Resource-use strategies of native and invasive plants in Eastern North American forests



  • Studies in disturbed, resource-rich environments often show that invasive plants are more productive than co-occurring natives, but with similar physiological tradeoffs. However, in resource-limited habitats, it is unclear whether native and invasive plants have similar metabolic constraints or if invasive plants are more productive per unit resource cost – that is, use resources more efficiently.
  • Using a common garden to control for environment, we compared leaf physiological traits relating to resource investments, carbon returns, and resource-use efficiencies in 14 native and 18 nonnative invasive species of common genera found in Eastern North American (ENA) deciduous forest understories, where growth is constrained by light and nutrient limitation.
  • Despite greater leaf construction and nitrogen costs, invaders exhibited greater instantaneous photosynthetic energy-use efficiency (PEUE) and marginally greater photosynthetic nitrogen-use efficiency (PNUE). When integrated over leaf lifespan (LL), these differences were magnified. Differences in efficiency were driven by greater productivity per unit leaf investment, as invaders exhibited both greater photosynthetic abilities and longer LL.
  • Our results indicate that woody understory invaders in ENA forests are not constrained to the same degree by leaf-based metabolic tradeoffs as the native understory flora. These strategy differences could be attributable to pre-adaptation in the native range, although other explanations are possible.


Nonnative plant invasions are common in environments of anthropogenic disturbance (Hobbs & Huenneke, 1992), which has led to the generalization that nonnative invaders (hereafter ‘invaders’) are most likely to outperform native species in disturbed habitats with high resource availability (e.g. Daehler, 2003). Mechanisms attributed to these disturbance-mediated invasions include broad physiological advantages of invaders over natives following episodic increases in resource availability (Davis et al., 2000). If these resource-based mechanisms are true, then invaders should exhibit advantages in functional traits that contribute to high productivity given ample resources, such as high specific leaf area, photosynthetic ability, and relative growth rate compared with native competitors (van Kleunen et al., 2010; Drenovsky et al., 2012). However, it remains unclear if invasion success in resource-limited ecosystems can be explained by mechanisms described for high-resource environments.

Over the past 15 yr, there has been substantial development of plant strategy theory and resource-use economics (Reich et al., 1997; Westoby et al., 2002; Wright et al., 2004, 2005). Wright et al. (2004) reported a global pattern of coordinated variation in leaf traits (‘worldwide leaf economic spectrum’ (LES)) that invokes general ecophysiological tradeoffs in resource economics as a global axis of variation in plant strategies. This spectrum of strategic variation describes species from those with slow returns on investments (possessing traits such as low specific leaf area, high construction costs, low photosynthetic rates, and high leaf lifespan) to those at the opposite extreme of quick returns on resource investments. Strategies that lie outside of this general LES are presumed to either be selected against (ecologically constrained by biotic interactions) or biophysically or genetically impossible (Reich et al., 1999; Donovan et al., 2011).

In an effort to understand invasion processes in light of these developments, studies have explicitly placed invasive plants along a spectrum of leaf trait variation that emphasizes coordinated variation among leaf traits (e.g. Leishman et al., 2007, 2010; Ordonez et al., 2010; Peñuelas et al., 2010; Ordonez & Olff, 2013). In particular, Leishman et al. (2010) argued that native and invasive plants share similar carbon (C) capture strategies, with invaders subject to the same tradeoffs between physiological investments and returns (i.e. constrained within the same LES). They concluded that, although invasive plants found in disturbed sites had traits that conferred greater productivity, they also experienced higher resource costs relative to natives. Therefore, invasive plants have strategies that correspond to the early successional, fast investment return portion of the LES, a conclusion used to mechanistically explain their dominance in disturbed, high-resource ecosystems (Leishman et al., 2007, 2010).

However, ecosystems subject to strong resource limitation are not immune to invasion (Martin et al., 2009), including Eastern North American (ENA) deciduous forests that experience very low light and nutrient levels during the growing season (Fridley, 2008). It is an open question as to whether invasion mechanisms described for high-resource environments, such as old fields, anthropogenic sites, and roadsides, are applicable to less disturbed ecosystems of low resource availability (Funk & Vitousek, 2007). It is generally understood that species adapted to resource-poor habitats follow strategies that place a higher premium on efficient use of resources (conservative strategies) at the expense of rapid growth (Aerts & Chapin, 1999).

Demographic studies of temperate forest tree invasions suggest that invaders do not necessarily follow demographic or life history tradeoffs evident in the native flora, such as that between low-light survivorship and high-light growth (Martin et al., 2010) and between classic r/K strategies of fast growth and reproduction versus persistence (Closset-Kopp et al., 2007). Select comparative studies, often in habitats of limited light or nutrients, report invasive plants with seemingly superior performance compared with natives at a given metabolic or resource cost, including increased growth rates (Osunkoya et al., 2010), greater mean performance or trait plasticity (Funk, 2008; Godoy et al., 2012; Paquette et al., 2012), greater photosynthetic rates at lower respiratory costs (Pattison et al., 1998; McDowell, 2002) and greater resource- or energy-use efficiencies (Baruch & Goldstein, 1999; Nagel & Griffin, 2004; Funk & Vitousek, 2007; Boyd et al., 2009). All else being equal, these findings imply that invasive species are not constrained by the same tradeoffs as natives, leading to greater production given similar resource investments. It remains unclear why these seemingly more efficient adaptations are not evident in neighboring native species. Phylogenetic constraints may exist, with certain floras never evolving certain trait combinations, which can explain how certain nonnative plants with novel resource-use strategies are superior competitors in a new range (Mack, 2003). A recent global analysis of leaf traits supports the possibility that evolutionarily distinct floras within similar biomes may have evolved different tradeoffs in resource capture strategies (Heberling & Fridley, 2012).

In ENA, the naturalized flora includes European forbs that inhabit open, managed, and disturbed sites. By contrast, invasive plants in ENA (i.e. those of highest management concern) are primarily woody species from Central and East Asia that are often invasive in forested habitats (Fridley, 2008). These shade-tolerant plants are particularly troublesome for management because their populations may increase as succession proceeds (Martin et al., 2009). In a recent common garden study of ENA forest species, Fridley (2012) found that invaders exhibit systematic differences in growth phenology, with significantly later leaf senescence for invasive species. It is unclear if any fitness advantage of an extended growing season for invasive species is equalized by tradeoffs at the leaf level such as shorter lifespan (i.e. more rapid leaf turnover) or lower daily productivity.

To test whether invasive plants in ENA forests exhibit different patterns of resource use from natives, we measured leaf-level C gains, energy and nitrogen (N) investments, and resource-use efficiencies (RUEs) of invasive and native shrubs and lianas found in ENA deciduous forests. All plants were grown in a common garden to concentrate on intrinsic trait differences, rather than those that might arise from environmental differences. We expanded upon other invasion studies (e.g. Leishman et al., 2010) to focus on phylogenetically related groups of species found in resource-limited habitats and considered both instantaneous and time-integrated traits (e.g. Funk & Vitousek, 2007). As ENA understory species are constrained by both light and N availability (Aber et al., 1993; Finzi & Canham, 2000), we hypothesized that ENA invaders should have greater C gains at lower resource costs. Therefore, we predicted that invasions in ENA forests are not attributable to greater resource use than natives per se, but rather, greater efficiency in the use of those limiting resources (i.e. greater C gains per unit resource cost).

Materials and Methods

Species selection and sampling protocol

We studied 32 shrub and liana species (14 native and 18 nonnative), with 14 nonnative species formally recognized as ‘invasive’ in ENA deciduous forests and the remaining four recognized as ‘naturalized’ (Fridley, 2008; Table 1). ‘Naturalized’ refers to nonnative species capable of maintaining natural populations without human intervention, whereas ‘invasive’ species refers to a subset of naturalized species with actively spreading populations that have been formally recognized by management agencies. Removing naturalized, but not (yet) invasive, species from the analysis did not affect conclusions (analysis not shown). Therefore, the nonnative group is hereafter referred to as ‘invasive’ for simplicity. Because invasive forest species in ENA are biased toward those of East Asian origin (Fridley, 2008), these species were the focus for comparison to native congeners. While the studied species are not equally abundant across ENA, many can co-occur and all species occur in the understory across the Eastern Deciduous Forest biome. Plant material was collected from an experimental garden in Syracuse, New York, USA (43°03′N, 76°09′W) established in 2006 in three replicate blocks and a seasonal shade treatment (80%) to simulate forest understory light regimes (see Fridley, 2012). Individuals were originally collected from the wild in the Syracuse region or, if unavailable, acquired commercially from nurseries in North America of similar latitude (Supporting Information Table S1).

Table 1. Woody, deciduous species measured in this study, including current status in Eastern North America (Fridley, 2008), general growth form, and biogeographic origin
FamilySpeciesInvasive statusGrowth formOrigin
  1. a

    Taxonomic classification is unclear. Studied individuals may be hybrids of Berberis canadensis and Berberis thunbergii.

BerberidaceaeBerberis canadensis Mill.aNativeShrub 
Berberis koreana Palib.NaturalizedShrubEast Asia
Berberis thunbergii DC.InvasiveShrubEast Asia
CaprifoliaceaeLonicera canadensis J. Bartram ex MarshallNativeShrub 
Lonicera fragrantissima Lindl. & PaxtonInvasiveShrubEast Asia
Lonicera hirsuta EatonNativeLiana 
Lonicera involucrata (Richardson) Banks ex Spreng.NativeShrub 
Lonicera japonica Thunb.InvasiveLianaEast/Central Asia
Lonicera maackii (Rupr.) Maxim.InvasiveShrubEast/Central Asia
Lonicera morrowii A. GrayInvasiveShrubEast Asia
Lonicera reticulata Raf.NativeLiana/shrub 
Lonicera sempervirens L.NativeLiana 
Lonicera standishii JacquesInvasiveShrubEast Asia
CelastraceaeCelastrus orbiculatus Thunb.InvasiveLianaEast/Central Asia
Celastrus scandens L.NativeLiana 
Euonymus alatus (Thunb.) SieboldInvasiveShrubEast Asia
Euonymus americanus L.NativeShrub 
Euonymus atropurpureus Jacq.NativeShrub/tree 
Euonymus  bungeanus Maxim.NaturalizedShrub/treeEast Asia
Euonymus  europaeus L.InvasiveShrub/treeEurasia
Euonymus  hamiltonianus Wall. ssp. sieboldianus (Blume) H. HaraNaturalizedShrub/treeEast/Central Asia
Euonymus obovatus Nutt.NativeShrub 
Euonymus phellomanus Loes.NaturalizedShrubEast Asia
ElaeagnaceaeElaeagnus angustifolia L.InvasiveShrub/treeEurasia
Elaeagnus commutata Bernh. ex Rydb.NativeShrub 
Elaeagnus multiflora Thunb.InvasiveShrubEast Asia
Elaeagnus umbellata Thunb.InvasiveShrubEast Asia
LauraceaeLindera benzoin (L.) BlumeNativeShrub/tree 
RhamnaceaeFrangula alnus Mill.InvasiveShrub/treeEurope/Central Asia
Frangula caroliniana (Walter) A. GrayNativeShrub/tree 
Rhamnus alnifolia L'Hér.NativeShrub 
Rhamnus cathartica L.InvasiveShrub/treeEurope/Central Asia

From July to mid-August 2011, two to six individuals per species were sampled from the common garden for subsequent measurements (mean ± SD: 3.22 ± 1.13 individuals per species). Gas exchange measurements were performed on cut branches, following the protocol of Niinemets et al. (2005). We used cut branches for logistical reasons and to ensure that measurements were made under consistent environmental conditions. On cool mornings between 06:00 and 08:00 h, two upper branches per individual were cut under water. To maintain xylem water potential, the severed ends were retained in water and transported to the laboratory within 1 h. The branches were then recut under water and covered in transparent polyethylene plastic to reduce transpiration. Branches were stabilized at room temperature under low light for 1–2 d before recording gas exchange measurements. Each morning, branches were recut under water and the foliage was misted. This pre-conditioning period minimizes temporal and species-level differences in stomatal openness for comparable, consistent photosynthetic measurements (Niinemets et al., 2005).

Leaf gas exchange

All gas exchange measurements were made in morning hours on recently expanded, mature leaves using an LI-6400 portable photosynthesis system equipped with CO2 and temperature control modules (Li-Cor, Lincoln, NE, USA). Leaf temperature was maintained at 25°C under ambient humidity throughout measurements with a sample chamber flow rate of 700 μmol s−1.

We measured leaf photosynthetic responses to irradiance (photosynthetic photon flux density (PPFD)) at 10 steps from 1000 to 0 μmol photons m−2 s−1 with sample chamber CO2 concentration at 400 μmol mol−1. All species were light saturated (without photoinhibition) at the highest light intensities. Net photosynthetic rate was recorded after equilibrating for at least 2 min at each PPFD and reaching defined stability parameters based on photosynthetic rate and stomatal conductance to water.

Light response curve parameters were estimated through nonlinear least squares regression of a nonrectangular hyperbola (Marshall & Biscoe, 1980):

display math(Eqn 1)

(Anet and Amax, the area-based net and maximum gross photosynthetic rates (μmol CO2 m−2 s−1), respectively; φ, the apparent quantum yield (mol CO2 mol photons−1); Rd, the daytime dark respiration rate (|Anet| at no light; μmol CO2 m−2 s−1); θ, curve convexity (dimensionless).) The light compensation point (LCP) was estimated from the x-axis intercept, and the light saturation point (LSP) was estimated as the PPFD when 75% of Amax (model asymptote) was achieved.

Similarly, we measured leaf photosynthetic responses to changing intercellular CO2 pressure (photosynthetic rate/intercellular CO2 partial pressure (A/Ci) curves) by varying sample chamber CO2 concentrations between 50 and 500 μmol CO2 mol−1 while maintaining saturating PPFD. A/Ci measurements were taken from the same leaves (or from adjacent nodes) as those for light response curves. Following Wullschleger (1993), maximum carboxylation rates (Vc,max) were estimated from CO2 response (A/Ci) curves when Ci was < 20 Pa (Rubisco limited). The biochemical photosynthetic model developed by Farquhar et al. (1980) was fitted using nonlinear least squares regression:

display math(Eqn 2)

(Ci, intercellular CO2 partial pressure (Pa); Kc and Ko, Michaelis–Menten constants for carboxylation and oxygenation (40.4 Pa and 24.8 kPa, respectively); O, the O2 concentration (21 kPa); Vc,max, the maximum carboxylation rate (μmol CO2 m−2 s−1); Rday, the daytime mitochondrial respiration rate (μmol CO2 m−2 s−1); Γ*, the CO2 compensation point in the absence of mitochondrial respiration (3.7 Pa).) Rubisco kinetic constants were obtained from von Caemmerer (2000). These constants were assumed to be similar among species (see Sharkey et al., 2007). Calculations using an independent set of literature-derived kinetic constants did not affect the reported conclusions (analysis not shown).

Leaf structural and biochemical characteristics

Following gas exchange measurements, the leaves were harvested, pressed, and oven-dried at 60°C for at least 48 h. Leaf area was measured using a leaf area meter (LI-3100; Li-Cor) and scanned leaf images. Specific leaf area (cm−2 g−1) was calculated as the leaf surface area per g dry mass. Ground leaf samples were placed in an ashing furnace at 500ºC for 4 h, and leaf ash concentration was calculated as ash mass divided by sample mass. Duplicate samples were averaged for each individual. Leaf N and C concentrations were determined using an elemental analyzer (CE Elantech, Lakewood, NJ, USA) for leaves collected for photosynthetic measurements and, for a limited species subset, leaf litter samples collected after leaf fall. Leaf lifespan (LL) was determined from 2008 to 2010 biweekly leaf censuses as described in Kikuzawa (1983) and averaged over the years monitored for each species (Fridley, 2012).

Leaf construction cost (CC) quantifies the amount of glucose equivalents required to construct a leaf in terms of C skeletons, reductant, and ATP, excluding additional costs for maintenance and substrate transport (Williams et al., 1989). Leaf CCmass (g glucose g−1) was estimated using the following equation (Vertregt & Penning de Vries, 1987; Poorter, 1994; Boyd et al., 2009):

display math(Eqn 3)

(Cmass, leaf carbon concentration; Ash, leaf ash concentration (proxy for mineral concentration; Vertregt & Penning de Vries, 1987); Nmass, leaf N concentration (all in g g−1).) We assumed leaf inline image accumulation is negligible compared with organic N forms, and nitrate is the dominant form of N uptake. The first part of the CC equation above takes into account the C costs (empirically determined from the relationship between glucose costs and C content of biochemical compounds; Vertregt & Penning de Vries, 1987). The second part of the first term (including Ash) subtracts the mineral component in organic tissue from the C cost, as the mineral fraction in organic tissue does not require C skeletons and energy required for their uptake is independent of costs for growth (Poorter, 1994). The last term of the CC equation above accounts for the additional, substantial costs required to reduce nitrate into organic N (proteins).

Metrics of leaf resource-use efficiencies

Resource-use efficiency (RUE) is broadly defined as the amount of C assimilated per unit resource (Funk & Vitousek, 2007). Potential photosynthetic energy-use efficiency (PEUE) was calculated as Amax,mass/CCmass. Time-integrated PEUE was calculated as PEUE × LL, which accounts for the duration of potential returns on initial leaf investment. Potential photosynthetic N-use efficiency (PNUE) was calculated as Amax,mass/[leaf N]. For 17 species, time-integrated N-use efficiency (NUE) was calculated as Amax,mass/[leaf litter N] × LL (Table S1; see Aerts & Chapin, 1999; Hirose, 2012). As leaf N resorption data were not collected for all species, an additional time-integrated metric of PNUE was calculated as PNUE × LL (‘lifespan PPNUE’ sensu Reich et al., 1992). The relationship between PNUE × LL (mol CO2 g−1 leaf N) and NUE (mol CO2 g−1 unresorbed N; incorporating N residence time) scaled in proportion with one another (i.e. isometric; Fig. S1). Therefore, we report our results using PNUE × LL as a robust estimate for time-integrated NUE, as this metric was calculated for the complete data set. In our data set, this tight relationship between metrics results from comparable rates of leaf N resorption between our study species (> 0.1; Table S2). However, N resorption can differ between native and invasive species (I. Jo, unpublished data). These potential differences caution that PNUE × LL may not always be a reliable proxy for time-integrated NUE, as N mean residence times can be important in determining overall efficiency (Berendse & Aerts, 1987).

Daily C gain was estimated using the light response curves and daily PPFD measured at 30-min intervals from June to August (see Fridley, 2012), assuming the nighttime respiration rate to be 75% of the dark respiration rate measured during the day (Williams et al., 1989). Payback time (days for net C gain to equal CC) was estimated as the ratio of CC to average daily C gain (Williams et al., 1989) multiplied by loge(2) to account for time-discounting effects (Poorter, 1994; Falster et al., 2012). This estimation of payback time does not account for age-related changes in photosynthesis or maintenance costs but can be considered a relative measure across species.

Data analysis

Where necessary, measurements were converted between area- (i.e. m−2 leaf) and mass-based estimates (i.e. g−1 leaf) through their corresponding specific leaf area. All statistical analyses were performed in R (R Development Core Team, 2013). We analyzed univariate data with phylogenetic generalized least squares (PGLS) regression models, which accounted for relatedness in the data set through the phylogenetic distance matrix (see Fig. S2 for tree). We accounted for variation within species by weighting values by their intraspecific precision (standard error−1). For each trait, we compared models with and without a fixed effect of native status (native or nonnative) using likelihood ratio tests, following a χ2 distribution with one degree of freedom. Because the phylogeny was approximate and based on estimated branch lengths, we also performed an analogous analysis using linear mixed-effect models (Bates et al., 2011) that accounted for intraspecific and genus-level variation through random effects. Results of this approach were qualitatively similar to PGLS and are excluded for brevity. When necessary, data were log-transformed to satisfy assumptions of variance homoscedasticity and normality of model residuals.

Bivariate trait relationships were analyzed with standardized major axis (SMA) line fitting implemented with the smatr package in R (Warton et al., 2012) using species means (see data Table S1). SMA line fitting minimizes residual variance in both x and y dimensions and is preferred in analyzing bivariate allometric relationships, as opposed to predicting y from x in classical regression (Warton et al., 2006). Testing in the SMA routine involves first testing for common slopes between groups. If the slopes do not differ, the lines fitted to the groups may represent a shift along their common slope and/or shifts in elevation (y-intercept).

Lastly, we implemented principal components analysis (PCA) to understand, in a multivariate context, how leaf RUE metrics, C assimilation, and leaf longevity were related and if coordinated group variation separated native and invasive plants along axes of physiological functioning. We specifically chose these traits to include in the PCA to summarize the multivariate relationships of our hypothesis that focused on C- and N-use efficiency.


Comparative functioning between native and invasive species

As a group, invaders had greater area-based net photosynthetic rates (Anet) than native species, at both saturating and relatively low PPFD (Fig. 1). However, area-based Anet was comparable between groups at and near 0 μmol photons m−2 s−1 (Fig. 1; Rd,area, Table 2). Mass-based dark respiration (Rd,mass) was moderately greater in invasive species (Table 2). Photosynthetic differences were further reflected through greater maximum photosynthetic rates (Amax,area and Amax,mass), maximum carboxylation rates (Vc,max), and average area- and mass-based daily C gain (C gainarea and C gainmass) in invasive species (accounting for phylogenetic relatedness; Table 2). Respiration efficiency (Amax,area/Rd,area) was also slightly greater in invasive species, suggesting that respiratory costs were lower per unit photosynthetic gain than in native species.

Table 2. Mean values (± 1 SE) of photosynthetic, biochemical, structural, and resource-use efficiency leaf traits among native and invasive species
Trait (units) aInvasiveNativeNative status (χ2)
  1. a

    Amax,area and Amax,mass, area- and mass-based light-saturated gross photosynthetic rates on an area and mass basis, respectively (Amax,mass = Amax,area × SLA); Rd,area and Rd,mass, area- and mass-based dark respiration rates at ambient [CO2], respectively; Amax/Rd, respiration efficiency; φ, apparent quantum yield; LSP, 75% light saturation point; LCP, light compensation point; Vc,max, maximum carboxylation rate; daily C gainarea and daily C gainmass, area- and mass-based average daily carbon assimilation; SLA, specific leaf area; Nmass and Narea, mass- and area-based leaf nitrogen concentrations; Ash, leaf ash concentration; Cmass and Carea, mass- and area-based leaf carbon concentrations; CCmass and CCarea, mass- and area-based leaf construction costs; LL, leaf longevity; PT, payback time, days to amortize leaf construction costs (Poorter, 1994); PNUE, photosynthetic nitrogen-use efficiency; PNUE × LL, ‘life-span PPNUE’ (sensu Reich et al., 1992), an index of time-integrated PNUE; PEUE, photosynthetic energy-use efficiency; PEUE × LL, time-integrated PEUE.

  2. The statistical significance of differences between native and invasive groups was determined using likelihood ratio tests (χ2 with 1 df) that compared PGLS regression models for each trait with and without a fixed effect of native status. Significantly greater mean values are indicated in bold. *, P < 0.05; **, < 0.01; ***, < 0.001.

Amax,area (µmol CO2 m−2 s−1) 11.19 ± 0.78 8.10 ± 0.719.53**
Amax,mass (µmol CO2 g−1 s−1) 0.2467 ± 0.0272 0.1802 ± 0.01816.84**
Rd,area (µmol CO2 m−2 s−1)0.79 ± 0.070.73 ± 0.062.04
Rd,mass (µmol CO2 g−1 s−1) 0.0176 ± 0.0021 0.0163 ± 0.00117.90**
Amax/Rd 17.37 ± 2.06 13.52 ± 1.646.09*
φ (µmol CO2 µmol−1 photons)0.076 ± 0.0030.074 ± 0.0032.35
LSP (µmol photons m−2 s−1)370.5 ± 92.4233.4 ± 55.2< 0.01
LCP (µmol photons m−2 s−1)10.8 ± 1.09.7 ± 0.83.53
Vc,max (µmol CO2 m−2 s−1) 46.10 ± 2.22 32.24 ± 3.0014.71***
Daily C gainarea (mmol CO2 m−2 d−1) 273.8 ± 9.0 223.9 ± 12.116.94***
Daily C gainmass (mmol CO2 g−1 d−1) 6.06 ± 0.53 5.00 ± 0.4010.15**
SLA (cm2 g−1)218.52 ± 12.75233.42 ± 8.030.25
Nmass (%) 2.97 ± 0.19 2.52 ± 0.1114.20***
Narea (g m−2) 1.40 ± 0.08 1.11 ± 0.0615.57***
Ash (mg g−1)91 ± 693 ± 84.22
Cmass (%)45.00 ± 0.2644.51 ± 0.401.46
Carea (g m−2)22.37 ± 1.4619.80 ± 0.713.44
C : N16.54 ± 1.05 18.60 ± 0.81 6.51*
CCmass (eq. g glucose g−1)1.328 ± 0.0181.384 ± 0.0201.61
CCarea (eq. g glucose m−2) 65.85 ± 4.16 56.81 ± 2.286.13*
LL (d) 145.1 ± 6.9 120.9 ± 6.47.72**
PT (d)6.12 ± 0.54 7.25 ± 0.98 6.95**
PNUE (µmol CO2 g−1 N s−1)8.28 ± 0.697.51 ± 0.673.57
PNUE × LL (mol CO2 g−1 N) 101.00 ± 8.61 77.73 ± 8.208.52**
PEUE (µmol CO2 kg−1 glucose s−1) 183.44 ± 19.07 137.95 ± 13.036.38*
PEUE × LL (kmol CO2 kg−1 glucose) 2.19 ± 0.19 1.41 ± 0.1315.29***
Figure 1.

Average modeled light response curves for 12 native (open symbols) and 18 invasive (closed symbols) species. Error bars (group mean ± 1 SE) show empirically measured peak season area-based net photosynthetic rates (Anet) at various irradiances (photosynthetic photon flux density (PPFD)). Statistical differences are between native and invasive groups evaluated from likelihood ratio tests of phylogenetic generalized least squares (PGLS) models (**, < 0.01; ***, < 0.001). Points indicate species means.

Invasive species had significantly greater mean CCarea, Nmass and Narea, similar leaf C and Ash, and, subsequently, lower leaf C : N (Table 2). With higher values of both Narea and Amax,area, invasive species were shifted further along a shared tradeoff in the Amax,areaNarea relationship (Fig. 2a). Leaf N was more closely associated with Vc,max than Amax,area, especially within invasive species (Fig. 2b). Among photosynthetic traits, the greatest correlation was between Amax,area and Vc,max (overall: r2 = 0.49; P < 0.001), which describes the recognized relationship between the maximum carboxylase activity of Rubisco and the realized net C assimilation rate at ambient CO2 concentrations. With significantly higher mean values in both traits (Table 2), invasive species were shifted further along a common slope (Fig. 2c).

Figure 2.

Standardized major axis (SMA) relationships for maximum carbon (C) assimilation parameters and leaf nitrogen (N). Points indicate species means. (a) Light-saturated gross photosynthetic rate (Amax)–N concentration (leaf N) (all species: r2 = 0.22, < 0.01; native: r2 = 0.01, = 0.76; invasive: r2 = 0.15, = 0.12), (b) maximum carboxylation rate (Vc,max)–leaf N (all species: r2 = 0.36, < 0.001; native: r2 = 0.07, = 0.42; invasive: r2 = 0.36, < 0.01), and (c) AmaxVc,max (all species: r2 = 0.49, P < 0.001; native: r2 = 0.36, < 0.05; invasive: r2 = 0.38, < 0.01). Significance tests are from SMA for each relationship for differences in slope, elevation (y-intercept), and shift along the common fitted slope (**, < 0.01; ***, < 0.001).

Considering investments in tissue construction along with the subsequent duration of photosynthetic function, the relationship between daily C gainmass and LL was negative, although the correlation was weak (overall: r2 = 0.07; = 0.15). With both shorter LL and lower C assimilation among natives (Table 2), a significant elevation (y-intercept) shift in fitted SMA lines was detected (< 0.01; Fig. 3a). Therefore, at a given LL, invaders had greater C gainmass (grouped in upper right of Fig. 3a). As expected, the payback time–LL relationship followed the inverse trend of daily C gainmass–LL, with a positive, weak association (Fig. 3b). At a given LL, invaders tended to have shorter payback time (SMA elevation test; < 0.05), even though invaders also had greater C costs in the form of glucose (CCarea) to compensate before achieving a net positive leaf C balance (Table 2).

Figure 3.

Standardized major axis (SMA) relationships between (a) mass-based daily carbon gain (daily C gainmass) and leaf longevity (LL) (all species: r2 = 0.07, = 0.15; native: r2 = 0.07, = 0.41; invasive: r2 = 0.24, < 0.05) and (b) payback time and LL (all species: r2 = 0.02, = 0.46; native: r2 = 0.04, = 0.49; invasive: r2 = 0.09, = 0.23). Points indicate species means. Significance tests are from SMA for each relationship for differences in slope, elevation (y-intercept), and shift along the common fitted slope (*, < 0.05; **, < 0.01).

Differences in instantaneous and time-integrated resource-use efficiencies

Although invaders exhibited both greater Amax and CCarea than natives, mean PEUE (Amax,area/CCarea) differed between groups, indicating that invaders possessed greater instantaneous C returns per unit energy investment. As a consequence of greater LL in invasive species, especially at a given payback time (Fig. 3b), this difference in PEUE was magnified when integrated over time (PEUE × LL; Table 2). By contrast, differences were marginal in PNUE (Table 2; = 0.059), which reflected a proportionately greater mean leaf N and Amax,area among invasive species (Fig. 2a). However, because of substantially greater LL in invasive species (Table 2), marginal PNUE differences became significant when integrated over time (PNUE × LL). Conclusions remained the same when considering leaf N resorption rates in a species subset (see the 'Materials and Methods' section; Fig. S1).

Multivariate trends in resource-use efficiencies

The first two axes of a PCA using RUE traits explained 86% of the variation in the data set, and separated native and invasive species (Fig. 4a; two-sided t-tests of species’ axis scores by native status: axis 1: = −2.25, df = 28, = 0.03; axis 2: = 1.89, df = 28, = 0.07). Axis 1 (64% of variance explained) was positively correlated with area-based daily C gain (daily C gainarea), PEUE, and PNUE, and negatively correlated with payback time, which matches univariate and bivariate testing (Table 2; Fig. 3). Axis 2 (22% of variance explained) was most strongly positively correlated with LL (Fig. 4b). LL was orthogonal to daily C gainarea, which echoes the bivariate tests that indicated that commonly held physiological tradeoffs involving LL were less constraining to productivity in invasive species relative to natives (Fig. 2).

Figure 4.

Ordination of native (= 12; open symbols) and invasive (= 18; closed symbols) species using principal components analysis (PCA) based on resource-use efficiency traits: payback time, leaf longevity (LL), indices of time-integrated photosynthetic nitrogen (N)-use efficiency (PNUE × LL) and energy-use efficiency (PEUE × LL), area-based daily C gain (Daily C Gainarea), PNUE, and PEUE. (a) Species scores along axes 1 and 2 and (b) trait loadings with vector length and direction corresponding to correlations between the trait and species scores. Axes 1 and 2 cumulatively explained 86% of the variance and separate native and invasive species’ resource-use efficiency trait syndromes (two-tailed t-tests: +, < 0.1; *, < 0.05). Point symbols follow Fig. 3.


Do invasive species follow different resource-use strategies?

Much of our current mechanistic understanding of plant invasions stems from studies in anthropogenic, disturbed habitats of high fertility (Martin et al., 2009), which show that invaders often exhibit greater resource acquisition rates, not necessarily differences in RUE or different constraints in leaf function (e.g. Leishman et al., 2010; Ordonez & Olff, 2013). In other words, native and invasive plants in a given community may follow the same general set of resource capture strategies and trait tradeoffs that have been generalized for species globally (e.g. Reich et al., 1997; Wright et al., 2004, 2005). We tested this premise in species common to ENA deciduous forest understories to investigate whether congeneric invasive and native species are similarly constrained to a common set of tradeoffs, or alternatively, whether species invasive in ENA exhibit more efficient resource-use strategies, suggesting greater C gain per unit resource invested. Among 32 native and invasive species common to ENA forests, we found evidence that invaders are both more productive and more efficient—they exhibit greater daily C gain and also greater C gain per unit C or N invested in leaf tissue relative to their native competitors (Table 2). Expanding upon past invasion studies (e.g. Baruch & Goldstein, 1999; Funk & Vitousek, 2007; Boyd et al., 2009; Leishman et al., 2010), these differences were most pronounced when integrated over time.

Invaders as a group exhibited significantly longer LL and greater CCarea – traits associated with high resource conservation, low growth rate, and lower competitive ability (Aerts & Chapin, 1999; Westoby et al., 2002; Wright et al., 2004). This finding of longer LL for invasive species builds upon the surprising recent finding that ENA forest invaders have greater relative C gains into the autumn, utilizing a temporal niche absent in the native flora (Fridley, 2012). Plants adapted to low-resource environments are expected to exhibit conservative resource-use strategies, but at the expense of fast growth (Aerts & Chapin, 1999). In low-resource ecosystems, the success of invasive species may therefore depend on greater resource conservation than native species, but with proportionately slower absolute growth rates. In ENA forests, we found that invaders had greater RUEs and similar light-use efficiencies compared with native species. Surprisingly, in addition to more conservative resource-use traits, invaders also exhibited traits associated with greater productivity. With both greater photosynthetic gains (Fig. 1) and greater mean LL, invaders as a group exhibited greater energy-use and N-use efficiencies through time (PEUE × LL; Table 2). LL was weakly related to photosynthetic functioning (Fig. 3) and separated invaders on an axis orthogonal to C gains (Fig. 4), which suggests that tradeoffs between resource investments and C returns are distinct between native and invasive species in ENA. Therefore, compared with native species, resource-use strategies of invasive species were not only more efficient but also more productive.

Past studies have found RUE differences between native and invasive species across a range of habitats (e.g. McDowell, 2002; Nagel & Griffin, 2004; Funk & Vitousek, 2007; Osunkoya et al., 2010), including in an ENA forest (Boyd et al., 2009). Funk & Vitousek (2007) found greater instantaneous PEUE and PNUE in invaders in Hawaii across light- and N-limited habitats. However, they found that natives had comparable RUEs when integrated over the lifespan of the leaf. In light of instantaneous differences and time-integrated similarities in RUEs, they reasoned that invasion was driven by dynamics on short (seasonal) timescales. We observed a different pattern in ENA temperate forests, which could be attributable to the deciduous environment and potential fitness advantages of extended leaf phenology (Fridley, 2012). In ENA, invasion could be explained by later autumn senescence among invasive species (Fridley, 2012) and longer LL that was largely independent of daily C gainarea (Fig. 3).

Other studies have stressed similarities in leaf resource economics among native and invasive plants (e.g. Leishman et al., 2007, 2010; Ordonez et al., 2010; Peñuelas et al., 2010; Ordonez & Olff, 2013). In a recent study comparing natives and invaders in Australia, Leishman et al. (2010) concluded that species from both groups followed metabolic tradeoffs consistent with leaf economics theory. Comparing the slopes of leaf trait relationships for native and invasive plants, they found that, across many trait relationships, natives had strategies at the slow returns end of a common, coordinated axis of plant strategies, with lower C assimilation rates proportionate with lower resource needs. Similarly, in ENA understory species, we found group shifts along common slopes for instantaneous photosynthesis- and N-related traits (Fig. 2). However, when considering other traits (e.g. CCarea and LL), our results suggest that ENA invaders are both more productive and more efficient than natives (Table 2; Figs 3, 4). Unlike Leishman et al. (2007, 2010), we found greater RUEs among invasive species. Because RUE is a ratio, differences can be found along a common slope in a bivariate relationship, suggesting that an RUE difference alone is not sufficient to conclude a fundamental difference in resource capture strategies (defined as differences in bivariate slope relationships; Leishman et al., 2010). Although N-related functional relationships in this study shared slopes between groups (Fig. 2), we found invaders to be more resource-use efficient as a product of greater cumulative C gains per unit invested, which is indicative of different physiological constraints between native and invasive species.

Our conclusions indicating resource-use strategy differences between native and invasive plants contrast with conclusions of these past studies for two possible reasons. First, many studies focused on instantaneous measures and did not incorporate temporal traits. Therefore, we cannot directly compare our results with those of past studies that do not consider LL. Although perhaps less important in disturbed ecosystems, the integration of time into functional comparisons plays a critical role when leaf duration ultimately determines whole-plant cumulative C gain (Reich et al., 1992; Westoby et al., 2002), especially in light- and N-limited ENA forests (Aber et al., 1993; Finzi & Canham, 2000). Additionally, the conclusions of Leishman et al. (2010) may be more representative of invasions in disturbed, fertile habitats, as few invasive species were found in undisturbed habitats in their study. However, Ordonez & Olff (2013) considered trait differences across resource and disturbance gradients and found that, compared with natives, invasive species in high-resource environments had greater mean trait values associated with fast growth, but trait differences between groups were similar across environments. Because individuals in our study were grown in a common environment, we tested inherent physiological differences, rather than in situ performance in different habitats.

Unexpectedly, we found that ENA invaders had significantly greater CCarea and, although not significant, lower mean specific leaf area (Table 2). Low CC has been invoked as a primary measure of invasion potential (Nagel & Griffin, 2001). However, studies have reported mixed results and interpretations, including invasive species with lower CC (Baruch & Goldstein, 1999; Nagel & Griffin, 2001; Boyd et al., 2009; Osunkoya et al., 2010) and greater CC (McDowell, 2002). Patterns in leaf CC alone may be insufficient to understand invasions across habitats, because invested costs are without ecological context unless viewed in light of C returns (Griffin, 1994). Williams et al. (1989) proposed payback time (days to amortize CC) as a trait to explain ecological variation in LL across habitats. Our measure of payback time can be considered a relative estimate across species because we did not directly account for leaf age-related declines in photosynthetic rates. In our data, the payback time–LL relationship was weak, although natives and invaders clearly occupied different portions of trait space (Fig. 3b). Further, at a given LL, invasive species tended to have greater C returns than natives (elevation shift; Fig. 3b). Because plants should be expected to retain leaves longer than their payback time to achieve a net positive C balance (Westoby et al., 2002; Falster et al., 2012), it is reasonable to expect the LL–payback time relationship to be weak within deciduous species. However, all else being equal, the tradeoff between LL and daily C gain among co-occurring plants should be expected to equalize fitness and maintain strategies along that continuum (Falster et al., 2012). Tradeoffs with LL were weak in our data, with ENA invaders achieving greater C gains (Figs 1, 2), a greater C assimilation rate at a given LL (Fig. 3a), and greater PEUE through time (Table 2).

Why are East Asian invaders less constrained by metabolic tradeoffs than ENA natives?

ENA invaders appear to be less constrained in their allocation to rapid growth versus long-lived tissues than the native flora, yet it remains unclear why native species would not also exhibit such strategies. One possibility is that we did not measure an important trait that, if measured or integrated over the whole plant, would equalize our reported differences in leaf resource economics. For instance, greater herbivore pressure in natives could lead to lower competitive performance (e.g. Keane & Crawley, 2002), thereby explaining why invasive plants in our study have longer LL without paying the expected metabolic costs. However, given that invasion between the Eurasian and ENA forest ecosystems is asymmetric, with East Asia contributing more invaders to ENA (Fridley, 2008) than vice versa (Weber et al., 2008), scenarios that invoke home range herbivory alone are unlikely to be general for this group of invasive species. Also, East Asian species coevolved with earthworms, which were largely absent from ENA until recent human introductions (Nuzzo et al., 2009). Therefore, the strategies of East Asian species may be better adapted than ENA natives to soil conditions maintained by earthworms, such as increased N cycling. It is uncertain if East Asian species evolved under lower soil nutrient availability to promote increased RUE. However, we found increased C gains for invasive species in addition to greater RUEs, which cannot be explained by historically different soil nutrient conditions alone.

Although tradeoffs in leaf functional traits are broadly consistent world-wide (e.g. Wright et al., 2004), there may be important differences in plant functional strategies between regions as a result of historical constraints (Heberling & Fridley, 2012). Fourteen of the 18 species measured in the current study have native distributions that include East Asia (Table 1). Despite both regions lying primarily in the temperate deciduous forest biome and composed of closely related taxa that diverged in the late Miocene (Donoghue & Smith, 2004), the flora of East Asia has experienced very different environmental conditions over the past several millennia, resulting in higher species diversity, endemism, and phylogenetic diversity (Qian & Ricklefs, 2000). Community-level properties, such as competitive pressure, disturbance, and time for resident species to adapt and fill niches, may make some communities more or less vulnerable to invasion (‘community maturity’ sensu Shea & Chesson, 2002). From this regional perspective, we speculate that invasive plants introduced from East Asia may have ‘pre-adapted’ traits in the native range that confer invasiveness in ENA (Fridley, 2011). However, ecophysiological comparisons in the home range would be needed to confirm this hypothesis.


We found both higher productivity and more efficient resource use in forest understory invaders in ENA compared with their native congeners, and lower metabolic constraints between LL and C gain in invaders. From a leaf-level perspective, we found invaders’ competitive success to probably be attributable to greater C gains, and, despite greater leaf N and energy resource investments, a greater duration of returns. This conclusion supports past findings emphasizing the importance of RUE in invasions in resource-limited ecosystems (Funk & Vitousek, 2007) and, in ENA forests, the functional importance of leaf phenology differences in the invasive flora (Fridley, 2012). If these findings are general, then such differences in the leaf function of invaders may be expected to drive large shifts in the productive capacities and nutrient budgets of deciduous forest ecosystems.


This material is based upon work supported in part by a US National Science Foundation Graduate Research Fellowship and a Syracuse University Graduate Fellowship to J.M.H. We are grateful to In Su Jo for sharing his N resorption data, and A. Craddock and Syracuse University for support of the experimental garden. We thank David Ackerly and three anonymous referees for comments that significantly improved the manuscript.