Plant ecology in recent years has focused on functional traits (Westoby & Wright, 2006; Swenson & Enquist, 2007; Chave et al., 2009; Cornwell & Ackerly, 2009) – phenotypic attributes that influence performance in a given environmental setting. As a starting point for cross-species comparisons, the traits selected have been easy to measure (Cornelissen et al., 2003). Examples include seed size, leaf size, maximum plant height, leaf mass per area, leaf area per stem area and wood density. But, even with large sample sizes across many species, the interpretation of results can be complex; most of these traits reflect multiple aspects of plant function. Wood density has attracted considerable interest. It has been related to growth rate, water storage, mechanical strength, efficiency and safety of hydraulic transport, and resistance to herbivory (Jacobsen et al., 2007; Sperry et al., 2008; Chave et al., 2009; Onoda et al., this issue, pp. 493–501). However, with so many potential competing functions influencing how densely a stem is built, it becomes hard to determine the utility of having denser or lighter wood in a particular setting. One way to make headway is to examine the harder-to-measure anatomical traits that contribute to lighter or denser wood and apply our understanding of the functional roles that these different anatomical tissues play. In the current issue of New Phytologist, Poorter et al., (pp. 481–492) have taken just such an approach. They quantified several of these harder-to-measure anatomical traits and examined their relationship to both wood density and various aspects of plant performance and life history for 42 abundant tree species in the rainforests of La Chonta, Bolivia.
‘…while plants have relatively little room to adjust the percentage of stem devoted to vessels, they seem to have ample leeway in deciding how this space is divided up.’
What drives variation in wood density across species?
Poorter et al. measured the stem cross-sections from each of the species as the fractions composed of vessel, fibre and parenchyma (Fig. 1). These three tissues have presumably evolved to meet the challenges faced by woody angiosperm stems. Vessels are the main conducting cells in angiosperms, with flow rates per vessel increasing with the cross-sectional area of the lumen (Tyree & Zimmermann, 2002). Fibres are the main support structures, and parenchyma serves for both storage and transport of resources between xylem and phloem (Martínez-Cabrera et al., 2009). Depending on selective pressures, evolution may alter the relative allocation to these different tissue types, or, alternatively, the individual densities (by shifts in chemical or structural composition) of the different tissues. Both types of change are reflected in an aggregate measure such as wood density. Partitioning variation in wood density into these components, however, should enhance our understanding of the functional significance of lighter or denser wood.
One exciting result arising from the study of Poorter et al. is their finding that variation in the density of the different tissue types, rather than the relative amount of each tissue, appears to underpin variation in wood density across these species (see also Zanne et al., 2010). As vessels take up volume but have limited mass (because of their large lumens), a relationship between wood density and fraction of cross-sectional area that is vessel might be expected. Similarly, because fibres are heavy as a result of thick walls, we might expect a relationship between wood density and the fraction of the cross-sectional area that is fibre. As it turned out, neither relationship was particularly strong. These results point to variation in tissue composition, rather than to relative amounts of vessel, parenchyma and fibre, as being the primary determinants of wood density. Supporting this idea, wood density has been related to fibre wall to lumen ratios, percentage allocation to fibre wall and fibre lumen size (Jacobsen et al., 2007; Martínez-Cabrera et al., 2009).
The lack of a relationship between total percentage vessel area and density is particularly significant, because it brings into question the suggested links between wood density and hydraulic conductivity. As well as making wood lighter, greater vessel area should improve conductivity, leading to predictions for relationships between wood density and conductivity (Meinzer, 2003). But, as Poorter et al. show, greater allocation to vessels does not necessarily lead to lighter wood, so links between conductivity and density should not be taken as a given.
Cross-species indicators of vascular strategy
The limited variation in percentage vessels (3–23%), observed by Poorter et al., contrasts with the nearly 500-fold variation observed in average vessel size (A) and number of vessels per unit area (or density of vessels, N) (Fig. 2). These two variables show a strong, negative correlation (see also Preston et al., 2006; Sperry et al., 2008; Zanne et al., 2010), indicating coordinated shifts in the mixture of vessel sizes and numbers across species. So, while plants have relatively little room to adjust the percentage of stem devoted to vessels, they seem to have ample leeway in deciding how this space is divided up. Furthermore, because potential conductivity increases to the square of A but only to the first power of N (Tyree & Zimmermann, 2002), altering the size–number mixture can have large influences on potential conductivity.
Poorter et al.– like other studies before them – adopt vessel size and number as the primary indicators of vascular strategy. We agree that vessel size and number are critical aspects of plant strategy, but note that two difficulties arise when using A and N in comparative studies. The first is that they are tightly correlated, and so do not provide independent information. The second is that neither A nor N distinguishes between the different ways in which species can adjust rates of water supply through sapwood. Cross-species variation in A or N could indicate changes to the total area of conducting tissue (vessel fraction), the mixture of vessel sizes and numbers, or both. These difficulties are overcome if one instead uses vessel fraction (F = AN, mm2) and the vessel size : number ratio (S; =A/N, mm4) as indicators of vascular strategy (Fig. 2).
We recently analyzed global patterns in S and F for over 2000 woody angiosperm species distributed worldwide (Zanne et al., 2010) and showed that the bulk of variation in A and N equates to variation in the size : number ratio S (95%), leaving only 5% accounted for by vessel fraction F. This decomposition indicates that shifts in the size : number ratio are therefore also responsible for most of the variation in potential conductivity observed across species. Analysis of the data of Poorter et al. recovered the same results (Fig. 2), demonstrating that such global patterns may also be observed within local studies. In fact, further analysis of the data of Poorter et al. reveal that all of the reported associations between vessel anatomy and plant performance or life history (see below) are related to shifts in the mixture of vessel sizes and numbers, and not to shifts in vessel fraction. Additionally, their data indicate that most of the cross-species variation in S and F observed worldwide can be found within a single community.
Does trait variation lead to variation in plant performance?
A consistent and outstanding feature of the publications from Poorter and colleagues in Bolivia and the Netherlands is their ongoing effort to link functional traits with plant performance (growth and survival) and life-history strategy (maximum adult height and sapling crown exposure). As with other studies of theirs (e.g. van Gelder et al., 2006; Poorter & Bongers, 2006), performance measures have been quantified across numerous individuals, allowing them to take the significant step of testing how functional these functional traits really are. Confirming previous findings, wood density was strongly related to both growth and survival (Chave et al., 2009). However, vessel size and number also had strong influences on growth rates. As relationships with the vessel fraction F were not significant, these must have been mediated through changes in the size : number ratio, S. The study of Poorter et al. thus supports two independent influences in stem design leading to faster growth rates, namely more efficient construction of stems (via shifts to low wood density) and higher hydraulic conductivity, enabling faster photosynthetic rates (via shifts to large S). To our knowledge, this is the first study to show that vessel traits related to high hydraulic conductivity form part of a spectrum of traits leading to fast growth; other traits that show this linkage include leaf mass per area, leaf nitrogen, wood density and stem allometry (Poorter & Bongers, 2006; Westoby & Wright, 2006; Chave et al., 2009).
A widely held view in ecology is that plants differentiate along successional niche and/or growth strategy axes, and that this differentiation has both generated and maintained diversity in tropical forests. However, it has been unclear how wood structure contributes to this differentiation. Thanks to detailed studies, such as that of Poorter et al., our understanding of these issues is improving; challenges, however, remain. The costs of fast growth, via high S or low wood density, are still poorly documented. The vessel-size number axis is presumed to represent a trade-off between conductivity and hydraulic safety (driven by embolism avoidance). But, does embolism avoidance limit success in shaded environments, and, if so, how? Similarly, low wood density is known to decrease the structural integrity of stems (van Gelder et al., 2006; Jacobsen et al., 2007; Onoda et al., 2010), but it remains unclear how this limits fast-growing, light-wooded species from recruiting in low light. These are important questions for future studies that seek to bridge the gap between functional traits, plant performance and life-history strategy (Fig. 1).