Dottorato in ENERGIA E AMBIENTE

Titolo della tesi: Tree Pests monitoring: Assessment of Remote Sensing technologies towards a worldwide perspective

According to FAO, in the period 1980-2002 more than 52 million hectares of forest in 37 countries were damaged by pests. Tree pests can cause rapid and widespread damage, reducing the economic value of plants, and in the case of fruit trees their production and, also, their role in mitigating climate change. Europe has roughly 43% of its land area covered with forest, amounting to close to 182 million hectares (Eurostat, 2018). Nowadays, several directives have been issued by EFSA (European Food Safety Authority) focused on the knowledge gaps and research priorities to address tree pests monitoring, as: • Delimitation of areas at risk, • Information about vectors, surveillance and detection, • Identification of areas of high disease risk where to target the surveys, • Standardized surveillance and detection methodologies and reporting procedures, • Protocols and guidelines for sharing real-time data, development of shared databases, • Early detection, monitoring and delimiting surveys after outbreaks. This study is part of the AMEOS (Assimilating Multi-source Earth Observation Satellite data for crop pests and diseases monitoring and forecasting) initiative, sponsored by an ESA-Dragon agreement, which aims to bring together cutting edge research to provide pest and disease monitoring and forecast information, integrating multi-source information to support decision making in the sustainable management of insect pests and diseases in agriculture. Project aims to improve pests monitoring and forecasting, by utilizing Earth Observation (EO) data, developing new algorithms, combining new and existing data from multi-source EO sensors to produce high spatial and temporal land surface information. The project focuses on the analysis of pests and diseases caused by the Pine Tree Nematode (PWN), the Pseudomonas syringae pv. Actinidiae (Psa), the Xylella fastidiosa (Xf), and the Halyomorpha halys (BMSB). The main objective of this work is the definition of a methodology to carry out a continuous monitoring of potential anomalies or symptoms cause by pests on trees and also their vectors and their climatic favorable conditions, via remote sensing technologies and models enriched with EO-based data. The work to achieve this goal has been divided according to the steps described below: • Analysis of state-of-art with reference to the pest and diseases of interest, vectors, hosts, global distribution, and current research projects. • Analysis and development of automatic feature extraction procedures capable to identify the objects of interest with available satellite and UAV imagery. • Analysis of the possibilities offered by available satellite imagery to detect orchards/trees growth anomalies. • Elaboration of risks maps of potential vector distributions and identification of climatic variables that could promote the spread of the pests via the application of species distribution models. • Identification of potential climatic risk zones for the appearance of new pest outbreaks, and environmental thresholds of invasion and persistence of the pathogens under different climate change scenarios. • Definition of a technological architecture for the manipulation of intermediate datasets and dissemination of results. As it was mentioned the current study is based on the analysis of different pests as the Pine Wood Nematode (PWN). The Pine wilt disease (PWD), caused by the pinewood nematode (PWN) Bursaphelenchus xylophilus, is one of the most serious biological invasions and damaging diseases that has affected conifer forests worldwide. Regarding Europe, PWN was introduced in Portugal in 1999 (Mota et al. 1999), has recently spread to Madeira island (Fonseca et al. 2012) and Spain (Robertson et al. 2011) and is listed as a quarantine pest in Europe (Ribeiro et at. 2012). And is estimated to have the potential to spread to 34% of Europe by 2030. EO capabilities to detect the Pseudomonas syringae pv. Actinidiae (Psa) was analyzed on kiwi trees, this disease has the potential to cause considerable production losses. In 2010, the New Zealand kiwifruit industry was hit by Psa, and the cost for this industry has been estimated to be approximately $126 million in 2012 with an on-going cost for the next 15 years of between $740 to $885 million (Greer & Saunders 2012). Following, the characteristics and possibilities to early detect the Xylella fastidiosa (Xf) was studied. This is a vector-transmitted bacterial plant pathogen associated with serious diseases in a wide range of plants. A significant difficulty that needs to be taken into account regarding this pest comes from its very wide host range (in September 2018 the list includes more than 500 plant species belonging to 82 families) (EFSA 2018), and since infections that do not cause symptoms in some host–strain combinations, despite the infected hosts continuing to act as inoculum sources. And finally, the capabilities to monitor the spread and impact of the Halyomorpha halys (or brown marmorated stink bug (BMSB)) was analyzed. Its damage is varied and may include scarring, depressions and internal tissue damage, premature flower and fruit drops, among others. In Italy was first detected in 2012, and it can result in significant economic loss. Farmers in northern Italy are also facing the problems of this pest that could wipe out fruit and vegetable crops. European Commission approved emergency support in 2020 to help farmers in six northern Italian regions. This pest caused an estimated €500 million of damage in 2019. Remote sensing is the science and art of obtaining information about an object, area, or phenomenon of interest through the analysis of data acquired by a device that is not in contact with the object, area, or phenomenon under investigation (Lillesand, 2004). Plants can be affected by abiotic stresses like shortage of water or nutrients; or by biotic stresses like insects, fungi, bacteria, or viruses. In many cases, the exact cause of the disease can only be discerned through destructive sampling, but knowing the symptoms in an environmental, spatial, and historical context can help pinpoint the cause (Hueni et al., 2018). In this context, multi-temporal acquisition of remote sensing data facilitates the possibilities to monitor the health status of the vegetation evaluating the impacts of any perturbations which affects the expected development. Ranging from in situ point measurements to drone, airborne, and satellite-based measurements of entire regional areas, remotely sensed data have been widely used to inform spatially implicit management decisions concerning agricultural production (Ashcroft et al., 2011; Hueni et al., 2018). In this study different available remote sensed imagery, as Sentinel, Rapideye, Prisma, Planetscope and also images captured by drone were analyzed. Depending on the spatial and spectral resolution of these datasets, these technologies were applied for different purposes as for tree detection and counting as well as for anomalies detection. The organization of the current study is based on three main sections: • Initially the analysis includes references to the pests under study, as: Xylella fastidiosa (Xf), the Pine Wood Nematode (PWN), the Pseudomonas syringae pv. Actinidiae (Psa), and Halyomorpha halys (or brown marmorated stink bug (BMSB)). It was also included a description of main hosts and vectors capable to transmit and spread the pests, and current areas affected in the world. • In the second part the available remote sensing technologies are introduced and, the possibilities and the methodology proposed towards the detection and monitor of tree pests are illustrated. Following with the application of feature extraction procedures to fulfill a geodatabase with the locations of trees under analysis, and a description of the models most used to detect and predict potential areas under risk of new outbreaks, as MaxEnt and Climex. • And finally, it is decribed the architecture implemented to disseminate the results and updated datasets that could be used to carry out new analysis, guaranteeing the accessibility towards annual EO based surveys, as suggested by EFSA.