Do trait–growth relationships vary with plant age in fire-prone heathland shrubs?

Abstract

1 INTRODUCTION

Field-measured plant growth rates vary considerably across species, influencing competitive interactions and shaping vegetation patterns across various ecological scales (Lambers & Poorter, 1992). Understanding the factors driving this variation is essential for predicting community dynamics and ecosystem responses to environmental change. Functional traits have been extensively studied as potential drivers of growth rate variability, with some traits generally showing consistent correlations with growth rates across different studies. For example, in numerous studies, faster growth tends to be correlated with lower wood density (WD) and with higher adult maximum height (Fajardo et al., 2024; Gray et al., 2019; Iida et al., 2023; Iida, Kohyama, et al., 2014; King et al., 2006; Kunstler et al., 2016; Poorter et al., 2008, 2018; Rubio et al., 2021; Rüger et al., 2012; Russo et al., 2010; Wright et al., 2010). By contrast, for leaf mass per area (LMA), a key trait in leaf ‘economics’ (Wright et al., 2004), the corresponding knowledge is variable: in some field studies, LMA is positively associated with growth rate (Bin et al., 2024; Gray et al., 2019), in others, the association is negative (Poorter et al., 2018; Poorter & Bongers, 2006), and commonly no relationship is found (Gibert et al., 2016; Hietz et al., 2017; Prado-Junior et al., 2017; Visser et al., 2016; Wright et al., 2019).

Increasingly, researchers in this area have explored the possibility that trait-growth relationships vary with plant size or age, or developmental stage (Gibert et al., 2016; Iida et al., 2023; Iida, Kohyama, et al., 2014; Iida, Poorter, et al., 2014; Iida & Swenson, 2020; Medeiros et al., 2019; Prado-Junior et al., 2017; Rüger et al., 2012; Visser et al., 2016; Wright et al., 2019; Yang et al., 2018). Plant size or age may help explain discrepancies among reported relationships, for example, for LMA. Generalising across studies is potentially complicated by differences in how growth rate is expressed, that is, as absolute growth rate or as relative growth rate RGR, which itself is strongly size-dependent (He et al., 2022). Generalisation is also complicated by methodological differences in how size-related effects are quantified, for example, by way of meta-analysis (Gibert et al., 2016), by including size as a covariate in a hierarchical Bayesian framework (Iida, Kohyama, et al., 2014; Iida, Poorter, et al., 2014; Rüger et al., 2012), or by running analyses separately for different size classes (Visser et al., 2016; Wright et al., 2010, 2019). Nonetheless, it seems that some generalities are beginning to emerge, the chief one concerning LMA. That is, LMA-growth relationships commonly shift with plant age, being clearly negative in early life stages and thereafter weakening such that, at later life stages, LMA and growth rate may be only weakly negatively related (Wright et al., 2010), unrelated (Gibert et al., 2016) or positively related (Iida et al., 2023; Iida, Kohyama, et al., 2014). Another seemingly emerging pattern is that crown width has a positive correlation with growth rate at small plant sizes but no longer at later life stages (Iida, Kohyama, et al., 2014; Prado-Junior et al., 2017). By contrast, little generality has emerged for size effects on wood density–growth relationships. For example, Iida, Poorter, et al. (2014) and Iida, Kohyama, et al. (2014) found WD more influential on growth rate at earlier life stages than later. Visser et al. (2016) found a stronger influence of WD on the growth of adults than of saplings, but within either life stage, they found that the influence of WD decreased with plant size. Then, in other studies, WD and growth rate have been reported as similarly negatively correlated at all plant sizes (Gibert et al., 2016; Iida et al., 2023; Wright et al., 2010).

Relatively few studies have explicitly quantified size effects, so even where results are reasonably consistent (as they are for LMA) the degree of generality is still unknown. This is especially true as most studies have concerned tree species growing in tropical or subtropical forests, ecosystems that tend to be characterised by steep vertical light gradients, meaning that the ecological niche of seedlings and saplings may be very different from those of adult plants (Grubb, 1977). It may be that there are weaker—or perhaps somehow different—age- and size-related trends in trait-growth relationships in vegetation types that are more open, or for which there are different main drivers of successional processes (e.g. fire). Perhaps, the key question for this research area should therefore be: why and in what situations should trait-growth relationships vary systematically with plant age or size?

Some theory exists for trait–GR relationships (Enquist et al., 2007; Hunt, 1978; Lambers & Poorter, 1992; Reich, 1998). The treatment that addresses size effects most directly is from Falster et al. (2018), which is a mathematically explicit elaboration and generalisation of cost: benefit theory previewed by Gibert et al. (2016). This approach starts from conventional assumptions about biomass allocation and trait trade-offs: biomass is divided into root, sapwood, leaf and reproductive components. Biomass growth in a given time step is given by the difference between income (photosynthesis) and losses (respiration and tissue turnover). An allometric model relates sapwood cross-sectional area to total leaf area. There is a fixed ratio of fine root mass to leaf area. Photosynthesis scales with leaf nitrogen. Leaf lifespan scales with LMA, etc. However, a key innovation was to focus on size-related changes in the tissue costs of producing new leaf area, meaning not only the cost of the leaf itself but also the per unit leaf area construction costs of new stem (vascular) and root tissues associated with the new leaf. This makes up an increasingly large proportion of biomass as plants increase in size. This shifts the relative influence of leaf versus wood traits on growth outcomes and the relative benefits and costs of key trade-offs between tissue construction costs and turnover rates, most importantly the trade-off between LMA (indexing leaf construction cost per unit area) and leaf turnover rate (the inverse of leaf lifespan). For example, the deployment of cheap, low LMA leaves always has the potential to enhance growth rate because a larger photosynthetic surface is constructed for a given dry mass invested. On the other hand, low LMA leaves have shorter lifespans (faster turnover); hence replacement costs can ameliorate or even overwhelm the photosynthetic benefits from low LMA, depending on the relative contribution of leaf to overall biomass and depending on other elements of the whole-plant carbon budget. Relevant to the empirical reports of size–trait–growth relationships reported above and to this particular study (see below), key predictions from Falster et al. (2018) include: (1) LMA and growth rate should be negatively correlated when plants are small, but unrelated or weakly positively related among large plants; and (2) Wood density should be negatively correlated to growth rate at all plant sizes.

Another unresolved question is how much trait-growth relationships vary according to how growth has been expressed, for example, on the basis of increments in stem diameter, plant biomass, or plant height (Bin et al., 2024; Falster et al., 2018; Hilty et al., 2021). For adult plants, stem diameter growth is the prevailing field measurement for growth assessment. This is for several reasons: stem diameters are relatively straightforward to measure; timber volume is important in many applied contexts and relates closely to stem diameter; and total above-ground biomass of forest trees scales with stem diameter, both within and among species (Chave et al., 2014). In two recent analyses, this scaling has been used to estimate biomass growth rates from primary data for stem diameters (Bin et al., 2024; Medeiros et al., 2019), but direct measurements of biomass growth for adult plants are usually lacking since it is usually impractical to destructively harvest whole trees. In many situations, height growth is also difficult to measure accurately for adult trees; hence this growth metric tends to feature more in studies concerning forest seedlings or saplings (Poorter & Bongers, 2006; Reich et al., 1992; Visser et al., 2016). Height, biomass, and stem diameter are not the only dimensions in which plants grow, crown dimensions and total canopy leaf area being others. Growth priorities may also change over an individual's life (Sumida et al., 1997). For example, height growth may be key early on for gaining access to light, and lateral canopy spread subsequently key for consolidating that advantage. Furthermore, canopy leaf area growth may plateau in older individuals while stem diameters continue to increase. Thus, when considering variation among species in their architectures and growth strategies across their whole life cycle, in principle, we might expect to see different trait–growth relationships depending on the types of growth measured. Indeed, in a study considering many different functional traits measured on 29 Hawaiian forest species, Medeiros et al. (2019) found almost no overlap between traits correlated with biomass RGR compared with those with stem diameter RGR. By contrast, among Chinese subtropical forest trees, Bin et al. (2024) did not find different traits correlated with biomass versus diameter growth, but rather that the relationships were generally stronger with biomass growth rate. We are unaware of any theoretical treatment of this issue. Falster et al. (2018) touched on it (expecting similar trait–growth relationships for different growth metrics) but, because fixed allometries were assumed, this is best thought of as an assumption rather than a conclusion from that study.

In this study, we investigated these issues among 14 common, co-occurring, woody shrub species from fire-prone, heathland vegetation occurring on infertile, sandstone-derived soils in the Sydney region of eastern Australia. Taking advantage of the well-characterised fire history of the area, and by focusing just on species that predictably regenerate from seed after fire, we were able to identify sites at which the vegetation was of different known ages (i.e., time since last major fire), ranging from ~1 to ~32 years. Growth increments were measured over a 12-month period at each of the sites; put together with trait data simultaneously collected, we were then able to characterise trait–growth rate relationships on a range of species, across plants ranging from ~1 to ~32 years in age. As detailed in Materials and Methods, numerous considerations went in to choosing sites for this space-for-time experimental design such that the chief difference between sites was most probably time since fire. Perhaps, most importantly we only used sites without any tree canopy cover, meaning that the top-of (shrub)-canopy light environment did not vary systematically with site age.

We measured and expressed annual growth rates in five ways, that is, in terms of annual increments in basal stem diameter (G_diam), height (G_height), total leaf area (G_area), ‘net’ above-ground biomass (G_net) and ‘total’ above-ground biomass (G_total), the latter meaning net increments plus estimated losses of leaf, stem material and reproductive tissue. This novel design allowed us to (1) assess the degree of consistency among different growth metrics (or, conversely, their lack of congruence—which would indicate they provide independent information about plant growth); (2) quantify age-related trends in growth rates and traits, this being an important precursor to (3) investigating the extent to which trait–growth relationships varied with plant age; and (4) identifying the extent to which relationships varied depending on what type of growth was considered. Simultaneous investigation of trait–growth relationships using five complementary growth metrics was a key advance of this study.

Most traits considered were standard in this research field: wood density (WD), leaf mass per area (LMA), leaf nitrogen concentration (per area and mass; that is N_area, N_mass), and leaf phosphorus concentration (per area and mass; P_area, P_mass). Another key advance was considering the influence on growth rates of above-ground leaf mass fraction (LMF), the ratio of total leaf dry mass to total above-ground dry mass. In principle (and all else being equal), we expected that species with higher LMF would attain faster growth rates—on the basis that they would have a more favourable ratio of canopy photosynthesis to whole-plant respiration. We are unaware of any previous tests of this proposition for field-sampled adult plants, but it follows on from reports of branch-scale LMF driving field growth rates in tropical savanna and rainforest vegetation (Gray et al., 2019; Wright et al., 2019), as originally predicted by Pickup et al. (2005), and the long history of measuring LMF in growth rate studies focusing on seedlings (Freschet et al., 2015; Poorter & Garnier, 2007; Quero et al., 2008; Umaña et al., 2021; Wright & Westoby, 1999)—in those cases, LMF being expressed on a whole-plant basis.

We oriented our work around several hypotheses:

• H1. WD and growth would be negatively correlated at all ages and (potentially) most strongly in older plants. This hypothesis stems from knowing that high density wood has a higher construction cost and often corresponds to lower hydraulic conductivity (via narrower conduits), both slowing growth (Chave et al., 2009; Wright et al., 2010), and from the idea that higher density wood has a slower turnover rate from sapwood to heartwood, which (all else equal) decreases stem diameter growth rate (Falster et al., 2018). Given that sapwood mass per unit leaf mass is higher on larger plants, the influence of wood traits (e.g. WD) on growth rates should in principle be stronger for larger and older plants (Falster et al., 2018; Gibert et al., 2016), potentially amplifying the aforementioned effects.
• H2. LMA and growth rate would be negatively correlated when plants were young, while in older plants either no relationship or a positive relationship would be observed. When plants are young, lower LMA connotes a higher light capture area constructed per unit leaf dry mass (Lambers & Poorter, 1992). However, as plants mature, and as wood increasingly contributes a larger fraction of biomass (and leaves, less), the cost of higher leaf turnover associated with lower LMA could outweigh these benefits, leading to no net advantage of lower LMA, or even a disadvantage in terms of growth rate (Falster et al., 2018; Gibert et al., 2016).
• H3. The relationship between growth rate and LMF would be positive across all ages. All else being equal, species with more leaf relative to wood mass were expected to have faster growth rates on the basis that higher allocation to leaf represents greater potential for photosynthetic benefits, and higher allocation to wood represents higher respiration costs (Pickup et al., 2005; Wright et al., 2019). Because LMF includes costs and benefits from leaf mass and also wood biomass, we predicted its effect on growth rates would not vary with plant age, although LMF itself is expected to decrease with plant age.
• H4. Leaf N and P would have positive effects on growth rates at all ages. This hypothesis assumes that higher nutrient concentrations are generally indicative of faster metabolic rates (Sterner & Elser, 2002) and, more specifically, indicative of higher photosynthetic rates (Domingues et al., 2010; Ellsworth et al., 2022) which, all else equal, should drive faster growth. With leaves contributing a decreasing fraction of plant biomass as plants age, one might predict trends with growth rates to weaken accordingly. By contrast, Falster et al. (2018) showed that the effect of leaf N per area (N_area) on growth rate was present across different plant sizes, as, unlike LMA and WD, the optimal N_area for the plant did not change with size. They did not report predictions for mass-based nutrient concentrations nor for leaf P.
• H5. Relationships between traits and growth would be qualitatively consistent across different expressions of growth rate, that is, with respect to G_diam, G_height, G_area, G_net, and G_total. This prediction, in essence a null hypothesis, is based on the assumption that the logic for trait–growth relationships (as described above) can be applied similarly to growth considered in terms of increments in height, stem diameter, total leaf area or biomass.