptive kernels, where the model is given a degree of freedom through a scale map which resizes the Gaussian kernels on-the-fly during training. They also weigh the regression loss to focus on misestimated pixels, similarly to the Focal loss²². Here, we applied the same method, with two important modifications: we removed the linear approximation used originally and instead drew exact Gaussian kernels on-the-fly, and we only allowed scale factors > 1, with a minimum standard deviation of 4.5 m for PlanetScope and 5 m for RapidEye. We found that these adjustments lead to smoother heatmaps with less artifacts, which ultimately made it easier to tune the important hyperparameters of scale regularization and base Gaussian standard deviation (6 m for PlanetScope, 6.7 m for RapidEye).

We trained the models over 2500 epochs, with a learning rate of 1e-5 until epoch 2000, after which we used a linearly decreasing learning rate. The training process was monitored using 20% of the training data as validation data. We trained 5 models with different random splits of the training data (but always used 80% for training and 20% for validation). The F1 scores on the heatmaps for all models ranged between 0.44-0.47 for a close radius of 5 m, and 0.67-0.69 for a wider radius of 50 m, which is close to the values from ref. ¹⁷, although they used aerial images of higher spatial resolution. We then predicted on the mosaics using an ensemble over the 5 models and averaged the results of the heatmaps. The local maxima of the ensemble heatmap were converted to a point file, reflecting the centers of tree crowns. The peak confidence at the crown center was saved as an attribute value associated with each centroid, reflecting the detection confidence. All peaks with a confidence above 0.35 were included in the following analyses. If the confidence was below 0.35, we considered this as no tree being detected.

A previously published tree crown segmentation model^14,15 was updated with 52,000 manually drawn crown labels from India, and a tree crown segmentation was conducted for the year 2021. Although this method is less reliable than the heatmap based detection, it provides an overview of the crown size distribution of the trees. Results showed that trees with a crown size below 20 m² were under-detected (Extended Data Table 1). We further located 22.7 million large trees with a crown size >100 m², which matches with the number of high confidence trees detected by RapidEye in 2010/2011 (22 million; confidence >0.8). Note that trees with a crown size >50 m² or 100 m² represent only a small fraction of the woody populations, with a proportion of about 20% and 4%, respectively (Extended Data Table 1).

We sampled trees from the heatmap predictions for the same area from both PlanetScope (n = 2.1 million) and RapidEye (n = 1.2 million) for 2019 (where RapidEye was still operational). We used the tree crown segmentation to determine the tree crown sizes of the point predictions from both PlanetScope and RapidEye, by overlaying the tree crown segments with the point predictions and associating the crown area as an attribute to the points. We found that the average tree crown size of the samples was 55 m² from PlanetScope and 62 m² from RapidEye. Trees detected with a confidence >0.8 had an average crown size of 96 m²for RapidEye. For PlanetScope, trees with a confidence >0.7 had a crown size of 67 m² and <0.7 the crown size was on average 59 m². This gives evidence that a higher confidence value also indicates a larger crown size.

Mapping changes and change confidence

This study focuses on tree losses rather than gains, as new trees are typically not growing large over a few years, and consequently they would not be detected in a reliable way. Moreover, the 2010 classification did not allow for wall-to-wall mapping.

We developed a framework that via crown center detection can track individual trees over an arbitrary number of temporal steps. We defined a circular buffer area of 15 m around the center of a detected tree and searched for centroid equivalents over the following images/years. If no tree was detected within the buffer area in the later years, the tree was counted a loss. If several trees were detected within the buffer, the closest tree was chosen.

For the long term comparison between the early and late epoch, we combined the results from RapidEye (2010 and 2011) to reflect the early period, and also PlanetScope (2018-2022) to reflect the later period. Tree crown centers found in the early period were then compared with the later period, and a tree was either classified as disappeared, which means it was only observed in 2010/2011 but not during the later period, or as remaining, which means the tree was observed in both periods.

For the tracking of trees over the period 2018-2022, we developed a metric we named change confidence, which quantifies the confidence of our tree detection into the following three classes: disappeared, tree, and low confidence tree/misclassification. A tree was marked as low confidence tree if it was only detected in one year with a low or medium confidence. The detection may be due to a misclassification, or if the tree is too small to be reliably monitored over several years and was thus only detected in one year, so it was not included in the reported statistics. A centroid was marked as “tree”, if a tree was detected in one year with a very high confidence, or several years with a low or medium confidence. We marked a tree as disappeared if it was detected in either 2018 or 2019 with a high confidence, or in both years with a medium confidence, and then was not detected in three consecutive years (2020, 2021, and 2022). Mapping of disappearing trees and also remaining trees with a low change confidence (<0.7) should be treated with caution. For example, if a tree mapped during 2018-2022 was only mapped in 2018 with a confidence of 0.5, it is likely that the tree is small or a false detection. If the tree is mapped only in 2018 with a confidence of 1.0, it is likely that the tree was then lost, as it was not mapped during 2019-2022. This assumes that a tree with a high confidence is a large tree that should be detected at least twice over 5 years. Contrastingly, if a tree was mapped with a lower confidence of 0.5 in 2018, but also in 2021 and 2022, the class is “tree” and the overall change confidence is higher, as the tree was detected in 3 years, which reduces the risk of a misclassification.

Sources of uncertainties

Our change confidence metric quantifies uncertainty at tree level. Only focusing on trees with a high change confidence reduces the uncertainty of the change instances reported, but also misses a number of cases of change that are real - but where the confidence is low due to image quality issues or the tree has a small crown. There are however, a number of remaining sources of uncertainty that are challenging to quantify:

A tree was only counted as disappeared during 2018-2022 if the combined confidence of 2018 and 2019 was above 0.7, and the tree was not detected between 2020-2022. While this gives a certain robustness, we observed cases where the image quality was excellent in both 2018 and 2019 resulting in very high confidence predictions, but the image quality was less ideal in all subsequent years 2020-2022, so some trees were missed in these three years, causing trees to be falsely classified as being disappeared. To account for this, we assumed that high quality images in general generate higher confidence predictions compared to lower quality images, so we calculated the average detection confidence of all trees within 1x1 km cells as a measure of image quality for each particular cell, which also accounts for quality variations within images. We then calculated a linear slope through the 1x1 km confidence grids, with time as the independent variable. A strong negative slope (< -0.05) implies that the average confidence in 2020-2022 was considerably lower than in 2018-2019, so we flagged this area as uncertain, and losses were not included in our reported numbers. The confidence slope map represents an additional layer (to the change confidence) that quantifies uncertainty that we provide with the database.

Some RapidEye scenes are spatially shifted, which leads to erroneous classifications of tree losses. When comparing the 2010/2011 results with the 2018-2022 results, we calculated the proportion of trees disappearing for each RapidEye scene footprint. If the proportion was above 40%, scenes were likely shifted and we masked them. We also manually masked scenes that were clearly shifted, visible by sharp edges along footprints.

A final source of uncertainty that cannot be quantified is the land cover map included in our analysis. We used the WorldCover map from 2020 to retain only croplands for the 2010-2018 comparison, and cropland, urban and bare for the 2018-2022 comparison. The quality of the landcover map impacts on the results, and it happens that large farmland trees or groups of trees are masked out as forest, or that the underlying classification is not correct. Future versions may include custom landcover maps which may lead to improved results. Using a landcover map from 2020 to study changes over 10 years can include areas that have been cleared for cropping and have not previously been farmland before, or were under fallow, or plantation forest. However, we observed that these areas rarely included large trees, but rather shrubs and small trees, and by reporting mainly the loss of high confidence trees, these are automatically excluded.

Evaluation of the tree detections and changes

We randomly selected 1000 points mapped as disappeared trees between 2010/2011 and 2018-2022 to evaluate the uncertainty on the reported long-term losses. The same 1000 points were used to evaluate if trees were correctly mapped in 2010/2011 or if it was a false detection. We further selected 1000 random trees that had been mapped as disappeared trees over 2018-2022 to evaluate the uncertainty related to disappeared trees over the PlanetScope period. Each point was manually and visually checked on the images and in Google Earth, also using the historic images where available. For the losses, we only considered high confidence changes, with values above 0.8 for RapidEye and 0.7 for PlanetScope. Uncertainties would be higher if different confidence thresholds are used. False detections of trees for 2010/2011 were found to be 3%. False losses for 2010-2018 were 2%. False losses for 2018-2022 were 21%.

We further conducted twelve qualitative interviews with villagers in the Telangana, Haryana, Kerala, Maharastra, Andrah Pradesh, Uttar Pradesh, Kashmir, and Jammu provinces during August 2023. The interviews were about 20-60 minutes each, and we asked about soils, management systems, water resources, changes in the number of trees, and reasons for possible changes. Participants were on average 59 years old.

Code availability. The tree detection framework is available from https://arxiv.org/abs/2208.10607. The crown segmentation framework is available from https://doi.org/10.5281/zenodo.3978185. A framework including both segmentation and counting can also be found at https://github.com/sizhuoli/TreeCountSegHeight. The PlanetScope mosaic generation is available from https://doi.org/10.5281/zenodo.7764360.

Data availability. The data will be made available upon publication of the manuscript.

Acknowledgements. MB, FR, WZ and XT are supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 947757 TOFDRY). MB also acknowledges funding from a DFF Sapere Aude grant (no. 9064–00049B). AK, RF acknowledge support by the Villum Fonden through the project Deep Learning and Remote Sensing for Unlocking Global Ecosystem Resource Dynamics (DeReEco).

Author Contributions. MB designed the study, prepared the data, sampled the training data, trained the models, conducted the analyses, designed the figures, and wrote the manuscript. FR wrote the codes for the data preparation and data analyses, supported by XT. DiG wrote the codes for the tree detection framework. The importance of the contributions of MB, FR and DiG can be considered as equal. AK and VG conducted the interviews. WZ analysed the climate data. PC, RF and DaG, DhG reviewed the manuscript.

Competing financial interest. The authors declare no competing financial interests.

Allen, C. D. et al. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecology and Management 259, 660–684 (2010).
Senf, C., Buras, A., Zang, C. S., Rammig, A. & Seidl, R. Excess forest mortality is consistently linked to drought across Europe. Nat Commun 11, 6200 (2020).
Hammond, W. M. et al. Global field observations of tree die-off reveal hotter-drought fingerprint for Earth’s forests. Nat Commun 13, 1761 (2022).
Zanaga, D. et al. ESA WorldCover 10 m 2020 v100. Preprint at https://doi.org/10.5281/zenodo.5571936 (2021).
Taylor, J. R. & Lovell, S. T. Designing multifunctional urban agroforestry with people in mind. Urban Agric & Regnl Food Sys 6, (2021).
Chinnamani, S. Agroforestry research in India: a brief review. Agroforest Syst 23, 253–259 (1993).
Basu, J. P. Agroforestry, climate change mitigation and livelihood security in India. New Zealand Journal of Forestry Science 44, S11 (2014).
S, D. & B. Teli, S. Millet-Based Agroforestry: A Nature-Positive Farming to Achieve Climate-Resilience and Food Security in India and Africa. SSRN Scholarly Paper at https://doi.org/10.2139/ssrn.4461034 (2023).
Johar, V. Socio-economic and Environmental Assets Sustainability by Agroforestry Systems: A Review. IJAEB 14, (2021).
Rizvi, R. H. et al. Mapping Agroforestry and Trees Outside Forest. World Agroforestry | Transforming Lives and Landscapes with Trees https://www.worldagroforestry.org/publication/mapping-agroforestry-and-trees-outside-forest (2020).
Islas, J. F. et al. An overview of Neem (Azadirachta indica) and its potential impact on health. Journal of Functional Foods 74, 104171 (2020).
Bhat, S. S., Sateesh, M. K. & Devaki, N. S. A new destructive disease of neem (Azadirachta indica) incited by Phomopsis azadirachtae. Current Science 74, 17–19 (1998).
Nagendra Prasad, M. N., Shankara Bhat, S., Charith Raj, A. P. & Janardhana, G. R. Detection of Phomopsis azadirachtae from dieback affected neem twigs, seeds, embryo by polymerase chain reaction. Archives Of Phytopathology And Plant Protection 42, 124–128 (2009).
Brandt, M. et al. An unexpectedly large count of trees in the West African Sahara and Sahel. Nature 587, 78–82 (2020).
Reiner, F. et al. More than one quarter of Africa’s tree cover is found outside areas previously classified as forest. Nat Commun 14, 2258 (2023).
Spiekermann, R., Brandt, M. & Samimi, C. Woody vegetation and land cover changes in the Sahel of Mali (1967–2011). International Journal of Applied Earth Observation and Geoinformation 34, 113–121 (2015).
Ventura, J. et al. Individual Tree Detection in Large-Scale Urban Environments using High-Resolution Multispectral Imagery. Preprint at https://doi.org/10.48550/arXiv.2208.10607 (2022).
Li, S. et al. Deep learning enables image-based tree counting, crown segmentation, and height prediction at national scale. PNAS Nexus 2, pgad076 (2023).
Mishra, V. & Aadhar, S. Famines and likelihood of consecutive megadroughts in India. npj Clim Atmos Sci 4, 1–12 (2021).
Jin, Q. & Wang, C. A revival of Indian summer monsoon rainfall since 2002. Nature Clim Change 7, 587–594 (2017).
Luo, Z. et al. Rethinking the Heatmap Regression for Bottom-up Human Pose Estimation. in 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 13259–13268 (2021). doi:10.1109/CVPR46437.2021.01306.
Lin, T.-Y., Goyal, P., Girshick, R., He, K. & Dollár, P. Focal Loss for Dense Object Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence 42, 318–327 (2020).
Anger, J., de Franchis, C. & Facciolo, G. Assessing the Sharpness of Satellite Images: Study of the Planetscope Constellation. in IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium 389–392 (2019). doi:10.1109/IGARSS.2019.8900526.
Singh, P. et al. Agroforestry improves food security and reduces income variability in semi-arid tropics of central India. Agroforest Syst (2023) doi:10.1007/s10457-023-00806-6.
Telwala, Y. Unlocking the potential of agroforestry as a nature-based solution for localizing sustainable development goals: A case study from a drought-prone region in rural India. Nature-Based Solutions 3, 100045 (2023).
Karlson, M. et al. Exploring the landscape scale influences of tree cover on crop yield in an agroforestry parkland using satellite data and spatial statistics. Journal of Arid Environments 218, 105051 (2023).
Skole, D. L., Mbow, C., Mugabowindekwe, M., Brandt, M. S. & Samek, J. H. Trees outside of forests as natural climate solutions. Nat. Clim. Chang. 11, 1013–1016 (2021).
Datta, P. & Behera, B. Climate Change Adaptation Through Agroforestry: Empirical Evidence from Indian Eastern Himalayan Foothills. in Strategizing Agricultural Management for Climate Change Mitigation and Adaptation (ed.
Fagan, M. E. et al. The expansion of tree plantations across tropical biomes. Nat Sustain 5, 681–688 (2022).
Du, Z. et al. A global map of planting years of plantations. Sci Data 9, 141 (2022).

There is NO Competing Interest.

ExtendedData.docx

Download PDF

Journal Publication

published 15 May, 2024

Read the published version in Nature Sustainability →

Version 1

posted

You are reading this latest preprint version

Severe decline in large agroforestry trees in India over the past decade

Status:

Journal Publication

Version 1

Abstract

Figures

Main

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1