Seed and Seedling Detection Using Unmanned Aerial Vehicles and Automated Image Classification in the Monitoring of Ecological Recovery

Eco-friendly structures for sustainable mangrove restoration

Mangrove forests around the world are under significant pressure from climate change (e.g., rising sea levels), and human-related anthropogenic activities (e.g., coastal infrastructure development). Mangrove restoration projects have increased over the past decades but seedling and propagule survival rates are reportedly low, while many projects have failed. There exists a need to assess the effectiveness of sustainable and cost-effective eco-friendly structures (EFS) for advancing the success of mangrove restoration and planting activities. Herein, by EFS, we refer to the frameworks made of biodegradable materials that help overcome establishment bottlenecks and thereby boost seedling survival and growth rates. In this study, we explored the effectiveness of EFS in aiding mangrove restoration success by enhancing seedling establishment and survival and tree growth rates. Furthermore, we examine the steps involved and the challenges limiting EFS implementation in mangrove restoration projects. EFS installed in coastal areas trap sediment and may provide protection for newly planted mangrove seedlings and propagules by providing a stable anchorage and attenuating water flow and waves. Additionally, once plants are established, these biodegradable structures would decompose and add to the soil nutrients stock, thereby improving its fertility and supporting mangrove growth. We emphasize that in sites with favorable biophysical conditions for mangrove growth (hydrology, soil, topography, climate, among others), using EFS can improve mangrove restoration success by enhancing seedling establishment, survival and growth. Mangrove restoration success may have add-on benefits such as increasing the provision of related ecosystem services, blue carbon credit financing and overall coastal environmental sustainability. Given the novelness of this topic in the scientific literature, this article aims to stimulate active discussions, including anticipation of potential challenges (e.g., cost-effectiveness, ability to scale and field limitations in a range of biogeographic settings), for bringing in improvements and scalable adoption strategies to the mangrove restoration approaches under consideration.

Plant Detection in RGB Images from Unmanned Aerial Vehicles Using Segmentation by Deep Learning and an Impact of Model Accuracy on Downstream Analysis

Crop field monitoring using unmanned aerial vehicles (UAVs) is one of the most important technologies for plant growth control in modern precision agriculture. One of the important and widely used tasks in field monitoring is plant stand counting. The accurate identification of plants in field images provides estimates of plant number per unit area, detects missing seedlings, and predicts crop yield. Current methods are based on the detection of plants in images obtained from UAVs by means of computer vision algorithms and deep learning neural networks. These approaches depend on image spatial resolution and the quality of plant markup. The performance of automatic plant detection may affect the efficiency of downstream analysis of a field cropping pattern. In the present work, a method is presented for detecting the plants of five species in images acquired via a UAV on the basis of image segmentation by deep learning algorithms (convolutional neural networks). Twelve orthomosaics were collected and marked at several sites in Russia to train and test the neural network algorithms. Additionally, 17 existing datasets of various spatial resolutions and markup quality levels from the Roboflow service were used to extend training image sets. Finally, we compared several texture features between manually evaluated and neural-network-estimated plant masks. It was demonstrated that adding images to the training sample (even those of lower resolution and markup quality) improves plant stand counting significantly. The work indicates how the accuracy of plant detection in field images may affect their cropping pattern evaluation by means of texture characteristics. For some of the characteristics (GLCM mean, GLRM long run, GLRM run ratio) the estimates between images marked manually and automatically are close. For others, the differences are large and may lead to erroneous conclusions about the properties of field cropping patterns. Nonetheless, overall, plant detection algorithms with a higher accuracy show better agreement with the estimates of texture parameters obtained from manually marked images.

Automated mapping of Portulacaria afra canopies for restoration monitoring with convolutional neural networks and heterogeneous unmanned aerial vehicle imagery

Ecosystem restoration and reforestation often operate at large scales, whereas monitoring practices are usually limited to spatially restricted field measurements that are (i) time- and labour-intensive, and (ii) unable to accurately quantify restoration success over hundreds to thousands of hectares. Recent advances in remote sensing technologies paired with deep learning algorithms provide an unprecedented opportunity for monitoring changes in vegetation cover at spatial and temporal scales. Such data can feed directly into adaptive management practices and provide insights into restoration and regeneration dynamics. Here, we demonstrate that convolutional neural network (CNN) segmentation algorithms can accurately classify the canopy cover of Portulacaria afra Jacq. in imagery acquired using different models of unoccupied aerial vehicles (UAVs) and under variable light intensities. Portulacaria afra is the target species for the restoration of Albany Subtropical Thicket vegetation, endemic to South Africa, where canopy cover is challenging to measure due to the dense, tangled structure of this vegetation. The automated classification strategy presented here is widely transferable to restoration monitoring as its application does not require any knowledge of the CNN model or specialist training, and can be applied to imagery generated by a range of UAV models. This will reduce the sampling effort required to track restoration trajectories in space and time, contributing to more effective management of restoration sites, and promoting collaboration between scientists, practitioners and landowners.

Different responses of growth and physiology to warming and reduced precipitation of two co-existing seedlings in a temperate secondary forest

Warming and precipitation reduction have been concurrent throughout this century in most temperate regions (e.g., Northeast China) and have increased drought risk to the growth, migration, or mortality of tree seedlings. Coexisting tree species with different functional traits in temperate forests may have inconsistent responses to both warming and decreased precipitation, which could result in a species distribution shift and change in community dynamics. Unfortunately, little is known about the growth and physiological responses of coexisting species to the changes in these two meteorological elements. We selected two coexisting species in a temperate secondary forest of Northeast China: Quercus mongolica Fischer ex Ledebour (drought-tolerant species) and Fraxinus mandschurica Rupr. (drought-intolerant species), and performed an experiment under strictly controlled conditions simulating the predicted warming (+2°C, +4°C) and precipitation reduction (-30%) compared with current conditions and analyzed the growth and physiology of seedlings. The results showed that compared with the control, warming (including +2°C and +4°C) increased the specific area weight and total biomass of F. mandschurica seedlings. These were caused by the increases in foliar N content, the activity of the PSII reaction center, and chlorophyll content. A 2°C increase in temperature and reduced precipitation enhanced root biomass of Q. mongolica, resulting from root length increase. To absorb water in drier soil, seedlings of both species had more negative water potential under the interaction between +4°C and precipitation reduction. Our results demonstrate that drought-tolerant species such as Q. mongolica will adapt to the future drier conditions with the co-occurrence of warming and precipitation reduction, while drought-intolerant species will accommodate warmer environments.

Machine Learning Regression Analysis for Estimation of Crop Emergence Using Multispectral UAV Imagery

Optimal crop emergence is an important trait in crop breeding for genotypic screening and for achieving potential growth and yield. Emergence is conventionally quantified manually by counting the sub-sections of field plots or scoring; these are less reliable, laborious and inefficient. Remote sensing technology is being increasingly used for high-throughput estimation of agronomic traits in field crops. This study developed a method for estimating wheat seedlings using multispectral images captured from an unmanned aerial vehicle. A machine learning regression (MLR) analysis was used by combining spectral and morphological information extracted from the multispectral images. The approach was tested on diverse wheat genotypes varying in seedling emergence. In this study, three supervised MLR models including regression trees, support vector regression and Gaussian process regression (GPR) were evaluated for estimating wheat seedling emergence. The GPR model was the most effective compared to the other methods, with R2 = 0.86, RMSE = 4.07 and MAE = 3.21 when correlated to the manual seedling count. In addition, imagery data collected at multiple flight altitudes and different wheat growth stages suggested that 10 m altitude and 20 days after sowing were desirable for optimal spatial resolution and image analysis. The method is deployable on larger field trials and other crops for effective and reliable seedling emergence estimates.

UAV-Supported Forest Regeneration: Current Trends, Challenges and Implications

Replanting trees helps with avoiding desertification, reducing the chances of soil erosion and flooding, minimizing the risks of zoonotic disease outbreaks, and providing ecosystem services and livelihood to the indigenous people, in addition to sequestering carbon dioxide for mitigating climate change. Consequently, it is important to explore new methods and technologies that are aiming to upscale and fast-track afforestation and reforestation (A/R) endeavors, given that many of the current tree planting strategies are not cost effective over large landscapes, and suffer from constraints associated with time, energy, manpower, and nursery-based seedling production. UAV (unmanned aerial vehicle)-supported seed sowing (UAVsSS) can promote rapid A/R in a safe, cost-effective, fast and environmentally friendly manner, if performed correctly, even in otherwise unsafe and/or inaccessible terrains, supplementing the overall manual planting efforts globally. In this study, we reviewed the recent literature on UAVsSS, to analyze the current status of the technology. Primary UAVsSS applications were found to be in areas of post-wildfire reforestation, mangrove restoration, forest restoration after degradation, weed eradication, and desert greening. Nonetheless, low survival rates of the seeds, future forest diversity, weather limitations, financial constraints, and seed-firing accuracy concerns were determined as major challenges to operationalization. Based on our literature survey and qualitative analysis, twelve recommendations—ranging from the need for publishing germination results to linking UAVsSS operations with carbon offset markets—are provided for the advancement of UAVsSS applications.

Incorporating rock in surface covers improves the establishment of native pioneer vegetation on alkaline mine tailings

BACKGROUND AND AIMS

Rates of tailings production and deposition around the world have increased markedly in recent decades, and have grown asynchronously with safe and environmentally suitable solutions for their storage. Tailings are often produced in regions harbouring biodiverse native plant communities adapted to old, highly-weathered soils. The highly-altered edaphic conditions of tailings compared with natural soils in these areas will likely select against many locally endemic plant species, making phytostabilisation, rehabilitation or ecological restoration of these landforms challenging.

METHODS

We established four substrate cover composition treatments on a dry-stacked magnetite tailings storage facility in semi-arid Western Australia, representative of standard industry practices for rehabilitating or restoring post-mining landforms in the region. Plots were seeded with a selection of locally native plant species and monitored for five years to determine whether different substrate cover treatments yielded different edaphic conditions (soil moisture, substrate surface temperature and substrate chemistry) and influenced soil development and the success of native vegetation establishment.

RESULTS

No vegetation established from seeds on unamended tailings with no surface cover, and substrate chemistry changed minimally over five years. In contrast, rock-containing surface covers allowed establishment of up to 11 native plant species from broadcast seeds at densities of ca. 1.5 seedlings m^-2, and up to 3.5 seedlings m^-2 of five native pioneer chenopods from capture of wind-dispersed seeds from surrounding undisturbed native vegetation. Greater vegetation establishment in rock-containing surface covers resulted from increased heterogeneity (e.g., lower maximum soil temperature, greater water capture and retention, surface microtopography facilitating seed capture and retention, more niches for seed germination). Soil development and bio-weathering occurred most rapidly under the canopy of native pioneer plants on rock-containing surface covers, particularly increases in organic carbon, total nitrogen, and organo-bound aluminium and iron.

CONCLUSIONS

Seed germination and seedling survival on tailings were limited by extreme thermal and hydrological conditions and a highly-altered biogeochemical environment. The design of surface cover layers appears crucial to achieving closure outcomes on tailings landforms, and designs should prioritise increasing surface heterogeneity through the incorporation of rock or other structure-improving amendments to assist the establishment of pioneer vegetation.

A geostatistical fusion approach using UAV data for probabilistic estimation of Xylella fastidiosa subsp. pauca infection in olive trees

Xylella fastidiosa is one of the most destructive plant pathogenic bacteria worldwide, affecting more than 500 plant species. In Apulia region (southeastern Italy), X. fastidiosa subsp. pauca (Xfp) is responsible for a severe disease, the olive quick decline syndrome (OQDS), spreading epidemically and with dramatic impact on the agriculture, the landscape, the tourism, and the cultural heritage of this region. An early detection of the infected plants would hinder the rapid spread of the disease. The main objective of this paper was to define a geostatistical approach of data fusion, which combines remote (radiometric), and proximal (geophysical) sensor data and visual inspections with plant diagnostic tests, to provide probabilistic maps of Xfp infection risk. The study site was an olive grove located at Oria (province of Brindisi, Italy), where at the time of monitoring (September 2017) only few plants showed initial symptoms of the disease. The measurements included: 1) acquisitions of reflected electromagnetic radiation with UAV (Unmanned Aerial Vehicle) equipped with a multi-spectral camera; 2) geophysical surveys on the trunks of 49 plants with Ground Penetrating Radar (GPR); 3) disease severity rating, by visual inspection of the proportion of canopy with symptoms; 4) qPCR (real time-quantitative Polymerase Chain Reaction) data from tests on 61 plants. The data were submitted to a set of processing techniques to define a "data fusion" procedure, based on non-parametric multivariate geostatistics. The approach allowed marking those areas where the risk of infection was higher, and identifying the possible infection entry routes into the field. The probability map of infection risk could be used as an effective tool for a preventive action and for a better organization of the monitoring plans.

Grammar Guided Genetic Programming for Network Architecture Search and Road Detection on Aerial Orthophotography

Photogrammetry involves aerial photography of the Earth’s surface and subsequently processing the images to provide a more accurate depiction of the area (Orthophotography). It is used by the Spanish Instituto Geográfico Nacional to update road cartography but requires a significant amount of manual labor due to the need to perform visual inspection of all tiled images. Deep learning techniques (artificial neural networks with more than one hidden layer) can perform road detection but it is still unclear how to find the optimal network architecture. Our main goal is the automatic design of deep neural network architectures with grammar-guided genetic programming. In this kind of evolutive algorithm, all the population individuals (here candidate network architectures) are constrained to rules specified by a grammar that defines valid and useful structural patterns to guide the search process. Grammar used includes well-known complex structures (e.g., Inception-like modules) combined with a custom designed mutation operator (dynamically links the mutation probability to structural diversity). Pilot results show that the system is able to design models for road detection that obtain test accuracies similar to that reached by state-of-the-art models when evaluated over a dataset from the Spanish National Aerial Orthophotography Plan.

Evaluation of Cotton Emergence Using UAV-Based Narrow-Band Spectral Imagery with Customized Image Alignment and Stitching Algorithms

Crop stand count and uniformity are important measures for making proper field management decisions to improve crop production. Conventional methods for evaluating stand count based on visual observation are time consuming and labor intensive, making it difficult to adequately cover a large field. The overall goal of this study was to evaluate cotton emergence at two weeks after planting using unmanned aerial vehicle (UAV)-based high-resolution narrow-band spectral indices that were collected using a pushbroom hyperspectral imager flying at 50 m above ground. A customized image alignment and stitching algorithm was developed to process hyperspectral cubes efficiently and build panoramas for each narrow band. The normalized difference vegetation index (NDVI) was calculated to segment cotton seedlings from soil background. A Hough transform was used for crop row identification and weed removal. Individual seedlings were identified based on customized geometric features and used to calculate stand count. Results show that the developed alignment and stitching algorithm had an average alignment error of 2.8 pixels, which was much smaller than that of 181 pixels from the associated commercial software. The system was able to count the number of seedlings in seedling clusters with an accuracy of 84.1%. Mean absolute percentage error (MAPE) in estimation of crop density at the meter level was 9.0%. For seedling uniformity evaluation, the MAPE of seedling spacing was 9.1% and seedling spacing standard deviation was 6.8%. Results showed that UAV-based high-resolution narrow-band spectral images had the potential to evaluate cotton emergence.

Multi-Sensor UAV Tracking of Individual Seedlings and Seedling Communities at Millimetre Accuracy

The increasing spatial and temporal scales of ecological recovery projects demand more rapid and accurate methods of predicting restoration trajectory. Unmanned aerial vehicles (UAVs) offer greatly improved rapidity and efficiency compared to traditional biodiversity monitoring surveys and are increasingly employed in the monitoring of ecological restoration. However, the applicability of UAV-based remote sensing in the identification of small features of interest from captured imagery (e.g., small individual plants, <100 cm2) remains untested and the potential of UAVs to track the performance of individual plants or the development of seedlings remains unexplored. This study utilised low-altitude UAV imagery from multi-sensor flights (Red-Green-Blue and multispectral sensors) and an automated object-based image analysis software to detect target seedlings from among a matrix of non-target grasses in order to track the performance of individual target seedlings and the seedling community over a 14-week period. Object-based Image Analysis (OBIA) classification effectively and accurately discriminated among target and non-target seedling objects and these groups exhibited distinct spectral signatures (six different visible-spectrum and multispectral indices) that responded differently over a 24-day drying period. OBIA classification from captured imagery also allowed for the accurate tracking of individual target seedling objects through time, clearly illustrating the capacity of UAV-based monitoring to undertake plant performance monitoring of individual plants at very fine spatial scales.