Leaf form and function – towards a radical new approach

One of the most striking ways in which plant ecology has matured during the past 40 years has been the emergence of a coherent theory of form and function for leaves in all of the world's vegetation-types. Working in Florida, where the climate is mild but seasonal, Monk (1966) was the first to establish a link between evergreenness in vegetation and nutrient-poverty in soil. His rationale, however, would never have got past an editor in the 1980s. Monk proposed a kind of group selection theory whereby the nutrient status of the already impoverished soil would decline more slowly if leaf fall was spread through the year, and so the roots would more easily 'grab' the mobile ions before they went off to the sea. Small (1972) rescued us by giving a proper explanation – perhaps the single most important idea in the whole field. Using measurements on many species of bogs and forests in Canada, he showed that while the evergreens had lower maximum rates of photosynthesis per unit of dry mass, they could be more efficient over the life-time of the leaf in terms of the amount of carbon fixed per unit nitrogen invested because their leaves last for two or even up to several years (the evergreenness was just incidental). It was easy to apply his idea to plants whose realised rates of photosynthesis were limited in other ways such as by deep shading in evergreen forest or by low temperature and shortness of the growing season – as on high mountains of the temperate zones, and in the arctic. Now, on pp. 403–416 of this issue, Wright & Westoby (2002) present an intriguing next step, highlighting the role of water rather than nutrients, shade or temperature. 'Classically we have thought of there being a fixed supply of available nitrogen in a given soil, and the plants doing what they can with it … attention now needs to be switched to co-ordinated studies of leaf and root properties' For plants from very dry places, you couldn't easily apply the Small approach because the amount of rain received in a given year varies hugely, and likewise the longevity of the leaves on many species. Nevertheless two ecologists who were not plant physiologists ventured to produce a very general model suggesting that in semiarid regions we should expect a negative relationship between maximum rate of photosynthesis and leaf longevity (Orians & Solbrig, 1977). The mechanistic basis of their hypothesis seemed, to many of us, to be totally implausible – that is, the idea that the longer-lived leaves should have inherently lower rates of photosynthesis because their mesophyll is composed of particularly small cells. In fact, other things being equal, smaller cells should provide a higher ratio of internal surface to external surface, and so higher rather than lower potential for photosynthesis. Incidentally, Orians & Solbrig (1977) referred to the longer-lived leaves of semidesert perennials as 'sclerophylls', which they certainly aren't – as Wright & Westoby note, they are notably lacking in fibrous tissues. A more plausible explanation for the lower maximum rates of photosynthesis in longer-lived leaves concerned the allocation of N within the leaf. In general, longer-lived leaves have lower concentrations of N, and it was hypothesized that somehow such leaves have a lower proportion of their total N in the enzymes catalyzing CO2-fixation (Mooney & Gulmon, 1979; Grubb, 1984; Field & Mooney, 1986). In the 1970s two other variables had begun to receive a lot of attention: the conductance of the whole epidermis for water vapour (gs) and the specific leaf weight (SLW, dry mass per unit of fresh area). The values for gs were found, in many different studies, to parallel the rate of photosynthesis (Cowan, 1981), and the maximum values for gs clearly paralleled stomatal densities (Koerner et al., 1979). SLW had been suggested by Stocker (1931) as a measure of 'sclerophylly', which in hindsight was regrettable because leaves with very different internal structures share high SLW (Grubb, 1986). In the 1970s, SLW (now usually called 'leaf dry mass per unit area, (LMA)) was picked up as a useful integrator of thickness and density – a measure of investment per unit area. One more variable, critical to the study by Wright & Westoby, fell into place in the 1980s. The work of Coley (1983) in tropical lowland rainforest established the correlation between the longevity of a leaf and its 'toughness' when subjected to a penetrometer test, and furthermore toughness was strongly negatively correlated with the rate of loss to invertebrate herbivores from mature leaves. There then developed a tendency to think of shorter-lived leaves as undefended, which was unfortunate, because actually they have a higher incidence of defences other than 'toughness', defences specifically effective against herbivores rather than providing protection against physical hazards as well – notably densely spaced arrays of short hairs, spines, stings, extrafloral nectaries and highly effective chemical deterrents (Grubb, 1992). Apart from the issue of defence, ecophysiologists across the world in the period 1985–2000 were teaching a consistent story about leaf form and function, with a wholly rational basis, as follows. In microhabitats where some key resource is in short supply, selection for longer-lived leaves has occurred because – over the leaf's life-time – such leaves fix more C per unit N invested than shorter-lived leaves. Longer-lived leaves are tougher, and have higher LMA (being thicker and/or denser) because that increases their generalized resistance to physical and biological hazards, and the longer the leaf-life the greater the economic justification for such 'investment in insurance'. The N concentration is lower in such leaves because it is diluted by the thick-walled tissues which provide the protection in general, and 'toughness' in particular. Maximum epidermal conductance to water vapour (and, more approximately, stomatal density) parallels maximum photosynthetic capacity as that minimizes both wastage of water and unnecessary investment in tissues for conductance of liquid water through the leaf (Grubb, 1984). The major outstanding issue all this time was why the maximum rate of photosynthesis per unit N was lower in longer-lived leaves. The work of Evans (1989) provided support for the view that the allocation of N to CO2-fixing enzymes was potentially an important partial explanation, but some critical studies (e.g. that of Hikosaka et al. (1998)) found no simple answer. A huge step forward, providing the context for the study of Wright & Westoby, was taken by Reich et al. (1999), who reported the results of comparative studies on numerous species in six major vegetation-types in the Americas, ranging from temperate alpine to tropical lowland, and from semidesert to rainforest. Their key contribution was to test critically whether or not the slopes (in log-log plots) relating the many variables so far discussed here were coincidental for the very different vegetation-types or were significantly different or were the same but with significantly different intercepts. As in many allometric studies, the key differences were mostly in intercepts. Reich et al. showed many cases of significantly different intercepts, several yet to be explored and understood, but one set of data stood out – that for semidesert shrubland in New Mexico. The set of species showed the expected correlations, but for a given life span had: the highest LMA; the equal-lowest maximum rate of net assimilation per unit dry mass (A mass ); the highest maximum rate of net assimilation per unit area (A area ); the lowest whole-epidermis conductance for water vapour; and the highest A area per unit of epidermal conductance for water vapour. Importantly these plants also had at a given N concentration a lower Amass. Wright et al. (2001, 2002 ) followed up this finding by making measurements of leaf form and function in a nicely designed comparison of species in wetter and drier parts of south-eastern Australia (long-term means for annual rainfall 1220 and 387 mm, respectively). In each area two sets of species were studied from soils relatively rich in phosphorus and relatively poor in phosphorus. An important feature of the study was that many families and even genera spanned the full range of rainfall, and it was possible for any trends in form and function emerging without consideration of phylogeny to be checked in numerous 'phylogenetically independent contrasts'. As in the American study it was found that at a given life span the dry-site species had higher LMA and lower Amass but higher Aarea as the LMA effect overwhelmed the Amass effect. At a given N concentration they had a higher LMA. At a given epidermal conductance they had a higher Aarea. Wright et al. interpreted these results as follows. At a given life span the leaves of dry-site plants are thicker and pack in more total N per unit area; potentially this allows them to have a higher water-use efficiency (WUE), but in practice the leaf-to-air water vapour deficit is so much greater that the WUE is much the same at the two sites (and then only when water is freely available in the soil at the dry site). In their article in this issue, Wright & Westoby take this study to the next intriguing stage by measuring and interpreting key mechanical properties of the leaves. Most present-day researchers concerned with leaf strength use an automated scissor- or guillotine-type of apparatus to measure work to shear (J m−1) and tissue toughness in the strict sense (J m−2), and Wright & Westoby have followed this course. Fascinatingly, they found that high-rainfall and low-rainfall species did not differ in work to shear at a given life span, but the low-rainfall species had a lower work to shear at a given LMA, the pattern being driven by a 50% lower toughness in the low-rainfall species. In other words, at a given life-span they are markedly less protected physically, almost certainly because of a low degree of development of fibrous tissues. Thus the tactic of piling the leaf high in N and getting a WUE (after rain) as good as that in high-rainfall plants comes at a potentially considerable cost in terms of attracting herbivores. Are the results likely to have a high degree of generality? The authors are cautious on this point, citing the great scatter in foliar N concentrations in a motley selection of semidesert shrubs and dry woodland trees, compiled by Killingbeck & Whitford (1996). I am more upbeat – provided we stick to the sort of comparison made by Wright & Westoby. Consider southern Africa first. T. Rooke (pers. comm.) has found that 53 out of 73 species in dry woodland in Botswana (approx. 400–600 mm year−1) have foliar N concentrations of 20–30 mg g−1, certainly higher than you would expect in most rainforests. I obtained surprisingly high values for LMA for Acacia species with tiny leaflets (both evergreen and deciduous) collected in Namibia and northern South Africa. Moving to Central and South America, the evidence is more equivocal, but studies at two sites in the Chamela Reserve in Mexico (approx. 750 mm year−1) obtained mean foliar N values of 30 and 36 mg g−1, respectively (Jaramillo & Sanford, 1995). Working in very dry woodland on the Coro Peninsula in Venezuela (approx. 400 mm year−1), Diaz (1999) obtained mean values of 23 and 30 mg g−1, respectively, for evergreen and deciduous-cum-semideciduous species, and found that the LMA values were high, relative to the N concentrations. Others working in South America have not found such high values for foliar N (Jaramillo & Sanford, 1995). Of course there is no one simple trend all the way from rainforest to semidesert in tropical and subtropical regions. Without a doubt the dominant leaf-types in semideserts in Australia and southern Africa are strongly correlated with, and apparently dependent upon, the soil concentrations of P (Australia; Specht & Specht, 1999) or N (Africa; P.J. Grubb & P.J. Carrick, unpublished). Huge areas of sandy soils very low in P and/or N are dominated by 'shrub-forming' grasses with hard, spiny, long-lived leaves. These leaves are not high in N, and judging by their effect on one's anatomy, are not lacking in toughness. And, of course, many semidesert areas have an abundance of succulents encased in a thick layer of hard tissues. These facts do not make the potential generality of the findings of Wright & Westoby for woodlands and forests on nonextreme soils any less interesting. As to understanding the properties of the dry-site leaves in mechanistic terms, we urgently need to know the area of chloroplast displayed on cell walls abutting air spaces per unit external area of leaf (Evans & von Cammerer, 1996). Presumably it will be higher than in other leaf types. But why is the maximum rate of photosynthesis per unit N lower in these plants? Is it a matter of investing less in enzymes catalyzing CO2-fixation, or are the mesophyll walls notably thicker (perhaps aiding desiccation-resistance by resisting collapse) so that there is a much greater resistance to diffusion of CO2 in the liquid phase? There is remarkably little information on mesophyll wall thickness, but it is known to vary between species from 0.07 to 0.3 µm in mesophytes, and to 2.5 µm in some high-altitude shrubs (Koerner et al., 1983). At all the sites used by Reich et al. (1999) and by Wright et al. (2001) there is a wide range of N concentrations (and associated properties) in species we assign to the same 'functional type' (e.g. semidesert shrub or very shade-tolerant rainforest tree). I have found the same in semidesert in South Africa (continuous variation in [N], and extremes differing by a factor of 8) and in rainforest in north-eastern Australia (factor of 5). Classically we have thought of there being a fixed supply of available N in a given soil, and the plants doing what they can with it. Surely the very wide array of foliar N concentrations within a functional type must mean that different species are investing to very different extents in uptake of N to 'run' a given amount of dry mass accumulated on a given soil; attention now needs to be switched to co-ordinated studies of leaf and root properties. Wright et al. (2001 ) recognize the problem, noting that very different foliar N concentrations (and associated properties) seem to be viable in any one species-rich system. But that won't satisfy a community ecologist interested in the basis of long-term coexistence of species, unless Hubbell's 'neutral model' theory is to be believed ( Hubbell, 2001 ). The problem is analagous to that of understanding variation in seed size ( Grubb & Metcalfe, 1996 ; Rees & Westoby, 1997 ). From 1940 to 1990 ecologists thought about why seeds of shade-tolerant or drought-tolerant plants were, on average, bigger than those lacking such tolerance. But the differences between the mean values for such functional groups are modest (generally 1–2 orders of magnitude in dry mass) while the range in dry mass within a functional group is typically much greater (3–6 orders of magnitude). Coming back to leaves, we need a wave of bright new ideas to explain the coexistence of species which might look to a nonspecialist very similar but which have very different machinery inside.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.