Adjustment of the age–height relationship for uneven-aged black spruce stands Hakim Ouzennou, David Pothier, and Fre ´ de ´ ric Raulier Abstract: Site index (SI) is commonly used in natural stands, even when their diameter distribution deviates from that of the monospecific, even-aged, fully stocked stands used to develop basic age–height relationships. Since deviations from basic age–height trajectories can be reflected in deviations of stand diameter distribution from a bell shape, we incorpo- rated different diameter diversity indices into an age–height equation to help improve height predictions and determine which index is best related to stand dominant height. This procedure was performed using black spruce (Picea mariana (Mill.) BSP) stands from a large network of permanent sample plots established across the province of Quebec, Canada. The age–height model that minimized the Akaike’s information criterion used the Shannon evenness index (E Sh ) as an equation modifier accounting for the diameter diversity variable. The model showed that for stands established on rela- tively poor sites (SI = 9), no substantial differences in dominant height were found between two contrasting ESh values. For SI = 15, however, the larger ESh value increased the dominant height by as much as 1 m at 80 years. These results suggest that introduction of E Sh into an age–height model can improve calculation of site index, particularly in regions characterized by the presence of numerous uneven-aged stands. Re ´sume ´: L’indice de qualite ´ de station (IQS) est souvent estime ´ pour des peuplements naturels, me ˆme si leur distribution de diame `tres s’e ´loigne de celle des peuplements mono-spe ´cifiques, de structure e ´quienne, et de densite ´ optimale qui sont utilise ´s pour mettre au point les relations a ˆge–hauteur de re ´fe ´rence. Puisque des de ´viations aux trajectoires a ˆge–hauteur de re ´fe ´rence peuvent se refle ´ter dans des de ´viations de la distribution des diame `tres des peuplements par rapport a ` une courbe normale, nous avons soumis diffe ´rents indices de diversite ´ diame ´trale a ` une e ´quation a ˆge–hauteur dans le but d’ame ´liorer la pre ´diction de la hauteur et de de ´terminer l’indice le mieux relie ´a ` la hauteur dominante des peuplements. Cette proce ´- dure a e ´te ´ applique ´e a ` des peuplements d’e ´pinette noire (Picea mariana (Mill.) BSP) a ` l’aide d’un vaste re ´seau de plac- ettes e ´chantillons permanentes e ´tablies dans la province de Que ´bec, au Canada. Le mode `le a ˆge–hauteur qui a minimize ´ le « Akaike’s information criterion » utilisait l’indice de re ´gularite ´ de Shannon (ESh) comme modificateur d’e ´quation relie ´a ` la diversite ´ diame ´trale. Dans le cas des peuplements e ´tablis sur des stations relativement pauvres (IQS = 9), ce mode `le n’a pas produit de diffe ´rences substantielles de hauteur dominante entre deux valeurs contraste ´es de E Sh . Cependant, pour une valeur d’IQS e ´gale a ` 15, la plus forte valeur de E Sh a produit une augmentation de la hauteur dominante de 1 m a ` 80 ans. Ces re ´sultats indiquent que l’introduction de E Sh dans un mode `le a ˆge–hauteur peut ame ´liorer l’estimation de l’IQS, particu- lie `rement pour re ´gions caracte ´rise ´es par la pre ´sence de nombreux peuplements de structure ine ´quienne. Introduction Site index is widely used to assess the potential productiv- ity of forest stands for its ease of computation from common forest inventory data, for its efficacy in predicting wood vol- ume, and for its ease of integration into forest growth and yield models. Site index seems to have appeared around the end of the 19th century with the first normal yield tables produced for even-aged and fully stocked stands (Assmann 1970). Accordingly, trees suitable for estimating site index are generally defined as free-growing, uninjured, dominant trees that are found in even-aged, well-stocked stands not affected by recent disturbances (Carmean and Lenthall 1989). While tree plantations and many natural stands com- posed of pioneer species correspond to these criteria, a large proportion of natural stands deviate from this definition. Nonetheless, site index continues to be used in the absence of other effective variables. This is the case with second- growth stands originating from advance regeneration, stands partially disturbed by climatic events or insect outbreaks, old-growth stands, etc. As such stands cover a large part of the forested land of Quebec, Canada, certain concerns arise. On the one hand, if site index models are developed only from the above-mentioned well-suited stands, using them with other types of stands can underestimate potential site productivity. On the other hand, if site index models inte- grate age–height data from stands of all structures, they could be associated with large prediction errors that would affect the accuracy of the merchantable volume projection at the landscape level, and might even produce substantial bias at the local scale. In stands deviating from an even-aged structure, trees of all crown classes may have undergone a suppression period during which height growth was reduced compared with that of a free-growing tree of an equivalent age. If a dominant tree selected for site index estimation has been suppressed, Received 23 November 2007. Accepted 20 March 2008. Published on the NRC Research Press Web site at cjfr.nrc.ca on 13 June 2008. H. Ouzennou, D. Pothier, 1 and F. Raulier. Centre d’e ´tude de la fore ˆt, De ´partement des sciences du bois et de la fore ˆt, Pavillon Abitibi- Price, 2405 rue de la Terrasse, Universite ´ Laval, Que ´bec, QC G1V 0A6, Canada. 1 Corresponding author (e-mail: david.pothier@sbf.ulaval.ca). 2003 Can. J. For. Res. 38: 2003–2012 (2008) doi:10.1139/X08-044 # 2008 NRC Canada