Chestnut ink disease, caused by Phytophthora spp., remains a major constraint on European chestnut orchards. We combined field‐based sanitary assessments with multitemporal UAV multispectral imagery to track disease dynamics at tree level in two orchards in the central Apennines (Pozza and Altino; 2024–2025). Flights with a DJI Mavic 3M (RGB + multispectral) produced orthophotos from which five vegetation indices (NDVI, GNDVI, ExNIR, NDExNIR, RVI) were extracted per crown. Differences among phytopathological classes (A–D) were tested with Kruskal–Wallis, followed by Dunn post-hoc with Bonferroni correction; we also reported effect sizes (r) to quantify the magnitude of pairwise contrasts. Across sites and years, spectral indices consistently separated the most severe class (D) from the others; separation among non-severe classes was strong at Altino in 2024 (A/B > C > D) and weaker in 2025. At Pozza (2025), GNDVI, ExNIR and NDExNIR did not discriminate class C from D after correction. Transition matrices (2024→2025) showed mostly stable populations (≈56% unchanged at Altino; ≈76% at Pozza), with changes mainly between contiguous classes. Treatment- stratified ΔExNIR (2025–2024) indicated site-dependent responses: at Altino all treatments had positive medians (MICROFORCE ≥ MICROLIFE > POLLINAMATURA), whereas at Pozza MICROFORCE and MICROLIFE were positive and POLLINAMATURA slightly negative. Spatial mapping of Δ-class and ΔExNIR revealed mosaic-like patterns rather than coherent spread fronts. Overall, UAV multispectral monitoring provides operational support for prioritizing severely affected trees and objectively evaluating management outcomes, while highlighting the need to reduce field-classification subjectivity and to extend multiyear, multimetric analyses.
Il mal dell’inchiostro del castagno, causato da Phytophthora spp., resta una criticità rilevante per i castagneti europei. In questo studio abbiamo integrato rilievi fitosanitari in campo e immagini multispettrali da UAV multi-temporali per seguire l’evoluzione della malattia a scala di pianta in due siti dell’Appennino centrale (Pozza e Altino; 2024–2025). Con un DJI Mavic 3M (RGB + multispettrale) sono state prodotte ortofoto da cui, per ogni chioma, sono stati estratti cinque indici di vegetazione (NDVI, GNDVI, ExNIR, NDExNIR, RVI). Le differenze tra classi fitopatologiche (A–D) sono state testate con Kruskal–Wallis e post-hoc di Dunn con correzione di Bonferroni; sono state inoltre riportate le dimensioni dell’effetto (r) per quantificare l’intensità dei contrasti a coppie. Complessivamente, gli indici separano in modo robusto la classe più grave (D); la distinzione tra le classi non gravi è marcata ad Altino nel 2024 (A/B > C > D) e più debole nel 2025. A Pozza (2025) GNDVI, ExNIR e NDExNIR non discriminano C rispetto a D dopo correzione. Le matrici di transizione (2024→2025) mostrano popolazioni per lo più stabili (≈56% invariati ad Altino; ≈76% a Pozza), con passaggi soprattutto tra classi contigue. Il ΔExNIR (2025–2024) stratificato per trattamento evidenzia risposte dipendenti dal sito: ad Altino mediane positive per tutti i trattamenti (MICROFORCE ≥ MICROLIFE > POLLINAMATURA), a Pozza segnali positivi per MICROFORCE e MICROLIFE e lievemente negativi per POLLINAMATURA. La mappatura spaziale di Δ-classe e ΔExNIR delinea pattern a mosaico, senza fronti di diffusione coerenti. Nel complesso, il monitoraggio multispettrale UAV offre un supporto operativo per prioritizzare le piante più compromesse e valutare in modo oggettivo gli esiti gestionali, sottolineando l’esigenza di ridurre la soggettività in campo e di estendere analisi multi- annuali e multi-indice.
Uso integrato di immagini da drone e osservazioni di campo per il monitoraggio del mal dell'inchiostro nei castagneti marchigiani
TURCHI, PIETRO
2024/2025
Abstract
Chestnut ink disease, caused by Phytophthora spp., remains a major constraint on European chestnut orchards. We combined field‐based sanitary assessments with multitemporal UAV multispectral imagery to track disease dynamics at tree level in two orchards in the central Apennines (Pozza and Altino; 2024–2025). Flights with a DJI Mavic 3M (RGB + multispectral) produced orthophotos from which five vegetation indices (NDVI, GNDVI, ExNIR, NDExNIR, RVI) were extracted per crown. Differences among phytopathological classes (A–D) were tested with Kruskal–Wallis, followed by Dunn post-hoc with Bonferroni correction; we also reported effect sizes (r) to quantify the magnitude of pairwise contrasts. Across sites and years, spectral indices consistently separated the most severe class (D) from the others; separation among non-severe classes was strong at Altino in 2024 (A/B > C > D) and weaker in 2025. At Pozza (2025), GNDVI, ExNIR and NDExNIR did not discriminate class C from D after correction. Transition matrices (2024→2025) showed mostly stable populations (≈56% unchanged at Altino; ≈76% at Pozza), with changes mainly between contiguous classes. Treatment- stratified ΔExNIR (2025–2024) indicated site-dependent responses: at Altino all treatments had positive medians (MICROFORCE ≥ MICROLIFE > POLLINAMATURA), whereas at Pozza MICROFORCE and MICROLIFE were positive and POLLINAMATURA slightly negative. Spatial mapping of Δ-class and ΔExNIR revealed mosaic-like patterns rather than coherent spread fronts. Overall, UAV multispectral monitoring provides operational support for prioritizing severely affected trees and objectively evaluating management outcomes, while highlighting the need to reduce field-classification subjectivity and to extend multiyear, multimetric analyses.| File | Dimensione | Formato | |
|---|---|---|---|
| 
									
										
										
										
										
											
												
												
												    
												
											
										
									
									
										
										
											TESI_TURCHI_PIETRO.pdf
										
																				
									
										
											 Open Access dal 09/10/2025 
										Dimensione
										70.95 MB
									 
									
										Formato
										Adobe PDF
									 
										
										
								 | 
								70.95 MB | Adobe PDF | Visualizza/Apri | 
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.12075/23100