A multi-trait test of the leaf-height-seed plant strategy scheme with 133 species from a pine forest flora


  • Daniel C. Laughlin,

    Corresponding author
    1. School of Forestry, Northern Arizona University, P.O. Box 15018, Flagstaff, Arizona 86011, USA
    2. Ecological Restoration Institute, Northern Arizona University, P.O. Box 15017, Flagstaff, Arizona 86011, USA
      *Correspondence author. E-mail: Daniel.Laughlin@nau.edu
    Search for more papers by this author
  • Jessica J. Leppert,

    1. Ecological Restoration Institute, Northern Arizona University, P.O. Box 15017, Flagstaff, Arizona 86011, USA
    Search for more papers by this author
  • Margaret M. Moore,

    1. School of Forestry, Northern Arizona University, P.O. Box 15018, Flagstaff, Arizona 86011, USA
    Search for more papers by this author
  • Carolyn Hull Sieg

    1. United States Department of Agriculture, Forest Service, Rocky Mountain Research Station, 2500 South Pine Knoll, Flagstaff, Arizona 86001, USA
    Search for more papers by this author

*Correspondence author. E-mail: Daniel.Laughlin@nau.edu


1. Westoby’s [Plant and Soil (1998), 199, 213] Leaf-Height-Seed (LHS) plant strategy scheme quantifies the strategy of a plant based on its location in a three-dimensional space defined by three functional traits: specific leaf area (SLA), height, and seed mass. This scheme is based on aboveground traits and may neglect strategies of belowground resource capture if root functioning is not mirrored in any of the axes. How then do fine roots fit into the LHS scheme?

2. We measured 10 functional traits on 133 plant species in a ponderosa pine forest in northern Arizona, USA. This data set was used to evaluate how well the LHS scheme accounts for the variation in above and belowground traits.

3. The three most important plant strategies were composed of multiple correlated traits, but SLA, seed mass, and height loaded on separate principle components. The first axis reflected the widely observed ‘leaf economics spectrum’. Species at the high end of this spectrum had high SLA, high leaf and fine root nitrogen (N) concentration, and low leaf dry matter content. The second axis reflected variation in seed mass and fine root morphology. Plants at the positive end of this spectrum were plants with large seeds and low specific root length (SRL). The third axis reflected variation in height and phenology. Plants at the positive end of this spectrum were tall species that flower late in the growing season.

4. Leaf N concentration was positively correlated with fine root N concentration. SRL was weakly positively correlated with SLA. SRL was not correlated with fine root N concentration. Leaf litter decomposition rate was positively correlated with the leaf economics spectrum and was negatively correlated with the height and phenology spectrum.

5. Leaf traits, seed mass, and height appear to be integrating properties of species that reflect much of the variation in plant function, including root function. Fine root N concentration was positively mirrored by the leaf economics spectrum, and SRL was inversely mirrored by seed mass. The leaf and height axes play a role in controlling leaf litter decomposability, indicating that these strategy axes have important consequences for ecosystem functioning.


Plants are multifaceted organisms that have evolved ecological strategies for sustaining populations in resource-limited environments (Grime 1979; Craine 2009). Plant strategies can be quantified by measuring functional traits (Grime et al. 1997; Reich et al. 2003), which are the properties of plants that impact plant fitness (Violle et al. 2008) and ecosystem processes (Lavorel & Garnier 2002). Comparisons of functional traits across taxa have provided insight into the primary functional gradients among plants (e.g. Grime et al. 1997; Reich et al. 1999; Craine et al. 2001; Díaz et al. 2004). One important gradient describes differences in resource acquisition (Reich, Walters & Ellsworth 1997), known as the ‘leaf economics spectrum’ (sensuWright et al. 2004a), which runs from plants with quick returns on investment in nutrients and dry matter [i.e. plants with leaves that have high photosynthetic rates, short life spans, high SLA, and high leaf nitrogen (N) concentrations] to plants with slower returns on their investments. This multi-trait spectrum (or strategy axis) is only one out of potentially many spectra important to plant growth, reproduction, and survival (Reich et al. 2003; Craine 2009).

Westoby (1998) proposed a simple ‘Leaf-Height-Seed’ (LHS) scheme that operationally quantifies the strategy of a plant species by its location in a three-dimensional space defined by three functional traits: specific leaf area, height, and seed mass. Specific leaf area (SLA, leaf area per unit dry mass) represents variation along the leaf economics spectrum and is indicative of a species’ ability to respond to opportunities for rapid growth (Reich et al. 1999). Plant height at maturity has been related to competitive ability and fecundity (Keddy & Shipley 1989; Aarssen & Jordan 2001). Seed mass reflects variation in dispersal capability and cotyledon-stage seedling survivorship (Westoby, Leishman & Lord 1996; Jakobsson & Eriksson 2000). This LHS plant ecology strategy scheme is potentially useful since it requires the measurement of only three easy-to-measure traits. However, plant strategies are thought to be gradients in multiple correlated traits rather than gradients in single traits (Reich et al. 2003). Moreover, any plant strategy scheme based entirely on aboveground traits may neglect strategies of belowground resource capture if root functioning is not mirrored in any of the three axes. How then do fine roots fit into the LHS scheme?

Root traits are harder to measure and have received far less attention than aboveground traits, despite the fact that most of the biomass and production in perennial-dominated ecosystems is belowground, and that many important ecosystem processes are tightly coupled with plant roots and rhizospheres (Aerts & Chapin 2000). There is some evidence for a ‘root economics spectrum’ analogous to tradeoffs seen in leaves (Eissenstat & Yanai 1997). For example, fine root N concentration scales positively with leaf N concentration (Craine & Lee 2003; Tjoelker et al. 2005; Kerkhoff et al. 2006). A recent analysis by Kembel et al. (2008) indicates that there are at least two gradients of root function. The strongest gradient suggests that species with fast relative growth rates have high leaf and root N concentrations, shorter-lived roots, and high SLA, indicating that roots and leaves are functionally coordinated. The weaker orthogonal axis described variation in specific root length. Specific root length (SRL, root length per unit root dry mass) of fine absorptive roots has been suggested to be the belowground analogue to SLA (Cornelissen et al. 2003). SRL is indicative of the potential rate of water and nutrient uptake and is considered to be a morphological index of belowground competitive ability (Lambers, Chapin & Pons 1998). SLA and fine root SRL were uncorrelated among a set of grassland species despite the positive association between leaf and fine root tissue chemistry (Tjoelker et al. 2005). Some studies have illustrated positive relationships between relative growth rate (which scales positively with SLA; Lambers & Poorter 1992) and SRL (Reich et al. 1998; Wright & Westoby 1999), whereas others have reported opposite trends (Boot 1989; Lambers & Poorter 1992) or even no relationship (Poorter & Remkes 1990; Huante, Rincon & Gavito 1992). If an independent root economics spectrum exists, then the LHS scheme may need additional dimensions (Westoby et al. 2002; Westoby & Wright 2006), but if leaf and root traits are functionally coordinated, then the LHS scheme will be supported because variation in root traits will be mirrored by aboveground traits.

Functional traits not only define plant strategies for survival, they are thought to influence important ecosystem processes (Chapin et al. 2000). Decomposition of leaf litter is one critical step in the internal recycling of limiting nutrients. Decomposition rates are partly controlled by tissue nutrient concentration and the density of structural material in the leaf (Cornwell et al. 2008), suggesting that the leaf axis in the LHS scheme controls leaf litter decomposition rates.

We quantified 10 functional traits on 133 plant species that commonly occur in southwestern USA Pinus ponderosa var. scopulorum P. & C. Lawson (ponderosa pine) forests. In addition to SLA, height, seed mass, SRL, leaf N and fine root N, we measured four additional traits that can influence plant fitness: leaf phosphorus concentration, which has been shown to be an important component of the leaf economics spectrum (Wright et al. 2004a); flowering date and duration, which summarize phonological aspects of a species’ life history (Grime et al. 1997); and leaf dry matter content (LDMC, ratio of leaf dry mass to fresh mass), which is indicative of the amount of structural material in a leaf (Garnier et al. 2001; Kazakou et al. 2006). We asked the following questions: (1) Is the LHS scheme supported when multiple traits, including root traits, are assessed simultaneously in a multivariate framework? (2) Are root traits correlated with leaf traits? (3) Do the LHS axes explain variation in leaf litter decomposition rates?

Materials and methods

Study system

This is the first study to document functional traits for the common plant species found in the widespread semi-arid ponderosa pine forest ecosystem in the southwestern USA. Water and nitrogen are the primary limiting resources in this ecosystem. Frequent low-intensity surface fire was historically the most important disturbance agent (Moore, Covington & Fulé 1999). These fires maintained a savanna-like system where clumps of large trees persisted amidst a matrix of a grassy understory, but fire suppression over the last century has caused tree densities to increase dramatically. In dense forest stands, the ponderosa pine overstory suppresses understory production and diversity (Laughlin et al. 2008). In open stands, the understory is dominated by both C3 and C4 bunchgrasses and forbs, which comprise the majority of the species pool.

This study was conducted on a ∼12 000 ha landscape in the Coconino National Forest near Flagstaff, Arizona, USA. The mean annual precipitation of Flagstaff is 56 cm and the mean annual temperature is 7·7 °C (Kohn & Welker 2005). Southwestern USA ponderosa pine forests occur across a broad edaphic gradient and therefore have a large species pool (Laughlin & Abella 2007). We chose to study 133 species that were detected either historically or currently on a set of long-term permanent plots. These species span 33 taxonomic families and 95 genera, and includes 88 C3 forbs, 13C3 grasses, nine C3 legumes, three C3 shrubs, two C3 trees, 14 C4 grasses, and four CAM forbs. All nomenclature follows the USDA NRCS Plants Database (http://plants.usda.gov/) accessed in 2009.

Functional traits

Plants were sampled from the same study sites from which the species list was generated. In cases where species occurred at more than one site, traits were measured at multiple sites. Following the recommendations of Cornelissen et al. (2003), we measured traits on robust, ungrazed plants grown in well-lit environments. When measuring traits on species that are adapted to shady conditions (e.g. Carex geophila, Thalictrum fendleri), we tried to sample individuals in canopy openings. Sampling began in May 2008 and ended in September 2008. The Flagstaff region received c. 47 cm of precipitation during the water year of 2008 (http://www.wrcc.dri.edu) and the Palmer Drought Severity Index for the same time period was c. 0·7 (http://www.ncdc.noaa.gov), each indicating that plants were sampled during a climatically average year.

We measured a core set of functional traits that reflect aspects of each species’ ability to disperse, establish, acquire water and nutrients, and photosynthesize (Weiher et al. 1999; Cornelissen et al. 2003). For all 133 species we measured SLA, height, seed mass, SRL, leaf N concentration, fine root N concentration, leaf phosphorus concentration, and LDMC. Julian flowering date and flowering duration were also obtained for each species. We measured leaf litter decomposition rates on 103 species. See Table S1 in electronic Supporting Information for mean trait values for every trait on all species in the study.

SLA is the ratio of leaf area to dry weight expressed as mm2 mg1. SLA was measured on ten to twenty individuals when the species was flowering. We non-randomly selected one fully expanded, healthy leaf from each individual for this measurement. One-sided leaf area was measured using the Agvis Imaging System (Decagon Devices, Pullman, WA, USA) within 5 h of harvesting. The leaf petiole and rachis on compound leaves were retained for this measurement. Leaves were oven-dried for 72 h at 55 °C prior to obtaining dry weights.

Canopy height is the height of the foliage (not the height of the inflorescence) of a species measured in cm. Height was measured on robust flowering individuals and therefore reflects an average maximum height for each species.

Seed mass is the oven-dry mass of an average seed expressed in mg. When possible, seeds were harvested from several individuals of each species. Seeds housed inside fleshy fruits (e.g., Rosa woodsii) were removed from the fruiting structures, and if a pappus was present it was removed (Weiher et al. 1999). Mean seed mass was determined by weighing the total mass of between 20 and 100 individual seeds (depending on the species), then dividing the total dry weight by the number of seeds in the sample.

Leaf dry matter content (LDMC) is the ratio of leaf dry mass to fresh mass expressed as mg mg1. LDMC was not measured on the same leaves that were used to measure SLA. LDMC was measured on leaves harvested from two individuals of each species, and variance of this trait within each species was very low (see also Weiher et al. 1999). Following Garnier et al. (2001), the shoots were harvested and placed in humid plastic bags in the field. Within 5 h, we re-cut each stem under water, and keeping the cut ends submerged, stored them in a refrigerator for 24 h to fully rehydrate the leaves. This procedure is necessary because LDMC is sensitive to hourly fluctuations in leaf water status (Garnier et al. 2001). After bringing the leaves back to room temperature, we weighed the rehydrated leaves to obtain ‘fresh weights’. Leaves were then oven-dried for 72 h at 55 °C prior to obtaining dry weights.

Foliar concentrations of nitrogen (Nmass) and phosphorus (Pmass) were determined on at least three (often five) individual replicates per species. Fully expanded, healthy leaves were harvested from the entire axis of a plant when it was flowering. Samples were oven-dried for 72 h at 55 °C, then ground to <0·5 mm using a Wiley Mill (Thompson Scientific, Swedesboro, NJ, USA). Foliar Nmass was analyzed on a Flash EA 1112 Elemental Analyzer (CE Elantech Inc., Lakewood, NJ, USA) at the United States Department of Agriculture (USDA) Forest Service Rocky Mountain Research Station (RMRS) Analytical Lab in Flagstaff, Arizona, USA. To determine Pmass, we digested ground leaf matter with sulfuric acid, potassium sulphate and copper sulphate in a block digestor to convert the phosphorus in the samples to orthophosphate. These samples were analyzed on a Lachat Quikchem 8000 (Lachat Instruments, Inc., Milwaukee, WI, USA) at the USDA Forest Service RMRS in Flagstaff, Arizona, USA.

Specific root length is the ratio of fine root length to dry mass expressed as m g1. Specific root length was measured on three individuals of each species during the latter part of the growing season. We excavated entire root systems from the soil with shovels and trowels, and gently washed soil from the roots in the laboratory with clean water rinses. Following the standardized protocol of Cornelissen et al. (2003), we used the fine (<2 mm) absorptive roots (i.e. unsuberized, often with evidence of root tips or hairs) in our determination of SRL. Only a subsample of fine roots from each individual was used in the measurement. This method yields a measure that better mirrors its aboveground analogue SLA (Cornelissen et al. 2003). The vast majority of the fine roots that we harvested were <0·5 mm. Root length was measured using the software winrhizo V. 2003a (Regent Instruments, Nepean, Ontario, Canada). Roots were then oven-dried for 72 h at 55 °C prior to obtaining dry weights.

Fine root Nmass was measured on the same roots used in the determination of SRL. After drying for 72 h at 55 °C, fine roots were ground to <0·5 mm in a Wiley Mill and were analyzed on a Flash EA 1112 Elemental Analyzer at the USDA Forest Service RMRS in Flagstaff, Arizona, USA.

Mean Julian flowering dates and flowering duration for each species were determined using regional floras that describe the first and last months that a species is in flower. For example, Artemisia carruthii flowers from August through October, which corresponds to the Julian days 213 through 304. Therefore, the mean Julian flowering day = (213 + 304)/2 = 259, and the flowering duration = 304−213 = 91 days. Data primarily came from McDougall (1973), but for some species we used data from the Intermountain Flora (Cronquist et al. 1986+) or the Flora of North America (Flora of North America Editorial Committee 1993+).

Species-specific leaf litter decomposition rates were quantified using litterbags (Harmon, Nadelhoffer & Blair 1999) inside an experimental exclosure at the G. A. Pearson Natural Area (located 10 km northwest of Flagstaff, Arizona, USA) from October 2007 to September 2008. Green (i.e. non-senesced) leaves were collected from living plants from 103 species. Litterbags had an inside area of 20 × 20 cm constructed with window screen material (1 × 2 mm mesh). Approximately 2 g of leaf dry mass were placed in these bags. Decomposition rates were determined for 103 of the 133 species in the study. The most dominant species (= 53) had two independent replicates, and the remainder of the species (= 50) were not replicated. Litterbags were placed on the surface of the forest floor in an open, well-lit environment. After the harvest, we removed any soil particles from the decomposed litter with brushes prior to drying at 55 °C for 72 h to obtain oven-dry weights. The decomposition rates reported here reflect the proportional mass loss over a 1-year period.

Data analyses

We evaluated Westoby’s (1998) LHS scheme by subjecting the large species-trait matrix that included root traits to a principal components analysis (PCA). If the LHS model is robust, then SLA, height, and seed mass will each load highly on the first three components, respectively, and root traits will either load on the first three axes, or they will contribute to additional dimensions. Following Westoby (1998), all variables were log10-transformed because trait values can vary by orders of magnitude, and because traits are often lognormally distributed between species. We calculated the eigenvalues and eigenvectors of the correlation matrix (see Table S2 in electronic Supporting Information) using sas-jmp version 8.0. Decomposition rate was not included as a variable in the matrix because rates were determined for only 103 species. We used anova and Tukey’s HSD (Honestly Significant Difference) post-hoc tests to determine how plant functional types (C3 forbs, C3 legumes, C3 woody plants, C3 graminoids, C4 graminoids, and CAM forbs) differed in each of the three major axis scores.

We evaluated the relationship between leaf and fine root traits. The results of the PCA informed our evaluation, but we also used regression analysis to determine the strength and sign of the relationship between: SLA and SRL; leaf Nmass and fine root Nmass; and fine root Nmass and the leaf economics spectrum (quantified by the axis scores of the first principal component, see Results). We also evaluated the relationship between SRL and fine root Nmass.

We used simple linear and backward multiple regression analyses to evaluate the relationships between leaf litter decomposition rate and four leaf traits (SLA, LDMC, leaf Nmass, leaf Pmass) to determine which trait or combination of traits were the best predictors of decomposition rate. In addition, we used linear and multiple regression to determine whether any of the first three principal components (i.e. plant strategy axes) were correlated with leaf litter decomposition rate.


The Leaf-Height-Seed strategy scheme

The 133 plant species exhibited broad ranges in each of the 10 functional traits (see Table S3 in electronic Supporting Information). Three principal components accounted for 58% of the total variance of the species-trait correlation matrix (Table 1).

Table 1.   Results of the principal components analysis of the species-trait correlation matrix. All variables were log10-transformed. Eigenvectors >|0·30| are highlighted in bold. Percents reflect the percent of total variance (i.e. the sum of the diagonal elements in the correlation matrix) accounted for by each principal component
Cumulative percent25·445·258·4
 Leaf Nmass0·420·400·11
 Fine root Nmass0·350·270·13
 Leaf Pmass0·250·17−0·05
 Seed mass−0·160·550·13
 Flowering date0·18−0·290·54
 Flowering duration0·110·170·47

The first principal component (PC1) accounted for 25% of the total variance and represented an axis of resource acquisition and turnover consistent with the leaf economics spectrum. Species found at the positive end of this spectrum exhibited high SLA, leaf Nmass, and fine root Nmass, and low LDMC (Fig. 1, Table 1). SLA exhibited the highest loading on PC1.

Figure 1.

 The Leaf-Height-Seed plant ecology strategy scheme is supported by a principal components analysis of 10 functional traits of 133 plant species that occur in southwestern USA ponderosa pine forests. Only the first two principal component scores are plotted here explicitly. Labels and arrows show the variables that exhibited eigenvector scores >|0·3| (see Table 1). Species symbols are coded by plant functional types. Species codes: ARPU, Aristida purpurea; BLTR, Blepharoneuron tricholepis; CAGE, Carex geophila; CHSE, Chamaesyce serpyllifolia; CHGR, Chenopodium graveolens; DRMO, Drymaria molluginea; DRLE, Drymaria leptophylla; HECO, Hesperostipa comata; IRMI, Iris missouriensis; LUAR, Lupinus argenteus; MUMO, Muhlenbergia montana; MURA, Muhlenbergia ramulosa; OXSP, Oxalis sp.; OXLA, Oxytropis lambertii; PIPO, Pinus ponderosa; POOR, Portulaca oleracea; QUGA, Quercus gambelii.

The second principal component (PC2) accounted for an additional 20% of the total variance and represented variation in seed mass, root morphology, and leaf Nmass. Species found at the positive end of PC2 had large seeds, low SRL and high leaf Nmass (Fig. 1, Table 1). Seed mass exhibited the highest loading on PC2 (Table 1).

The third principal component (PC3) accounted for an additional 13% of the total variance and represented variation in height, flowering date and flowering duration. Species found at the positive end of PC3 were tall species that flowered late in the season for only a short duration (Fig. 1, Table 1). Height exhibited the highest loading on PC3 (Table 1).

Plant functional types accounted for some of the variation in axis scores (Fig. 2). C3 forbs, C3 legumes, and CAM forbs were located at the high end of the leaf economics spectrum (PC1), whereas C3 woody plants and both C3 and C4 graminoids were located at the low end (Fig. 2a). Plant functional types accounted for 31% of the variation of PC1. C3 legumes and C3 woody plants were located at the high end of the seed mass spectrum (Fig. 2b). C3 graminoids had larger seeds and lower SRL than C4 graminoids (Fig. 2b). Plant functional types accounted for 40% of the variation in PC2. Other than C3 woody plants being taller than most other species in the flora, plant functional types accounted for only 11% of the variance in PC3 (Fig. 2c).

Figure 2.

 Box and whisker plots illustrate plant functional type variation along the (a) leaf economics spectrum (PC1), (b) seed mass and SRL spectrum (PC2), and (c) height and phenology spectrum (PC3). The three shrubs and two trees were combined into a ‘C3 woody plants’ group for simplicity. Each of the three anova tests were significant (< 0·05), and significantly different pairwise contrasts (Tukey’s HSD) are indicated by different lowercase letters.

Leaf and fine root traits

Leaf Nmass was positively correlated with fine root Nmass (Fig. 3a). Similarly, the leaf economics spectrum, as represented by the first principal component, was positively correlated with fine root Nmass (Fig. 3b). Fine root Nmass loaded most strongly on PC1 (Table 1).

Figure 3.

 Simple linear relationships between fine root Nmass (%) and (a) leaf Nmass (%), (b) the leaf economics spectrum (represented by the first principal component), and (c) specific root length (g m−2), and between (d) specific leaf area (mm2 mg−1) and specific root length.

Specific root length was not correlated with fine root Nmass (Fig. 3c, = 0·65), and SRL was only weakly positively correlated with SLA (Fig. 3d). Specific root length loaded most strongly on PC2 (Table 1).

Plant strategies and leaf litter decomposition

Using data measured on 103 species, leaf litter decomposition rate was positively correlated with SLA (R2 = 0·07, = 0·0078), leaf Nmass (Fig. 4a, R2 = 0·18, < 0·0001), and leaf Pmass (R2 = 0·04, = 0·0339), and was negatively correlated with LDMC (Fig. 4b, R2 = 0·22, < 0·0001). A multiple regression model with LDMC (partial r= 0·10, < 0·0001) and leaf Nmass (partial r= 0·07, = 0·0026) as predictors explained 27% of the variance in decomposition rate (proportion mass loss = 0·34 + logLeafN × 0·29 – logLDMC × 0·31).

Figure 4.

 Simple linear relationships between litter decomposition rate (measured as the proportion of litter mass loss after decomposing in the forest for 1 year) and (a) leaf Nmass (%), (b) leaf dry matter content (mg mg−1), (c) the leaf economics spectrum (represented by the first principal component), and (d) the height and phenology spectrum (represented by the third principal component). The outlier in panels b and d represents ponderosa pine, and both relationships were still significant if this species was removed from the regression analysis.

The leaf economics spectrum (i.e. PC1) was significantly positively correlated with decomposition rate (Fig. 4c). The seed mass and root morphology spectrum (i.e. PC2) was not significantly correlated with decomposition rate (R= 0·04, = 0·053). The height and phenology spectrum (i.e. PC3) was weakly negatively correlated with litter mass loss (Fig. 4d). A multiple regression model with PC1 (partial r= 0·18, < 0·0001) and PC3 (partial r= 0·08, = 0·0020) as predictors explained 26% of the variance in leaf litter decomposition rate (proportion mass loss = 0·64 + PC1 × 0·04 – PC3 × 0·04).


Westoby’s (1998) Leaf-Height-Seed plant ecology strategy scheme was well supported in this multi-trait analysis of 133 species. The three principle strategies were composed of multiple correlated traits, but SLA, seed mass, and height loaded on separate axes. The LHS scheme also accounts for belowground plant function since root traits were correlated with aboveground traits.

The leaf economics spectrum is one important axis of specialization in this flora, and it likely plays a role in controlling productivity, litter decomposition, and N cycling. The leaf economics spectrum represents a fundamental tradeoff between species with traits that confer rapid acquisition of resources and rapid turnover of biomass vs. species that efficiently conserve their slowly acquired resources. This axis reflects a gradient in litter decomposition rate because species with high SLA, low LDMC and high leaf Nmass (e.g. Chenopodium graveolens, Oxalis sp., Chamaesyce serpyllifolia) decompose more rapidly than species at the opposite end of the spectrum (e.g., trees, such as Pinus ponderosa and Quercus gambelii).

Leaf Nmass is well known as a predictor of litter decomposition rate (Cornwell et al. 2008), but the ability to predict decomposition rates from LDMC is relatively new. Quested et al. (2007) found that community-aggregated LDMC was the best predictor of community litter decomposition rates. Our results show that LDMC was negatively correlated with decomposition rate, and that LDMC is an important component of the leaf economics spectrum (Wilson, Thompson & Hodgson 1999). LDMC is likely a good predictor of decomposition rate because it has been shown to be positively correlated with lignin concentration (Kazakou et al. 2006), which is slow to break down. Predicting decomposition rate from LDMC alone may be especially useful since LDMC is simpler to measure and less expensive than measuring leaf nutrient concentrations. These results suggest that plant strategies, such as the leaf economics spectrum, can have significant consequences for ecosystem processes, such as carbon and nutrient cycling (Cornwell et al. 2008).

There was a clear positive link between leaf Nmass and fine root Nmass. The positive relationship is one of the few consistent linkages between leaf and root traits (Craine & Lee 2003). Because leaf Nmass is known to influence decomposition rates (Cornwell et al. 2008), species with rapidly decomposing leaf litter will likely also have rapidly decomposing fine roots. Species with high fine root Nmass also have short root lifespans (Kembel et al. 2008), and species with high root tissue density also have high LDMC (Wahl & Ryser 2000). These results imply that there is functional coordination between roots and shoots (Grime 1979), but SRL does not reflect this coordination.

The second axis of specialization in this flora reflects a gradient in seed mass and SRL. Variation in seed mass reflects the fundamental tradeoff between seed output and seed size (Henery & Westoby 2001) and between seed size and persistence in the seed bank (Thompson, Band & Hodgson 1993). Annual plants, such as the two Drymaria spp., are examples of small-seeded species that produce many seeds that emerge from the soil seed bank each year during the wet season. This axis also represents a gradient from species with high SRL and low seed mass to low SRL and high seed mass. Reich et al. (1998) found that seed mass and SRL were also inversely correlated among nine boreal tree species, but the relationship was less pronounced in other studies (Gross, Maruca & Pregitzer 1992; Huante, Rincon & Gavito 1992). Perhaps small-seeded species require high SRL in order to rapidly obtain water and mineral nutrients in the absence of endosperm reserves. Seedlings may have different SRL than adult plants, so this interpretation is somewhat confounded by ontogeny. Several studies have suggested a negative relationship between seed mass and relative growth rate (Hunt & Cornelissen 1997; Reich et al. 1998; Wright & Westoby 1999), but the pattern is not universal (Shipley & Peters 1990). Species with high SRL exhibit a root morphology that is conducive to more rapid acquisition of water and mineral nutrients (Lambers, Chapin & Pons 1998), which should allow the species to grow faster in productive environments (Grime et al. 1997). If this interpretation is correct, then SRL should have been positively correlated with the leaf economics spectrum. Other studies have shown strong correlations between SRL and SLA (Reich et al. 1998; Wright & Westoby 1999) for seedlings grown in pots. However, SRL and tissue Nmass were uncorrelated in this study and in another comparative analysis (Tjoelker et al. 2005).

What can explain the lack of covariance between SRL and the leaf economics spectrum? SRL is a composite metric of both fine root diameter and tissue density. Species with identical SRL may have different diameters and tissue density, indicating different functional responses to gradients in resources. Unfortunately, data on tissue density is not available for comparison in this study. Furthermore, SRL is but one simple aspect of belowground resource capture and does not take into account total root mass, root-to-shoot ratio, root turnover, distribution of fine roots throughout the soil profile, root architecture, or foraging ability into resource-rich soil patches, all of which are important determinants of belowground resource capture (Kembel et al. 2008; Hodge 2009). The lack of covariance between SRL and leaf traits does not necessarily mean that root and leaf function are uncoordinated. Given the methodological problems with measuring SRL (Ryser 2006), other metrics may be better indicators of root function. Root tissue density and nutrient concentration, for example, scale positively with leaf traits (Wahl & Ryser 2000; Craine et al. 2001).

The third axis represented a gradient from tall plants that flower late in the growing season to short plants that flower early in the growing season. Pinus ponderosa is the tallest species in the forest and suppresses understory diversity and productivity (Laughlin et al. 2008). Tall herbs, such as Bahia dissecta, require the full growing season to attain their maximum height, whereas species like Noccaea montana and Antennaria rosulata flower early in the season and are short in stature. Height was not completely independent of seed mass because tall species generally had large seeds (Díaz et al. 2004). The height axis was negatively correlated with leaf litter decomposition rate perhaps because tall species require more structural material and may have higher leaf lignin content (not measured in this study) than short species.

Plant functional types are important categories used by ecologists to determine how plants respond to perturbations and environmental gradients. Plant functional types accounted for much less than half of the variance in each of the major axes of specialization, and although the mean values within types can differ from each other, there is large overlap in trait ranges. Given this considerable overlap, we concur with Wright et al. (2004b), and urge caution when assigning mean trait values to plant functional type categories in vegetation models. Given recent advances in our ability to model continuous trait variation among species (e.g., Garnier et al. 2004), future research should evaluate how the functional composition of communities, especially with respect to the leaf-height-seed spectrums, influences ecosystem function.

Plant strategy axes are gradients of multiple correlated traits (Craine 2009). However, leaf traits, seed mass, and height at maturity appear to be fundamental and integrating properties of species that reflect much of the total variation in plant function, including root function (Westoby 1998; Westoby & Wright 2006). Though root morphology could not adequately be predicted from leaf morphology, root tissue chemistry aligned with the leaf economics spectrum (Wahl & Ryser 2000; Kembel et al. 2008). The leaf and height axes can be used to predict leaf litter decomposition rate, indicating that plant strategies have important consequences for ecosystem functioning.


We thank P. Fulé, M. Kearsley, S. Hart, I. Wright, L. Poorter and the anonymous reviewers for their helpful comments on the manuscript. This research was supported by a Joint Venture Agreement (#08-JV-11221633-233) with the Rocky Mountain Research Station (RMRS), McIntire-Stennis appropriations to the Northern Arizona University (NAU) School of Forestry, and the Ecological Restoration Institute (ERI). J. Leppert was supported by a NAU Hooper Undergraduate Research Award.