Higher water use efficiency (WUE) in C4 plants may allow for greater xylem safety because transpiration rates are reduced. To evaluate this hypothesis, stem hydraulics and anatomy were compared in 16 C3, C3–C4 intermediate, C4-like and C4 species in the genus Flaveria. The C3 species had the highest leaf-specific conductivity (KL) compared with intermediate and C4 species, with the perennial C4 and C4-like species having the lowest KL values. Xylem-specific conductivity (KS) was generally highest in the C3 species and lower in intermediate and C4 species. Xylem vessels were shorter, narrower and more frequent in C3–C4 intermediate, C4-like and C4 species compared with C3 species. WUE values were approximately double in the C4-like and C4 species relative to the C3–C4 and C3 species. C4-like photosynthesis arose independently at least twice in Flaveria, and the trends in WUE and KL were consistent in both lineages. These correlated changes in WUE and KL indicate WUE increase promoted KL decline during C4 evolution; however, any involvement of WUE comes late in the evolutionary sequence. C3–C4 species exhibited reduced KL but little change in WUE compared to C3 species, indicating that some reduction in hydraulic efficiency preceded increases in WUE.
In angiosperms, water use efficiency (WUE) of photosynthesis has been proposed to influence the structure and function of the hydraulic pathway (Kocacinar & Sage 2003, 2004, 2005; Franks & Brodribb 2005). Plants with greater WUE have lower water requirements per unit carbon gained, and thus can lower the demand placed upon the conducting pathway by a transpiring leaf canopy. Within similar environments, plants with greater WUE should have reduced selection pressure for efficient xylem and hence might evolve features enhancing hydraulic safety. Safety features include short, narrow and thick-walled vessels, which enhance redundancy in the flux pathway (Tyree & Zimmermann 2002). Safety could also be enhanced by fewer pits, smaller pits and thicker pit membranes that increase the resistance to air seeding and cavitation (Choat, Cobb & Jansen 2008). Alternatively, plants with greater WUE might maintain the same degree of hydraulic safety, and instead boost photosynthetic potential by increasing the size of the leaf canopy served by a unit of xylem tissue (Kocacinar & Sage 2003). Increasing hydraulic safety or canopy size per unit xylem, or both, would reduce the leaf-specific conductivity (KL, stem hydraulic conductivity per unit leaf area supported by the stem). Thus, if WUE affects the trade-off between hydraulic safety and efficiency, WUE and KL could be inversely related.
C4 plants have markedly higher WUE than C3 plants (Osmond, Winter & Ziegler 1982; Pearcy & Ehleringer 1984; Long 1999) and also consistently exhibit lower KL than C3 plants of similar functional type or ecological habitat (Kocacinar & Sage 2003, 2004, 2005). In C4 plants from arid regions, a lower KL is associated with changes in xylem structure that increase hydraulic safety, whereas in annual C4 species from mesic habitats, a lower KL is associated with greater leaf area per unit xylem (Kocacinar & Sage 2003, 2004). These patterns are present in both herbaceous and woody species, and between species of similar ecological habit and close taxonomic affinity. However, the ability to draw strong evolutionary inferences regarding WUE and KL is limited by poor understanding of phylogenetic relationships in the C3 and C4 groups compared by Kocacinar & Sage (2003, 2004). As a result, a statistically robust comparative analysis of the evolutionary pattern with respect to WUE and hydraulic efficiency has not been possible, because xylem hydraulics vary with both phylogenetic distance and ecological habitat (Pockman & Sperry 2000; Tyree & Zimmermann 2002; Preston & Ackerly 2003); a phylogenetically based comparison of plants from similar ecological settings is essential to rule out spurious relationships (Maherali et al. 2004).
The genus Flaveria (Asteraceae) has been widely used as a model system for studying the physiology and molecular biology of the C4 plant evolution of C4 photosynthesis (Sage 2004). The genus includes C3 and fully expressed C4 species, as well as numerous C3–C4 and C4-like intermediate species (Table 1). Recently, a well-resolved phylogeny based on chloroplastic trnl-F, nuclear ITS (internal transcribed spacer) and ETS (external transcribed sequence) has been published for 21 of the 23 known species in the genus (McKown, Moncalvo & Dengler 2005). C3 photosynthesis is unequivocally the ancestral condition, and multiple origins of intermediate and C4 photosynthesis are present within the genus. In Flaveria, C3–C4 intermediate photosynthesis has been subdivided into three categories type I, type II and C4-like (Table 1; Edwards & Ku 1987; Vogan, Frohlich & Sage 2007). Type I C3–C4 intermediate species exhibit enlarged bundle sheath cells and reduce the photosynthetic CO2 compensation point to half that of C3 species by localizing the release of photorespired CO2 into the bundle sheath tissue. A C4 metabolic cycle involving phophoenolpyruvate carboxylation is not pronounced in type I intermediates. Type II intermediates have similar traits as type I intermediates but also have a limited engagement of a C4 cycle, as indicated by increased PEP carboxylase activities and CO2 compensation points that are between values exhibited by type I intermediates and C4 species. C4-like species have a well-developed C4 cycle, fully developed Kranz anatomy and exhibit CO2 compensaton points that are similar to or a few parts per million higher than C4 plants; however, they exhibit slight levels of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) activity in the mesophyll cells and thus retain residual patterns of C3 photosynthesis (Ku et al. 1991). In Flaveria, two major lineages of C3–C4 and C4-like intermediate species are present; clade A contains the type II C3–C4 species Flaveria ramosissima, the C4-like species Flaveria palmeri and Flaveria vaginata and all C4Flaveria species. Clade B contains seven type I and type II C3–C4 species, and the C4-like species Flaveria brownii. The C3 species Flaveria robusta is located at a node basal to both clades, and is closely related to a third independently evolved lineage of C3–C4 intermediacy represented by the type I intermediate in Flaveria sonorensis (McKown et al. 2005). These multiple origins of C3–C4 and C4-like photosynthesis offer a unique opportunity to examine whether there are convergent relationships between WUE and KL during the evolution of the C4 carbon-concentrating mechanism.
Table 1. Photosynthetic type, growth habit, habitat and distribution of Flaveria species in this study
Gypseous soils, open slopes, scrub vegetation, roadsides
Flaveria palmeri J. R. Johnston
Gypseous or clay-loam soils, lake shores, cultivated fields, roadsides
Flaveria pringlei Gandoger
Arid and gypseous soils, cultivated fields, roadsides
Flaveria ramosissima Klatt
Type II C3–C4 (A)
Gypseous or sandy-clay soils, scrub vegetation, cultivated and disturbed areas, roadsides
Flaveria robusta Rose
Gypseous, slate and sandy soils, deciduous woodland, open slopes, roadsides
Flaveria sonorensis A. M. Powell
Type I C3–C4
Warm mineral springs
Flaveria trinervia (Spreng.) C. Mohr
Africa, Caribbean, Central/South America, Mexico, USA
Saline, gypseous and moist soils, disturbed areas
Flaveria vaginata B. L. Robinson & Greenman
Arid and gypseous soils, scrub vegetation, cultivated fields
Flaveria also represents an evolutionary model where differences in habitat variation are minimized. Flaveria species are small- to medium-sized shrubs and herbs, and most species occur in similar habitats in Mexico and in the southern USA (Powell 1978; McKown et al. 2005; Sudderth et al. 2007). Numerous Flaveria species of different photosynthetic pathway also occur in the same regions, and some, such as Flaveria pringlei (C3) and Flaveria kochiana (C4) grow adjacent to each other on the same site (Sudderth, Espinosa & Holbrook 2008). Flaveria species exhibit similar morphologies and are frequently found in disturbed conditions on alkaline or gypseous soils, and they also occur in periodically arid habitats, often with some level of salinity (Powell 1978; Sudderth et al. 2008). A major difference between photosynthetic types is the lifespan of the respective species. Most C4 species are classified as annuals, while C3 species are perennial (Powell 1978); however, the C4F.kochiana is perennial (Sudderth et al. 2007). C4Flaveria species have 60–85% higher WUE than C3Flaveria (Monson, Schuster & Ku 1987; Monson 1989), which is consistent with the general view that C4 plants have two to four times the WUE of C3 plants (Larcher 2003). The Flaveria species expressing intermediate traits between the C3 and C4 conditions can also show intermediate patterns of gas exchange, anatomy and WUE (Monson 1989; McKown & Dengler 2007). C3–C4Flaveria species have intermediate WUE values between C3 and C4Flaveria species when photosynthetic rates are limited by nitrogen and are measured below current atmospheric CO2 levels (Monson 1989). In comparisons with C3 species, the type II C3–C4Flaveria floridana has up to 83% higher WUE than the C3 species (Schuster & Monson 1990; Monson & Jaeger 1991).
In this study, our objective was to test the hypotheses that an inverse relationship is present between WUE and KL in the genus Flaveria, and that this relationship is associated with the evolution of the C4 pathway. We compared WUE and xylem hydraulics in a broad sampling of C3, C3–C4, C4-like and C4Flaveria species. We also compared the stem xylem anatomy in numerous Flaveria species of different photosynthetic types. Using the Flaveria phylogeny of McKown et al. (2005), we tested KL and WUE data for phylogenetically independent contrasts to examine if any differences in xylem hydraulics are associated with evolutionary diversification of the C3–C4 intermediate and C4 photosynthetic pathways.
MATERIALS AND METHODS
Photosynthetic type, habit, general distribution and habitat of the Flaveria species used in this study are given in Table 1. All Flaveria plants were established from seeds or cuttings. Seeds of Flaveria bidentis (L.) Kuntze and F.pringlei Gandoger were courtesy of Dr Nancy G. Dengler at the University of Toronto, Toronto, Ontario, Canada. Seeds and/or cuttings of Flaveria angustifolia (Cav.) Persoon, Flaveria anomala B. Robinson, Flaveria australasica Hooker, F.brownii A. M. Powell, Flaveria chloraefolia A. Gray, F.floridana J. R. Johnston, F.robusta Rose, F.sonorensis A. M. Powell, Flaveria trinervia (Spreng.) C. Mohr and F.vaginata B. L. Robinson & Greenman were obtained from Dr Gerald Edwards at Washington State University, Pullman, WA, USA. Cuttings of Flaveria cronquistii Powell were obtained from Dr Maurice S. B. Ku from Washington State University. Cuttings of F.kochiana B.L. Turner were courtesy of Dr Erika A. Sudderth at Harvard University, Boston, MA, USA.
Flaveria plants were grown in 6 L pots on an outdoor rooftop garden during summer months at the University of Toronto in Toronto, Ontario, Canada. The potting medium was a mix of 25% Promix (Sun Gro Horticulture Canada Ltd., Seba Beach, Alberta, Canada), 25% sand and 50% triple-mix loam soil. Triple-mix loam soil contained 60% top soil, 20% compost and 20% sand, and was purchased from a commercial landscape company (J. Jenkins & Sons Co., Gormley, Ontario, Canada). All plants were watered regularly and fertilized once a week with a full-strength Hoagland's solution.
Stem hydraulic properties
Stem hydraulic conductivity (Kh) was measured on the 16 Flaveria species listed in Table 1 as described previously (Kocacinar & Sage 2003). For all Flaveria plants sampled, the youngest branches with a complete leaf canopy were measured to eliminate effects of leaf loss on KL. In each replicate, samples were taken far enough down the stem to include secondary xylem and to minimize pith tissue. Kh, xylem-specific conductivity (KS) and KL were assessed on stem segments 0.5–1.5 cm in diameter and 5–15 cm in length. A comparison of seven representative C3, C3–C4 intermediate and C4Flaveria species showed that segment lengths between 7.5 and 15.0 cm did not significantly affect estimates of Kh and KS (data not shown) nor KL (Fig. 1). In six of seven species, KL was independent of segment length between 2.5 and 15.0 cm. Chiu & Ewers (1993) also showed that shortening segment length from 20 to 2 cm did not affect hydraulic conductance (Kh) in the shrub Lonicera fragrantissima.
Degassed and filtered (0.2 µm) water was used as a perfusion solution in all hydraulic measurements. To avoid microbial growth, perfusion solutions were changed every 3 d. Stems were first perfused under elevated pressure (150–175 kPa) to remove any embolisms, following which flow rates were gravimetrically measured at 5–20 kPa (Dryden & Van Alfen 1983). For each sample, a plot of flow versus pressure was constructed using at least three independent measurements taken under different pressures. Hydraulic conductivity (Kh) was calculated as the slope of the pressure–flow plot multiplied by segment length. Data from these measurements were used to calculate leaf-specific conductivity (KL, which equals Kh divided by leaf area distal to the segment). Total leaf area distal to each measured stem was determined with a Li-Cor 3100 area meter (Li-Cor Inc., Lincoln, NE, USA). From the stems measured for Kh, xylem cross-sectional area was determined using an Olympus AX70 light microscope (Olympus America Inc., Melville, NY, USA). Measurements from stem cross sections were used to calculate xylem-specific conductivity (KS, which equals Kh divided by total xylem cross-sectional area). Using these data, Huber values (HV, xylem cross-sectional area divided by leaf area distal to the segment) were also calculated to compare patterns of sapwood to leaf area allocation.
Gas exchange measurements were made on 16 Flaveria species with a Li-Cor 6400 open gas exchange system (Li-Cor Inc.). Measurements were started at 1000 h (mid-morning) and lasted until 1600 h (mid-afternoon). All measurements were conducted at a leaf temperature of 30 °C, a photosynthetic photon flux density (PPFD) of 1500 µmol photons m−2 s−1, 370 µbar ambient CO2 and leaf-to-air water vapour concentration gradient (ΔW) of 20 mmol H2O mol−1 air. This corresponded to 50–60% relative humidity in the cuvette depending on the photosynthetic type and species. WUE (mmol CO2 mol−1 H2O) was calculated as WUE = [(Ci/Ca) − 1]Ca/1.6ΔWwhere Ci is the intercellular partial pressure of CO2 and Ca is the ambient partial pressure of CO2 (Farquhar & Sharkey 1982).
The vulnerability of xylem to cavitation was evaluated for two C3 species (F.pringlei, F.robusta) and two C4 species (F.bidentis, F.trinervia). For each species, 15–25 plants were sampled to generate a vulnerability curve (Pockman & Sperry 2000). Because of the large numbers of plants needed and the limited growth space, we could only measure xylem vulnerability on a few species. Consequently, we compared the two C3 and C4 species with the most similar growth form. Xylem embolism was induced by withholding water from potted plants or by air seeding under positive pressure with a pressure chamber (PMS Instruments, Corvallis, OR, USA). If embolism was induced by dehydration, xylem pressure potential was measured on three different leaves or branches and values were averaged for each replicate plant using the pressure chamber (Sperry, Donnelly & Tyree 1988). If air seeding was used, samples were inserted into the pressure chamber and the pressure was increased by approximately 0.01 MPa s−1 until the desired pressure was reached (Cochard, Cruiziat & Tyree 1992). Samples were maintained under this pressure for 15–20 min and then the pressure was released by the same rate at 0.01 MPa s−1. Samples were kept overnight in double plastic bags to reach pressure equilibrium. On the following day, stem segments were cut under water and initial hydraulic conductivity was measured at 4–8 kPa. Stem segments were then perfused under elevated pressure (up to 175 kPa) for 30 min until maximum conductivity was reached (Sperry et al. 1988). At least three initial and three maximum conductivities were measured for each sample at different pressures.
Pressure values were plotted against flow for initial and maximum conductivities in each sample. A linear relationship was present between initial measurements if refilling did not occur during measurements. A curvilinear regression would result if some refilling occurred. In this case, only the measurements taken under lower pressures before refilling were used for measuring loss of Kh. The percentage loss of conductivity was determined as
Stem xylem anatomy
Anatomical features were measured for nine Flaveria species (two C3, two C3–C4, two C4-like, three C4). Stem sections were prepared and embedded in paraffin from stem segments used for hydraulic conductivity measurements as previously described in Kocacinar & Sage (2003). All observations on stem cross sections were made using an Olympus AX70 light microscope (Olympus America Inc.). Each stem segment was subsampled in three randomly chosen sectors. Together, these subsamples represented at least 20% of the stem xylem area. All xylem conduits in each section subsample were measured with Image Pro Plus software (Media Cybernetics, Silver Spring, MD, USA) to determine vessel mean diameter (VMD) and maximum vessel diameter (MVD). From these data, vessel frequency (VF, the number of vessels per unit xylem area) was calculated. The mean diameter of conduits that facilitate 95% of the flow (D95) was obtained as described by Pockman & Sperry (2000). Hydraulic mean diameter (HMD) was estimated as HMD = 2 (Σr5/Σr4), where r is the radius of a conduit (Sperry et al. 1994).
Further measurements included the conduit efficiency (CE), which represents the capacity of the wood anatomy to hydraulically supply the leaf canopy (Kocacinar & Sage 2003). This value was obtained by dividing the sum of the fourth power of all conduits extrapolated to the whole xylem area with the leaf area supplied by these conduits. Leaf-specific porosity (LSP) represents the porosity of the xylem indexed to the leaf area and was calculated as the total vessel lumen area divided by total leaf area (Kocacinar & Sage 2003). To obtain vessel lengths, three to six stems per species were infiltrated using the paint-infusion method (Zimmermann & Jeje 1981; Kocacinar & Sage 2003). All paint-filled vessels were then counted every 2 cm, and these were used to estimate conduit lengths.
Data for all Flaveria species were grouped into six photosynthetic categories (C3, type I C3–C4, type II C3–C4, C4-like, perennial C4 and annual C4). This does not represent common evolutionary lineages in Flaveria (see McKown et al. 2005), but is useful for interpreting overall patterns among photosynthetic types. Data were statistically tested using SigmaStat software (SYSTAT Software, Richmond, CA, USA). Mean values of Kh, KS, KL, WUE and the Huber value were compared for each photosynthetic category using a one-way analysis of variance (anova) followed by Tukey's or Dunn's (when equal variance failed) post hoc tests. Mean species values for KL and WUE were analysed for significant correlation using the least-squares regression routine in the SigmaPlot programme (Systat Software, Point Richmond, CA, USA).
The relationship between KL and WUE was re-tested with correction for species relationships using Mesquite v. 2.0 software (Mesquite Software Inc., Austin, TX, USA) (Maddison & Maddison 2006). This method calculated ‘phylogenetically independent contrasts’ for each node in the phylogenetic tree by estimating the difference between trait values of each species and/or nodes and by correcting these differences for distance between species and/or nodes in the phylogeny (Felsenstein 1985; Midford, Garland & Maddison 2005). In this study, distance was based on branch lengths from a three-gene marker analysis (McKown et al. 2005). Contrasts were generated using the independent contrast PDAP:PDTREE module v. 1.09 (Midford et al. 2005) with Mesquite. The Pearson product–moment correlation test was used to test for significant correlation between KL contrasts and WUE contrasts. A least-squares regression of the scatterplot of the independent contrast pairs (KL and WUE contrasts) was generated to report the relationship between both variables corrected for phylogenetic history. A second analysis using Mesquite reconstructed KL and WUE variables on the Flaveria phylogeny and predicted ancestral states of these traits (simulated by parsimony).
Hydraulic properties and water use efficiency
Stem hydraulic conductivity (Kh) was two to five times greater in the C3 and annual C4Flaveria species than in the C3–C4, C4-like and perennial C4Flaveria species (Fig. 2a, Table 2). Little difference in Kh was observed among the C3 species, whereas the annual C4 species had almost fourfold greater Kh than the perennial C4F.kochiana (Table 2). Trends in stem xylem-specific conductivity (KS) were similar to Kh (Fig. 2b, Table 2). KS was greatest in the C3 species and the annual C4F.bidentis, and lowest in the C4-like species. The intermediate and perennial C4 species had KS values that were in between the C3 and C4-like species. Leaf-specific conductivity (KL) showed significant differences between C3 species and the intermediate and C4 species (Fig. 2c, Table 2). The C3 species had the greatest KL values, while the C4-like species and perennial C4F.kochiana had the lowest KL values. C3–C4 species and annual C4 species had similar KL values that fell between the C3 and the perennial C4 and C4-like species. Mean KL was 1.7 times higher in the C3 species than in the annual C4 species, 3.1 times greater in the C3 species than in the perennial C4F.kochiana and 5.5 times greater in the C3 species than in the C4-like species. Huber values varied little between the groups of species, with the exception of the type II intermediates which had higher values than the C4-like and C4 species (Table 2).
Table 2. Stem hydraulic parameters, water use efficiency (WUE) and Huber values (HV) of the C3, type I C3–C4, type II C3–C4, C4-like, perennial C4 and annual C4 groups of Flaveria species
Kh (kg m s−1 MPa−1 × 10−5)
KS (kg m−1 s−1 MPa−1)
KL (kg m−1 s−1 MPa−1 × 10−4)
WUE (mmol CO2 mol−1 H2O)
Values are means ± SE. Means for each group were calculated from the pooled raw values measured for each species, using data from Figs 2, 3 and 9. See Table 1 for the photosynthetic group designation. Letters represent statistical groupings at P < 0.05 determined using one-way analysis of variance (anova) and post hoc Tukey's test, or when the data failed an equal variance test, Dunn's test.
C3 species and both types I and II intermediates had WUE that were about half the values observed in the C4-like and C4 species (Fig. 3, Table 2). Unlike stem hydraulic properties, WUE in Flaveria species did not vary between annuals and perennials.
Xylem vulnerability to cavitation
C3Flaveria species lost 50% of conductivity at less negative xylem pressure potential than the C4 species (Fig. 4a,b). C3F.robusta lost 50% Kh at −3.3 MPa and C3F.pringlei at −4.5 MPa, compared with the C4F.trinervia and C4F.bidentis, each of which lost 50% conductivity near −5 MPa.
Stem anatomical properties
Stems of the C3 species tended to have wider vessels than the intermediate and C4 species (Fig. 5, Table 3). Vessel frequency (VF) was also lowest in the C3 species compared with the other Flaveria species. Vessels were widest and least frequent in C3F.pringlei compared with all other Flaveria species. Among C3F.robusta, intermediate species and C4 species, C4F.bidentis had the widest vessels and the C4-like species the narrowest vessels. Among the intermediate, C4-like and C4 species, the highest vessel frequency (VF) was observed in C3–C4F.sonorensis and the lowest VF in the C4-like F.brownii and in the C4F.australasica and F.trinervia; however, there were no general trends distinguishing the intermediate species from the C4 species.
Table 3. Hydraulic and anatomical parameters for stems of C3, C3–C4, C4-like and C4Flaveria species
VF (no. of vessels/unit xylem, mm−2)
CE (m2) × 10−14
LSP × 10−5
Values are means (±SE) of five to nine samples for each species.
CE, conduit efficiency value; D95, mean diameter of conduits that account for 95% of the flow; HMD, hydraulic mean diameter; LSP, leaf-specific porosity; MVD, maximum vessel diameter; MVL, maximum vessel length; VF, vessel frequency; VMD, vessel mean diameter.
43.5 ± 1.1
79.7 ± 1.9
86 ± 6
56.3 ± 0.9
50.3 ± 1.1
8.28 ± 0.4
2.39 ± 0.1
14.3 ± 1.8
29.9 ± 1.1
65.7 ± 1.5
105 ± 5
43.9 ± 0.9
38.7 ± 0.9
3.01 ± 0.5
1.48 ± 0.2
13.3 ± 1.5
Type I C3–C4
27.3 ± 1.1
60.1 ± 1.9
149 ± 10
36.3 ± 1.5
32.3 ± 1.6
3.47 ± 0.4
2.43 ± 0.2
7.8 ± 1.4
Type I C3–C4
26.2 ± 0.7
60.3 ± 3.0
168 ± 8
39.2 ± 0.7
33.4 ± 0.9
2.80 ± 0.4
1.75 ± 0.2
24.2 ± 3.5
22.7 ± 0.7
42.4 ± 2.5
132 ± 9
28.1 ± 0.7
25.9 ± 0.6
0.88 ± 0.1
0.99 ± 0.1
9.8 ± 2.0
20.7 ± 0.7
48.4 ± 3.3
155 ± 11
29.6 ± 1.2
25.4 ± 0.9
0.75 ± 0.1
0.85 ± 0.1
7.4 ± 0.9
26.6 ± 1.3
63.4 ± 2.2
133 ± 9
37.4 ± 0.9
32.8 ± 1.6
2.21 ± 0.3
1.47 ± 0.2
33.0 ± 0.8
66.1 ± 1.3
154 ± 10
42.9 ± 0.5
38.5 ± 0.8
3.94 ± 0.2
1.95 ± 0.1
8.6 ± 0.7
26.4 ± 1.2
55.4 ± 1.7
130 ± 5
36.4 ± 1.9
32.6 ± 1.6
2.82 ± 0.5
1.87 ± 0.2
6.6 ± 0.5
Values for the hydraulic mean diameter (HMD) and the mean diameter of conduits that account for 95% of the flow (D95) show the same trend as VMD in all Flaveria species (Table 3). C3F.robusta, C3–C4 species and C4 species generally had similar HMD and D95 values, and the C4-like species had the lowest HMD and D95 values. Conduit efficiency (CE) was more than twice as great in C3F.pringlei compared with the other Flaveria species (Table 3). CE was substantially lower in both C4-like species compared with C3F.robusta, C3–C4 or C4 species. In contrast to the trends in CE, leaf-specific porosity (LSP) was not as substantially different among Flaveria species of different photosynthetic types; however, similar to CE, the lowest LSP values were observed in the C4-like species (Table 3).
The maximum vessel length (MVL) was observed in C3–C4F.sonorensis and was far greater than any MVL observed in the other Flaveria species (Table 3). With the exception of C3–C4F.sonorensis, the C3 species had greater MVL than the intermediate and C4 species. There were no appreciable differences in MVL between intermediate and C4 species (except for F. sonorensis). Frequencies of different vessel lengths showed that in the C3 species, the majority of vessels was 6 cm or less (Fig. 6). In contrast to this, vessels in the intermediate and in the C4 species were generally shorter with over 80% of the vessels being 4 cm or less and at least 60% of the vessels falling in the 2 cm class. Type I C3–C4F.sonorensis showed a substantially different pattern compared to all the other Flaveria species and had a larger range of vessel lengths.
Correlation of KL and WUE
KL and WUE are negatively correlated (P < 0.001, Fig. 7a). With correction for phylogenetic relationships among Flaveria species, these variables remained correlated as indicated by the significant slope of the regression of WUE contrasts on KL contrast (P < 0.001, Fig. 7b). Reconstruction of these variables on the phylogenetic tree using parsimony estimations shows that WUE and KL changed together in the evolutionary diversification of C4-like and C4 species from C3–C4 ancestors; however, changes in KL without concurrent changes in WUE are apparent in the transition from C3 species to the C3–C4 intermediate species (Fig. 8).
It has long been hypothesized that an evolutionary trade-off exists between hydraulic safety and efficiency (Tyree & Sperry 1989). Hypothetically, the optimal balance point between safety and efficiency reflects habitat variation with drier habitats favouring safety traits, and wetter, resource-rich environments favouring efficiency traits that allow for sustained carbon gain and competitive success. Increased WUE would allow a hypothetical optimum to shift towards a reduced KL by either increasing hydraulic safety or enhancing leaf area per stem (Kocacinar & Sage 2005). Increasing hydraulic safety would be of value in dry habitats, while increased leaf area per stem would be advantageous in competitive situations.
In general, the results from this study examining Flaveria are consistent with this ‘WUE trade-off’ hypothesis. All C4 and C4-like Flaveria species had higher WUE and lower KL than the C3Flaveria species, with the differences being greatest between C3 perennials versus C4 and C4-like perennials. Compared with the C3 species, the C4 and C4-like plants had proportionally more of their vessels in the shorter-size classes and had smaller maximum vessel lengths. Perennial C4-like species also had narrower vessels and lower conduit efficiency values than C3Flaveria species. We also observed that two C4 species exhibited 50% conductivity loss at more negative xylem pressure potential than two C3 species, indicating greater cavitation resistance is also present in the C4Flaveria species. These results are in accordance with results of previous comparisons with 35 C3 and C4 species of close taxonomic affinity, ecological habitat or growth form (Kocacinar & Sage 2003, 2004). KL was consistently lower in the C4 species of a similar life form and taxonomic group, and in woody species, KS was generally greater in the C3 species. Associated with the differences in KL, C4 species in these previous comparisons often exhibited vessels that were similar to, or narrower and shorter than, the C3 species in a comparison group (Kocacinar & Sage 2003, 2004). Wood density, a trait closely associated with increased hydraulic safety (Hacke et al. 2001), is often greater in woody C4 than in C3 shrub species as well (Kocacinar & Sage 2004). In sum, the results from Flaveria and the earlier studies by Kocacinar & Sage (2003, 2004) demonstrate strong support for the hypothesis that the C4 photosynthetic pathway leads to lower KL and safer xylem anatomy.
In Flaveria, multiple origins of C3–C4 and C4-like intermediacy are associated with parallel evolution of stem hydraulic traits. For example, the type II intermediates F.anomala and F.ramosissima occur in independent lineages and exhibit similar KL and WUE values (Fig. 8). F. brownii and F.vaginata represent at least two independent origins of C4-like photosynthesis and show similar conductivities and anatomical traits, including low KL, high WUE and small conduit diameters. A comparison of closely related species with different photosynthetic pathways is also possible (Fig. 8). In clade B, C4-like F.brownii is closely related to the C3–C4F.floridana, and both species have similar morphology and grow in saline coastal areas (Powell 1978; McKown et al. 2005). The presence of C4-like characteristics in F.brownii corresponds to 75% lower KL and nearly twice the WUE as seen in F.floridana. A similar comparison between C3–C4F.ramosissima and the closely related C4-like F.palmeri in clade A shows that a 60% lower KL is associated with twice the WUE in the C4-like species. (These WUE and KL differences between F.brownii and F.floridana, and F.ramosissima and F.palmeri, were significant at P < 0.05 by a one-way anova and Tukey's post hoc tests comparing all Flaveria species.)
The mechanism by which C4 photosynthesis promotes reductions in KL is hypothesized to be the greater WUE in the C4 species. The main supporting evidence for this hypothesis is the large, parallel increases in WUE that accompanied the large reductions in KL in the transition from C3–C4 to C4-like species. There is, however, a significant decline in KS and KL in the transition from C3 to the C3–C4 species that is not associated with WUE enhancement. Among the C3–C4 species, both KS and KL fall between the values observed in the C3 and C4-like species; however, WUE in the C3–C4 intermediate species is similar to the values of the C3 species. This reduction in KL in the C3–C4 intermediates may be an evolutionary vestige from a KL shift in a common ancestor, or it may result from adaptation by the intermediates to a common and possibly more extreme set of conditions than what their C3 relatives experience. Alternatively, KL reduction may somehow have functional significance for C3–C4 intermediacy.
The evolutionary vestige hypothesis is not supported by the results, because all C3 species show a higher KL than each independent line of C3–C4 intermediacy. For example, a lower KL relative to the C3 relatives is present in F.ramosissima (the basal intermediate in clade A), F.angustifolia (the basal intermediate in clade B) and F.sonorensis, an intermediate that is independently derived from an F.robusta-like ancestor (Fig. 8; McKown et al. 2005). Adaptation to a common habitat is also not supported at this time, because habitat aspects that would favour lower KL relative to C3 ancestors are not obvious. All Flaveria intermediates are summer active in hot deserts of North America, disturbed sand surfaces along the Caribbean coast or in semi-arid monsoon regions of Mexico (Powell 1978; Monson 1989; McKown et al. 2005; Sudderth et al. 2008). F.sonorensis is described by Powell (1978) as occurring near warm mineral springs in semi-arid regions of subtropical Mexico. F.angustifolia occurs on disturbed, sandy or loamy soils in sclerophyllous scrub zones of south-central Mexico, and F.ramosissima occurs on disturbed, sandy-clay or rocky soils along dry washes, roads and fields in thorn-scrub regions of southern Mexico (Powell 1978; Sudderth et al. 2008). All three of these species occur in similar regions and habitat types as their C3 relatives, and there are no obvious climate or site differences to indicate more severe growing conditions for the intermediates (Sudderth et al. 2008). The C3 species F.robusta, for example, grows on gypseous soils in the semi-arid hills and disturbed fields along the Pacific coast of central Mexico, while F.pringlei and F.cronquistii are C3 perennials growing on gypseous soils and disturbed roadsides in the same region in south-central Mexico where F.ramosissima and F.angustifolia occur (Powell 1978; Sudderth et al. 2008). These environments are episodically harsh enough that a low KL and any associated increase in xylem safety may be adaptive, but this would be the case for both C3 and C3–C4 species. It is possible the C3–C4 species are somehow specialized for more severe microclimates within these habitats; however, in a recent examination of their site characteristics and ecophysiology, Sudderth et al. (2008) could find no obvious functional differences between the C3 species and C3–C4 species occurring in a common region of south-central Mexico. Thus, there is little evidence to support habitat difference as a cause for the lower KL in the C3–C4 intermediates.
The final hypothesis, that the low KL is functionally linked to C3–C4 intermediacy, is an intriguing possibility. C3–C4 intermediacy is thought to be an evolutionary adaptation to high rates of photorespiration, as would occur in hot environments where intercellular CO2 levels are depressed (e.g. from low stomatal conductance relative to photosynthesis and low atmospheric CO2; Sage 2004). The functional trait consistently associated with C3–C4 intermediacy is localization of glycine decarboxylase into the bundle sheath cells and shuttling of photorespiratory metabolites to the bundle sheath for decarboxylation (Monson 1999). The released photorespiratory CO2 is then trapped in the bundle sheath, where it accumulates and supports high efficiency of CO2 re-fixation by bundle sheath Rubisco. C3–C4 intermediacy is thus a modest CO2-concentrating mechanism that enhances net photosynthesis under photorespiratory conditions (Monson 1989, 1999; Vogan et al. 2007). Relative to C3 plants, C3–C4 intermediate species have their greatest photosynthetic advantage at lower intercellular CO2 levels and hot conditions (Vogan et al. 2007). Consistently, most are found in semi-arid to arid environments of low latitudes where stomatal limitations can be substantial. Schuster & Monson (1990), for example, observed that the C3–C4F.floridana has twice the photosynthetic capacity as the C3F.cronquistii at 40 °C, while at 30 °C, the C3–C4 advantage was only 1.5 times as great. Given this understanding, we hypothesize that the lower KL of the intermediates contributes to the C3–C4 advantage by restricting water flow to the leaf canopy during periods of high potential evapotranspiration. It is possible the low KL reflects the evolution of a safe xylem to protect the intermediate species from hydraulic failure during conditions of extreme evapotranspiration (e.g. above 40 °C, when vapour pressure differences between leaf and air can exceed 6 kPa). A low KL would favour stomatal closure (Sperry 2000), leaf warming and more photorespiration, which in turn could increase the selective advantage of glycine localization to the bundle sheath cell. Higher photorespiration and the resulting increase in glycine production would then lead to greater CO2 production and greater activity of Rubisco in the bundle sheath. In short, a reduced KL may strengthen natural selection for a photorespiratory CO2 pump.
Plants in hot or dry environments often have a greater ratio of conducting xylem to leaf area (a higher Huber value), thereby enhancing the capacity of the vascular system to supply leaves with water (Choat et al. 2008). Among the Flaveria species, the Huber values show little relationship to KL (Fig. 9a). By contrast, KS is significantly correlated with KL (Fig. 9b), indicating that qualitative changes in xylem structure rather than altered allocation to leaf area are the primary drivers of the changes in KL in most Flaveria species. This observation is consistent with the observed shift in anatomical features from long, wide vessels in C3Flaveria to short, narrow vessels in C4-like and two C4Flaveria species. The exception to this pattern is observed in the weedy C4 annual F.bidentis, which has a C3-like value of KS but reduced KL, and thus stands as an outlier in the KL versus KS plot (Fig. 9b).
Comparisons of weedy C3 and C4 annuals in the Chenopodiaceae, Amaranthaceae and Aizoaceae showed each photosynthetic type to have similar xylem characteristics in terms of KS and xylem anatomy, but the C4 species had greater leaf area per stem and thus lower KL than the C3 annuals within a taxonomic family (Kocacinar & Sage 2003, 2005). In the resource-rich sites where these weeds occur, allocation to greater leaf area per unit xylem was interpreted by Kocacinar & Sage (2003) to be more advantageous for the C4 annuals than increasing hydraulic safety. Among the Flaveria species, the annual C4 species (F.australasica, F.bidentis, F.trinervia) have the widest distribution as pan-tropical weeds, compared with more limited distributions of C3 and C3–C4 intermediate Flaveria species (Powell 1978; McKown et al. 2005; Sudderth et al. 2007). As noted by Kocacinar & Sage (2003) for C3 and C4 weeds, we propose that F.bidentis exploits higher WUE by increasing canopy photosynthesis through greater leaf area. By contrast, the annual F.trinervia shows a reduction in KS that parallels the reduction in KL, indicating that it exploits the greater WUE by modifying the xylem, rather than by increasing canopy area per stem.
Hydraulic safety and resistance to cavitation has often been suggested to directly reflect resistance to air seeding by the pit membranes (Tyree & Sperry 1989; Choat et al. 2008). If true, then decreases in KL are hypothesized to increase the tension at which 50% of the hydraulic conductivity is lost. Our results with four Flaveria species, and previous vulnerability curves comparing C3 and C4 annuals (Kocacinar & Sage 2003), are consistent with a trade-off between KL and 50% loss of conductivity, but the support is limited by the similarity of the vulnerability curves between F.pringlei and the two measured C4 species. Recent meta-analyses and phylogenetically independent contrasts in woody C3 species generally show that the trade-off between KL and 50% loss of stem conductivity is weak, if present at all (Maherali et al. 2004; Bhaskar et al. 2007). Weakness in this relationship raises some doubt regarding the link between KL and hydraulic safety, and thus whether an increase in WUE might alter selection pressure on xylem efficiency and safety. In defence of the hypothesis that increased WUE allows for greater hydraulic safety in C4 taxa, we note that hydraulic safety is more than resistance to air seeding at the pit membrane. Resistance to cavitation also includes increased conduit redundancy, hydraulic capacitance, xylem density, stomatal control and root-to-shoot allocation, each of which can reduce the likelihood that xylem tension will reach the point of catastrophic xylem failure (Tyree & Zimmermann 2002; Pratt et al. 2007). The low KL values in the C3–C4 intermediates and C4Flaveria species are associated with enhanced redundancy via shorter and often narrower vessels. We therefore hypothesize that the safety features that lower KL enhance fitness in the C4-like and C4Flaveria species by reducing the chance that xylem tension will approach values where the catastrophic xylem failure occurs.
In summary, the results demonstrate an inverse relationship between WUE and hydraulic efficiency in Flaveria. Our results with the multiple lineages of C3–C4 intermediates also show that KL changes precede the evolution of enhanced WUE, indicating KL reduction is somehow associated with CO2 concentration via a photorespiratory CO2 pump. The substantial changes observed in KL and WUE in the distinct transitions from C3–C4 to C4-like photosynthesis strongly indicate that WUE enhancement further contributes to a sizable reduction in KL, consistent with our hypothesis that the C4WUE advantage allows for greater xylem safety or canopy allocation.
Our results from Flaveria are significant for understanding how large WUE differences may have influenced xylem evolution in C3 plants. In particular, these results may help explain changes in xylem anatomy in the fossil record that have occurred during periods of substantial atmospheric CO2 decline. In the past 100 million years, many angiosperms lineages evolved xylem features indicative of greater hydraulic efficiency, notably, wider vessel elements with open-end walls (Wheeler & Baas 1991; McElwain, Willis & Lupia 2005). These trends parallel a large reduction in atmospheric CO2 from above 1000 mg g−1 before 60 million years ago to below 350 mg g−1 25 million years ago (Zachos et al. 2001; Pagani et al. 2006), leading to speculation that low CO2 promoted the evolution of more efficient xylem (Kocacinar & Sage 2004; McElwain et al. 2005). In modern species, plants grown at a range of CO2 levels that are equivalent to the variation observed in the past 60 million years show similar differences in WUE as C3 and C4 species (Kimball & Bernacchi 2006). If the long-term changes in atmospheric CO2 resulted in equivalent WUE changes as seen experimentally, then the results from Flaveria would support the hypothesis that declining atmospheric CO2 reduced WUE, and may have contributed in part to selection for more efficient xylem features. In the future, intrinsic WUE will again increase in C3 species with the rise in atmospheric CO2 (Baker et al. 1990; Kimball & Bernacchi 2006). It is logical to hypothesize that species and genotypes with greater hydraulic safety may be pre-adapted to better exploit future high CO2 atmospheres, as they might realize the benefits of greater hydraulic safety while avoiding limitations on carbon gain caused by reduced hydraulic efficiency. Such hypotheses are easily testable, and should be examined in the near future as hydraulic properties have an important control over biogeographic distributions.
We thank Thomas Fung, Mathew Lam and Kathy Sault for their assistance with the anatomy and microscopy. We also thank Dr Taylor Feild and Lawren Sack for valuable advise on the project. This research was supported by a scholarship from the Ministry of Education of Turkey and TUBİTAK to F. Kocacinar and by National Science and Engineering Research Council (NSERC) grant no. OGP0154273 to R.F. Sage.