Agricultural and Forest Meteorology 290 (2020) 108031 Contents lists available at ScienceDirect Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet Inter-individual variability in spring phenology of temperate deciduous trees T depends on species, tree size and previous year autumn phenology ⁎ Lorène Julia Marchanda,b, , Inge Doxa, Jožica Gričarc, Peter Prislanc, Sebastien Leysa, Jan Van den Bulcked, Patrick Fontie, Holger Langef, Erik Matthyseng, Josep Peñuelash,i, Paolo Zuccarinih,i, Matteo Campiolia a Centre of Excellence PLECO (Plants and Ecosystems), Department of Biology, University of Antwerp, 2160Wilrijk, Belgium b UMR 6553 ECOBIO (Ecosystèmes, Biodiversité, Evolution), Université de Rennes 1, CNRS, 263 Av. du Général Leclerc, 35042Rennes, France c Department of Yield and Silviculture, Slovenian Forestry Institute, Vecna pot 2, 1000Ljubljana, Slovenia d Laboratory of Wood Technology, UGent-Woodlab, Department of Environment, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000Gent, Belgium e Swiss Federal Research Institute WSL, Zürcherstrasse 111, 8903Birmensdorf, Switzerland f NIBIO (Norwegian Institute of Bioeconomy Research), Division of Environment and Natural Resources, Høgskoleveien 8, 1433Ås, Norway g Evolutionary Ecology Group, Department of Biology, University of Antwerp, 2160Wilrijk, Belgium h Centre for Research on Ecology and Forestry Applications (CREAF), Cerdanyola del Vallès, Barcelona, Spain i CSIC, Global Ecology Unit CREAF-CSIC-UAB, Bellaterra 08193, Spain A R T I C LE I N FO S UM M A R Y Keywords: We explored the inter-individual variability in bud-burst and its potential drivers, in homogeneous mature stands Leaf unfolding of temperate deciduous trees. Phenological observations of leaves and wood formation were performed weekly Wood formation from summer 2017 to summer 2018 for pedunculate oak, European beech and silver birch in Belgium. The Coloration variability of bud-burst was correlated to previous’ year autumn phenology (i.e. the onset of leaf senescence and Pedunculate oak the cessation of wood formation) and tree size but with important differences among species. In fact, variability European beech Silver birch of bud-burst was primarily related to onset of leaf senescence, cessation of wood formation and tree height for oak, beech and birch, respectively. The inter-individual variability of onset of leaf senescence was not related to the tree characteristics considered and was much larger than the inter-individual variability in bud-burst. Multi- species multivariate models could explain up to 66% of the bud-burst variability. These findings represent an important advance in our fundamental understanding and modelling of phenology and tree functioning of de- ciduous tree species. 1. Introduction among individuals at the same site, where meteorological conditions are close to identical. Still, there is an inter-individual variability of Phenology is the discipline that studies ‘the timing of detectable bud-burst (Fig. 1), which can be up to 20 days (Puchałka et al., 2017). reciprocal events in the life cycle of plants and animals in connection This variability in spring phenology is important for the trees, as it can with the environment, ranging from an individual- to an ecosystem influence tree performance, resilience and gives competitive advantage level’ (Forrest and Miller-Rushing, 2010). In the deciduous tree species during the current and the following year (Barbaroux et al., 2003; of the temperate zone, primary growth starts with bud-burst, which is Pérez-de-Lis et al., 2016). Moreover, inter-individual variability in therefore an essential determinant of seasonal photosynthetic uptake, spring phenology can crucially affect the function and structure of the energy and water balance but also of tree functional traits (Piao et al., forest understory. For instance, variations in the flowering phenology of 2019). forest herbs is related to variation in the canopy phenology of primary Even if trees in forest stands experience gradients of micro- growth, with fitness advantages (better germination seed rate) for early meteorological conditions,the environmental factors affecting tree flowering individuals (Baeten et al., 2015). Finally, inter-individual phenology in spring (e.g.temperature, photoperiod) (Čufar et al., 2008; variability of spring phenology can impact leaf herbivores dynamics Prislan et al., 2013; Vitasse and Basler, 2013) do not vary significantly and their spatial distribution (Forkner et al., 2008) ⁎ Corresponding author. E-mail address: lorene.marchand@univ-rennes1.fr (L.J. Marchand). https://doi.org/10.1016/j.agrformet.2020.108031 Received 2 March 2020; Received in revised form 5 May 2020; Accepted 6 May 2020 Available online 29 May 2020 0168-1923/ © 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 Fig. 1. Inter-individual variability of bud-burst at the beech stand of the Park of Brasschaat. 17/04/2018. Photo Lorène J. Marchand. It is generally thought that inter-individual variability in spring understand fundamental aspects of tree functioning and to model the phenology is mainly due to genetic differences among individuals. For individual tree phenology. instance, the temperature sum requirement for birch bud-burst differs among genotypes (Possen et al., 2014). However, relating variability in 2. Material and methods spring phenology to genetic variability among individuals is technically very difficult to integrate into the current generation of population-, 2.1. Study species and study sites forest- and terrestrial ecosystem models or meta- and global analyses (Müller et al., 2017). On the other hand, the task of elucidating and Three of the most common temperate deciduous tree species in modelling inter-individual variability of spring phenology would be Europe were studied in Northern Belgium: Pedunculate oak(Quercus greatly facilitated if such variability could be related to differences in robur L.), European beech (Fagus sylvatica L.) and silver birch (Betula tree development and growth among individuals or to variability in tree pendula Roth.). Oak and beech are late successional species with a de- functional traits that are influenced by genetic variability but are easy terminate growth (i.e. they produce leaves in 1–2 flushes). Birch is a to measure and model. For instance, it is known that age and size dif- pioneer species (Dyderski et al., 2018) with an indeterminate growth ferences affect inter-individual variability in phenology, as younger and (i.e. it produces leaves continuously throughout the season). The wood smaller trees need an earlier start of the growing season to compensate anatomy of these species is also different: oak is a ring-porous species for the over-shading from taller and older trees (Augspurger and whereas beech and birch are diffuse-porous species. Northern Belgium Bartlett, 2003). Inter-individual variability of phenological events in is characterized by a maritime temperate climate optimal for our stu- spring can also be related to inter-individual variability of phenology in died tree species, with precipitation (778 mm year−1) distributed reg- autumn. For instance, Delpierre et al. (2017) found evidence that later ularly over the year and average monthly temperatures ranging from leaf senescence can also induce later leaf unfolding the following year +3 °C (January) to +17 °C (July) (Campioli et al., 2012) with an on mature oak trees. Relationships between inter-individual variability average annual temperature of 10.1 °C. The study comprises two forest in spring phenology and inter-individual variability in tree growth and areas (51°12′ – 51°21′ N, 4°26′ – 4°37′ E, 18–22 m a.s.l.). The first one is developmental characteristics (comprising age, dendrometric data, the ‘Park of Brasschaat’ (PB), the second one is the military domain autumn phenology, annual ring increment and tree competition) might ‘Klein Schietveld’ (KS), a Natura2000 reserve. The study areas are at the help elucidating the drivers of such variability, its potential modelling distance of 9,5 km, in a region with topography flat and without any and how much of spring phenology variability is dependent from pre- features possibly causing significant climatological differences vious’ year phenology variability. However, up to date, data on this (Mariën et al., 2019). Both areas encompass planted (oak and beech) topic is lacking. and unplanted stands (birch) on sandy soil, with low management ac- We investigated whether, and to what extent, inter-individual tivity. A thick organic layer (60–90 cm) is present at PB, making it more variability of bud-burst is related to inter-individual variability of tree fertile than KS. A total of six stands were investigated, three of which characteristics and autumn phenology of the previous year (onset of are located in the park: PB-Q (park oak), PB-F (park beech) and PB-B leaf senescence and cessation of wood formation). Therefore, we: (i) (park birch) and three others in the military reserve: KS-Q (military assessed the timing of bud-burst and of previous’ year onset of senes- reserve oak), KS-F (military reserve beech) and KS-B (military reserve cence and their inter-individual variability for three deciduous species birch). In each stand, 8–16 (co)dominant healthy trees without win- (Quercus robur L., Fagus sylvatica L. and Betula pendula Roth.) in two dows in the crown and dead branches, were selected in the summer of nearby forest areas in Antwerp Province, Belgium; (ii) studied the im- 2017. Leaf phenological observations were performed on all trees (56 pact of individual tree (growth) characteristics (age, height, autumn trees), whereas phenology of wood formation (secondary growth) was phenology etc.) on the inter-individual variability in spring phenology, only followed on half of the trees due to the labor-intensive nature of and (iii) built statistical models to determine the importance of the these analyses. explanatory variables on the inter-individual variability of bud-burst. In addition, a comparison between the inter-individual variability of bud- 2.2. Leaf phenological observations burst and of leaf senescence onset, and their relationships with tree characteristics, was performed. Our approach will help to better Phenological observations of bud-burst in 2018 were conducted 2 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 weekly for the first half of March and then twice a week from mid- and centered at the study tree, were considered as possible competitors. March till the end of May. Binoculars (Tokura explorer bak4, Tokura, Small understory trees (trees <10 cm DBH for birch stands and Japan) with a magnification of 8–27 × 25 were used to observe five <20 cm DBH for beech and oak stands) were excluded. The tree density randomly chosen buds in the upper third of the tree crown and five index was taken as equal to the total basal area of the competitor trees. randomly chosen buds in the lower third of the crown (the crown was divided in three equal parts) to account for the possible variability in 2.5. Tree age and tree-ring width the timing of bud-burst within the crown due to shadow effects (the top of the crown receives more direct light than the bottom (Davi et al., In total 12 beech, 8 oak and 8 birch trees were cored with a Pressler 2011). Phenology of primary growth (leaves) was monitored according corer, with two increment cores per tree, one in N and one in S direc- to a 5-stage scale adapted from Vitasse et al. (2009) and tion. The cores were conditioned in a climate chamber at 20 °C and 65% Gričar et al. (2017): 0: dormant bud; 1: swelling bud; 2: bud-burst; 3: relative humidity for two weeks and then glued on wooden sample emerging leaves, and 4: one leaf at least is completely detached from holders. The cores were sanded and tree-ring width measurements were the bud. The date of bud-burst was defined as the date when 50% of the performed with a Lintab measuring stage (0.01 mm precision) con- buds in a tree reached stage 2, combining together top and bottom nected to the TSAP-Win software for data acquisition. crown buds to have an indication at the tree level (Vitasse et al., 2009). Leaf autumn phenology was monitored by observing the canopy col- 2.6. Data analysis oration at day t (Xt) (Dox et al., 2020). In late summer and early au- tumn, canopy coloration was estimated directly, as a percentage. Later All analyses were conducted in the open source programming lan- in the season, when the process became more intense, Xt was estimated guage R, version 3.4.2. A p-value of 0.05 was chosen as the level of through a combined rating of the percentage of leaves that had changed significance. Graphs were made with the package ggplot2 color (αt) and the percentage of leaves that had fallen (βt), as described (Wilkinson, 2005). Three different sets of statistical analyses were in (Vitasse et al., 2011): conducted. (i) The timing of bud-burst and onset of leaf senescence αt *(100 − βt ) were compared, separately, among stands with a two-way ANOVA xt = + βt 100 followed by a Tukey's HSD (honestly significant difference) test, with species and site as factors. Normality of residuals and homogeneity of Observations were carried out weekly from late August until Xt was variances were verified with Shapiro test (Royston, 1982) and ncvTest 100% in mid-late November. To determine the onset of leaf senescence (package car,(Fox and Weisberg, 2019), respectively. (ii) Univariate a breakpoint analysis, performed with the R package ‘segmented’ linear model between bud-burst (or leaf senescence onset) and the tree (Muggeo, 2008), was conducted on the seasonal time series of canopy characteristics or previous’ year phenological events were constructed, coloration. The onset of leaf senescence was defined as the point in time normality of residuals and variance homogeneity were done as above. (breakpoint) when the seasonal canopy coloration sharply increased When the conditions of normality were not met, a generalized linear due to the start of the natural process of leaf senescence (before this model was performed instead of the linear model. (iii) Multivariate point slight coloration already happened due to slow build-up of leaf linear models were constructed after checking for collinearity between stress damage). Details about the determination of the leaf senescence variables (Table SI in appendix). Normality of residuals and variance onset at the study sites can be found in (Mariën et al., 2019). homogeneity were performed as above. ANOVA analysis (package car) was done on each best model including species as factor to test the 2.3. Wood phenological observation and analysis model qualitative characteristics as a whole. To assess autumn phenology of wood formation in 2017, stem micro-cores were collected weekly from late August till late November. 3. Results Sampling took place between a height of 1.3 m and 2 m. Samples were collected using a Trephor (Rossi et al., 2006). An upward spiral sam- 3.1. Timing of bud-burst pling was followed over the season to avoid wound reactions. The preparation following Prislan et al. (2014) and the analysis of the Bud-bursttook place between DOY 96 and 117 across sites and inter- micro-cores was done at the Slovenian Forestry Institute (Ljubljana). species, with stand average bud-burst dates between DOY 100 and 109 Samples were observed under a LEICA DM 4000B light microscope (Table 1). Stand average bud-burst date differed between the two sites (Leica Microsystems, Wetzlar, Germany) using transmission and po- (p = 0.03) and inter-species (p<0.01; post-hoc test with p<0.05 for all larised light. Polarised light induced brightness in the secondary walls species combinations) (Fig. 2). Inter-individual variability in bud-burst (characteristic of mature cells). Histometric analyses were performed varied among species and, for oak, between sites. Across the six stands, with a LEICA DMC 4500 camera and LAS image analysis software inter- individual bud-burst variability varied from 2 to 13 days (LEICA Application Suite version 4.9.0, copyright 2003–2006) to de- (Table 1). termine the number of cambial cells in the dormant/active stage. Ces- sation of wood formation was recognizable by a completely red cell Table 1 wall of all cells due to a safranin-astra blue staining, whereas cells that Inter-individual variability of bud-burst and of leaf senescence onset for stands were still in the wall-thickening phase had blue colored inner parts of of oak (Q), beech (F) and birch (B) at two forest areas (‘Park of Brasschaat’ (PB) and ‘Klein Schietveld’ (KS)) close to Antwerp, Belgium, expressed as minimum the cell walls. We defined the end of wood formation as the moment in (min), maximum (max), their difference (range) and mean ( ± standard error) time when latewood differentiation was completed (i.e. the proportion of bud-burst and leaf senescence onset date for the trees at each stand. of cells still in the wall-thickening phase was <0.5%) (Dox et al., 2020). Stand Bud-burst (DOY) Onset of senescence (DOY) min max mean ± SE range min max mean ± SE range 2.4. Tree competition index PB-Q 106 113 109 ± 0.77 7 277 318 303 ± 5.38 41 To examine the effect of canopy competition, which is present when KS-Q 108 110 109 ± 0.31 2 280 317 297 ± 4.92 37 the crown of a neighboring tree is in contact with the crown of the study PB-F 104 117 109 ± 1.5 13 266 309 292 ± 5.3 42 KS-F 96 109 105 ± 0.7 13 280 307 293 ± 1.7 27 tree, we determined a tree competition index (or tree density index). All PB-B 98 102 100 ± 0.5 4 254 305 278 ± 7.9 51 trees inside a circular area, with a diameter of 4 m for birch and 8 m for KS-B 99 103 101 ± 0.6 5 260 286 281 ± 3.12 26 beech and oak respectively, proxy to the projection of the tree crown, 3 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 Fig. 2. Average date ( ± standard error) of onset of senescence in 2017 (orange) and bud-burst in 2018 (green) for beech, oak and birch at two forest areas (Park of Brasschaat, PB, and military reserve Klein Schietveld, KS) in the vicinity of Antwerp, Belgium. p- values represents the significance level of the factor ‘species’ and ‘site’ when a two-way ANOVA was ap- plied to onset of leaf senescence and budburst, sepa- rately, at the stand level. 3.2. Univariate models of inter-individual variability of spring phenology taller trees presenting later bud-burst) (Table 2, Fig. 3e). When the species were pooled, bud-burst was significantly and positively corre- For oak individuals, bud-burst in 2018 was significantly and posi- lated to the onset of leaf senescence, cessation of wood formationand tively correlated to the onset of leaf senescence in the previous year tree diameter (Table 2, Fig. 3b, d, f). Weaker multi-species relationships (Table 2, Fig. 3a). Thus, for this species under these conditions, in- were also found between bud-burst and tree height (negative) and age dividuals with a later onset of leaf senescence in the previous year (positive). presented bud-burst later in the current year. For beech, bud-burst was positively correlated with the timing of cessation of wood formation in the previous year (Table 2, Fig. 3c). In other words, trees with delayed 3.3. Comparison between bud-burst and onset of leaf senescence cessation of wood formationpresented later bud-burst. For beech, weak negative relationships were found between bud-burst and tree height Leaf senescence onset presented a substantially larger inter-in- (p = 0.065) and tree diameter (p = 0.065), with taller and larger trees dividual variability across species and sites (DOY 255–318, 63 days) presenting earlier bud-burst (Table 2). For birch a correlation between than bud-burst (21 days) (Table 1). Average onset of leaf senescence bud-burst and tree height was also found, but of opposite sign (thus varied across species (p<0.01), except for the comparison oak and beech (p = 0.11 with post-hoc test), but it did not vary between sites Table 2 Characteristics of univariate linear models between bud-burst in 2018 and tree characteristics or autumn phenology in 2017, for oak (n = 15), beech (n = 23) and birch (n = 16), separately and pooled together, from two forest areas close to Antwerp, Belgium. (a) indicates variables with smaller sample size (for oak n = 7, for beech n = 12 and for birch n = 8); the bold font indicates when the p-value is significant (<0.05); the asterisk indicates the use of a generalized linear model quasipoisson instead of a linear model. Potential drivers OAK BEECH BIRCH ALL SPECIES TOGETHER T-value R² P-value T-value R² P-value T-value R² P-value T-value R² P-value Onset of senescence 2017 (DOY) 2.221 0.2028 0.0433 0.563 −0.0321 0.5796 0.366 −0.0659 0.72 3.841 0.2O21 <0.01 Tree diameter (m) 1.22 0.0315 0.243 −1.939 0.1071 0.0654 0.176 −0.0691 0.863 3.355 0.1647 0.02 Tree competition index (m²/ha) 1.355 0.0527 0.197 −1.141 0.0564 0.266 −0.786 −0.0261 0.445 0.303 −0.018 0.763 Tree height (m) −0.416 −0.0627 0.684 −1.94 0.1073 0.0653 2.164 0.1971 0.0482 −1.791 0.0410 0.079 cessation of wood formation 2017 (DOY) (a) 0,197 −0.1592 0.850 2.373 0.2963 0.039 −1.866 0.1667 0.111 4.727 0.4508 <0.01 Age (year) (a) 0.527 −0.1151 0.617 −0.969 −0.006 0.355 −0.354* 0.0204 0.735 2.028 0.1069 0.0534 10 year average ring width (μm) (a) −0.021 −0.1666 0.984 0.894 −0.0186 0.392 1.715 0.1329 0.137 −1.038 0.003 0.309 4 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 Fig. 3. Significant (p<0.05) relationships between bud-burst and onset of senescence (for oak (a) and all species pooled together (b)), cessation of wood formation (for beech (c) and all species pooled together (d)), tree height (for birch (e)) and tree diameter (for all species pooled together (f)). The black line represent the linear model with its significance value. Thegray bands represent the confidence interval of 95%. (p = 0.99). The relationships between onset of leaf senescence and tree Table 3 diameter, tree height and the tree competition index were not sig- Characteristics of multivariate and multi-species linear models of inter-in- nificant when the species were considered separately (Suppl. Table SIII) dividual variability of bud-burst, for individuals of oak, beech and birch pooled but a significant (p<0.01) positive relationship between onset of leaf together (Model 1 n = 53, Model 2 n = 27) from two forest areas close to senescence and tree diameter was found when the species were pooled Antwerp, Belgium. Bold font indicates when the p-value is significant (<0.05). “NA” indicates no available variables and “ns” indicates no significant vari- (Suppl. Table SIII). ables. 53 trees 27 trees 3.4. Multivariate models of inter-individual variability in bud-burst F-value P-value F-value P-value When consideringthe independent variables examined, plus the Specie 25.15 <0.01 10.88 <0.01 species and site factors, we obtained a model with five significant Site 7.66 <0.01 2.286 0.146 Tree height ns ns 0.275 0.606 variables (height, species, site, onset of senescence, cessation of wood Onset of senescence 2017 nc nc 29.95 <0.01 formation) but with onset of senescence and species as the most im- Cessation of wood formation 2017 NA NA 2.473 0.131 portant variables (Table 3). This model could explain 66% of the inter- Age NA NA ns ns individual variability. When considering only the independent variables 10 year average ring width NA NA ns ns related to the tree size and tree competition, we obtained a simpler Tree competiton index ns ns ns ns Tree diameter 27.07 <0.01 ns ns model (three variables) but still able to explain 60% of the bud-burst R² adjusted 60% 66% variability. In this case, species and diameter were the key variables. Compared to the previous model, diameter substituted the effect of height (see the correlation between diameter and height; Suppl. Table SI) and of leaf senescence onset (see the correlation between diameter and leaf senescence onset; Suppl. Table SIII). Note that for the onset of 5 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 Table 4 Characteristics of single species multivariate linear models for inter-individual variability of bud-burst considering tree diameter, tree height, tree competition index and with or without leaf senescence onset, for oak (n = 15), beech (n = 23) and birch (n = 16), separately. Bold font indicates when the p-value is significant (<0.05). “nc” indicates no considered variables on our analyses. “ns” indicetes no significant variables. Oak Beech Birch With onset of Without onset of With onset of Without onset of With onset of Without onset of scenescence senescence scenescence senescence scenescence senescence Potential drivers F-value P-value F-value P-value F-value P-value F-value P-value F-value P-value F-value P-value Site 3.076 0.107 3.284 0.095 6.194 0.022 6.382 0.02 0.0167 0.900 ns ns Diameter 0.015 0.906 0.3485 0.565 3.926 0.062 4.57 0.04 1.269 0.289 1.00 0.34 Tree competition index ns ns ns ns ns ns ns ns 5.593 0.042 2.04 0.18 Height ns ns ns ns ns ns ns ns 7.036 0.026 10.89 <0.01 Onset of senescence 2017 3.831 0.08 ns ns 0.439 0.516 ns ns 0.2367 0.638 ns ns R² adjusted 22% 20% 26% 29% 40% 44% leaf senescence the best model could explain only 24% of the variability (Fu et al., 2014), but age differences might play a confounding role. For (Supp. Table SII). birch, inter-individual variability in spring phenology was positive re- Single species models could explain much less variability in bud- lated to height. Taller and more dominant trees have probably less need burst (ca. 20–40%) than multi-species models (Table IV) even if the to start canopy development earlier. However, this relationship seems relevant model variables were the same. As for the multiple-species species-specific, as a negative correlation between bud-burst timing and models, onset of leaf senescence could be substituted without loss of tree height was found for beech. As suggested above, carbohydrates performance also in the single-species models (Table 4). might play a role in advancing bud-burst for beech and taller, larger beech trees might have more carbohydrates reserve. 4. Discussion To the best of our knowledge, no modelling work has yet been done to capture inter-individual variability in spring phenology, whereas The inter-individual variability in bud-burst observed in this study most phenological modelling has focused on determining the environ- varied among species (in particular, beech seems to have a higher mental cues (e.g. temperature) of spring phenology for a stand or po- variability than birch). However, it was smaller than the variability pulation (Delpierre et al., 2016; Fu et al., 2012; Linkosalo et al., 2008). reported in other studies (e.g. between 16 and 20 days for beech and up Our results showed that tree diameter, species and site can explain to 10 days for oak (Prislan et al., 2013; Puchałka et al., 2017)). Site together the majority of the variability (66%) of bud-burst among trees differences in inter-individual variability was relevant for oak. Maybe at the landscape level. An explained variance of 66% is high for an this was partially related to the inter-individual variability of leaf se- experimental design such as the one used here and for a dynamics (i.e. nescence onset, which was strongly linked to bud-burst variability for inter-individual variability of spring phenology) not yet comprehen- oak (Table 2). sively explored (see for comparison the forest growth efficiency models In fact, for oak individuals, timing of bud-burst was positively re- in (Campioli et al., 2015). At single species level, our study shows im- lated to the timing of onset of leaf senescence in the previous year. portant relationships between inter-individual variability of bud-burst Similar results were found for oak in France (Delpierre et al., 2017) and and intra-individual variability of variables related to tree size and young oak trees in Belgium (Fu et al., 2014). An earlier leaf senescence autumn phenology of the previous year. However, multivariate models might induce an earlier endodormancy. Thus the chilling requirement showed that these relationships could explain only a minor portion of in winter (and heat requirement in spring) could be reached earlier in the inter-individual variability in bud-burst for each species, separately. case of earlier leaf senescence, advancing bud-burst in the following It is possible that variables not considered in our study will play a role year (Fu et al., 2014). For beech, bud-burst was directly related to the (e.g. carbon and nutrients reserves, fine root dynamics). timing of the previous year’ cessation of wood formation with an earlier As bud-burst and onset of leaf senescence occurred during a time bud-burst related to an earlier cessation of wood formation. To the best window, it is possible that trees of the same stand have experienced of our knowledge, this has never been detected before. As cessation of different meteorological conditions before each phenological event (e.g. wood formation occurred earlier than leaf senescence onset (280 ± 7 trees with later onset of senescence have experienced colder conditions and 290 ± 3 DOY, respectively) (data not shown), this observation before senescence than trees with earlier onset of senescence). Because could be explained by a larger reserve storage achieved through the of the complex interactions among tree size, annual phenological events accumulation of photoassimilates produced by the leaves before se- and seasonal meteorological conditions, we have not considered the nescence and not allocated to wood growth (Čufar et al., 2015). In fact, influence of the latter on inter-individual phenology variability. there is evidence that soluble carbohydrate concentration is highly However, the impact of seasonal meteorological conditions on inter- correlated to bud-burst and also than buds richer in non-structural individual phenology variability will be an interesting point for further carbohydrates open earlier (Maurel et al., 2004; Pérez-de-Lis et al., research, particularly for onset of senescence that showed a large 2016). For beech, bud-burst is not correlated with the onset of leaf variability. The importance of autumn phenology in affecting bud-burst senescence, in agreement with the study of Delpierre et al. (2017) in stresses the relation between the tree life cycle events across the dor- France. The different relationship between bud-burst and previous-year mant season and the legacy of the previous year's growth conditions on onset of leaf senescence for beech (non-significant) and oak (sig- current growth. However, the comparison between bud-burst and onset nificant) might be due to the higher chilling requirement of beech of leaf senescence showed that the inter-individual variability of the (Delpierre et al., 2017). In fact, for the latter species, earlier start of latter is much larger than the former, respectively, and that leaf se- endodormancy probably has no positive effect in reaching the chilling nescence onset shows less correlation with tree characteristics than requirement earlier and, thus, to trigger earlier bud-burst bud-burst. This indicates that inter-individual variability of different (Delpierre et al., 2017). Our results differed from the ones obtained phenophases has different underlying mechanisms. Onset of leaf se- from an experiment on young beech trees where bud-burst appeared nescence might be more variable than bud-burst because of the larger earlier in case of an earlier leaf senescence in the previous year impact on the former of inter-individual variability in nutrient status 6 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 and canopy architecture (and thus light environment). Both, nutrient biogeochemistry in the temperate oceanic European zone: review and prospects for status (Estiarte and Peñuelas, 2015; Weih, 2009) and light environment Belgium as a case study. J. For. Res. 17 (1), 1–18. https://doi.org/10.1007/s10310- 011-0255-8. (Brelsford et al., 2019; Michelson et al., 2018) affect the autumn leaf Čufar, K., De Luis, M., Prislan, P., Gričar, J., Črepinšek, Z., Merela, M., Kajfež-Bogataj, L., senescence. In general, onset of leaf senescence is a much more subtle 2015. Do variations in leaf phenology affect radial growth variations in Fagus syl- process than bud-burst and we still lack a clear understanding of this vatica. Int. J. Biometeorol. 59 (8), 1127–1132. https://doi.org/10.1007/s00484-014- 0896-3. process (Gallinat et al., 2015). Čufar, K., Prislan, P., de Luis, M., Gričar, J., 2008. Tree-ring variation, wood formation and phenology of beech (Fagus sylvatica) from a representative site in Slovenia, SE 5. Conclusion Central Europe. Trees 22 (6), 749–758. https://doi.org/10.1007/s00468-008-0235-6. Davi, H., Gillmann, M., Ibanez, T., Cailleret, M., Bontemps, A., Fady, B., Lefèvre, F., 2011. Diversity of leaf unfolding dynamics among tree species: new insights from a study Our results showed that inter-individual variability of bud-burst along an altitudinal gradient. Agric. For. Meteorol. 151 (12), 1504–1513. https://doi. significantly related to tree diameter, height and previous’ year autumn org/10.1016/j.agrformet.2011.06.008. Delpierre, N., Guillemot, J., Dufrêne, E., Cecchini, S., Nicolas, M., 2017. Tree phenolo- phenology (onset of leaf senescence and cessation of wood formation) gical ranks repeat from year to year and correlate with growth in temperate decid- with differences between deciduous species. Based on these variables, uous forests. Agric. For. Meteorol. 234–235, 1–10. https://doi.org/10.1016/j. multi-species statistical models explained 60–66% of the inter-in- agrformet.2016.12.008. dividual variability of bud-burst. These results extend our under- Delpierre, N., Vitasse, Y., Chuine, I., Guillemot, J., Bazot, S., Rutishauser, T., Rathgeber, C.B.K., 2016. Temperate and boreal forest tree phenology: from organ-scale processes standing of tree phenology, tree functioning and the relation between to terrestrial ecosystem models. Ann. For. Sci. 73 (1), 5–25. https://doi.org/10.1007/ spring and autumn dynamics. Moreover, these finding could improve s13595-015-0477-6. simulations of single-tree phenology in landscape or regions with var- Dox, I., Gričar, J., Marchand, L.J., Leys, S., Zuccarini, P., Geron, C., Prislan, P., Mariën, B., Fonti, P., Lange, H., Peñuelas, J., Van den Bulcke, J., Campioli, M., 2020. Timeline of ious forest stands, and forest models when inter-individual variability is autumn phenology in temperate deciduous trees. Tree Physiol tpaa058. https://doi. sought for. org/10.1093/treephys/tpaa058. Dyderski, M.K., Paź, S., Frelich, L.E., Jagodziński, A.M., 2018. How much does climate change threaten European forest tree species distributions. Glob. Chang. Biol. 24 (3), Author contribution 1150–1163. https://doi.org/10.1111/gcb.13925. Estiarte, M., Peñuelas, J., 2015. Alteration of the phenology of leaf senescence and fall in L.J.M., I.D., S.L and M.C., collected data. J.V.D.B. and I.D. processed winter deciduous species by climate change: effects on nutrient proficiency. Glob. Chang. Biol. 21 (3), 1005–1017. https://doi.org/10.1111/gcb.12804. the tree age data. L.J.M., P.P and J.G. performed secondary growth Forkner, R.E., Marquis, R.J., Lill, J.T., Corff, J.L., 2008. Timing is everything? analyses, while L.J.M. performed all other analyses. L.J. M. and M.C. Phenological synchrony and population variability in leaf-chewing herbivores of wrote the text. All authors contributed to discussions and revisions. Quercus. Ecol. Entomol. 33 (2), 276–285. https://doi.org/10.1111/j.1365-2311. 2007.00976.x. Forrest, J., Miller-Rushing, A.J., 2010. Toward a synthetic understanding of the role of Declaration of Competing Interest phenology in ecology and evolution. Philos. Trans. R. S. B Biol. Sci. 365 (1555), 3101–3112. https://doi.org/10.1098/rstb.2010.0145. Fox, J., Weisberg, S., 2019. An R Companion to Applied Regression, 3th edn. SAGE pub- The authors declare that they have no known competing financial lication. interests or personal relationships that could have appeared to influ- Fu, Y.H., Campioli, M., Van Oijen, M., Deckmyn, G., Janssens, I.A., 2012. Bayesian ence the work reported in this paper. comparison of six different temperature-based budburst models for four temperate tree species. Ecol. Modell. 230, 92–100. https://doi.org/10.1016/j.ecolmodel.2012. 01.010. Acknowledgements Fu, Y.S.H., Campioli, M., Vitasse, Y., De Boeck, H.J., Van den Berge, J., AbdElgawad, H., Asard, H., Piao, S., Deckmyn, G., Janssens, I.A., 2014. Variation in leaf flushing date influences autumnal senescence and next year’s flushing date in two temperate tree This research was funded by the ERC Starting Grant LEAF-FALL species. Proc. Natl. Acad. Sci. 111 (20), 7355–7360. https://doi.org/10.1073/pnas. (714916). We would like to thank the Belgian institutions that gave 1321727111. permission to conduct research in the study areas: Agency for Forest Gallinat S., Amanda, Primack B., Richard, Wagner, David, 2015. Autumn, the neglected season in climate change research. Trends in Ecology & Evolution 30 (3), 168–176. and Nature of the Flemish Government (ANB), the Military Defence of https://doi.org/10.1016/j.tree.2015.01.004. Belgium (Defensie), and the City of Brasschaat. Special thanks are due Gričar, J., Lavrič, M., Ferlan, M., Vodnik, D., Eler, K., 2017. Intra-annual leaf phenology, to Stijn Willen (UGent) for sampling and analysis of the tree cores and radial growth and structure of xylem and phloem in different tree parts of Quercus pubescens. Eur. J. For. Res. 136 (4), 625–637. https://doi.org/10.1007/s10342-017- to Dirk Leyssens (ANB). 1060-5. Linkosalo, T., Lappalainen, H.K., Hari, P., 2008. A comparison of phenological models of Supplementary materials leaf bud burst and flowering of boreal trees using independent observations. Tree Physiol. 28 (12), 1873–1882. https://doi.org/10.1093/treephys/28.12.1873. Mariën, B., Balzarolo, M., Dox, I., Leys, S., Marchand, L.J., Geron, C., Portillo-Estrada, M., Supplementary material associated with this article can be found, in AbdElgawad, H., Asard, H., Campioli, M., 2019. Detecting the onset of autumn leaf the online version, at doi:10.1016/j.agrformet.2020.108031. senescence in deciduous forest trees of the temperate zone. New Phytologist. https:// doi.org/10.1111/nph.15991. Maurel, K., Leite, G.B., Bonhomme, M., Guilliot, A., Rageau, R., Petel, G., Sakr, S., 2004. References Trophic control of bud break in peach (Prunus persica) trees: a possible role of hexoses. Tree Physiol. 24 (5), 579–588. https://doi.org/10.1093/treephys/24.5.579. Augspurger, C.K., Bartlett, E.A., 2003. Differences in leaf phenology between juvenile and Michelson, I.H., Ingvarsson, P.K., Robinson, K.M., Edlund, E., Eriksson, M.E., Nilsson, O., adult trees in a temperate deciduous forest. Tree Physiol. 517–525. Jansson, S., 2018. Autumn senescence in aspen is not triggered by day length. Baeten, L., Sercu, B., Bonte, D., Vanhellemont, M., Verheyen, K., 2015. Intraspecific Physiol. Plant. 162 (1), 123–134. https://doi.org/10.1111/ppl.12593. variation in flowering phenology affects seed germinability in the forest herb Primula Muggeo, V.M.R., 2008. segmented: an R Package to Fit Regression Models with Broken- elatior. Plant Ecol. Evol. 148 (2), 283–288. https://doi.org/10.5091/plecevo.2015. Line Relationships. R. News 8, 7. 1089. Müller, M., Seifert, S., Finkeldey, R., 2017. Comparison and confirmation of SNP-bud Barbaroux, C., Bréda, N., Dufrêne, E., 2003. Carbohydrate reserves in adult trees of two burst associations in European beech populations in Germany. Tree Genet. Genomes contrasting broad-leaved species (Quercus petraea and Fagus sylvatica). New 13 (3). https://doi.org/10.1007/s11295-017-1145-9. Phytologist 11. Pérez-de-Lis, G., Rossi, S., Vázquez-Ruiz, R.A., Rozas, V., García-González, I., 2016. Do Brelsford, C.C., Nybakken, L., Kotilainen, T.K., Robson, T.M., 2019. The influence of changes in spring phenology affect earlywood vessels? Perspective from the xylo- spectral composition on spring and autumn phenology in trees. Tree Physiol. 39 (6), genesis monitoring of two sympatric ring-porous oaks. New Phytologist 209 (2), 925–950. https://doi.org/10.1093/treephys/tpz026. 521–530. https://doi.org/10.1111/nph.13610. Campioli, M., Vicca, S., Luyssaert, S., Bilcke, J., Ceschia, E., Chapin III, F.S., Ciais, P., Piao, S., Liu, Q., Chen, A., Janssens, I.A., Fu, Y., Dai, J., Liu, L., Lian, X., Shen, M., Zhu, X., Fernández-Martínez, M., Malhi, Y., Obersteiner, M., Olefeldt, D., Papale, D., Piao, 2019. Plant phenology and global climate change: current progresses and challenges. S.L., Peñuelas, J., Sullivan, P.F., Wang, X., Zenone, T., Janssens, I.A., 2015. Biomass Glob. Chang Biol. 25 (6), 1922–1940. https://doi.org/10.1111/gcb.14619. production efficiency controlled by management in temperate and boreal ecosystems. Possen, B.J.H.M., Rousi, M., Silfver, T., Anttonen, M.J., Ruotsalainen, S., Oksanen, E., Nat. Geosci. 8 (11), 843–846. https://doi.org/10.1038/ngeo2553. Vapaavuori, E., 2014. Within-stand variation in silver birch (Betula pendula Roth) Campioli, Matteo, Vincke, C., Jonard, M., Kint, V., Demarée, G., Ponette, Q., 2012. phenology. Trees 28 (6), 1801–1812. https://doi.org/10.1007/s00468-014-1087-x. Current status and predicted impact of climate change on forest production and Prislan, P., Gričar, J., Čufar, K., 2014. Wood Sample Prep. Micros. Anal. 8. 7 L.J. Marchand, et al. Agricultural and Forest Meteorology 290 (2020) 108031 Prislan, P., Gričar, J., de Luis, M., Smith, K.T., Čufar, K., 2013. Phenological variation in Vitasse, Y., Delzon, S., Dufrêne, E., Pontailler, J.-.Y., Louvet, J.-.M., Kremer, A., Michalet, xylem and phloem formation in Fagus sylvatica from two contrasting sites. Agric. For. R., 2009. Leaf phenology sensitivity to temperature in European trees: do within- Meteorol. 180, 142–151. https://doi.org/10.1016/j.agrformet.2013.06.001. species populations exhibit similar responses. Agric. For. Meteorol. 149 (5), 735–744. Puchałka, R., Koprowski, M., Gričar, J., Przybylak, R., 2017. Does tree-ring formation https://doi.org/10.1016/j.agrformet.2008.10.019. follow leaf phenology in Pedunculate oak (Quercus robur L.). Eur. J. For. Res. 136 Vitasse, Y., François, C., Delpierre, N., Dufrêne, E., Kremer, A., Chuine, I., Delzon, S., (2), 259–268. https://doi.org/10.1007/s10342-017-1026-7. 2011. Assessing the effects of climate change on the phenology of European tempe- Rossi, S., Menardi, R., Anfodillo, T., 2006. Trephor: a New Tool for Sampling Microcores rate trees. Agric. For. Meteorol. 151 (7), 969–980. https://doi.org/10.1016/j. from tree stems. IAWA J. 27 (1), 89–97. https://doi.org/10.1163/22941932- agrformet.2011.03.003. 90000139. Weih, M., 2009. Genetic and environmental variation in spring and autumn phenology of Royston, P., 1982. Algorithm AS181: the W test for Normality. Appl. Stat. 31, 176–180. biomass willows (Salix spp.): effects on shoot growth and nitrogen economy. Tree Vitasse, Y., Basler, D., 2013. What role for photoperiod in the bud burst phenology of Physiol. 29 (12), 1479–1490. https://doi.org/10.1093/treephys/tpp081. European beech. Eur. J. For. Res. 132 (1), 1–8. https://doi.org/10.1007/s10342-012- Wilkinson, L., 2005. The Grammar of Graphics, 2sd edition. Springer-Verlag. 0661-2. 8