Statistical modelling and RCS detrending methods provide similar estimates of long-term trend in radial growth of common beech in north-eastern France

Jean-Daniel Bontemps (1) , Jan Esper (2)

1 LERFoB - Laboratoire d'Etudes des Ressources Forêt-Bois
2 Department of Geography [Mainz]

Dendrochronological methods have greatly contributed to the documentation of past long-term trends in forest growth. These methods primarily focus on the high-frequency signals of tree ring chronologies. They require the removal of the ageing trend in tree growth, known as 'standardisation' or 'detrending', as a prerequisite to the estimation of such trends. Because the approach is sequential, it may however absorb part of the low-frequency historical signal. In this study, we investigate the effect of a sequential and a simultaneous estimation of the ageing trend on the chronology of growth. We formerly developed a method to estimate historical changes in growth, including a careful control of site fertility in the sampling design and a simultaneous separation of the site fertility, developmental stage, and calendar year effects on growth, using a statistical modelling (SM) approach. The method has been applied to the radial growth of dominant trees in even-aged stands of common beech in north-eastern France. We compare the SM method to the regional curve standardisation (RCS) method, which is widely used in dendrochronology, and has been proven to retain more long-term signals than other detrending techniques. We also test a variant of the RCS in which the developmental stage is measured by size rather than age, as is the case in the SM approach. The SM and RCS methods produce similar long-term chronologies, showing an increase of approximately 50% in radial growth over the 20th century and a recent decline of around 18% in magnitude. The negative bias induced by the sequential estimation of ageing and date effects on growth in the RCS is identified, but remains minor. The chronology estimated by using the RCS variant (regional size curve) is lower in magnitude than that estimated by using the RCS, but the difference is moderate (5%). These results highlight the conservative properties of the RCS with regard to the low-frequency historical signals in growth. They also suggest that singularly high estimates of growth trends reported in the dendrochronology literature are not method-dependent. We hypothesize that they may be caused by a negative site-age linkage in sampling designs. The decline reported for common beech in this temperate area of France is lower in magnitude than that evidenced in Mediterranean contexts.

Mots clés: growth trends; dendrochronology; standardisation; statistical modelling; Fagus sylvatica

Domaines: Milieux et Changements globaux; Ecosystèmes; Sylviculture, foresterie; Applications [stat.AP]

Jean-Daniel Bontemps, Jan Esper. Statistical modelling and RCS detrending methods provide similar estimates of long-term trend in radial growth of common beech in north-eastern France. Dendrochronologia, 2011, 29 (2), pp.99-107. ⟨hal-00948667⟩