Understanding ecological variation across species: area-based vs mass-based expression of leaf traits


  • Mark Westoby,

    1. Department of Biological Sciences, Macquarie University, Sydney, NSW, Australia
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Peter B. Reich,

    1. Department of Forest Resources, University of Minnesota, St Paul, MN, USA
    2. Hawkesbury Institute for the Environment, University of Western Sydney, Penrith, Australia
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Ian J. Wright

    Corresponding author
    • Department of Biological Sciences, Macquarie University, Sydney, NSW, Australia
    Search for more papers by this author
    • These authors contributed equally to this work.

(Author for correspondence: tel +61 2 9850 4228; email ian.wright@mq.edu.au)

This article discusses the paper by Lloyd et al. (2013, in this issue of New Phytologist, pp. 311–321), a paper that sets out to criticize the leaf economic spectrum (LES) described by Wright et al. (2004). Table 1 identifies the two key arguments from Lloyd et al. (2013) and responds to them.

Table 1. The two main arguments put by Lloyd et al. (2013, in this issue of New Phytologist, pp. 311–321) (left-hand column) as to why mass-based representation of leaf trait data should be regarded as misleading, together with our comments or responses (right-hand column)
Argument from Lloyd et al. (2013)Our comments or responses
Argument 1: Photosynthesis is naturally an area-based process. When it is expressed on a mass basis, the dominant axis of variation then mainly reflects leaf mass per area (LMA) variation rather than leaf-area (Aarea) variation.

Response 1.1: Indeed it is true that the leaf economic spectrum (LES) reflects variation in LMA and leaf lifespan. The LES expresses differences across species in the cost of investing in a unit leaf area and in the duration of the revenue stream that arises from the investment. That is the reason why we called it the leaf economic spectrum rather than, say, the photosynthetic metabolism spectrum.

Response 1.2: More generally we (Wright et al., 2004) advocate looking at variation across species both on an area basis and on a mass basis, since both can be enlightening in complementary ways. For example, both leaf structure (LMA and associated traits) and biochemistry (e.g. nitrogen (N)) affect leaf carbon exchange processes, with these effects being partially or entirely independent of one another (Reich & Walters, 1994; Reich et al., 1998; Peterson, 1999). Gas exchange occurs through surfaces (area basis), while growth over time is most often expressed on a mass per mass basis, and there are good reasons for both choices.

Response 1.3: There is no reason to regard Amass as a derived measurement vs Aarea as a primary measurement. When assimilation A is measured in a gas cuvette, typically mass of the leaf sample and area of the leaf sample are each measured separately. Amass is just as fundamental a measurement as Aarea is. There are three primary measurements (assimilation, sample leaf mass, sample leaf area), but since the size of the sample is not informative about the biology, the variables reported are always the three ratios Aarea, Amass, LMA. Any two of these ratios can be used to calculate the third. This point extends to further measurements involving leaf N, which typically is measured on a dry mass basis in the first instance, but can be expressed as Nmass or Narea.

Argument 2: If there were random or ‘error’ variation in LMA, then this would propagate during conversion of Aarea into Amass, giving rise to negative correlation across species between Amass and LMA even when there is no correlation between Aarea and LMA.

Response 2.1: Certainly it is true that much if not all of the negative correlation between Amass and LMA can be seen as a result of dividing Aarea through by LMA. To put that another way, what is biologically enlightening is that Aarea is not much correlated across species with LMA; at least as enlightening as that Amass is strongly negatively correlated with LMA.

Response 2.2: In any event, high r2 among photosynthetic and nutrient-concentration traits and LMA is not a primary basis for identifying the LES. The primary basis is the logic of investment costs and returns on investment, and the wide spread of LMA and leaf lifespan across species, including species coexisting within the same habitat.

Response 2.3: One of the points made by Lloyd et al. (2013) is that these scaling relationships depend strongly on the variances of the (log-scaled) variables relative to each other. Indeed it was for this reason that Wright et al. (2004) expressed relationships between variables via the standardized major-axis slope, which is the ratio of the standard deviations of the two variables, rather than via ordinary least-squares regression, where slopes are different depending which variable is treated as x and which as y.

The LES is a dimension of ecological variation across plant species. It concerns construction and maintenance costs of leaves, and duration of photosynthetic returns from those investments (Reich et al., 1997; Wright et al., 2004). The LES is underpinned by a positive relationship between leaf mass per area (LMA, reflecting investment costs) and leaf lifespan (LL, reflecting duration of the revenue stream arising). Variation among species in LMA reflects underlying variation in quantitative anatomy such as thickness of cuticle, mesophyll and other cell layers, and the degree of structural reinforcement (Niinemets, 1999).

Photosynthetic rate on a leaf-area basis (Aarea) does not vary nearly so widely across species as does leaf mass per area LMA (c. 40-fold vs 100-fold, in the glopnet dataset provided by Wright et al., 2004 and re-analysed by Lloyd et al., 2013). Also, photosynthetic rate on a mass basis (Amass; 140-fold variation) decreases with increasing LMA across species. These two statements are equivalent to each other because Aarea divided by LMA is the same as Amass. Hence, if Aarea graphed against LMA is a horizontal relationship or nearly so, then a relationship between Amass and LMA will be sloped downwards. The two statements are inter-convertible.

Lloyd et al. (2013) are concerned with the equivalence between these two statements. They carry out several statistical manoeuvres, but they all boil down to the fact that mass-based and area-based data contain the same evidence and can be inter-converted. They seem especially concerned that high r2 between LMA and mass-based assimilation rate or nitrogen (N) content can equally be interpreted as low correlation between LMA and area-based assimilation or N. But high r2 among traits is not the main pillar supporting the LES concept. Rather the LES rests on the logic of the trade-off between construction cost and duration of revenue from a unit of leaf area deployed.

Mass-based vs area-based data presentation has actually been under discussion over decades (Field & Mooney, 1986; Reich & Walters, 1994; Reich et al., 1997, 1998, 1999; Peterson, 1999). Wright et al. (2004), in the paper proposing the LES, provided a page-long section titled ‘Area vs mass basis of expression’, elaborating what can be learned from the different presentations of equivalent data.

Central to Lloyd et al. (2013) is their view that area-based statements are genuine and mass-based statements are in some sense derived or artefactual. They write ‘… the relevant question is which of the two is the “parent” correlation: that is, the one justified on theoretical or logical grounds’. They call mass-basis data spurious (though with a degree of ambivalence): ‘correlations [between Amass and LMA] may be considered “spurious”: Yet only spurious in the sense … the correlations themselves are real … [yet] requiring careful insights for their interpretation’. They coin a new term ‘lulu-effect’, seemingly with the intention this may sound less derogatory than ‘spurious’.

The crux of the argument put forward by Lloyd et al. (2013) is this: ‘Given that the main function of leaves is to intercept light for which the rate of arrival of photons into the plant canopy has a natural dimension of flux density (i.e. flux per unit area), then an area-based metric seems to us to be the more logical one’. In response and surely with equal logic, it can be argued that a main function of leaves is to deliver a profitable return on the investment that has been made in constructing the leaf. Since investment and return are naturally expressed in units of mass, then a mass-based representation will correspond to this argument. Perhaps the first argument will appeal more to people interested in photosynthetic physiology, and the second more to people interested in plant growth and economics. But both points of view deserve respect.

Looking at leaf metabolism and gas exchange as rates at a point in time is complementary to looking at leaves as investments, made for the purpose of a growth process over time. They are two views of the same elephant. It makes biological sense that assimilation rates on an area basis would vary across species within somewhat constrained bounds. There must be an upper bound set by the supply of light and by the problems of exchanging water for CO2 across a surface. Low assimilation rates on an area basis are presumably disadvantaged by competition. By contrast, LMA and LL associated with a leaf surface can and do vary more widely across species. Low LMA together with short LL (and high Nmass and Pmass) is a trait-constellation capable of producing competitive returns on investment over a leaf's lifetime, but so also is high LMA together with long LL (and low Nmass and Pmass). It is this spread of LMA-strategies that is expressed by the LES. Clearly, leaves are under natural selection to meet two key criteria: considered day-by-day, they need to intercept light and carry on photosynthesis at a competitive rate; considered over the leaf's lifetime, they need to provide a positive return on the investment costs for the leaf. Investment and return are measured most obviously in units of dry matter, but it is also relevant to think about returns per unit N or phosphorus (P) invested in the leaf.

We agree with the main statistical point of Lloyd et al. (2013), that strong r2 between LMA and mass-based leaf traits can correspond with the same leaf traits on an area basis being uncorrelated with LMA, or only weakly so. This point is not new. That mass- and area-bases for expression are complementary and can be interconverted via LMA is well known, as is the use of multiple regression approaches to quantify assimilation–N relationships independent of LMA (Reich et al., 1998; Peterson, 1999; Wright et al., 2004). Consequently the position of Lloyd et al. (2013), that mass basis is somehow misleading and should not be adopted, is, in our view, a step backwards rather than forwards. And we do not give credit to Lloyd et al.'s complaint that Wright et al. (2004) has somehow hindered researchers from looking at data on a leaf-area basis. That is simply untrue. Wright et al. (2004) explicitly advocated looking at leaf traits from both perspectives.