Plant and Soil

, 348:7

Plant mineral nutrition in ancient landscapes: high plant species diversity on infertile soils is linked to functional diversity for nutritional strategies


    • School of Plant BiologyThe University of Western Australia
  • Mark C. Brundrett
    • School of Plant BiologyThe University of Western Australia
  • John A. Raven
    • School of Plant BiologyThe University of Western Australia
    • Division of Plant SciencesUniversity of Dundee at SCRI, Scottish Crop Research Institute
  • Stephen D. Hopper
    • School of Plant BiologyThe University of Western Australia
    • Royal Botanic Gardens, Kew
Marschner Review

DOI: 10.1007/s11104-011-0977-6

Cite this article as:
Lambers, H., Brundrett, M.C., Raven, J.A. et al. Plant Soil (2011) 348: 7. doi:10.1007/s11104-011-0977-6


Ancient landscapes, which have not been glaciated in recent times or disturbed by other major catastrophic events such as volcanic eruptions, are dominated by nutrient-impoverished soils. If these parts of the world have had a relatively stable climate, due to buffering by oceans, their floras tend to be very biodiverse. This review compares the functional ecophysiological plant traits that dominate in old, climatically buffered, infertile landscapes (OCBILS) with those commonly found in young, frequently disturbed, fertile landscapes (YODFELs). We show that, within the OCBILs of Western Australia, non-mycorrhizal species with specialised root clusters predominantly occur on the most phosphate-impoverished soils, where they co-occur with mycorrhizal species without such specialised root clusters. In global comparisons, we show that plants in OCBILs, especially in Western Australia, are characterised by very low leaf phosphorus (P) concentrations, very high N:P ratios, and very high LMA values (LMA = leaf mass per unit leaf area). In addition, we show that species in OCBILs are far more likely to show P-toxicity symptoms when exposed to slightly elevated soil P levels when compared with plants in YODFELs. In addition, some species in OCBILs exhibit a remarkable P-resorption proficiency, with some plants in Western Australia achieving leaf P concentrations in recently shed leaves that are lower than ever reported before. We discuss how this knowledge on functional traits can guide us towards sustainable management of ancient landscapes.


Ancient landscapesBiodiversityCluster rootsLMAMycorrhizaNitrogenOCBILPhosphorusSclerophyllousYODFEL


There is a trend in plant ecology to emphasise global patterns in large data sets relating plant ecophysiological traits to climatic and latitudinal information (Reich and Oleksyn 2004; Wright et al. 2005, 2004). Whilst such global trends undoubtedly provide pivotal insight into the evolution of the world’s vegetation and are crucial for parameterising vegetation–climate models, it is also important to acknowledge deviations from general trends and acknowledge effects of edaphic factors, in particular landscape, substrate and soil age and soil phosphorus availability (Beadle 1953; Lambers et al. 2008b). Edaphic factors, in particular nutrient availability, as related to plant species distribution will be the focus of this review.

Excluding epiphytes, parasites and carnivorous species, terrestrial plants acquire most essential mineral nutrients from soil, primarily via two pathways: (i) direct absorption via roots, and (ii) indirectly via symbiotic mycorrhizal fungi. The vast majority of plants have the capacity for phosphorus (P) uptake via both pathways, but are considered to depend primarily on symbioses with mycorrhizal fungi to acquire P (Smith and Read 2008). In fact, between 86% and 94% of plants are mycorrhizal on a global scale (Brundrett 2009). Many non-mycorrhizal plants either grow in nutrient-impoverished habitats and have specialised nutritional strategies, such as carnivores, parasites and cluster-rooted species, or they lack specialised root structures and grow in disturbed, wet or arid habitats (Fig. 1).
Fig. 1

A summary of functional diversity in nutritional strategies in plant families globally (A), in the United Kingdom (UK) (B), Guiana Shield Region of South America (C), the Cape Floristic Region of South Africa (D) and in Western Australia (E). D and E are regions with a relatively high proportion of ancient (OCBIL) landscapes. Data are the relative diversity of plants in each functional category by allocating species at the family or genus level using data from Brundrett (2009). The diversity of species in families is provided by the Natural History Museum Postcode Plants Database (, accessed 17–12-2009), (Funk et al. 2007), (Goldblatt and Manning 2000) and Florabase (, accessed 1–7-2009)

South-western Australia has a wide diversity of landscapes, soils and climates, which support a highly diverse flora with many endemic species (Gibson et al. 2004; Hopper and Gioia 2004). It is a centre of exceptionally high plant species richness and endemism and one of the world’s 25 biodiversity hotspots, which are conservation priorities (Myers et al. 2000). The high biodiversity in this region is associated with a long evolutionary history without major tectonic disturbance or glaciation which has allowed Gondwanan and Tertiary relics to coexist with more recently evolved species adapted to highly infertile soils and periodic disturbances such as fire (Hopper and Gioia 2004). The significance of species and communities is often poorly known due to the lack of regional surveys in these areas (Anonymous 2003).

Biodiversity hotspots, i.e. areas of very high species richness, endemism and accentuated threat, are recognised at several different scales in south-western Australia. The entire Southwest Australian Floristic Region (SWAFR) is an internationally recognised hotspot (Myers et al. 2000), which also contains smaller, nationally recognised hotspots within it (Hopper and Gioia 2004). Use of a more rigorous approach, using plant occurrence data in the Western Australian Herbarium’s Florabase (, has identified seven subregional centres of biodiversity and endemism within the SWAFR (Hopper and Gioia 2004). The Southwest Interzone, a transitional region that occurs immediately to the east of the SWAFR (Beard 1990) also has a high level of plant endemism and rarity (Anonymous 2007).

Here, we review existing literature and some unpublished information on the mineral nutrition of plants in natural habitats in old, climatically buffered, infertile landscapes (OCBILs), sensu Hopper (2009). We compare plants in OCBILs with those found in young, frequently disturbed, fertile landscapes (YODFELs) and identify significant gaps in our understanding of the ecophysiological functioning of species in these old systems. Links between plant functional traits, plant diversity and soil properties in ancient landscapes are discussed, with particular reference to the Southwest Australian Floristic Region. These links are examined by comparing data at both global (ancient vs. younger landscapes) and local scales (due to variations in soil fertility). While we note that the age of a given landscape, its substrates and its soils may differ, the concept of OCBILs explicitly focuses on landscapes where substrates and soils are also relatively old with very slow rates of erosion and rejuvenation. Hence OCBILs are old with deeply weathered infertile soils, contrasting with YODFELs, which are young and fertile, with occasional rejuvenation of soils and substrates by disturbance events such as glaciation and volcanic ash deposition.

Historical perspective

Hypotheses concerning soil fertility and landscape age have had a chequered and controversial history. For example, in south-western Australia, after a long transoceanic journey from Cape Town and inspecting Rottnest Island offshore from future Perth on the 31st December 1696, Dutch commander Willem de Vlamingh recorded “… I had great pleasure in admiring this island, which is very attractive, and where it seems to me that nature has denied nothing to make it pleasurable beyond all islands I have ever seen, being very well provided for man’s well-being, with timber, stone and lime for building him houses, only lacking ploughmen to till these fine plains. There is plentiful salt, and the coast is full of fish. Birds make themselves heard with pleasant song in these scented groves. So I believe that of so many people who seek to make themselves happy, there are many who would scorn the fortunes of our country for the choice of this one here, which would seem a paradise on earth.” (Playford 1998: p. 84). However, a more prosaic account of the “fine plains” of the island appeared in the log of uppersurgeon Mandrop Torst, who stated “the ground is covered with little or no soil, mostly white sand and rocky, and in my opinion not suitable for cultivation.” (Playford 1998: p. 29). The presence of nutrient-deficient sands and rock, so prevalent in south-western Australia, was here recorded for the first time by Europeans.

Subsequent explorers and settlers similarly voiced contrasting impressions as the coast was mapped and parties ventured inland (Hopper 2003, 2004). Deep sand, subdued to flat topography and prevalence of salt lakes well inland were interpreted by many observers as signifying a young landscape recently emerged from the sea. Charles Darwin advocated such a view for south coastal headlands of King George Sound (Hopper and Lambers 2009), echoing early British maritime explorers George Vancouver and Matthew Flinders. Yet evidence to the contrary would ultimately dispel the young landscape hypothesis. The anomalous richness and high endemism of the flora of this region of supposedly young landscapes drew early comment (e.g., Hooker 1860). Marine fossils were conspicuously absent across most of the southwest inland. The notion that most south-western Australian landscapes may be very old and infertile gradually supplanted the young landscape hypothesis, and remains the prevailing view (Hopper et al. 2009).

Joseph Hooker first considered that high local endemism in the south-western flora was due to infertile soils and harsh climate, inhibiting seed production and thereby allowing many species to coexist in a small area (Darwin Correspondence Project, Letter 2358, 12th November 1858). While other factors, including genetic compatibility, influence seed production, Hooker’s hypothesis opened up a rich field of enquiry pertaining to the complex but subtle soil mosaics of the OCBILs of south-western Australia and elsewhere, and their rich endemic floras. The ancient soils of Western Australia have had a major impact on historical patterns of settlement as local variations in soil quality, especially fertility, determined where agriculture could succeed before the large-scale application of fertilisers was feasible (McArthur 1991).

N or P limitation in OCBILs?

Levels of N and P vary in rock types across the Earth (Mason 1958; McBirney 1993; Vinogradov 1962; Vitousek et al. 2010), but by no means as much as they do in soils of OCBILs and YODFELs. The mean P content of the Earth’s crust is 1.2 g per kg, with a variation of over two orders of magnitude: the quartzite sandstone underlying the OCBIL fynbos biome has less than 40 mg P per kg. The mean N:P ratio by mass of the Earth’s crust is <0.1, with a range of 0.01–0.8, with ratios in igneous rocks (felsic and mafic) of <0.1 and a mean of 0.8 in sedimentary rocks (Mason 1958; McBirney 1993; Vinogradov 1962; Vitousek et al. 2010). Research on chronosequences in both New Zealand (Crews et al. 1995; Richardson et al. 2004; Walker and Syers 1976) and Hawaii (Chadwick et al. 1999; Crews et al. 1995) has made it clear that, whilst N is a major limiting soil nutrient in young landscapes (and hence in YODFELs), P becomes increasingly limiting when soils age (and hence in OCBILs) (Lambers et al. 2008b). Work on a chronosequence in a drier habitat in Arizona shows a similar sequence (Selmants and Hart 2010). This leads to the hypothesis that leaf P levels are typically lower, and N:P ratios higher, in landscapes where OCBILs predominate, such as in south-western Australia and in the Cape Region in South Africa, when compared with younger landscapes on Earth. This hypothesis will be tested here by compiling available literature data.

In addition to the ecological interest in N and P as key limiting factors, there is a growing interest in the strategies identified in species in OCBILs to efficiently acquire and use P, because of dwindling global P reserves (Cordell et al. 2009; Lambers et al. 2006; Vance et al. 2003). Following a recent review (Lambers et al. 2008b), we explore if species with root clusters (Lambers et al. 2006; Roelofs et al. 2001) are relatively more prominent and mycorrhizal species relatively less prominent (Brundrett 2009) on the most P-impoverished soils in OCBILs. In addition to P acquisition, we compare P use in OCBILs with that in other regions of the world, focusing both on photosynthetic P-use efficiency and P-remobilisation efficiency and proficiency. Working towards more sustainable cropping systems that acquire and use P more efficiently is becoming increasingly important, and knowledge on native plants in old landscapes may guide us towards that outcome (Ryan et al. 2009).

Plant species richness per plot is strongly correlated with decreasing total and “plant-available” soil P concentrations (Fig. 2); “plant-available” P amounts to 5% of total P, across all of the soils sampled in this region (Fig. 3a). A similar dependence of plant distribution and species diversity on soil P levels has been found before (Adam et al. 1989; Beadle 1962, 1966; Huston 1980). Preliminary analysis of plant alpha diversity and mineral nutrition strategies in south-western Australia suggests that a relatively high proportion of species are non-mycorrhizal (Brundrett 2009; Lambers et al. 2008b). The data used for Fig. 2 can only be used to investigate trends in alpha diversity, but beta diversity also tends to be higher in habitats such as proteaceous heathland vegetation where the overall diversity in plots is highest (unpublished data). Figure 2b illustrates the relative proportion of non-mycorrhizal species belonging to different ecological and nutritional categories. Non-mycorrhizal species with proteoid or dauciform roots represent a significant proportion of all species on the poorest soils only (Fig. 2c). Pate and Bell (1999) found that on similar nutrient-poor soils the biomass of species with proteoid roots represent 65% of the total biomass, as opposed to 14% for mycorrhizal species; the remainder of the biomass included carnivorous species and Restionaceae (19%) as well as N2-fixing species (2%). On soils with greater P availability, species with these specialised nutrient-acquisition strategies are very uncommon. Mycorrhizal species, including Myrtaceae, are common on these soils (Lambers et al. 2006). In Western Australia, the strong association between Eucalyptus- (Myrtaceae) dominated vegetation and soils with a subtending clay layer, that tend to be more fertile than soils with deep sand or laterite, has led to the theory that dominant vegetation changes soil structure over many years (Pate and Verboom 2009). Similarly, the strong association of Proteaceae with soils over laterite has been ascribed to the activity of cluster roots, which are thought to have mobilised iron in the topsoil, allowing it to move down the profile and giving rise to laterite (Verboom and Pate 2003).
Fig. 2

Correlations between plant diversity and soil phosphorus. A. Plant species richness (numbers of species in 100-m2 quadrats) as dependent on soil P status, in the Southwest Australian Botanical Region (SWAFR), a Global Biodiversity Hotspot in which OCBILs predominate. a. Total number of species. b and c. Numbers of non-mycorrhizal species in specific functional groups. b. Plant diversity and soil data are from a comprehensive floristic survey of over 1000 quadrats in the wheatbelt of Western Australia by Gibson et al. (2004). Each point is an average of all plots for each floristic group at the 25 group level (n = 7 to 53 plots, mean = 27)
Fig. 3

Nutrient status for soils sampled in the wheatbelt of south-western Australia, a Global Biodiversity Hotspot in which OCBILs predominate. a. “Plant-available soil P” as dependent on “Total soil P”. b. The correlation between Total soil N and Total soil P. Data are averages for floristic groups as explained in the legend for Fig. 2

Several data sets exist on N and P concentrations in leaves of a range of plants, worldwide. Here we bring these data together, accepting that many are inherently biased. For example, Rundel et al. (1988) focused on plants in Mediterranean ecosystems. Wright et al.’s (2004) data on species from south-western Australia are biased towards Lamont’s large data set on Hakea species (Proteaceae) (Wright et al. 2004). The only relatively large data set we have deliberately left out is that of Foulds (1993), as this shows an average N:P ratio of 14.3 (range 4.2 for Lamiaceae to 22.4 for Rutaceae), well outside the range that is common in the literature (Güsewell 2004) or common for plants of the same families in the same region (Table 1). Quite likely, there is something wrong with the P analyses in this dataset (W. Stock, pers. comm.). Even if we acknowledge that the data compiled in Table 1 were deliberately collected with a certain bias, a clear message emerges. First, leaf P concentrations in south-western Australia are lower than global average values and also lower than those in any other region listed in Table 1. Leaf P concentrations for fynbos vegetation in the Cape Region of South Africa are also low, albeit slightly higher than those for south-western Australia. Average values for Australia are appreciably higher than those for south-western Australia and the Cape Region in South Africa, but values for China and global averages are considerable higher. Second, leaf N concentrations show a similar trend as leaf P concentrations, but leaf N concentrations are not quite as low, leading to higher leaf N:P values in regions of the world where P is clearly severely limiting (Table 1). Third, leaf N:P ratios are highest where leaf P concentrations are lowest, pointing to severe P limitation. Plant productivity in Mediterranean California, Chile, France and Greece as well as in China, is most likely limited more by N than by P. The world at large shows N:P values that suggest that plant productivity in large parts of the world is limited by P, and not (just) by N (Table 1).
Table 1

Concentrations of N and P and N:P ratios in leaves of plants in different regions of the world. Values for Australia are for various regions on this continent as obtained from data sets including several species and may include values of species in south-western Australia. Values for south-western Australia are predominantly for kwongan vegetation, which predominates in OCBILs. Rundel et al.’s data set pertain to Mediterranean regions only, including fynbos vegetation in the Cape Region of South Africa. Fynbos similarly predominates on OCBILs underlain by Table Mountain Sandstone and Cape Granite, but also may occur on younger landscapes (YODFELs). Denton et al. values refer to 1-year old leaves; values for 3-year old leaves are 15% lower. Sources: (Denton et al. 2007a; Grigg et al. 2008a, b; Han et al. 2005; Herppich et al. 2002; Rundel et al. 1988; Westman and Rogers 1977; Wright et al. 2004; A.M. Grigg, (unpubl.); A. Boonman and E.J. Veneklaas, (upubl.))
Leaf N concentrations are also lower when leaf P levels are low and N:P ratios are higher. The low concentrations of both leaf P and leaf N are partly accounted for by “dilution”, as discussed in the next section on leaf mass per unit area (LMA). Alternatively, low leaf N concentrations might indicate that soil N is also in short supply (but not quite as severely as P, as evidenced by higher N:P ratios). A further possibility, as predicted by the “Growth-Rate Hypothesis”, which arose from data on non-photosynthetic microbes and metazoa (Acharya et al. 2004; Elser et al. 1996), the low leaf P (or, better, meristem P) would lead to low ribosomal RNA (rRNA) amounts, and thus limit the rate of protein synthesis. The “Growth-Rate Hypothesis” predictions of a direct proportionality of growth rate with rRNA content, and a declining protein:rRNA with increasing growth rate, are in less good agreement with data from photosynthetic organisms than with those from non-photosynthetic organisms (Flynn et al. 2010; Matzek and Vitousek 2009; Sterner and Elser 2002). However, this lack of fit to some predictions of the “Growth-Rate Hypothesis” by photosynthetic organisms does not undermine the argument that a very low P content in biomass necessitates a low rRNA content and hence, even with the maximum known catalytic activity of ribosomes, a low rate of protein synthesis. Given that the availability of N declines in old soils, albeit not as pronounced as that of P (Parfitt et al. 2005; Richardson et al. 2004), this is a likely explanation. In fact, in the very old soils in the wheatbelt of south-western Australia, soil N and P levels are closely correlated (Fig. 3b). In addition, plants might reduce their N uptake when their growth is severely limited by P. There is, indeed, ample evidence that plants down-regulate their rate of N uptake when P is in short supply: soybean (Rufty et al. 1993); bean (Gniazdowska and Rychter 2000). The rate of N assimilation and leaf N concentrations decline when P is increasingly limiting for plant growth: soybean (Rufty et al. 1993); tomato (De Groot et al. 2003). De Groot et al. (2003) found an increase in N:P ratio from 8 to 14, when comparing tomato plants grown with abundant N and P supply with plants supplied with sufficient N but severely limiting P supply. The increase in plant N:P when P limits the growth rate also occurs in symbiotic diazotrophs that are relying on N2 as their N source. An example is the work on the cluster-rooted white lupin, where the N:P ratio increases from just below 4 to 16 (Schulze et al. 2006). Thus, with the rather poorly defined requirement for additional P in diazotrophs growing with N2 compared with combined N (Vitousek and Field 1999; Vitousek et al. 2010), N2 fixation occurs at a slower rate, but probably not at a significantly higher rate per unit plant P (Tables 1 and 2 in Schulze et al. 2006). Note that these N:P values are lower than those in Table 1 because they pertain to a fast-growing crop species; fast-growing ruderals tend to have lower N:P values (Güsewell 2004). Low N:P ratios for fast-growing plants are in support of the “Growth-Rate Hypothesis” in that such plants require a high activity of rRNA to sustain rapid rates of protein synthesis.
Table 2

Leaf mass per unit area (LMA) of plants in different regions of the world. Values for Australia are for various regions on this continent as obtained from data sets including several species and may include values of species in south-western Australia. Values for south-western Australia are predominantly for kwongan vegetation. Rundel et al.’s data set pertain to Mediterranean regions only, including fynbos vegetation in the Cape Region of South Africa. Sources: (Denton et al. 2007a; Herppich et al. 2002; Mitchell et al. 2008a; Rundel et al. 1988; Stock et al. 1997; Wright et al. 2004; P.J. Mitchell et al. (unpubl).; A.M. Grigg et al. (unpubl.))

In conclusion, the low leaf N concentrations in plants in P-limited landscapes are largely due to their very high LMA values (Table 2), essentially “diluting” all nutrients. Conversely, given the high N:P ratios, low P concentration are not simply the result of “dilution” by LMA. Low leaf N concentrations may also reflect down-regulation of their rate of N uptake and assimilation. Given that the availability of N also declines in very old soils, it is likely that this decreased N availability offers an additional explanation for low leaf N levels.

Leaf nutrient concentrations and leaf mass per unit area (LMA)

Low concentrations of both N and P in plants in old landscapes (Table 1) reflect a high degree of scleromorphy and a high leaf mass per unit area (LMA) of these leaves. Comparing large data sets on LMA, each with the inevitable biases discussed above, again a clear pattern emerges. Plants in south-western Australia clearly exhibit the highest LMA values, followed by “average” Australian plants and plants from the Cape Region in South Africa (Table 2). However, leaf P concentrations in south-western Australia are about four times lower than global average values, whereas LMA values show a three-fold difference. That is, low leaf P concentrations of the scleromorphic leaves are partly, but not entirely accounted for by “dilution” by scleromorphic structures; leaf P concentrations per unit area are also lower, despite the more scleromorphic leaves being thicker (Hassiotou et al. 2009a, b). The thick, scleromorphic leaves have long diffusion pathways for CO2 from the leaf surface to the mesophyll cells. It is very likely that this trait has been the driving force for the evolution of stomatal crypts, whose depth correlates strongly with the thickness of the leaves in which they are found (Hassiotou et al. 2009b; Roth-Nebelsick et al. 2009).

High LMA values are typical for plants from nutrient-poor habitats (Beadle 1962, 1966; Lambers and Poorter 1992; Wright et al. 2002). Accumulation of fibre, thick cell walls, sclerenchyma and quantitatively important secondary plant compounds increase the lifespan and defence against herbivores and abiotic stress (Wright and Cannon 2001). The extremely high LMA values for south-western Australian plants are in tune with the severely nutrient-impoverished status of the soils in this region. The driving forces to reduce nutrient losses due to herbivory must have been strong over prolonged periods in this region dominated by ancient landscapes. The gradual depletion of plant-available phosphorus in Australia has been linked to increased dominance by sclerophyllous plants over the past 25 million years (Beadle 1953, 1954, 1962, 1966). However, this trend is strongest in Western Australia, where the Australian centres of diversity for Proteaceae and Restionaceae occur (Crisp et al. 2004).

Photosynthetic N- and P-use efficiency (PNUE and PPUE)

Given the low concentrations of both N and P in leaves of plants in ancient landscapes, are these leaves capable of achieving rates of photosynthesis similar to plants elsewhere in the world (Wright et al. 2004)? Average rates of area-based photosynthesis in south-western Australia, mostly measured during the wet season, are very similar to those in the rest of Australia and to global averages (Table 3).
Table 3

Rates of photosynthesis per unit leaf area, leaf N (PNUE) and leaf P (PPUE) of plants in different regions of the world. Values for Australia are for various regions on this continent as obtained from data sets including several species and may include values of species in south-western Australia. Values for south-western Australia are predominantly for kwongan vegetation. Rundel et al.’s data set pertains to Mediterranean regions only, including fynbos vegetation in the Cape Region of South Africa. Sources: (Denton et al. 2007a; Grigg et al. 2008a; b; Mitchell et al. 2008a; Wright et al. 2004; P.J. Mitchell et al. (unpubl.); A.M. Grigg et al. (unpubl.))

Rates of photosynthesis per unit leaf N (PNUE) are also very similar, with one report of lower values for fynbos plants in South Africa’s Cape Region. PNUE tends to decrease with increasing LMA (Poorter and Evans 1998; Poorter et al. 1990), and it has been suggested that this is associated with greater allocation of N to cell-wall material, rather than the photosynthetic apparatus (Field and Mooney 1986). High N allocation to leaf cell walls would decrease the N:P of OCBIL plants, assuming there is little P in cell walls. However, no tradeoff was found between N allocated to cell walls as opposed to Rubisco in a range of Australian sclerophyllous species grown in a common garden (Harrison et al. 2009). This would explain why PNUE values of scleromorphic leaves of Australian plants in Table 3 are similar to global averages (Table 3).

Rates of photosynthesis per unit leaf P (PPUE) are particularly high in Banksia species in south-western Australia (Denton et al. 2007a), and tend to be higher for Australian plants in general when compared with global averages (Wright et al. 2004) (Table 3). Grigg’s data set (Grigg et al. 2008a; 2008b) pertains to plants in the Great Sandy Desert in north-western Australia. These plants perform like plants from other regions in Australia (Tables 1, 2 and 3), and very differently from plants in the southwest of Western Australia. The dune systems of the Great Sandy Desert are much younger than the OCBILs of the SWAFR (e.g., Bowler et al. 2001; Hesse et al. 2004). The present results on high PPUE values in environments with severely limiting P availability where plants tend to have a higher LMA agrees with a global comparison of 340 tree and shrub species (Hidaka and Kitayama 2009).

How do some plants from the most phosphorus-impoverished regions in the world achieve relatively high area-based rates of photosynthesis, and extremely high P-based rates (PPUE)? We do know that some of these species, e.g., Hakea prostrata, preferentially allocate P to their photosynthetically active leaf cells, rather than to epidermal and vascular bundle cells (Shane et al. 2004c). However, there is insufficient evidence that this is typical for species from severely P-impoverished sites, and a similar preferential allocation of P to mesophyll has been found in barley (Fricke et al. 1996; Karley et al. 2000; Williams et al. 1993). Since a large fraction of P in cells is in inorganic form in the vacuole (Lauer et al. 1989; Lee and Ratcliffe 1993), an effective way to increase PPUE would be to exclude P from vacuoles. However, soybean leaves (Lauer et al. 1989) and maize roots (Lee et al. 1990), also deplete vacuolar P well before cytoplasmic pools are affected by a low P supply and hence this response is not typical for plants from P-impoverished soils either. Moreover, P concentrations in vacuoles of Hakea prostrata leaves, which have a low total leaf P concentration of 0.4 mg g−1 DW, is about 5 mM (measured using cryoSEM) (Shane et al. 2004c). That vacuolar P concentration is very similar to that in leaves of maize grown under P-starved conditions, measured using 31P NMR (Loughman et al. 1989). Therefore, less accumulation of inorganic P does not provide a likely and complete explanation for a high PPUE.

Is it possible that major P-containing compounds, e.g., phospholipids, have been replaced by, e.g., galactolipids? The discovery of extraplastidic digalactosyldiacylglycerol biosynthesis in Arabidopsis thaliana induced by phosphate deprivation has revealed a biochemical mechanism for plants to conserve phosphate (Härtel et al. 2000). Apparently, plants can reversibly replace phospholipids with non-phosphorus galactolipids if phosphate deprivation requires this (Andersson et al. 2003; Calderon-Vazquez et al. 2008; Cruz-Ramírez et al. 2006; Gaude et al. 2008; Tjellström et al. 2008; Yamaryo et al. 2008). Similarly, replacement of phospholipids by sulfolipids might contribute to the high P-use efficiency of OCBIL plants (Raven 2008). However, phospholipids are only a minor fraction of the total amount of P in cells (Chapin and Bieleski 1982), and, so far, experimental evidence for replacement of phospholipids by non-phosphorus lipids in OCBIL plants is lacking.

Plaxton and co-workers explored how plants respond to phosphate deprivation through the induction of alternative pathways of glycolysis and mitochondrial electron transport, which act as respiratory bypasses to allow respiration to proceed in P-deficient plant cells (Duff et al. 1989; Theodorou and Plaxton 1993). The bypasses effectively negate the necessity for adenylates and inorganic phosphate, whose pools are severely decreased during P starvation. However, they also make respiration less efficient in term of ATP production per unit of glucose respired (Lambers et al. 2008a). There is no information on the significance of these bypasses in leaves of plants from P-impoverished soils that show high PPUE. Such savings are potentially important, since there is very little P cost of the machinery, which is being used less efficiently. The high PPUE emphasises that organic C is not in short supply, and a lower ATP production per glucose respired is therefore not a problem.

So far, there is no evidence that any of the above suggestions to enhance the use of P in leaves pertain to species from the most severely P-impoverished landscapes which exhibit very high P-use efficiencies. Does this mean that we need to look further, beyond what we currently know about Arabidopsis thaliana and crop plants? There may be a way forward to address this problem. The species investigated so far all induce the P-saving strategies in senescing organs when their growth is severely limited by P. We hypothesise that the most P-use efficient native species deploy these strategies constitutively, and not only when their leaves are senescing. In the North Pacific Subtropical Gyre the synthesis of membrane lipids accounts for 18–28% of the phosphate taken up by the total planktonic community (Van Mooy et al. 2006). Interestingly, Prochlorococcus, the cyanobacterium that dominates the phytoplankton in this environment, primarily synthesises sulfoquinovosyldiacylglycerol, a lipid that contains sulfur and sugar instead of phosphate. In Prochlorococcus, <1% of the total phosphate uptake is incorporated into membrane lipids. Thus, evolution of this “sulfur-for-phosphorus” strategy set the stage for the success of picocyanobacteria in oligotrophic environments (Van Mooy et al. 2006). Similar substitutions of phospholipids have been observed in the phytoplankton in the Sargasso Sea (Van Mooy et al. 2009). More research is required to test this “sulfur-for-phosphorus” hypothesis for OCBIL plants.

Do high rates of photosynthesis per unit leaf P also imply high rates of growth per unit leaf P? Little is know on growth rates of P-efficient Proteaceae, but a careful growth analysis with Banksia grandis determined a rate of 26 mg g−1 day−1 at the lowest P supply, with rates decreasing at increasing P supply (Barrow 1977). For comparison, maximum relative growth rates of a range of Pinus species range from 15 to 60 mg g−1 day−1 and leaf P concentrations from 1.4 to 4.6 mg P g−1 DW (Matzek and Vitousek 2009). Given the very low leaf P concentrations of about 0.2 mg P g−1 DW (Denton et al. 2007a), this comparison suggests that the P efficiency of growth of Banksia species is an order of magnitude greater than that of Pinus species. Detailed growth analyses, combined with P-fractionation studies, are required to explore if, indeed, OCBIL plants are also efficient in their use of P for growth and what underlying mechanisms are to account for such high efficiency.

Mycorrhizal associations in landscapes

Data presented in Fig. 2 support the hypothesis that plants with various types of root clusters are more likely to be successful on severely P-impoverished soils, whereas mycorrhizas are more common on soils containing somewhat higher P concentrations (Lambers et al. 2008b, 2006). A simple explanation for this pattern is that arbuscular mycorrhizas act as “scavengers” for phosphate in solution that is not accessible for roots, because it is too far away from the roots or inside soil particles that are too small for roots or root hairs to enter (Smith and Read 2008). Conversely, root clusters effectively “mine” soil phosphate, mobilising scarcely available sources that are bound to iron, aluminium or calcium (Lambers et al. 2008a, b). In agreement with this, the diversity of arbuscular mycorrhizal plants was not correlated with soil fertility in ancient landscapes in Western Australia (data not presented). Ectomycorrhizas probably play an intermediate role, with some capacity to mine as well as a great capacity to scavenge phosphate (Lambers et al. 2008b). On the most severely P-impoverished soils, not only the plant species diversity but also the functional diversity in terms of P-acquisition strategies is highest, since non-mycorrhizal plants coexist with mycorrhizal species.

Figure 2 presents results of a comparative study of the relative diversity of plants with different root types as dependent on soil P levels. These data are from a very large survey of plant communities in the semiarid Mediterranean wheatbelt region of Western Australia (Gibson et al. 2004). Figure 2 summarises plant diversity and soil data from over 1000 dryland sites, by averaging these data at the floristic group level for 25 plant community types. Results of this analysis show statistically significant correlations between the diversity of plants with non-mycorrhizal roots, especially members of the Proteaceae with cluster roots, and soil P. There were also significant negative correlations between diversity of plants with ericoid mycorrhizas (Ericaceae) and ectomycorrhizas (some Myrtaceae, Rhamnaceae and some Papilionaceae) but trends for mycorrhizal plant diversity were not as strong as for non-mycorrhizal plants (Fig. 2c). This confirms the frequent observation that members of this family are more likely to be dominant on soils with low P availability (Allsop and Stock 1993; Lambers et al. 2008b, 2006; Pate and Bell 1999; Pate and Dawson 1999). However, a number of other correlations were also apparent, as soil texture and soil type are also correlated with soil P concentrations (data not presented). Total soil P data are used here because there was a very strong correlation between available and total P across the samples (R2 = 0.803) and total P could be measured with greater accuracy, due to detection limits when dealing with severely impoverished soils.

Perhaps as impressive as the story about cluster roots is the amazing diversity of carnivorous plants in ancient landscapes. In fact, Western Australia is the global centre of diversity for carnivorous plants with ¼ of all species in the world being found here (>150 out of ∼600). Parasitic plants have a peak in diversity in humid tropics, but their co-dominance in ecosystems is also another iconic feature of ancient landscapes. This is especially the case for trees in the genera Nuytsia (Loranthaceae) and Santalum (Santalaceae) in many habitats in Western Australia (Hopper 2010).

It is interesting to note that there is also a much higher frequency of ectomycorrhizal plants in Western Australian habitats than global averages might suggest. These include the dominant (Eucalyptus) trees and mallees (small multi-stemmed trees) in many habitats, as well as shrubs in the Myrtaceae, Fabaceae, Casuarinaceae and Rhamnaceae (Brundrett 2009). These woody plants tend to be dominant or co-dominant in soils that are not extremely infertile by Western Australian standards, so competition between plants may be more important than nutrient deficiency in these habitats.

Figure 1 is one of the first attempts to show the diversity of plants with different nutritional strategies for all vascular plant species on global and regional scales. Allsop and Stock (1993) provide similar data for South Africa, and Brundrett (2009) for Western Australia. These regional comparisons, which are possible because the majority of plant families have been allocated a consistent mycorrhizal status (Brundrett 2009), provide a powerful new tool for investigating the relative success of strategies without the sampling biases associated with small-scale studies. Figure 1 compares the functional diversity of different categories of non-mycorrhizal plants in three biodiversity hotspots (Guiana Shield in South America, Cape Region of South Africa, Western Australia) with global totals, and includes a temperate region for comparison (UK). The Guiana Shield area includes both ancient (Tepui) and younger habitats. The young landscape with relatively fertile soils (UK) has relatively few non-mycorrhizal plants, most of which are disturbance opportunists. By comparison, ancient landscapes in Australia and South Africa have exceptionally high diversity of non-mycorrhizal plants with specialised roots. The Guiana shield region data have fewer plants with specialised roots and its non-mycorrhizal plants are dominated by epiphytes, as would be expected in tropical forests.

P remobilisation

Some of the species discussed above that exhibit high PPUE values are also extremely efficient at remobilising P from senescing leaves. In particular some Banksia species are capable of remobilising over 80% of their leaf P (Denton et al. 2007a). Given their leaves have very low P concentrations when mature, their P proficiency, i.e. the minimum level to which P is depleted, is extremely high; in fact, higher than values reported before in the literature (Killingbeck 1996; Lambers et al. 2008a). Remobilisation from cluster roots in Hakea prostrata is equally efficient and proficient (Shane et al. 2004b). A high P-resorption proficient from senescing clusters is bound to be ecologically important, given the very high turnover rates of these structures (Playsted et al. 2006; Shane et al. 2006; 2004b).

The high remobilisation efficiency and proficiency of some species in OCBILs contrast markedly with global average values, showing that approximately half of the P content of leaves is resorbed during senescence (Aerts 1996; Killingbeck 1996).

P toxicity

Whilst many species in ancient landscapes are very effective at acquiring scarcely available P using root clusters (Shane and Lambers 2005), many of them, in different families, are also very sensitive to slightly elevated soil P levels (Handreck 1991a, b; Hawkins et al. 2008; Lambers et al. 2002; Parks et al. 2000; Shane et al. 2004a). This sensitivity is associated with extremely high leaf P concentrations, due to a very low capacity to down-regulate P uptake in their roots (Shane et al. 2008a, b, 2004a, c).

On a global scale, P toxicity is rare in plants, compared with the number of incidences reported for plants in phosphorus-impoverished habitats in Australia and South Africa. One explanation for the relatively high frequencies in OCBILs is that there is no selective force against a low capacity to down-regulate P uptake. However, could there also be distinct advantages associated with a low capacity to down-regulate P uptake? To answer this question it would help to know both the phylogeny of P sensitivity and the P status of the soils at the time P sensitivity evolved. We do have a good understanding of the phylogeny of genera in which P sensitivity is common, e.g., Banksia (Mast et al. 2005). Unfortunately, the information on the phylogeny of P sensitivity within this genus is still too scanty to allow robust conclusions linked to phylogeny. Further research to address this issue is in progress.

Are P sensitivity and proficient P remobilisation linked?

The question as to a distinct advantage associated with a low capacity to down-regulate P uptake would be easier to address if we understood the molecular mechanism of P toxicity. Here, recent information on the P sensitivity of several mutants of Arabidopsis thaliana might turn out to be valuable, provided the mechanism discovered for this model species is similar to what has evolved in P-sensitive species in OCBILs. Recent studies have demonstrated the novel functions of microRNAs (miRNAs) in regulating adaptive responses to nutrient stresses (Chiou 2007). Plant miRNAs usually down-regulate the abundance of their target mRNAs by post-transcriptional cleavage of the targeted mRNA. For example, miR399 is up-regulated during P deficiency which results in down-regulation of PHO2, a gene than encodes an E2 ubiquitin conjugase-related enzyme (UBC24). Ubiquitin is involved in targeted protein degradation (Doerner 2008). Plants over-expressing miR399 or defective in the gene involved in targeted protein degradation display P toxicity because of increased P uptake, enhanced root-to-shoot translocation, and retention of P in their old leaves (Chiou 2007). This suggests that the miR399-mediated regulation of PHO2 expression is critical in P homeostasis. The existence and conservation of miRNAs and their target genes involved in P uptake among many plant species points to the evolutionary importance of these miRNA-mediated nutrient-stress responses. It is possible that a similar modification in the signal-transduction pathway as described for Arabidopsis thaliana plays a role in the P sensitivity of species in severely P-impoverished landscapes.

Having speculated what the molecular mechanism of P sensitivity in OCBIL species might be, we return to the question if there might be a distinct advantage associated with a low capacity to down-regulate P uptake. We propose that the low capacity to down-regulate P uptake, i.e. constitutive expressions of specific genes involved in P transport, might be essential to ensure highly efficient and proficient P mobilisation from senescing tissues, as we discussed above. Efficient remobilisation is obviously a major advantage in P-impoverished habitats, whereas P sensitivity would not be a disadvantage. Conversely, P sensitivity would be a serious disadvantage in soils with higher P levels, and, if we are correct that constitutive expression of genes encoding specific P transporters is mechanistically linked to efficient P remobilisation from senescing tissues, this trait would preclude efficient remobilisation on soils with higher P levels. Clearly, more research is required to test our hypothesis.

P accumulation in seeds

Contrary to the very low P concentrations in leaves (Table 1), seed P concentrations are often very high in species naturally occurring in severely P-impoverished landscapes (Denton et al. 2007a; Grundon 1972; Kuo et al. 1982; Milberg and Lamont 1997; Pate and Dell 1984). Seed P can contribute up to 48% of the total above-ground P, e.g., in Banksia hookeriana, which grows on severely P-impoverished soils in south-western Australia (Witkowski and Lamont 1996). This provides further evidence for very strong selection pressures related to P availability.

South-western Australian species store on average 4.7 times more P per seed at twice the concentration when compared with species from the Cape Region in South Africa (Groom and Lamont 2010). This agrees with the contention that the P availability in south-western Australia, on average, is less than that in the Cape Region. However, all these concentrations are all much higher than global averages values (Marschner 1995).

Phosphorus reserves stored in seed must be important for successful recruitment in P-impoverished habitats. In the Mediterranean environments of south-western Australia and the Cape Region, recruitment typically occurs after a fire, with very few seeds germinating afterwards (Orians and Milewski 2007).

Ecosystem-level impacts of P-impoverishment on N content

The constraint on plant biomass accumulation in P-impoverished habitats as well as the generally low availability of P to biota can constrain the N accumulation in the ecosystem. P limitation can constrain the N content of plant biomass (see above) and hence in newly formed litter so that the limited amount of biomass and necromass that can accumulate has a low N content as well as a low P content. This effect is additional to any limitation of diazotrophy as a result of the restricted availability of P, and possibly of the trace metals used in relatively large amounts in nitrogenase compared to other means of N acquisition (Kustka et al. 2003a, b; Sañudo-Wilhelmy et al. 2001; Vitousek and Field 1999; Vitousek et al. 2010). Restrictions on N accumulation in the system relative to that of P also come about from loss processes that are outside the influence of plants in terms of natural selection although they may represent emergent properties of plant evolution. Such processes include leaching of dissolved organic nitrogen and of nitrate, release of gaseous N in denitrification and nitrification, and combustion of organic N to gaseous products in fire (Bowden and Bormann 1986; Bowden 1986; Vitousek and Field 1999; Vitousek et al. 2010).

Ecological consequences of functional plant diversity in OCBILS

Factors with the potential to limit effective restoration of vegetation in Western Australia are listed in Table 4. These factors are not universally applicable, as they vary between bioregions and locations and some also vary seasonally. There typically also is a high diversity in the structure of plant communities in Western Australian ecosystems (Anonymous 2006). In general, restoration of areas with highly diverse biotic and abiotic ecosystem components will be more challenging than restoration of areas of lower diversity. It needs to be acknowledged that, in general, ecosystem restoration is likely to be more difficult in OCBILs than in other parts of the world, because of factors listed in Table 4 (Hopper 2009; Standish and Hobbs 2010). Constraining factors should not be used to justify lesser standards of restoration, as they can be largely overcome with adequate experience, effective training, sufficient support from scientific research and sufficient commitment (Anonymous 2006).
Table 4

Major factors that constrain vegetation restoration successes in Western Australia (after Anonymous 2006)

A. Abiotic factors

1. Altered Landforms and soils

2. Saline groundwater

3. Groundwater-dependent vegetation

4. Soils or rocks that become acidic or toxic

5. Unreliable rainfall and climate especially in arid regions

B. Biodiversity factors

6. A high degree of plant species biodiversity and endemism in many bioregions

7. High variability in plant spatial diversity (ecosystem diversity); inadequate knowledge of plant and animal biodiversity, reproduction

8. Complex interactions and functions of biodiversity components in ecosystems

9. Threatened species and communities of species

10. Genetic diversity within taxa

11. Problems with seed germination or acquisition

C. Impacts on biodiversity

12. Environmental weeds that rapidly invade natural ecosystems after disturbance

13. Grazing by alien and indigenous animals

14. Diseases and pests, especially Phytophthora dieback

15. Changes to fire regimes

16. Plants and animals with highly specific habitats

17. Plants intolerant to changes in soil fertility

Managing natural systems in OCBILs is also challenging when it comes to dieback due to Phytophthora cinnamomi. At this stage, the only chemical that is available to stop the spread of this major threat to biodiversity in south-western Australia in phosphite (=phosphonate) (Fairbanks et al. 2000; Shearer et al. 2009; Shearer and Fairman 2007). Phosphite is an analog of phosphate and rapidly converted to phosphate by soil microorganisms (White and Metcalf 2007). It is sprayed from small aircraft on a regular basis at rates similar to what local farmers use (,com_docman/task,doc_details/Itemid,/gid,315/). In fragile, low-P OCBILs, such fertilisation with P offers a major threat, potentially suppressing the typical P-starvation responses (Lambers et al. 2006; Lee et al. 2005; Ticconi et al. 2001, 2004; Varadarajan et al. 2002). Clearly, there is an urgent need to develop alternative strategies to deal with Phytophthora cinnamomi in OCBILs.

Surface soils are critical biological repositories for recovery of native vegetation following disturbance in OCBILs. Rokich et al. (2000) demonstrated that the top 5 cm of soil in south-western Australian Banksia woodlands contain 90% of all seed and micro-organisms that sustain the above-ground plant communities. The next 5 cm contains a further 5%. Removal of this thin layer of topsoil compromises the ability of the community to persist and recover from other disturbances for very long periods of time (Hopper 2009).

Plant distribution in OCBILs appears to be intimately linked with edaphic factors, as is found for plants in other landscapes (Lambers et al. 2008a). However, whereas plant distribution patterns elsewhere are often determined by major edaphic factors (e.g., salinity, pH, heavy metals, water availability), in OCBILs these patterns can be far more subtle. For example, endemism in ironstone communities in south-western Australia is closely associated with rooting strategies (Poot and Lambers 2003, 2008). Plant species distribution in OCBILs is also linked with their sensitivity to soil P, as outlined above (Shane et al. 2008b; Shane and Lambers 2006). More subtle nutritional strategies, which remain to be elucidated, probably explain endemism in some species in the genus Banksia (Denton et al. 2007b). Whilst it is widely acknowledged that plants depend on soil and that soil factors determine plant distribution, there is also increasing evidence that plants substantially change the properties of soil (Lambers et al. 2009; Pate and Verboom 2009; Pate et al. 2001; Verboom and Pate 2003; Verboom and Pate 2006). The best examples are the tendency for texture-contrast soils (clay under sand) to occur under eucalypt woodlands and laterite under proteaceous heathland in Western Australia, but causal relationships between vegetation and soils are still being worked out (see Pate and Verboom 2009).

OCBILs are clearly very complex landscapes, and not uniformly ancient and nutrient impoverished, with functional plant diversity strongly depending on location in the landscape (Lambers et al. 2006; Mitchell et al. 2008a, b; Shane et al. 2008b). As illustrated in Fig. 2, soils in the SWAFR are not uniformly nutrient impoverished, but show a range in nutrient status, with associated functional diversity. Given time, soils develop due to the activities of plants and associated microorganisms (Lambers et al. 2009; Pate and Verboom 2009; Taylor et al. 2009). The available evidence suggests that variation in edaphic factors and associated functional plant diversity increases over time, so is typically far greater in OCBILs than in YODFLs. This functional diversity pertains equally to nutritional strategies and strategies involving plant water relations, as climates in the SWAFR became more arid as soils also become more infertile. Many of the characteristic features of OCBIL plants, especially sclerophylly, may be linked both to both water and nutrient conservation. Further research is required to unravel the evolutionary influences of soil infertility and aridity in OCBILs. However, we believe that soil infertility has been one of the strongest influences driving the evolutionary diversity of plants in ancient soils due to the increased diversity of plants with specialised nutrition relative to other categories of plants in the least fertile soils. On a global scale, the majority of non-mycorrhizal plants are found in marginal habitats such as extremely cold, wet, dry, saline or disturbed habitats (Brundrett 2009). We suggest that the extremely infertile soils of ancient landscapes represent another category of severely stressed plant habitat where benefits from mycorrhizal symbioses are less than elsewhere.


OCBILs are complex, nutrient-impoverished landscapes. Due to their age, species have been able to locally evolve in or migrate to edaphic niches in OCBILs. As such, age of the landscape accounts in part for the high species diversity in OCBILs (Linder 2008; Sauquet et al. 2009). Their low nutrient availability and associated low productivity offers an additional explanation for high species diversity in OCBILs. Because of the low productivity in OCBILs, it takes a long time for any individual with a superior strategy to outcompete its neighbours. By the time that is about to be achieved, the rainfall pattern may have shifted, disadvantaging the species that was about to win, a fire may reset the clock, or other forms of disturbance may have occurred (Orians and Milewski 2007).

The functional diversity of plant species in OCBILs needs to be taken into account when managing these landscapes. Management practices that enhance the availability of nutrients, e.g., high fire frequency (Fisher et al. 2006), or frequent applications of phosphite (Shearer and Fairman 2007), which is microbially converted into phosphate in soil (Adams and Conrad 1953), will decrease the competitive ability of species adapted to severely impoverished sites and allow invasion of weeds without specialised adaptations (Fisher et al. 2009). Competition for soil nutrient is likely to be more intense than elsewhere, especially in periods following severe disturbance.

Species in OCBILs offer a great potential to study desirable traits to be pursued in agricultural systems, taking into account that we are about to reach “peak phosphorus” (Cordell et al. 2009). Some of these traits may be desirable to be introduced in crop species (Pang et al. 2010; Ryan et al. 2009); others might be useful in systems taking advantage of the traits in crop rotations or intercrops (Adiku et al. 2008; Jemo et al. 2006; Kamh et al. 1999).

Future research to enhance knowledge of plant nutrition in ancient landscapes is important to underpin conservation strategies, as it is becoming increasingly evident that management strategies that work well to conserve biota in young, often disturbed, fertile landscapes (YODFELs) are inadequate to protect OCBILs (Hopper 2009; Standish and Hobbs 2010). Of particular interest would be unravelling links between nutrient strategies and competitive outcomes relative to soil fertility, especially after disturbance.


This work was funded by the Australian Research Council (ARC). We are very grateful to Neil Gibson and Greg Keighery for access to floristic survey data, and to Ian Wright and colleagues for permitting access to their global dataset. Mark Brundrett is supported by Lotterywest funding for the Wheatbelt Orchid Rescue Project. We thank Etienne Laliberté for his thoughtful comments on this manuscript. The University of Dundee is a registered Scottish charity No 015096.

Copyright information

© Springer Science+Business Media B.V. 2011