Étude de quelques mécanismes écophysiologiques
associés aux changements temporels de croissance de
la forêt boréale du nord-est du Canada
Thèse
Matthieu Baret
Doctorat en sciences forestières
Philosophiæ Doctor, (Ph.D.)
Québec, Canada
Étude de quelques mécanismes écophysiologiques
associés aux changements temporels de croissance
de la forêt boréale du nord-est du Canada
Thèse
Matthieu Baret
Sous la direction de :
David Pothier, directeur de recherche
Steeve Pepin, codirecteur de recherche
iii
Résumé
Le développement temporel de la plupart des peuplements forestiers est caractérisé par une période de croissance rapide jusqu’à la fermeture de la canopée, laquelle est généralement suivie d’une baisse de croissance dont les mécanismes sous-jacents restent encore mal expliqués. L’objectif général de cette thèse était ainsi d’explorer certains mécanismes qui pouvaient partiellement expliquer ce déclin en croissance des forêts lié à l’âge dans la zone boréale du nord-est du Canada. Le premier chapitre visait à tester l’hypothèse que la proportion des ressources allouées aux racines augmente avec le temps en réaction à une diminution de la disponibilité des nutriments du sol causée par une accumulation de matière organique, diminuant ainsi la croissance aérienne des arbres. Les résultats basés sur des mesures de respiration du sol et de croissance annuelle des arbres supportaient partiellement l’hypothèse, suggérant que d’autres mécanismes étaient impliqués dans le déclin en croissance des vieilles forêts. Le second chapitre a permis d’améliorer la compréhension de la dynamique temporelle de croissance des arbres selon leur classe sociale et de son effet sur la croissance globale du peuplement. Des composantes fonctionnelles du taux de croissance relatif des arbres et la dominance de croissance des peuplements (une mesure de la contribution relative des arbres selon leur taille à la croissance globale du peuplement) ont été quantifiées le long des stades de développement des peuplements. Les résultats ont montré que d’une manière générale, le déclin de la dominance de croissance observé avec le temps était principalement causé par une diminution de l’acquisition et d’utilisation des ressources des arbres dominants plutôt que par une augmentation des conditions de croissance des arbres non-dominants. Le dernier chapitre a permis d’affiner ce résultat en testant l’hypothèse de limitation hydraulique qui stipule qu’une augmentation des contraintes hydrauliques liée à la taille des arbres explique leur diminution en croissance. Des mesures de flux de sève dans les troncs et de composition en isotopes de carbone et d’oxygène dans les aiguilles de l’année courante ont mis en évidence une augmentation des contraintes hydrauliques chez les arbres dominants au cours du temps, expliquant ainsi en partie leur baisse de croissance et celle des peuplements liés à l’âge. Ces trois chapitres ont ainsi permis d’identifier certains mécanismes impliqués dans les changements temporels de croissance de la forêt boréale du nord-est du Canada, information capitale pour une gestion durable des forêts.
iv
Abstract
The development over time of most forests is characterised by an early fast growing period until canopy closure, which is generally followed by a decreasing growth rate whose underlying mechanisms remain unclear. The general objective of this thesis was thus to explore some mechanisms that could partially explain the observed age-related productivity decline of northeastern boreal Canadian forest stands. The first chapter aimed to test the hypothesis that the proportion of resources that are allocated to roots increases with stand age as a response to a decrease in nutrient availability related to the long-term accumulation of organic matter, thus decreasing stem growth. Results based on soil respiration and annual tree growth measurements partly supported the hypothesis, suggesting that other mechanisms were involved in the growth decline of old-growth forest stands. The second chapter improved our understanding of the temporal tree growth dynamics based on their social class and its effect on total stand growth. Some functional components of the relative growth rate of trees and stand growth dominance (a measure of the relative contribution of different sized trees to total stand growth) were quantified along development stages. Results showed that overall, the observed decrease in stand growth dominance with increasing age was explained mainly by declining resource acquisition and utilization in dominant trees rather than through improved growth conditions of non-dominant trees. The last chapter refined these results by testing the hydraulic limitation hypothesis, which states that an increase of hydraulic constrains in tall trees is related to their decreasing growth with time. Tree-level sap flow and foliar C and O isotope abundance measurements highlighted an increasing hydraulic constrains in dominant trees over time, thus partly explaining their declining growth and the aged-related stand growth decline. These three chapters allowed us to identify some of the mechanisms involved in the temporal growth changes of the northeastern boreal forest, a crucial information for a sustainable forest management.
v
Table des matières
Table des matières ... v
Liste des tableaux ... vii
Liste des figures ... viii
Chapitre 1 Long-term changes in belowground and aboveground resource allocation of boreal forest stands ... 7
Résumé ... 8
Abstract ... 9
Introduction ... 10
Material and methods ... 13
Study area ... 13
Sites characteristics ... 13
Soil CO2 flux measurements ... 16
Statistical analysis ... 18
Results ... 19
Ra/ANPPwood ratio ... 19
Heterotrophic respiration rate ... 20
Changes as a function of the organic layer thickness ... 22
ANPPwood, leaf area index and ANPPwood per unit leaf area across the chronosequence ... 24
Organic layer thickness, foliar nitrogen concentration and C/N ratio ... 24
Discussion ... 26
Ra/ANPPwood ratio and Rh as respective proxies for belowground-to-aboveground resource allocation and decomposition rate ... 26
Temporal changes in Ra/ANPPwood ratio ... 28
Conclusion... 31
Acknowledgements ... 32
Chapitre 2 Long-term changes in stand growth dominance as related to resource acquisition and utilization in the boreal forest ... 33
Résumé ... 34
Abstract ... 35
Introduction ... 36
Material and methods ... 38
Study area ... 38
Site and tree characteristics ... 38
Growth dominance coefficient ... 41
vi
Tree allocation to leaf biomass ... 43
Statistical analysis ... 43
Results ... 44
Temporal changes in growth dominance coefficient and relative growth rate ... 44
Relative growth rate components as a function of time-since-last fire ... 46
Leaf biomass partitioning as a function of time elapsed since the last fire ... 48
Discussion ... 50
Temporal changes in stand growth dominance ... 50
Leaf morphology ... 51
Leaf biomass partitioning ... 51
Leaf efficiency ... 52
Conclusion... 54
Acknowledgements ... 55
Chapitre 3 Hydraulic limitations in dominant trees as a contributing mechanism to the age-related growth decline of boreal forest stands ... 56
Résumé ... 57
Abstract ... 58
Introduction ... 59
Material and methods ... 61
Study area ... 61
Site and tree characteristics ... 61
Leaf carbon and oxygen isotope composition ... 62
Water flux estimations ... 65
Statistical analysis ... 67
Results ... 68
Transpiration of different-sized trees and growth dominance ... 68
Leaf isotope composition ... 72
Discussion ... 76
Conclusion... 79
Acknowledgements ... 80
vii
Liste des tableaux
Table 1-1Characteristics of the study sites ... 15 Table 1-2. Parameter estimates (± SE, standard errors of the estimates) and related statistics for the prediction equation of log10ln(Ra/ANPPwood (Ra / ANPPwood) as a function of the time since last fire.
The value of R2 associated with this model is 0.45. ... 19
Table 1-3. Parameter estimates (± SE) and related statistics for the prediction equation of log10 Rh as
a function of the time since last fire. The value of R2 associated with this model is 0.79. ... 21
Table 1-4. Parameter estimates (± SE) and related statistics for the prediction equation of log10 (Rh)
as a function of the thickness of organic layer. The value of R2 associated with this model is 0.90. 24
Table 2-1. Proportion (% of the total stand basal area) of black spruce (bS), balsam fir (bF), white spruce (wS), jack pine (jP), white birch (wB) and trembling aspen (As) along the chronosequence. ... 38 Table 3-1. Characteristics of the stands sampled for measurements of leaf carbon and oxygen isotope composition. ... 64 Table 3-2. Diameter at breast height (DBH) and height of the five black spruce trees selected per stand for sap flow measurements. ... 66 Table 3-3 Values of SEL-DBH and the related statistics for the five sampled stands over each sampling
viii
Liste des figures
Figure 1. Changement temporel d’accumulation la biomasse vivante (a) et de l’accroissement en biomasse (b) pour des pessières noires de la forêt boréale du nord-est du Canada (Garet et al., 2009) en fonction des stades de développement décrits par Bormann et Likens (1979) ... 2 Figure 1-1. Changes in autotrophic respiration rate/ANPPwood ratio as a function of time since the last
fire. Each point represents the season average value calculated from each plot and bars correspond to standard errors. ... 20 Figure 1-2. Changes in heterotrophic respiration rate as a function of time since the last fire. In order to account for variation in soil water content, this variable was set to its mean value for each forest stand. Each point represents the season average value calculated from each plot and bars correspond to standard errors. The line was drawn from Eq. 1 using the average values of the covariate. ... 22 Figure 1-3. Changes in (a) Ra/ANPPwood ratio (expressed in μmol CO2 m-2 s-1/ kg C ha-1 yr-1) a) and
(b) heterotrophic respiration rate (Rh, expressed in μmol CO2 m-2 s-1) b) as a function of the thickness
of organic layer. Each point represents the season average value calculated from each plot and bars correspond to standard errors. The line was drawn from Eq. 1 using the average values of the covariate. ... 23 Figure 1-4. Changes in a) ANPPwood, b) LAI, and c) ANPPwood per unit leaf area (ANPPwood/LA) as a
function of time since the last fire. ... 25 Figure 1-5. Relationship between Ra/ANPPwood ratio and TBCF/ANPPwood + foliage ratio. Each point
corresponds to a stand as described in Vogel et al. (2008) ... 27 Figure 2-1. Tree diameter distribution of black spruce (bS), balsam fir (bF) and minority species (Other) according to the age classes <100 yr. (a), 101-500 yr. (b) and >501 yr. (c) of the chronosequence. ... 40 Figure 2-2. Growth dominance coefficient as a function of time-since-last fire. ... 44 Figure 2-3. Relative growth rate (a) and differences in relative growth rate between dominant and non-dominant trees (b) as a function of the time-since-last fire. ... 45 Figure 2-4. Growth dominance coefficient as a function of differences in relative growth rate between dominant and non-dominant trees. ... 46 Figure 2-5. Differences in specific leaf area (a) and net assimilation rate (b) between dominant and non-dominant trees as a function of the time that had elapsed since the last fire. The number of observations for SLA was lower than for NAR, because only data from a sub-sample of the 13 plots were available for SLA measurements. ... 47 Figure 2-6. Differences in nitrogen leaf area (a) and nitrogen use efficiency (b) between dominant and non-dominant trees as a function of the time that had elapsed since the last fire. ... 48 Figure 2-7. Differences in leaf weight ratio between dominant and non-dominant trees (a), mean stand residuals (b) and differences in mean stand residuals between dominant and non-dominant trees (c) of the relationship of leaf biomass vs aboveground woody biomass (Eq. 5) as a function of the time that had elapsed since the last fire. ... 49 Figure 3-1. Whole-tree transpiration per unit leaf area (EL) in July 2012 as a function of tree diameter
ix data point represents a daily EL value and the dashed lines correspond to the linear regressions
between EL and tree DBH. ... 69
Figure 3-2. Slope of the relationship between whole-tree transpiration per unit leaf area and tree diameter (SEL-DBH) in June, July and August 2012, as a function of the time that had elapsed since the
last fire. Error bars indicate standard errors and the dotted line corresponds to a zero slope value. . 70 Figure 3-3. Growth dominance coefficient (GD) as a function of the slope between whole-tree transpiration per unit leaf area and stem diameter (SEL-DBH) that was calculated for July 2012. The
dashed line represents a simple linear regression. ... 72 Figure 3-4. Leaf C isotope abundance (δ13C) of dominant and non-dominant trees as the
time-since-last-fire. Each data point corresponds to a foliar δ13C value of an individual tree. ... 73
Figure 3-5. Leaf C isotope abundance (δ13C) as a function of leaf O isotope composition (δ18O) of
dominant trees in the chronosequence. Each data point corresponds to foliar δ13C and δ18O values of
an individual dominant tree (up to three trees per stand were sampled) in stands that were aged < 120 y (ntrees = 10), between 120 and 200 y (ntrees = 7), and > 200 y (ntrees = 8). ... 74
Figure 3-6. Growth dominance coefficient as a function of leaf C isotope abundance (δ13C) in
dominant trees. Each data point represents the mean values of δ13C in dominant trees (n
max = 1 to 3)
and the growth dominance coefficient in the corresponding stand. The dashed line represents the linear regression. ... 75
x Liste des abréviations
δ13C: isotopic ratio of 13C to 12C
δ18O: isotopic ratio of 18O to 16O
ANPPwood : aboveground net primary productivity of wood
C/N: carbon-nitrogen ratio of the organic layer DBH : diameter at breast height
EL: transpiration rate per unit leaf area
GD: growth dominance LA: projected leaf area LAI: leaf area index LWR: leaf weight ratio NAR: net assimilation rate NLA: nitrogen per unit leaf area NUE: nitrogen use efficiency Ra: autotrophic respiration
RGR : relative growth rate Rh: heterotrophic respiration
Rs : soil respiration
SA: tree sapwood area
SEL-DBH: slope of the relationship between whole-tree transpiration per unit leaf area and tree diameter
SLA: specific leaf area
TBCF: total belowground carbon flux TOL: thickness of organic layer TSF : time since the last fire
xi
Remerciements
L’aboutissement de cette thèse n’aurait pas été possible sans l’encadrement de mon directeur de recherche, David Pothier. Je le remercie chaleureusement pour son soutien, ses conseils, sa disponibilité et pour la confiance qu'il m'a accordée durant toutes ces années. Merci également à mon codirecteur, Steeve Pepin, pour ses conseils toujours pertinents et plus particulièrement pour son assistance dans les parties techniques.
Mes remerciements vont aussi à la chaire de recherche industrielle CRSNG-Université Laval en sylviculture et faune qui a financé ce projet.
Merci à tous les aides de terrain et professionnels de recherche de l’Université Laval qui m’ont apporté leur aide à un moment ou un autre: Frédéric Atshit, Alice Bernier-Banville, Alain Brousseau, Thomas Bourbonne, Marie Coyea, Audrey-Anne Després, Philippe Goulet et Ivan Savard. J'aimerais aussi exprimer ma reconnaissance envers William F.J. Parsons pour la révision des articles en anglais. Merci également à tous mes collègues de l’Abitibi-Price qui ont rendu cette expérience très agréable, particulièrement à Aude, Arnaud et Sébastien. Je leur suis reconnaissant de cette bonne ambiance et de nos soirées de détente. J’adresse un salut particulier à Charles Ward pour son aide et pour avoir répondu à mes multiples questions sur sa base de données d’inventaire forestier sans laquelle je n’aurais rien pu faire.
Merci à mes parents et à ma sœur Séverine qui m’ont souvent rendu visite Québec malgré la distance. Enfin, merci à Astrid pour son soutien, ses encouragements et pour toutes les autres choses qu’il serait trop long de lister ici.
xii
Avant-propos
Ce document est présenté sous la forme d’une thèse par articles, rédigés en anglais. En tant qu’auteur principal j’ai élaboré les hypothèses de recherche, mis en place le dispositif expérimental, collecté les données, fait les analyses statistiques, interprété les résultats et rédigé les manuscrits. Chacune de ces étapes ont été supervisées par mon directeur de recherche David Pothier et mon codirecteur, Steeve Pepin. Le troisième coauteur des deux premiers articles, Charles Ward, a réalisé les inventaires forestiers dont j’utilise les données dans ces deux articles. Les trois articles ont les références suivantes :
Chapitre I
Baret, M., Pepin, S., Ward, C. and Pothier, D. 2015. Long-term changes in belowground and aboveground resource allocation of boreal forest stands. Forest Ecology and Management 350: 62-69
Chapitre II
Baret, M., Pepin, S., Ward, C. and Pothier, D. 2017. Long-term changes in stand growth dominance as related to resource acquisition and utilization in the boreal forest. Forest Ecology and Management 400: 408-416
Chapitre III
Baret, M., Pepin, S. and Pothier, D. 2017. Hydraulic limitations in dominant trees as a contributing mechanism to the aged-related growth decline of boreal forest stands. L’article sera soumis prochainement à Forest Ecology and Management.
Coauteurs des chapitres :
Steeve Pepin : Département des sols et de génie agroalimentaire, Université Laval, Pavillon de l’Envirotron, 2480 boulevard Hochelaga, Québec, QC, G1V 0A6, Canada.
steeve.pepin@fsaa.ulaval.ca
David Pothier : Centre d’étude de la forêt, Département des sciences du bois et de la forêt, Pavillon Abitibi-Price, 2405 rue de la Terrasse, Université Laval, Québec, QC, G1V 0A6, Canada.
xiii Charles Ward : Direction de la recherche forestière, Ministère des Forêts, de la Faune et des Parcs du Québec 2700 rue Einstein, Québec, QC, G1P 3W8, Canada.
1
Introduction générale
La forêt boréale canadienne représente 28 % de la zone boréale mondiale et s’étend du Yukon et du nord de la Colombie-Britannique jusqu’à Terre-Neuve-et-Labrador (Ressources naturelles Canada, 2017). De par son importance sur le plan environnemental (stockage du carbone à l’échelle mondiale, régulation du climat), culturel (tourisme), économique et social (exploitation forestière, minière, hydroélectrique), un aménagement à long terme est capital afin de préserver les services fournis par la forêt boréale. La mise en évidence du lien important entre la biodiversité et le bon fonctionnement des écosystèmes (Johnson, 1992; Chapin et al., 1997; Swartz et al., 2000) et d’un réchauffement climatique à long terme (Kattenberg et al., 1996) à la fin du siècle dernier a permis d’accélérer la prise de conscience collective de tenir compte des valeurs environnementales dans l’économie. Cette prise de conscience impliquait notamment de trouver des méthodes alternatives dans la manière d’exploiter la forêt boréale afin de s’éloigner du dogme du rendement soutenu centré sur la production continue de bois (Gauthier et al., 2008). Au cours des deux dernières décennies, les connaissances scientifiques sur la structure et le fonctionnement des écosystèmes forestiers ont énormément progressé, permettant la mise en place de plans d’aménagement forestier s’inscrivant dans un contexte d’aménagement durable des forêts (Gauthier et al., 2008). Dans cet ordre d’idée, le concept d’aménagement écosystémique qui consiste à pratiquer un aménagement forestier apte à maintenir la diversité biologique et la viabilité des écosystèmes (Ministère des Ressources naturelles et de la Faune du Québec, 2017) a émergé. Le principe est de maintenir les forêts aménagées dans un état proche de celui des forêts naturelles, réduisant ainsi les pertes de biodiversité et de fonctionnalité des écosystèmes. Au Québec, à la suite de la commission d’étude sur la gestion de la forêt publique québécoise (Commission Coulombe, 2004), ce principe est maintenant au cœur des politiques de gestion des forêts publiques. Les pratiques sylvicoles doivent s’inspirer de la dynamique naturelle de ces forêts afin de récolter le bois sans compromettre le capital forestier à long terme.
Or, la dynamique forestière est soumise aux perturbations naturelles et anthropiques, principaux responsables des changements de structure, de composition et de fonctions des écosystèmes forestiers (Shugart et al., 1992; Bergeron et al., 1998). Bormann et Likens (1979) ont proposé un modèle qui décrit les changements temporels d’accumulation de la biomasse vivante d’une forêt après une coupe à blanc (en supposant qu’aucune autre perturbation n’intervienne par la suite), dont les résultats sont présentés à la Figure 1 en fonction d’ajustements apportés par Garet et al. (2009) pour la forêt boréale du nord-est du Canada :
2 Figure 1. Changement temporel d’accumulation la biomasse vivante (a) et de l’accroissement en biomasse (b) pour des pessières noires de la forêt boréale du nord-est du Canada (Garet et al., 2009) en fonction des stades de développement décrits par Bormann et Likens (1979)
Après une perturbation majeure, une phase de réorganisation s’installe sur une période de 10 à 20 ans pendant laquelle les pousses d’arbres vont commencer à grandir jusqu’à dominer graduellement la station. Ensuite, la biomasse des arbres vivants augmente rapidement pendant la phase d’accumulation pour atteindre une valeur maximale (Figure 1a, point B) à un âge correspondant approximativement à la longévité moyenne de la cohorte d’établissement. Cette phase est suivie d’une phase de transition au cours de laquelle la biomasse vivante diminue jusqu’à l’atteinte d’un plateau. Cette phase d’équilibre a été observée par Bormann et Likens (1979) dans les forêts feuillues du nord-est américain, là où les sols sont plus riches et où la décomposition de la matière organique nord-est plus
3 rapide qu’en forêt boréale. Des variations de ce schéma d’accumulation de biomasse ont été observées en forêt boréale, le dernier stade pouvant présenter un déclin continu (Paré and Bergeron, 1995) à cause notamment d’une baisse de la disponibilité des nutriments du sol (Van Cleve and Yarie, 1986; Pastor et al., 1987). Cependant dans notre région d’étude, cet état d’équilibre a été observé (Garet et al., 2009) et peut théoriquement durer jusqu’à une prochaine perturbation majeure, le maintien de bonnes conditions de drainage du sol au cours du temps expliquant en partie ce schéma (Ward et al., 2014).
Les changements temporels d’accumulation de biomasse (Figure 1a) s’accompagnent de changement du taux de croissance (Figure 1b): celui-ci va augmenter au début du développement, puis atteindre un maximum (Figure 1b, point A) qui coïncide généralement avec l’atteinte de la valeur maximale de surface foliaire du peuplement, pour ensuite décroitre graduellement jusqu’au point B (Figure 1b) qui marque la fin de la phase d’accumulation de biomasse (Assmann, 1970; Ryan et al., 1997). Ensuite, le couvert dominant meurt progressivement alors que les espèces en régénération le remplacent peu à peu. Le taux d’accroissement devient donc négatif, puis retrouve un équilibre lorsque la majorité des arbres de la première cohorte a disparu et qu’une ou plusieurs cohortes composent l’étage dominant. Bien que le déclin du taux de croissance du peuplement une fois la canopée fermée soit un phénomène observé dans de très nombreuses études (Turner and Long, 1975; Grier et al., 1989; Taylor and MacLean, 2005; McMahon et al., 2010; Xu et al., 2012), les mécanismes sous-jacents sont toujours sujets à débat (Ryan et al., 2004). La mise en évidence des causes associées au déclin de l’accroissement en biomasse et au déclin de la biomasse vivante sont d’autant plus importantes en forêt boréale qui, par son vaste territoire, a une influence énorme sur le bilan du carbone à l’échelle mondiale avec environ 20 % du carbone total piégé annuellement (Pan et al., 2011). Pour établir une gestion durable des forêts s’appuyant sur des prévisions à long terme de la croissance et de la production forestière qui permettent de déterminer un taux annuel de récolte sans compromettre le capital forestier, une meilleure connaissance des mécanismes responsables de ces variations de productivité est nécessaire afin de mieux prévoir ces déclins et pouvoir atténuer leurs effets.
La première hypothèse qui tentait d’expliquer ce déclin en croissance était que la respiration de l’aubier augmentait avec la taille des arbres (Yoda et al., 1965; Kira and Shidei, 1967; Whittaker and Woodwell, 1967). Cette hypothèse a depuis été réfutée (Ryan and Waring, 1992; Ryan et al., 2004), le taux de respiration de maintenance des tissues du bois n’utilisant seulement que de 5 à 12 % de la fixation annuelle du carbone (Ryan et al., 1995). De très nombreuses hypothèses ont alors été avancées pour expliquer la baisse temporelle de croissance des peuplements, ce qui inclut un contrôle génétique du changement de morphologie et de physiologie du feuillage (Day et al., 2001; Day et al.,
4 2002), une diminution de la disponibilité des nutriments du sol (Binkley et al., 1995), une augmentation de l’allocation du carbone à la production racinaire (Gower et al., 1992; Haynes and Gower, 1995), un changement dans les relations de dominance entre les arbres au cours du temps (Binkley, 2004; Binkley et al., 2006), une diminution de la photosynthèse causée par une diminution de la conductance hydraulique spécifique foliaire liée à la taille des arbres (Ryan et al., 1997; Ryan et al., 2006) et une réduction de la turgescence des cellules des bourgeons et des feuilles des grands arbres (Woodruff et al., 2004). Tous ces mécanismes ont été au moins partiellement validés par des résultats expérimentaux. Il est donc possible qu’aucun mécanisme seul puisse expliquer le déclin en croissance des forêts lié à l’âge, mais plutôt une combinaison de mécanismes spécifiques à un peuplement agissant seul ou en interaction sur différentes échelles spatiales.
L’objectif général de cette thèse est d’identifier certains mécanismes responsables des changements temporels de croissance et de productivité de la forêt boréale du nord-est du Canada. Les trois mécanismes suivants ont été étudiés :
- Augmentation de l’allocation du carbone à la production racinaire au cours du temps (chapitre 1);
- Changement dans les relations de dominance entre les arbres au cours du temps (chapitre 2); - Diminution de la photosynthèse causée par une diminution de la conductance hydraulique
spécifique foliaire liée à la taille des arbres au cours du temps (chapitre 3).
Ces mécanismes ont été choisis car en agissant sur une longue période, ils avaient une forte probabilité d’expliquer le déclin en croissance des forêts compte tenu des particularités régionales de la forêt boréale du nord-est du Canada. En effet, contrairement au reste de la forêt boréale qui présente un intervalle de retour de feu qui se situe généralement entre 100 et 300 ans (Heinselman, 1981), le cycle de cette perturbation majeure peut être allongé jusqu’à 500 ans (Bouchard et al., 2008) à cause d’un climat plus humide et des précipitations plus importantes qu’à l’ouest (Foster, 1985). Par conséquent, cette région est composée de 65 à 70 % de vieilles forêts (Côté et al., 2010), i.e. de peuplements qui ont atteint la phase d’équilibre décrite dans le modèle de Bormann et Likens (1979). Les vieilles forêts sont structurellement différentes des peuplements forestiers équiens (i.e. un peuplement où les arbres sont dans la même classe d’âge) communément retrouvés en forêt boréale et régulés par des cycles de feux plus courts (Boucher et al., 2003). Un cycle de feu de 500 ans étant largement supérieur à la durée de vie des deux espèces qui dominent ces forêts, l’épinette noire - Picea mariana (Mill.) BSP- et le sapin baumier - Abies balsamea (L.) Mill. -, les vieilles forêts sont composées d’arbres n’appartenant pas à la cohorte initialement établie après la perturbation (peuplements inéquiens, i.e. les arbres appartiennent à plus de deux classes d’âge). L’abondance de vieilles forêts dans la région
5 permet de vérifier les mécanismes de croissance sur une période beaucoup plus longue que les précédentes études qui portaient sur la baisse d’accroissement en biomasse qui suit la fermeture de la canopée de peuplements équiens à un âge relativement peu avancé (Murty et al., 1996; Ryan et al., 1997; Binkley, 2004; Ryan et al., 2006; Fernández and Gyenge, 2009). De plus, les vieilles forêts boréales n’ont jamais été prises en compte dans les précédentes études portant sur les processus de croissance, l’âge maximal observé étant d’environ 200 ans (Bond-Lamberty et al., 2004; Vogel et al., 2008; Xu et al., 2012).
Dans le chapitre 1, nous avons exploré un mécanisme relié aux changements temporels des conditions édaphiques lors du développement d’une forêt. Après un feu, les conditions édaphiques sont favorables à une mise en place rapide de la régénération car en réduisant substantiellement l’épaisseur de la couche de matière organique du sol, le feu est associé à une hausse subite des nutriments disponibles pour la régénération (Simard et al., 2001). Ensuite, l’épaisseur de la couche de matière organique a tendance à augmenter avec le temps. Cette accumulation peut conduire au phénomène de paludification observé par Fenton et al. (2005) en Abitibi-Témiscamingue ou se stabiliser plus rapidement, comme observé par Ward et al. (2014) dans notre zone d’étude. Une épaisse couche de matière organique est ainsi souvent observée dans les vieilles forêts (Simard et al., 2009). Puisqu’une épaisse couche de matière organique diminue la température du sol et la disponibilité en nutriments (Bonan and Shugart, 1989), les conditions de croissance devraient diminuer au cours du temps. Comme il a déjà été observé qu’une diminution de la fertilité du sol modifiait l’allocation du carbone à la partie racinaire (Gower et al., 1992; Haynes and Gower, 1995; Vanninen and Mäkelä, 2005), ce mécanisme pourrait expliquer les changements temporels de croissance de la partie aérienne des arbres de la forêt boréale du nord-est du Canada. Afin de caractériser ce mécanisme sur le long terme, une approche descriptive par chronoséquence a été utilisée. Ce type d’approche consiste à sélectionner une gamme de peuplements d’âges différents afin d’avoir un aperçu des changements des paramètres d’intérêt au cours du temps. Ainsi, à l’aide d’une chronoséquence après feu couvrant une période d’environ 1000 ans, les changements temporels d’allocation des ressources entre les parties aériennes et souterraines des arbres ont été étudiés.
Le chapitre 2 porte sur un mécanisme impliquant les changements temporels de dominance de croissance d’un peuplement, c’est-à-dire la contribution relative des arbres de différentes tailles à la croissance totale du peuplement. Au cours de son développement, une forêt est caractérisée par une différenciation des arbres en différentes classes sociales (Olivier and Larson, 1996). Cette différenciation, reliée à des taux de croissance différents entre les arbres de tailles variées, est notamment contrôlée par l’approvisionnement, l’efficacité d’utilisation et la compétition entre les
6 arbres pour les ressources (eau, lumière, nutriments du sol). Dans cet ordre d’idée, Binkley et al. (2002) et Binkley (2004) ont proposé que les changements de structure observés au cours du développement d’une forêt contribueraient au déclin de la croissance du peuplement en augmentant les différences d’efficacité d’utilisation des ressources (définie comme la production de bois par unité de ressources utilisées) entre des arbres dominants et non-dominants. En utilisant la chronoséquence après feu couvrant une période d’environ 1000 ans, nous avons vérifié si ce changement temporel de dominance de croissance des peuplements était un mécanisme qui participait au déclin de croissance des forêts boréale du nord-est du Canada, via une différence dans l’acquisition et l’utilisation des ressources entre les arbres dominants et non-dominants.
Dans la continuité du chapitre 2, nous nous sommes intéressés à un mécanisme lié à la taille et à l’âge des arbres dans le troisième chapitre. Chez les grands et vieux arbres, une baisse de photosynthèse (et donc de croissance) serait causée par une augmentation des contraintes hydrauliques de l’arbre (Yoder et al., 1994; Ryan and Yoder, 1997; Ryan et al., 2006). En effet, la conductance stomatique de ces arbres serait diminuée par une réduction de la conductance hydraulique spécifique foliaire, en réaction à différents facteurs tels que l’augmentation de la longueur du chemin parcouru par l’eau (dans les racines, le tronc et les branches), une augmentation de la densité de l’aubier (Bowman et al., 2005), une diminution de la conductance de l’aubier (Pothier et al., 1989) ou l’augmentation du potentiel hydrique gravitationnel. Cette réduction de conductance hydraulique en fonction de la hauteur des arbres ayant notamment été mise en évidence chez de nombreuses espèces d’arbres (Delzon et al., 2004; Martinez-Vilalta et al., 2006; Drake et al., 2010), l’hypothèse de limitation hydraulique à la photosynthèse est devenue populaire pour expliquer le déclin en croissance des forêts lié à l’âge. L’approche par chronoséquence nous a permis de déterminer dans quelle mesure une limitation hydraulique à la photosynthèse causée par la taille des arbres pouvait expliquer les changements temporels de croissance en forêt boréale du nord-est du Canada. Tester cette hypothèse dans notre zone d’étude est d’autant plus important car étant donné la présence de vieux et grands arbres dans les vieilles forêts, ce mécanisme a une forte probabilité de participer au déclin en croissance des peuplements avec le temps.
7
Chapitre 1 :Long-term changes in belowground and
aboveground resource allocation of boreal forest
stands
1
1 Version intégrale d’un article publié/ Integral version of a published paper:
Baret, M., Pepin, S., Ward, C. and Pothier, D. 2015. Long-term changes in belowground and aboveground resource allocation of boreal forest stands. Forest Ecology and Management 350: 62-69.
8
Résumé
Cette étude examine un mécanisme possiblement impliqué dans la baisse de croissance d’un peuplement après la fermeture de la canopée. Nous avons émis l’hypothèse que la proportion des ressources allouées aux racines augmentait avec l´âge du peuplement, en réaction à une diminution de la disponibilité des nutriments du sol causée par l’accumulation d’une couche de matière organique dans la forêt boréale du nord-est du Canada. Des indices basés sur la respiration du sol et la biomasse du tronc des arbres ont mis en évidence une augmentation de l’allocation des ressources aux racines au cours des 200 premières années après un feu, suivi d’une diminution dans les vieilles forêts. Le schéma inverse fut observé pour le taux de décomposition de la matière organique du sol. Le déclin en croissance pouvait ainsi être attribué à une plus grande allocation des ressources aux racines seulement dans les stades de développement précédant celui des vieilles forêts, leur faible productivité n’étant pas expliquée par ce mécanisme.
Mots-clés: Allocation relative des ressources, développement d’un peuplement, productivité forestière, forêt boréale, respiration du sol, disponibilité des nutriments du sol
9
Abstract
Age-related decline of forest stand growth is a common phenomenon, but the associated physiological causes remain uncertain. This study investigated a possible mechanism that could explain stand growth decline observed after canopy closure. We hypothesised that the proportion of resource allocation to roots increases with stand age as a response to a decrease in nutrient availability, which is related to the long-term accumulation of organic matter in boreal forests. Proxies based on soil respiration measurements and stem biomass production were used to describe temporal changes in the proportion of carbon allocated to belowground and aboveground stand components along a 1067-year post-fire chronosequence. The proportion of resources that were allocated belowground increased in the first 200 years following fire and declined thereafter. The inverse pattern was observed for the organic matter decomposition rate. Stand-level decline in wood productivity that was observed during the first 60-year period after fire can be attributed to a greater proportion of carbohydrates being allocated to roots in response to a decrease in nutrient availability. However, the relatively low productivity of old-growth stands was not associated with high belowground allocation, suggesting that other mechanisms operating at the tree- or stand-level may be involved.
Keywords: Relative resource allocation, Stand development, Forest productivity, Boreal forest, Soil respiration, Soil nutrient availability
10
Introduction
As has been observed in several studies, forest growth rate generally increases during early developmental stages, reaches a maximum that corresponds to a peak in stand leaf area, and then decreases gradually (Gholz and Fisher, 1982; Ryan and Yoder, 1997). A decline in growth rate after canopy closure is a well-known phenomenon (Assmann, 1970) but the underlying mechanisms remain unclear (Ryan et al., 2006). Because the hypothesis of an imbalance between photosynthesis and respiration was not supported by experimental results (Ryan and Waring, 1992), a decline in gross primary productivity or a shift in allocation to stem production are strong candidates for inclusion in a process that would explain the age-related decline in net primary production (Ryan et al., 2004; Drake et al., 2011). Accordingly, several authors have suggested lines of research that involve increases in nutrient limitation or hydraulic resistance (Ryan and Yoder, 1997; Weiner and Thomas, 2001), or changes in stand growth dominance (Binkley, 2004). Consequently, it is possible that no single mechanism can explain this decline which could be caused by site-specific processes that are acting alone or in interaction (Ryan et al., 2006).
In this study, we investigated a region of the boreal forest that is characterised by fire return intervals as long as 500 years (Bouchard et al., 2008), which are due to a cold climate and abundant precipitation (Foster, 1985). Consequently, this region is dominated by old-growth stands that are structurally different from even-aged stands that are commonly found in the boreal forest and regulated by short fire cycles (Boucher et al., 2003). To our knowledge, such old-growth stands have never been sampled in previous studies of boreal stand growth processes, given that the maximum observed stand age was generally less than 200 years (Bond-Lamberty et al., 2004; Vogel et al., 2008; Xu et al., 2012). Including old-growth boreal forests in these studies is important because they are relatively abundant and often characterised by a thick organic layer that results from continuous accumulation of organic matter over time (Fenton et al., 2005). Thick organic layers are known to immobilise large quantities of nutrients and to decrease soil temperature (Bonan and Shugart, 1989), thereby leading to poor growth conditions. Because a decrease in soil fertility has been shown to modify tree carbon allocation to roots (Gower et al., 1992; Haynes and Gower, 1995; Vanninen and Mäkelä, 2005), a comparison of old-growth and younger stands could highlight key processes that would be otherwise overlooked.
Tree carbon allocation to roots is difficult to measure directly. For example, excavation of entire tree root systems is time-consuming, and even if done properly, a large quantity of fine roots is usually destroyed by the hydraulic excavation technique that are used to separate roots from the soil matrix.
11 Fine root dynamics (biomass, production, mortality and turnover) may be studied using various approaches such as sequential soil coring (indirect) or the ingrowth bag (direct) method, yet both methods have their strengths and limitations (Yuan and Chen, 2012). Yet, it is possible to estimate the pattern of the proportion of resource allocation to roots at the stand level by using measurements of soil respiration (Rs), which in turn can be partitioned into autotrophic and heterotrophic respiration.
Autotrophic respiration corresponds to the metabolic activities of roots, rhizosphere and mycorrhizae which occur in all forest ecosystems, regardless of soil conditions (Hanson et al., 2000). During the growing season, autotrophic respiration is mainly controlled by concentrations of carbohydrates that have been recently fixed through photosynthesis (Ekblad and Högberg, 2001; Bowling et al., 2002) and, thus, is related to gross primary productivity (Litton et al., 2007). As autotrophic respiration is strongly related to root production (Litton et al., 2007), the proportion of resource allocation to the root system compared to that of the aboveground component of a forest stand can be approximated by the ratio of soil autotrophic respiration to tree stem biomass increment. We used this ratio as a proxy for a “root respiration to stemwood production” ratio; from a physiological perspective, this ratio would be interpreted as reflecting the differential proportion of resource allocation to roots vs. stemwood.
Heterotrophic respiration represents the microbial respiration that depends upon substrate quality and quantity (Ryan and Law, 2005). Heterotrophic respiration can thus be used as a proxy for the decomposition rate of the organic layer, which constitutes an important component of nutrient availability in the boreal forest (Bonan and Shugart, 1989). By using measurements of autotrophic and heterotrophic respiration, together with aboveground biomass increment in forest stands of different ages, it would be possible to investigate temporal changes in the proportion of resources that are allocated to roots in relation to soil resource availability. In the humid boreal forest we investigated, nutrient availability was observed to decrease with time elapsed since the last fire (Ward et al., 2014) and can thus be considered as the soil resource most likely to affect forest productivity. We investigated a possible mechanism that could explain the decline in stand growth that was observed following canopy closure in the northeastern Canadian boreal forest (Ward et al., 2014). We hypothesised that the proportion of resources that are allocated to roots increases with stand age as a response to a decrease in nutrient availability, which is related to the long-term accumulation of organic matter in boreal forests. To achieve this goal, we sub-sampled 15 stands from a chronosequence previously used by Ward et al. (2014), which covered a post-fire period of over 1000 years in the northeastern Canadian boreal forest. This study is therefore complementary to that of
12 Ward et al. (2014) who studied the temporal changes in organic layer thickness, soil temperature, nutrient availability and stand productivity.
13
Material and methods
Study area
Sites were located north of Baie-Comeau (49°07’N, 68°10’W), Quebec, Canada, in the black spruce-feather moss bioclimatic subdomain (Robitaille and Saucier, 1998). The regional climate is cold maritime, with a mean annual temperature of 1.5 °C and mean annual precipitation of 1014 mm. Snow generally represents 35% of yearly total precipitation and the growing season lasts for about 155 days. The fire return interval of the study region was estimated at 270 years (Bouchard et al., 2008).
Black spruce (Picea mariana (Mill.) BSP) and balsam fir (Abies balsamea (L.) Mill.) are the dominant canopy species in these forests, with relatively minor components of white spruce (Picea glauca (Moench) Voss.), paper or white birch (Betula papyrifera Marsh.), jack pine (Pinus banksiana Lamb.), tamarack or eastern larch (Larix laricina (Du Roi) K. Koch), and trembling aspen (Populus
tremuloides Michx.). The low frequency of fire in the area led to the creation of a forest landscape
that was composed of 65-70% old-growth, uneven-aged stands (Côté et al., 2010).
Sites characteristics
To investigate the effect of stand age on the proportion of resource allocation to roots, we used a sub-sample of a post-fire chronosequence (see Ward et al. 2014 that included 15 sites, which were aged from 17- to 1067-years-since- fire. The original chronosequence was composed of 30 stands, selected to be as similar as possible in terms of surface deposits, topographic position, exposure and drainage. A particular attention was given to select sites characterized by deep glacial tills with good drainage, which are the dominant biophysical features of the study area (Bouchard et al., 2008). We randomly selected three stands in each of five age classes (0-50 y, 51-100 y, 101-150 y, 151-200 y, > 200 y), and established one 0.04 ha circular plot in each stand. Within each plot, the number of stems per species was determined for all trees with a diameter at breast height (DBH, 1.3 m) greater than 9.0 cm. Trees with a DBH between 6.0 and 9.1 cm were inventoried in 20 sub-plots of 4 m2 that had been
established systematically within the 400 m2 plot, with five sub-plots per cardinal point.
The principal characteristics of the 15 sites are presented in Table 1-1. The thickness of the organic layer (cm), including living bryophytes, was measured at 16 locations within each plot, i.e., four locations in each of the four cardinal directions with a minimum distance of 3 m between locations.
14 In addition, determinations of the organic layer thickness were made at least 50 cm from any major obstacle (tree stump, rocks, etc.).
Time since last fire (TSF) for stands that were < 200-year-old were determined according to the historical fire map of the region, which had been prepared by Bouchard et al. (2008). They extracted basal discs from fire-scarred trees or cored several dominant trees, generally being extracted from trembling aspen and jack pine because of their rapid initial growth. TSF was calculated by subtracting the year of inventory from the year of the last fire event. For the older stands, the lifespan of individual black spruce and balsam fir trees was exceeded (Burns and Honkala, 1990); therefore, extraction of tree increment cores would have not given a precise TSF measurement, as individuals in the first cohort had likely disappeared. In such cases, 14C dating of charcoal samples from the last fire was
performed (for more details, see Barrette et al. 2013).
Aboveground net primary productivity of wood (ANPPwood) was estimated as annual production of
live stemwood biomass. Five-year wood biomass production was estimated from increment cores that had been taken at 1.3 m and oriented toward the plot centre for all trees with DBH greater than 9.0 cm, using the equations of Lambert et al. (2005) that are valid for a large part of the Canadian territory, including the study area. These equations estimate wood biomass from DBH and were constructed with trees whose DBH ranged from 1.5 to more than 40 cm in the case of balsam fir and black spruce. For saplings with DBH between 5.1 and 9.0 cm, a sub-sample of at least three individuals per tree species was selected for increment coring. Leaf area index (LAI) in each plot was calculated using local species-specific allometric relationships between projected leaf area and sapwood area for each tree that had been inventoried in the plot using the equations presented by Ward et al. (2014). On one site (S-26), we were not able to determine ANPPwood or LAI because there were only saplings with
DBH < 5 cm. For all other sites, the estimated values of ANPPwood and LAI allowed us to compute
the annual aboveground wood biomass produced per unit of projected leaf area (ANPPwood/LA)
15
Table 1-1Characteristics of the study sites
*Results originally published in Ward et al. 2014Note: Ra is the mean values of autotrophic respiration, Rh the mean value of heterotrophic respiration, LAI the
leaf area index, ANPPwood the aboveground net primary productivity of wood, TOL the thickness of the organic layer, Foliar N the foliar nitrogen concentration
and C/N the carbon-nitrogen ratio of the organic layer.
Station Time since last fire Class
age Slope Aspect Elevation
Mean
DBH Density Basal area Ra Rh LAI* ANPPwood* TOL*
Foliar N* C/N*
years years % cardinal points m cm trees.ha-1 m2 ha-1 µmol CO
2 m-2 s-1 µmol CO2 m-2 s-1 m2 m-2 kg C. ha-1 yr-1 cm mg.kg-1 S-08 17 30 NE 381 6.09 250 0.7 0.16 3.28 0 5.3 9 11279 36.1 S-45 17 0-50 30 NW 372 6.02 700 3.4 0.14 2.43 0.2 28.06 12 12523 45.3 S-26 17 35 N 393 0 0 0 1.67 1.81 0 0 20.5 10433 41.5 S-20 67 10 S 340 13.21 5275 52 1.44 1.60 8.3 67.6 22.6 8316 63.4 S-21 67 51-100 13 NE 337 13.04 3025 38.3 1.30 1.84 5.6 78.6 19.7 10819 53.7 S-25 90 26 S 185 13.5 2825 32.1 3.20 2.80 3.7 78.8 11.9 10112 44.0 S-38 112 21 NW 230 17.98 1250 34.7 1.57 1.91 4.3 49.7 18.4 8195 43.8 S-41 112 101-150 0 E 157 19.16 1150 36 1.08 1.26 3.4 59 20.6 8379 61.3 S-42 112 20 SW 307 18.86 1425 41.4 1.42 1.32 4.8 45 23.5 9167 62.8 S-14 153 5 S 353 15.32 2050 36.5 1.32 1.41 3.9 36.6 22.9 8787 67.2 S-28 198 151-200 0 NE 313 16.05 1925 25.1 3.05 0.99 2 47.6 25.6 8641 51.2 S-39 198 0 NW 324 14.76 1400 26.3 3.10 1.26 2.7 47.6 19.5 8976 52.5 S-19 553 25 W 327 15.53 2100 32.9 1.47 1.54 3.8 68.2 26.6 9158 47.4 S-18 676 > 250 30 NW 320 14.34 1875 25.9 1.15 1.54 2.9 47.7 22 9135 43.8 S-02 1067 38 NW 334 17.98 850 25.7 0.87 1.84 2.4 53.3 18.9 n.d. 46.5
16 From the end of September to the end of October 2007, foliage samples were collected to determine foliar nitrogen concentrations. Within each plot, trees were selected to collect three branches of the upper side of the crown facing south. Current year needles were collected and grouped together to form a single sample per plot. The samples were oven-dried at 65 °C during 48 hours, then grounded prior to a wet digestion performed with a H2SO4H2O2 solution (Parkinson and Allen, 1975). Nitrogen
concentrations were determined by colorimetry with the Quick Chem method (Zellweger Analytic, Inc. Lachat Instruments Division, Milwaukee, Wisconsin, USA).
At the end of the 2007 and 2008 growing seasons, the organic and mineral layers were sampled in each plot. Four 1,000 cm3 (10 × 10 × 10 cm) samples of the organic layer were taken at all sites 6 m
from the plot center and eight 27 cm3 (3 x 3 x 3 cm) mineral soil samples were taken to a depth of 20
cm, four at 4 m from the plot center and four at a 10-m distance at the cardinal points. Samples were combined per plot and layer, air dried and sieved at 2 mm. A subfraction was ground for total analysis. Then, total N was determined by colorimetric determination of ammonium following a Kjeldahl oxidation on a Lachat Quick chem (Zellweger Analytic, Inc. Lachat Instruments Division, Milwaukee, Wisconsin, USA). Total C in organic layers was determined by loss on ignition at 550 °C and converted to C by dividing by a factor of 1.8 (Nelson and Sommers, 1996). For the mineral layer, total C was determined by the Walkley Black procedure (Yeomans and Bremner, 1988). Additional details on the determination of the nutrient concentration in the organic layer and the mineral soil are provided by Ward et al. (2014).
Soil CO
2flux measurements
We used the trench plot technique (Hanson et al., 2000) to estimate autotrophic (Ra) and heterotrophic
(Rh) contributions to soil CO2 flux (Rs = Ra + Rh). This technique is based on the establishment of
pairs of adjacent plots, with one being trenched and the other left undisturbed. Within the trenched plots, CO2 flux measurements should only correspond to Rh, since all tree roots are theoretically dead.
Conversely, within non-trenched plots, CO2 flux measurements correspond to Rs. Thus, Ra can be
estimated by the difference between measurements made in adjacent trenched and un-trenched plots (Ewel et al., 1987). However, such Ra estimates may still include a minor part of Rh due to the
mycorrhizosphere activity. During the first months following plot trenching, CO2 flux measurements
in trenched plots were likely to be larger than those in un-trenched plots because of the decomposition of the cut roots. To avoid this problem, plot trenching must be performed at least 9-10 months prior to the beginning of CO2 flux measurements (Bowden et al., 1993; Vogel and Valentine, 2005). By
17 between the trenching method and the total belowground carbon allocation (TBCA) method according to a study performed in North American boreal forests (Vogel et al., 2008).
Therefore, we delimited five pairs of subplots (1 × 1 m) within each of the 15 stands in June and July 2009. The perimeter of one subplot within each pair was trenched to the depth of the C horizon to sever all roots, while the other subplot was left intact. Sheets of landscaping fabric were inserted into the trenched perimeter of the subplots to prevent ingrowth of new tree roots during subsequent measurements. Water will pass through the material, but the mesh size is too small for roots to grow through the fabric. The trenched perimeter of each plot was then backfilled to re-establish hydraulic continuity between trenched plots and the surrounding soil. In addition, the vascular understory vegetation within each trenched plot was removed at the beginning of the experiment and thereafter when necessary.
In June 2011, two soil collars were installed in each trenched and un-trenched plot to facilitate CO2
flux measurements. These soil collars were installed about two weeks before the first measurements were taken to give the plots time to recover from the disturbance caused by collar insertion into the soil. Soil CO2 fluxes were measured with an LI-6400 infrared gas analyser, equipped with a soil
respiration chamber (model 6400-09, Li-Cor, Lincoln, NE, USA). At each CO2 flux measurement,
we measured soil temperature with a 16.5-cm long temperature probe (provided with the soil respiration chamber) and soil moisture content with a FieldScout TDR 300 (Spectrum Technology, Plainfield, IL, USA). These measurements were made to detect potential differences in soil CO2 flux
between trenched and control plots caused by disturbance effects (Edwards, 1975; Blet-Charaudeau et al., 1990) and differences in soil water regime (Hanson et al., 1993; Thierron and Laudelout, 1996). Each pair of plots was measured three times during summer 2011, i.e., in mid-June, mid-July and mid-August. Within each sampling period, we were generally able to measure two stands per day between 9 and 12 AM, for a total of 7 to 9 days to complete the 300 measurements in the experimental design (2 soil collars/plot x 10 plots/stand x 15 stands). All measurements have been taken during days without precipitation as rain could affect the measurements and damage the equipment.
Because the stand is the experimental unit, autotrophic and heterotrophic respiration rates were averaged at the plot level. Due to the limitation of our methodology, the Ra/ANPPwood ratio was
calculated regardless of the different temporal scales of measurements between autotrophic respiration rates (μmol CO2 m-2 s-1) and ANPPwood values (kg C ha-1 yr-1). Shifts in the proportion of
resource allocation to roots were determined from the relative differences in stand ages, rather than their absolute values.
18
Statistical analysis
Linear models were used to test the effects of time that had elapsed since the last fire (TSF) and forest floor thickness of organic layer on the Ra/ANPPwood ratio and on heterotrophic soil respiration (Rh).
A linearised form of a non-linear model was used to estimate the relationship between both autotrophic and heterotrophic respiration and TSF:
log10Yi= b0+ b1log10TSF + b2TSF + ⋯ + bnXn [1]
where Yi is the rate of Rh (μmol CO2 m-2 s-1) or the Ra/ANPPwood ratio, TSF is time since last fire (year),
b0 to bn are parameters to estimate, and Xi are other explanatory variables that were included in the
model, i.e., soil temperature, soil water content, or their transformed values, together with their interactions. As a first step, Equation 1 was fitted with only TSF and log10TSF as explanatory
variables. Then, we included all possible combinations of the other explanatory variables in the model. Those that did not play a significant role in explaining the variation of Yi (P > 0.05) were successively eliminated by using a stepwise procedure. The same method was applied to explain the variation in Yi with organic layer thickness (TOL), which was introduced in Equation 1 with the
replacement of TSF.
To take into account the bias that was introduced by logarithmic transformation of Yi when
transformed back to its original units (Beauchamp and Olson, 1973), we used the correction factor suggested by Sprugel (1983). Model assumptions (normality of residuals and homogeneity of variance) were validated using a Shapiro-Wilk test and a graphical analysis of the residuals. Statistical analyses were performed in R, version 3.0.1 (R Core Team, 2016) with results considered significant at α = 0.05.
19
Results
Ra/ANPP
woodratio
The descriptive statistics of the model that explained variation in the Ra/ANPPwood ratio with TSF are
summarised in Table 1-2.
Table 1-2. Parameter estimates (± SE, standard errors of the estimates) and related statistics for the prediction equation of log10ln(Ra/ANPPwood (Ra / ANPPwood) as a function of the time since last fire.
The value of R2 associated with this model is 0.45.
Parameter Variable Estimate SE t-value P-value
b0 Intercept -6.3937 0.9457 -6.760 < 0.0001
b1 log10(TSF) 0.6864 0.2304 2.980 0.0125
b2 TSF -0.0025 0.0009 -2.735 0.0194
Note: TSF is the time-since-fire (year)
Across the chronosequence, the Ra/ANPPwood ratio increased with increasing TSF up to a maximum
value of ~0.044 µmol CO2 m-2 s-1/kg C ha-1 yr-1 at ~300 years following the last fire, and then started
to decline progressively (Figure 1-1) until the oldest stand that had been sampled, i.e., TSF = 1067 (~0.016 µmol CO2 m-2 s-1/kg C ha-1 yr-1). Adding soil temperature and soil water content to Equation
20 Figure 1-1. Changes in autotrophic respiration rate/ANPPwood ratio as a function of time since the last
fire. Each point represents the season average value calculated from each plot and bars correspond to standard errors.
Heterotrophic respiration rate
Parameter estimates for the model explaining variation in Rh as a function of TSF are presented in
21 Table 1-3. Parameter estimates (± SE) and related statistics for the prediction equation of log10 Rh as
a function of the time since last fire. The value of R2 associated with this model is 0.79.
Parameter Variable Estimate SE t-value P-value
b0 Intercept 2.2204 0.3174 6.995 < 0.0001
b1 log10(TSF) -0.2578 0.0902 -2.857 0.0170
b2 TSF 0.0008 0.0003 2.594 0.0268
b3 SWC -0.0268 0.0086 -3.049 0.0123
Note: TSF is the time- since- fire (year) and SWC the soil water content (%)
Based on this model, heterotrophic respiration rate decreased with increasing TSF to a minimum value of 1.45 μmol CO2 m-2 s-1 at TSF = 305 years (Figure 1-2). Afterward, there was a slight increase
in Rh, to 1.99 μmol CO2 m-2 s-1 for the oldest stand that was sampled. Soil water content had a highly
22 Figure 1-2. Changes in heterotrophic respiration rate as a function of time since the last fire. In order to account for variation in soil water content, this variable was set to its mean value for each forest stand. Each point represents the season average value calculated from each plot and bars correspond to standard errors. The line was drawn from Eq. 1 using the average values of the covariate.
Changes as a function of the organic layer thickness
No significant relationship was found between the Ra/ANPPwood ratio and the thickness of the organic
layer (Figure 1-3a). However, a negative relationship was found between the heterotrophic component of soil respiration and the thickness of the organic layer (Figure 1-3b).
23 Figure 1-3. Changes in (a) Ra/ANPPwood ratio (expressed in μmol CO2 m-2 s-1/ kg C ha-1 yr-1) a) and
(b) heterotrophic respiration rate (Rh, expressed in μmol CO2 m-2 s-1) b) as a function of the thickness
of organic layer. Each point represents the season average value calculated from each plot and bars correspond to standard errors. The line was drawn from Eq. 1 using the average values of the covariate.
Parameter estimates for the model explaining variation in Rh as a function of the organic layer
24 Table 1-4. Parameter estimates (± SE) and related statistics for the prediction equation of log10 (Rh)
as a function of the thickness of organic layer. The value of R2 associated with this model is 0.90.
Parameter Variable Estimate SE t-value P-value
b0 Intercept 3.0109 0.2920 10.311 < 0.0001
b1 log10(TOL) -0.6821 0.1189 -5.737 < 0.0001
b2 SWC -0.0197 0.0057 -3.484 0.0051
Note: TOL is the thickness of organic layer, SWC the soil water content (%)
ANPP
wood, leaf area index and ANPP
woodper unit leaf area across the chronosequence
Temporal changes in ANPPwood, LAI and ANPPwood per unit leaf area (ANPPwood/LA) of the sampledstands are depicted in Figure 1-4. Contrary to Ward et al. (2014), we were not able to fit robust linear models for ANPPwood and LAI as a function of TSF. However, LAI seems to increase up to around 60
years, followed by a decline until age 200 years at which LAI reached a minimum value that was maintained up to 1067 years. The change of ANPPwood as a function of TSF seems to follow the same
pattern as that of LAI. Finally, ANPPwood/LA clearly decreased from 17 to 67 years and remained
approximately constant thereafter (Figure 1-4c).
Organic layer thickness, foliar nitrogen concentration and C/N ratio
As already described by Ward et al. (2014), the organic layer thickness and the C/N ratio of the organic layer increased from 17 to 67 years at which both variables remained approximately constant (Table 1-1). However, the foliar nitrogen concentration followed the inverse pattern with a decrease observed during the first decades that was followed by a minimum plateau up to the end of the chronosequence (Table 1-1).
25 Figure 1-4. Changes in a) ANPPwood, b) LAI, and c) ANPPwood per unit leaf area (ANPPwood/LA) as a
26
Discussion
Ra/ANPPwood ratio and Rh as respective proxies for belowground-to-aboveground
resource allocation and decomposition rate
At the tree or stand scale, determining resource allocation to belowground and aboveground parts is difficult, due to the limitations of measurement approaches and to the structural and functional complexity of plant organs. Although the root-shoot ratio has been used for many decades in plant ecology (Mokany et al., 2006), it corresponds to the ratio of accumulated belowground to aboveground biomass during stand development, rather than reflecting biomass allocation at a particular stand age (Litton et al., 2007). Resource allocation would be better estimated by quantifying the carbon fluxes from different stand components, as in the case of the total belowground carbon flux (TBCF) index (Raich and Nadelhoffer, 1989; Giardina and Ryan, 2002).
Accordingly, we derived a proxy that adequately reflects the proportion of resources that are allocated to belowground and aboveground stand components, viz., Ra/ANPPwood. First, the autotrophic
respiration (Ra) is strongly related to the TBCF index and thus to root production (Litton et al., 2007).
As reported by these authors, the TBCF index was linearly and significantly related to Ra (R2 = 0.60),
indicating that although Ra is only one component of the TBCF, this component is sufficiently
important to adequately represent the variation in TBCF (Fischer et al., 2007). Second, the rate of wood production (ANPPwood) can effectively reflect the patterns of change in aboveground carbon
flux because it is a very large component of aboveground carbon flux and is sensitive to changes in stand age and resource availability (Litton et al., 2007).
To validate the use of Ra/ANPPwood as a proxy for representing the ratio of carbon allocated to
belowground and aboveground components, we used data that were published by Vogel et al. (2008), who estimated various components of carbon allocation in nine well-drained black spruce stands of different ages growing in the western USA and Canada. Their results allowed us to calculate and compare values of Ra/ANPPwood with a commonly used proxy, the TBCF index. Because of the close
relationship between the two indices (R2 = 0.95; Figure 1-5), we concluded that R
a/ANPPwood could
be used to describe temporal changes in the proportion of carbon that was allocated to belowground and aboveground stand components in our study.
27 Figure 1-5. Relationship between Ra/ANPPwood ratio and TBCF/ANPPwood + foliage ratio. Each point
corresponds to a stand as described in Vogel et al. (2008)
As suggested by Olsson et al. (2005), we used the heterotrophic respiration rate as a proxy for organic matter decomposition rate, more specifically the availability of labile C (Ahn et al., 2009). The labile C fraction corresponds to carbon that is easily decomposed by microorganisms. Accordingly, Vogel et al. (2005) observed a strong positive correlation (R2 = 0.89) between R
h and the organic matter
decomposition rate in black spruce stands. The decrease in Rh observed between 17 and 287 years
post-fire (Figure 1-2) was consistent with decreases in microbial activity that have been observed in other long-term chronosequence studies (Williamson et al., 2005).
