Large-scale estimation of xylem phenology in black spruce through remote sensing

There is a growing need for understanding the timing of wood formation in relation to the dynamics of bud phenology at wide geographical scale. This study analysed the relationships between long-term chronologies of xylem growth and the timing of plant phenology detected by Normalized Difference Vegetation Index (NDVI) in five permanent stands across the latitudinal distribution of black spruce [Picea mariana (Mill.) BPS] in the boreal forest of Quebec, Canada. Xylogenesis was studied weekly from April to October for thirteen years (2002-2014) on anatomical sections derived by wood microcores. The timing of the growing season detected by remote sensing was extracted from MODerate resolution Imaging Spectroradiometer (MODIS) 250 m 16-days NDVI data. The NDVI time-series were fitted using a double-logistic curve. Phenological chronologies from remote sensing and xylem phenology showed a latitudinal trend. The models correlating the data inferred from satellite sensors and the spring observations of xylem phenology were significant (p < 0.0001). The length of NDVI growing season and the duration of xylogenesis showed a close correlation. This study demonstrated that the timing of xylem formation could be suitably estimated at wide geographical scale using remote sensing data. However, the inter-annual phenological variability remained unexplained, which might limit the application of the models only to the years considered by our chronologies.