Global biogeography of plant chemistry: filling in the blanks

Authors


It would perhaps come as a surprise to many nonbiological scientists (or even some biologists) to learn that despite our ability to characterize a number of environmental variables, such as climate, along regional or continental gradients, until recently we have had almost no basis for doing so for plant and soil chemistry. New work, including a paper by Han et al. in this issue (pp. 377–385), is beginning to fill in the blanks on this otherwise empty slate. It is well known that long-term climate records exist in a relatively well-distributed network across much, but not all, of the globe. Hence, we are able to quantify the difference in climate between, for example, central Saskatchewan, Canada and central Nebraska, USA but not the differences in plant or soil nutrient concentrations or contents between these two regions. Given the importance of nitrogen (N) and phosphorus (P) to plant function, to production of agricultural and unmanaged ecosystems and to global biogeochemical cycles, including the carbon (C) cycle, one could argue that knowledge of biogeography of their biochemistry is as useful as knowledge of many other kinds, yet it has been little emphasized. Why?

‘… global heterogeneity in leaf N and P is likely substantial enough that we will require sweepingly comprehensive data sets before we will be able to reconcile differences that may arise owing to differences in intensity of sampling in different ‘ecoregions’ of the Earth.’

Why do we know so little about the biogeography of plant chemistry?

Several factors likely contribute to our lack of understanding of the biogeography of plant chemistry. First, whether one was interested in agricultural crop yield, ecological physiology or biogeochemistry, research in plant chemistry – here represented by the simple stoichiometry of C, N and P – has historically been process-oriented or site-based. (As a reminder, given slight variation in leaf C concentration, leaf N and P concentrations are excellent indicators of C : N and C : P ratios.) For example, the emphasis has traditionally been on understanding how biochemistry fundamentally regulates plant physiological function, and on potential consequences for interactions with competitors, consumers and decomposers. Spatially, we might have asked what regulates leaf N or P concentration (hereafter just leaf N or P, for brevity) at the microsite scale, such as in a forest gap or in a low or high spot in an agricultural field. I would guess that less than 1, 0.1 or even 0.01% of all publications that reported plant nutrient contents have been concerned with spatial patterns at regional to global scales. Second, plant chemistry is phenomenally heterogeneous both temporally and spatially, with many factors playing a role in generating these patterns.

Among these important factors are climate, geomorphology, vegetation type and site history (Reich & Oleksyn, 2004; Wright et al., 2005). Temperature and moisture gradients can directly influence leaf chemistry and can indirectly influence soil biogeochemical processes and vegetative composition, each of which can influence the average foliar N or P. Geomorphology influences the kinds of mineral substrate from which soils develop and thus soil characteristics such as N cycling, P availability and cation exchange capacity, all of which influence plant composition and nutrient status. For a given type of mineral substrate, time since major geological disturbance (i.e. soil age) also influences soil nutrient supply and hence plant chemistry (Vitousek et al., 1995; Richardson et al., 2005). Finally, given the close coordination of leaf N and P with other leaf traits such as leaf life span and specific leaf area (Reich et al., 1997; Wright et al., 2004), communities or biomes dominated by certain kinds of species (e.g. deciduous or evergreen; those having a short leaf life span vs those having a long leaf life span; those having a high specific leaf area (SLA) vs those having a low SLA) will differ in leaf N and P. For example, even when growing on similar soils, evergreen trees always have lower N and P on average than deciduous ones. Spatial heterogeneity in this set of factors (climate, geomorphology, site history, vegetation type) at regional, continental and global scales is impressive, and so far has swamped our ability to develop predictive models of leaf N or P. However, as the work of Han et al. demonstrates via a comprehensive assessment of plant foliar N and P across all of China, we are beginning to characterize quantitatively and increase our understanding of these issues at local, regional or global scales.

China: one gap filled, several to go

There has been a recent increase or renewal of interest in the biogeography and the stoichiometry of ecological chemistry (e.g. Sterner & Elser, 2002; McGroddy et al., 2004; Reich & Oleksyn, 2004). These studies have highlighted the general importance of stoichiometry to ecology across the range of biota and ecosystem types (Sterner & Elser, 2002), identified biogeographic patterns in terrestrial foliar stoichiometry across local, regional and global gradients (Vitousek et al., 1995; McGroddy et al., 2004; Reich & Oleksyn, 2004; Richardson et al., 2005), and tested for global convergence in the relationships between foliar stoichiometry and other foliar metabolic and morphological characteristics (Reich et al., 1997; Wright et al., 2004). However, we likely know more about the processes involving links between ecosystem physiology and biogeochemistry than we do about its spatial patterns, especially at large scales.

Despite advances, even the most comprehensive studies published to date have had gaping ‘holes’ in their biogeographic coverage (Fig. 1). One area for which there have been few data reported in the peer-reviewed international literature is China. For example, in the Reich & Oleksyn (2004) study, 5086 records of 1287 species were used, and only 11 of the 5086 records were from China. The new report by Han et al. goes a long way towards filling this gap, and in so doing identifies some differences with past studies that illuminate the need for a more comprehensive global data base. The findings presented by Han et al. are consistent in some but not all respects with findings from earlier studies based on data from other regions. As seen in previous studies, Han et al. found that leaf N and P were significantly greater in herbs than in woody plants and in deciduous than in evergreen species. Han et al. also reported that leaf N and P increase with increasing mean annual temperature (MAT) and latitude, as also recently shown by McGroddy et al. (2004), Reich & Oleksyn (2004) and Kerkhoff et al. (2005).

Figure 1.

Map showing sites of the Reich & Oleksyn (2004) (triangles) and Wright et al. (2004) (circles) global studies. Many sites are not visible owing to their proximity to other sites.

Equally or more interesting are the ways in which the Chinese data differ from previously published data. Although Han and colleagues report a mean leaf N similar to that in two other recent broad studies (all ≈ 20.1–20.6 mg g−1; Elser et al., 2000; Reich & Oleksyn, 2004) that they use as benchmarks, they note that the mean leaf P in the Chinese data (1.46 mg g−1) is significantly (P < 0.05) lower than in the Reich & Oleksyn (1.77) or Elser et al. (1.99) data sets. As a result, the mean leaf N : P ratio is also higher in the Chinese data (16.3) than in the two other data sets (≈ 13). Han et al. logically interpret these differences as follows.

‘Because leaf N : P mass ratio is a good indicator of the relative limitation of N vs P (N : P ratios < 14 often indicate N limitation and N : P ratios > 16 frequently signifying P limitation … ), the higher N : P ratio of this study than in others … might imply that China's flora are relatively more limited by P than the world flora analysed by Reich & Oleksyn (2004)’.

They go on to suggest that low soil P content may be the cause of low leaf P and high leaf N : P ratio in Chinese flora, given that leaf P is related, albeit loosely, to soil P content, and that data compilations suggest that soil P in China is on average lower than the global average. Their conclusion, although tentative, seems appropriate, given our current state of knowledge.

It is also of interest to compare these results to those of another almost entirely independent global-wide data survey, the Glopnet study of Wright et al. (2004), which Han et al. did not refer to. The Glopnet data have a lower mean leaf P (1.11 mg g−1; 58 sites, n = 752) than even the Chinese data; a slightly lower leaf N (19.3 mg g−1; 143 sites, n = 2061) than the Elser et al. (2000), Reich & Oleksyn (2004) or Han et al. data; and a higher N : P ratio (18.2; 58 sites, n = 745) than all three of these data sets. Although a formal analysis needs to be done, it appears that the Glopnet data set (Wright et al., 2004) contains a greater fraction of data from Australia and other regions known to have predominantly infertile soils with low P contents than do the Reich & Oleksyn (2004) or Elser et al. (2000) data sets. If so, this is consistent with the explanation of Han et al. for lower leaf P and higher N : P ratio for the Chinese data than for the Reich & Oleksyn (2004) or Elser et al. (2000) data. However, these comparisons suggest that global heterogeneity in leaf N and P is likely substantial enough that we will require sweepingly comprehensive data sets before we will be able to reconcile differences that may arise owing to differences in intensity of sampling in different ‘ecoregions’ of the Earth.

The Chinese data of Han et al. also differ from other recent studies in the nature of the relationships of leaf N, leaf P and the leaf N : P ratio with latitude or MAT (McGroddy et al., 2004; Reich & Oleksyn, 2004; Kerkhoff et al., 2005). Although both leaf N and leaf P increase with latitude and MAT for the Chinese data as in the earlier publications, the correlations differ (slopes differ significantly: P < 0.001) when comparing Chinese data to the Reich & Oleksyn (2004) global data (Fig. 2). The relationship of leaf N with MAT has a steeper slope and better fit in the Chinese than the global data set, with the reverse true for the relationship of leaf P with MAT. Moreover, there is virtually no relationship of leaf N : P ratio with MAT in the Chinese data, whereas Reich & Oleksyn (2004) reported 31% of total variation in leaf N : P ratio could be associated with variation in MAT, and McGroddy et al. (2004) and Kerkhoff et al. (2005) also noted a positive relation between the two. Although a number of factors could lead to such differences, what those are is not clear at present.

Figure 2.

Regression of leaf N (mg g−1), leaf P (mg g−1) and leaf N : P ratio in relation to mean annual temperature (MAT, °C) for Chinese (open symbols, solid lines; Han et al., 2005) and global (filled symbols, dotted lines; Reich & Oleksyn, 2004) data compilations. For consistency with the published record, for the Chinese data, the data were at the species level within sampling areas (as in Han et al., 2005, fig. 4) and for the global data set, the data were species averages (as in Reich & Oleksyn, 2004, fig. 1). However, to normalize the data and enable statistical comparison of the data sets, the Han et al. data were converted to logarithmic values and the relationships with MAT were considered to be linear. Slopes of the two data sets were significantly different (P < 0.001) in all three cases. Relations of (a) leaf N vs MAT for the Chinese data (r2 = 0.14, P < 0.001, n = 813) and the global data (r2 = 0.03, P < 0.001, n = 1251); (b) leaf P vs MAT for the Chinese data (r2 = 0.10, P < 0.001, n = 1177) and the global data (r2 = 0.37, P < 0.001, n = 923); and (c) leaf N : P ratio vs MAT for the Chinese data (r2 < 0.01, P = 0.53, n = 786) and the global data (r2 = 0.31, P < 0.001, n = 894).

What do we need to know in the future?

Knowledge of broad biogeographic patterns of leaf N and P not only is important for and contributes to understanding of continental- to global-scale issues, but also to local and process-oriented questions. The latter may seem counterintuitive, but in fact, if we can make sense of how multiple global drivers collectively influence plant N and P, this will provide a foundation and context within which to view patterns and processes at local scales. Thus, in conclusion, we simply need more data about leaf and soil chemical attributes for as many ecosystem types in as many geographic regions as possible, especially when those attributes can be linked to quantitative information about vegetation type and history, geomorphology, soils, land use history, etc. I look forward to the day when an accurate global contour map of plant N or P can be made. You would not likely plan a picnic around it (although in an N- and P-rich site, the ants might be less likely to hone in on your greens), but it would provide an important information layer for both ecological science and global environmental management that could help us better predict responses of terrestrial ecosystems to disturbances such as elevated atmospheric CO2, N deposition, pest outbreaks or alternative land management scenarios.

Ancillary