Changing juvenile growth patterns in tropical trees: selective effects, history, or both?


A tree seed germinates in a forest and begins to grow. Given the right combination of environmental conditions, and a bit of luck, the seedling might one day become a fully grown tree. The quiet process of seed dispersal, germination and establishment occurs trillions of times each year across the world’s forests, but only a vanishingly small fraction of seedlings ever survive to become adult trees. Because individual germination events occur on such large numerical and spatial scales, and because trees are long-lived, understanding what makes some seedlings successful (and most not), may have important ramifications for regional and global-scale carbon cycling for decades or centuries to come. Several studies have demonstrated increased growth and biomass accumulation in tropical and temperate forests in recent decades and have suggested that these changes may be caused by increased atmospheric CO2 concentrations (e.g. Phillips et al., 1998; Johnson & Abrams, 2009). The focus, however, has been squarely on mature trees (usually > 10 cm diameter at breast height (dbh)). Long-term trends in the early growth rates of trees are poorly documented. This represents a double-barrelled gap in our knowledge because early growth rates determine both the composition and future structure of the forest (and therefore future carbon storage potential). In large part, this problem is caused by methodological challenges. Evaluating temporal variability in the early growth of trees is a tricky business. On the one hand, longitudinal studies that directly measure seedling growth are typically short (i.e. years to (rarely) decades) relative to the potential life span of the trees. On the other hand, retrospective studies based on tree-ring analyses, which estimate early growth of trees that were established decades or centuries in the past, represent a biased sample – the trees that survived to the present. Conducted independently, neither provides a complete picture of long-term variability in early growth rates.

In tropical forests, where few tree species have annual growth rings, we know little of long-term forest dynamics and disturbance history’.

In this issue of New Phytologist, Rozendaal et al. (pp. 759–769) directly address this knowledge gap. They ask the question: is the early growth of trees that have recently established in a Neotropical forest different from the early growth of trees that established decades or centuries earlier in the same forest? They build on recent work by Landis & Peart (2005) in the northern temperate zone, in which observational studies of current tree growth were combined with a retrospective tree-ring analysis approach to compare early growth rates between extant seedlings and adult trees for species of differing shade tolerance. Landis & Peart hypothesized that early growth rates in shade-intolerant species would be higher in the adults because only the fastest growing seedlings and saplings would survive the early period of stand development (i.e. the juvenile selection effect). And, this is precisely what they found. By contrast, they expected that the early growth of adults of shade-tolerant species would not differ from the growth of the current seedling and sapling crop, as fast growth was not a prerequisite for success among species that could survive in shady conditions. However, to their surprise, they found that the slow-growing, shade-tolerant species also showed a strong juvenile selection effect.

Rozendaal et al. have modified Landis & Peart’s (2005) approach and applied it to a suite of tropical tree species growing in Bolivia. This is not a trivial undertaking. In temperate forests, there is a long history of using tree-rings to reconstruct stand development patterns and assess the impacts of past disturbances (Oliver & Larson, 1996). In tropical forests, where few tree species have annual growth rings, we know little of long-term forest dynamics and disturbance history. That is beginning to change, however. Several recent studies have demonstrated the potential for successful dendroecological reconstructions in tropical forests in Asia (Baker et al., 2005; Buckley et al., 2007; Sano et al., 2009), Africa (Worbes et al., 2003) and the Americas (Brienen & Zuidema, 2006). While these studies have all been conducted in seasonally dry forests with many fewer species than their hyper-diverse aseasonal analogues, they provide new – and important – insights into the temporal modes of change in species-rich tropical forest communities. They have, however, focused primarily on adult trees. Rozendaal et al. have lowered their gaze to the small trees and examined growth patterns during the critical period of establishment. They took advantage of several planned selective logging coupes in forests in northeastern Bolivia to collect stem cross-sections from individuals across the full range of size classes for five species differing in shade tolerance. From each section they calculated the total cross-sectional area of new wood added each year across a range of size classes. So, for example, they calculated how fast an individual was growing when it was 1–2 cm dbh, 2–3 cm dbh, etc. An important advantage of the approach of Rozendaal et al. is that it provides insight into not only growth patterns at the earliest ontogenetic stages of trees of different ages, but across the entire developmental history of these trees. And, this, it turns out, is where things get interesting…

Unlike Landis & Peart (2005), Rozendaal et al. found that the juvenile selection effect was not ubiquitous. Some species showed evidence of it, whereas others did not. Surprisingly, the most shade-intolerant species, Cedrela odorata, which was expected to exhibit the strongest juvenile selection effect, did not show any evidence of faster early growth in the older trees – its growth seemed independent of tree size and age. By contrast, Cedrilinga catenaeformis, the next most shade-intolerant species, showed strong juvenile selection effects across nearly all size classes. The three shade-tolerant species, Clarisia racemosa, Peltogyne cf. heterophylla and Pseudolmedia laevis, showed the most curious pattern, however. In the smallest size class (0–1 cm dbh), early growth rates were higher in older trees (although not significantly so for Pseudolmedia). But, from 1 to 10 cm dbh, all three species showed evidence of historical growth increases; that is, saplings and poles were growing faster now than they had in the past.

The results of Rozendaal et al. raise two interesting questions. First, why does an extremely shade-intolerant tree species not show evidence of a juvenile selection effect? Second, why would older trees of relatively shade-tolerant species show faster growth (relative to more recent recruits) in the smallest size class and then slower growth for much of the rest of their youth? The first question may simply reflect the extreme high growth of the Cedrela– in the 0–1 cm dbh size class, its basal area growth was two (!) orders of magnitude greater than that of the shade-tolerant species. Any juvenile selection effects may have already played out in an earlier ontogenetic stage. However, the second question is the more intriguing. While CO2 fertilization would be the facile explanation, Rozendaal et al. are appropriately cautious in pushing that link. They provide several other suggestions to explain the observed pattern. In particular, they note that there is no evidence of large disturbances at their sites in the tree-ring data. The observed pattern suggests a more subtle change in the underlying disturbance regime, though. All of the significant differences in growth (both positive and negative) have occurred in relatively small size classes (< 10 cm dbh), which is indicative of small-scale, rather than large-scale, disturbances. In the wake of small-scale disturbances (such as treefall gaps), shade-tolerant trees often establish contemporaneously with shade-intolerant trees, growing quickly for a brief period before being overtopped. Oliver Phillips and colleagues have documented greater rates of biomass turnover (and therefore small gap formation) in Neotropical forests in recent decades as a result of higher rates of mortality and recruitment (Phillips & Gentry, 1994; Phillips et al., 1998). In principle, this should lead to more gaps in the forest canopy (relative to the past), but those gaps should be smaller. If the gaps are smaller, then the earliest growth of new recruits should be slower than in the past, when less frequent, but larger, canopy gaps would have formed. However, an increase in biomass turnover and gap formation in modern times would also lead to higher light intensities in the subcanopy, which would favour growth of established saplings and poles (e.g. 1–10 cm dbh). The result would be the observed historical increases in growth.

Like all good research, the study of Rozendaal et al. raises more questions than it answers. Two obvious sets of questions stand out. First, are historical growth changes nonlinear? Is there evidence of threshold events or baseline shifts in growth? Second, are these historical growth changes occurring elsewhere? In a recent study of adult individuals of eight tree species from the eastern USA, Johnson & Abrams (2009) showed that younger trees are growing faster at a given size than older trees did at the same size (i.e. historical growth increase) and that this was consistent across all levels of shade tolerance. These results provide an interesting counterpoint to Rozendaal et al.’s study by showing that larger trees are capable of benefitting from changes to growing conditions. Clearly, there is a need for a broader survey of comparative dendroecological studies in which past growth patterns are examined in the light of current growth dynamics. A key point from this study is that placing current processes into their historical context is necessary for understanding the organisms and the communities in which they grow. Such studies provide useful insights into long-term variability in tree growth, the relative contributions of disturbance regimes and climate variability on moderating this variability, and potential avenues for future growth.