Tree phenology - observer intercalibration and individual tree phenological scoring
Delpierre, Nicolas; Chuine, Isabelle; Cole, Eleanor (2021), Tree phenology - observer intercalibration and individual tree phenological scoring, Dryad, Dataset, https://doi.org/10.5061/dryad.1g1jwstvx
We report data documenting (1) the dynamics of budburst and leaf senescence in three European natural forest tree populations over the period od 2012-2015 and (2) the variability among phenological observers as documented from seven observer inter-calibration experiments conducted in France over 2007-2017 for both the budburst and leaf senescence period.
We report three sets of ground phenological observations, conducted in different forests: (1) observations of two species (Quercus petraea, Carpinus betulus) in the Fontainebleau-Barbeau forest (France) from 2013-2017; (2) observations of three species (Quercus petraea, Carpinus betulus, Castanea sativa) in the Orsay forest (France) from 2012-2015; (3) observations of six species (Quercus robur, Fraxinus excelsior, Fagus sylvatica, Betula pendula, Corylus avellana, and Acer pseudoplatanus) in Wytham Woods (Oxfordshire, UK) in 2013 and 2014. The three datasets share the particularity of documenting the whole sequence of spring budburst or leaf senescence at the individual tree scale. This means that each tree was monitored from winter state (all buds closed) to 100% budburst and from mature green summer leaves to all leaves colored or fallen. In Fontainebleau-Barbeau and Orsay, the same tree individuals were monitored during spring and autumn. The three sets of observations were each collected consistently by a small group (less than five) of observers who have been trained and inter-calibrated before data collection, such that the observer variability in this dataset is minimal.
At the Fontainebleau-Barbeau and Orsay forests, the percentage (on a 10%-step scale) of open leaf buds (in the upper-third of the tree crown), leaf coloration and leaf fall (over the whole tree crown) for individual trees were monitored with binoculars 2-3 times a week from March to June and once a week from September to December.
At Wytham Woods, the development of buds was scored over whole tree crowns using a key of phenological stages different from BBCH, at a 3-day interval from March to May. A seven-stage key (stage 1 = small dormant buds, 2 = larger, slightly elongated buds, 3 = larger, loosened greenish brown buds, 4 = further elongated buds with leaves starting to erupt, 5 = leaves emerging but still tight, 6 = leaves loosening and extending outwards, 7 = leaves fully emerged and unfurled) was used for Quercus robur and a five-stage key (stage 1 = dormant bud, 2 = elongated/swollen buds, 3 = budburst, 4 = leaves emerging and extending outwards, 5 = leaves fully emerged and unfurled) used for the other species.
The scale used for the observation of budburst in the Fontainebleau-Barbeau and Orsay forests differed with that used in Wytham Woods. Indeed, bud observations in the Fontainebleau-Barbeau or Orsay forests started from the first signs of buds opening (0% buds opened), and ended when 100% buds were open, corresponding to the interval from stage 4 to stage 7 using the seven-stage key and stage 3 to stage 5 using the five-stage key in Wytham Woods. In order to homogenize protocols, we scaled the Wytham Woods observations of budburst to a 0-100% scale similar to the one used in Fontainebleau-Barbeau and Orsay, setting stage 3 (or 4 depending on the species) of the Wytham Woods scale to 0% and setting stage 5 (or 7, depending on the species) to 100%.
We ensured that the tree individual phenological sequences collected at Fontainebleau-Barbeau, Orsay and Wytham woods were monotonically increasing with time. If a phenological observation record was less than the previous record, we corrected it to be equal to the previous record (affects 0.25% / 7.8% of all budburst/leaf senescence data).
The observer inter-calibration dataset was obtained during seven inter-calibration sessions of observers conducted in France during 2007-2017, three of which concerned budburst observations, and four of which concerned observation of leaf senescence. The sessions took place in different locations each year and concerned different tree species. An average of 30 individual trees (from 12 to 49 trees) were observed by 10 to 37 observers each time following the same protocol and scoring scale (BBCH scale). The observers were scientists, technicians and doctoral students in biological sciences, and most of them participated all the sessions, which lasted one and a half day each. The inter-calibration sessions were organized as follows. At the first session, all participants were trained to the BBCH scoring scale and protocol. At all sessions, all participants first trained altogether on a set of tree individuals. Participants subsequently also scored another set of tree individuals independently. They used binoculars for observing tree crowns. Observers were instructed not to exchange information or discuss during this scoring in order to guarantee the independence of the scoring. The data obtained with this independent scoring were used for this study.