Structure–function constraints of tracheid-based xylem: a comparison of conifers and ferns


Author for correspondence:
Jarmila Pittermann
Tel: +1 831 459 1782


  • The ferns comprise one of the most ancient tracheophytic plant lineages, and occupy habitats ranging from tundra to deserts and the equatorial tropics. Like their nearest relatives the conifers, modern ferns possess tracheid-based xylem but the structure–function relationships of fern xylem are poorly understood.
  • Here, we sampled the fronds (megaphylls) of 16 species across the fern phylogeny, and examined the relationships among hydraulic transport, drought-induced cavitation resistance, the xylem anatomy of the stipe, and the gas-exchange response of the pinnae. For comparison, the results are presented alongside a similar suite of conifer data.
  • Fern xylem is as resistant to cavitation as conifer xylem, but exhibits none of the hydraulic or structural trade-offs associated with resistance to cavitation. On a conduit diameter basis, fern xylem can exhibit greater hydraulic efficiency than conifer and angiosperm xylem.
  • In ferns, wide and long tracheids compensate in part for the lack of secondary xylem and allow ferns to exhibit transport rates on a par with those of conifers. We suspect that it is the arrangement of the primary xylem, in addition to the intrinsic traits of the conduits themselves, that may help explain the broad range of cavitation resistance in ferns.


The evolution of tracheid-based xylem in the Lower Devonian led to profound shifts in plant size and structure, and marked the first appearance of tracheophytes, the so-called true vascular plants (Pittermann, 2010; Kenrick & Crane, 1997; Niklas, 1992; Bateman et al., 1998; Sperry, 2003). Tracheids preceded the widespread appearance of vessels by an estimated 150 million yr and served as the fundamental water transport tissue for some of the earliest land plants, including Rhynia and Psilophyton, horsetails, ferns and the extinct arborescent lineages of the Late Devonian such as the lycopod Lepidodendron and pro-gymnosperms such as Archaeopteris (Cichan, 1985; Stewart & Rothwell, 1993; Taylor et al., 2009). Generally, the evolution of tracheids is characterized by increasing length and diameter (particularly during the Devonian), greater deposition of secondary cell wall material and progressive specialization in the inter-tracheid pit membranes, the complexity of which peaked with the appearance of the torus-margo pit membrane found in conifers, Gingko and some angiosperms (Niklas, 1985; Sperry, 2003; Pittermann, 2010; Pittermann et al., 2005; Jansen et al., 2004). The subsequent specialization of tracheids into fibers and vessels that characterized the evolution of angiosperm wood allowed for a division of labor whereby short, narrow fibers provide mechanical support while multicellular vessels function solely for water transport (Bailey & Tupper, 1918; Carlquist, 1988). Although vessels may confer water transport efficiencies that are well over three orders of magnitude greater than those of conifers (Tyree & Zimmermann, 2002; McCulloh et al., 2010), it is remarkable that tracheid-based xylem continues to serve as the primary transport tissue for two abundant and diverse plant lineages, the conifers and the ferns. We know that, on a xylem area basis, conifers and angiosperms can exhibit similar hydraulic efficiencies (Pittermann et al., 2005), but how does the performance of the tracheid-based xylem of ferns compare with the more derived xylem of conifers?

Despite the extraordinary diversity and world-wide abundance of terrestrial and epiphytic ferns (Moran, 2008; Schuettpelz & Pryer, 2009), our understanding of the vascular performance of these primitive plants is just gaining momentum (Calkin et al., 1985; Veres, 1990; Brodribb et al., 2005; Watkins et al., 2010). Early work on water transport in several fern species showed that the main axis of the frond exhibits progressively lower hydraulic conductivities along its length as a result of a decrease in conduit abundance and conduit size, especially from the start of the leafy rachis to the tip of the frond (Gibson et al., 1985; Schulte et al., 1987). These seemingly low rates of water transport were again reported in a broad sampling of tropical pteridophytes, a finding that was mirrored in the concurrently low rates of gas exchange (Brodribb & Holbrook, 2004, Brodribb et al., 2007, Watkins et al., 2010). Low transpiration rates are consistent with the preference of these tropical ferns for the low-light habitats characteristic of forest understories and dense canopies: because understory plants may only experience brief photosynthetic peaks during sunflecks, selection places a lower premium on the evolution of high vascular and gas-exchange capacity in favor of a smaller sized, nonwoody, slow-growing life form with reduced metabolic costs. There are, however, some exceptions to these generalizations, most notably in the form of tree ferns, desert-dwelling ferns and the many temperate species that appear to thrive in a broad variety of temperate high-light habitats, such as Pteridium aquilinum and Blechnum chilense (Page, 2002; Saldana et al., 2007). Considering that several species of ferns, such Lygodium microphyllum and P. aquilinum, can be highly invasive (Robinson et al., 2010), it is not unreasonable to hypothesize that ferns are capable of high rates of water transport and photosynthesis.

The vascular system in both fern fronds and rhizomes consists of tracheids that tend to be longer and wider than those of conifers, with scalariform pitting extending along the entirety of at least one side of the tracheid wall (Gibson et al., 1985; Veres, 1990; Carlquist & Schneider, 2001). Cryptic vessels have been reported in the rhizomes of several species on the basis of what appeared to be scalariform perforation plates in the terminal ends of vessel elements, but their frequency in the ferns is now presumed to be much lower than originally thought (Carlquist & Schneider, 2001, 2007). Because water transport in ferns occurs exclusively through primary vascular tissue, the xylem and phloem of ferns are encased in discrete bundles that may span the length of the frond and bifurcate at each pinna. The bundles are arranged in a variety of stelar patterns ranging from the simple protostele of whisk fern (Psilotum nudum) to the more complex siphonostele- and dictyostele-like arrangements found in more derived fern species. Secondary xylem is absent in extant ferns though extinct plants such as Medullosa, which bear some resemblance to tree ferns, produced secondary xylem from a polystelic arrangement of several bifacial vascular cambia (Cichan, 1986; Wilson et al., 2008; Taylor et al., 2009). Interestingly, a recent examination of two Botrychium species excluded the possibility that these ferns exhibit true cambial-derived secondary growth, although the developmental pattern of their rhizomes is regarded as a departure from standard definitions of primary and secondary growth (Rothwell & Karrfalt, 2008).

Tracheid-based xylem is common to both conifers and ferns, but key differences in xylem architecture have a profound effect on the overall structure of these plants, as well as the physical principles that guide the shape and size of the xylem conduits. Most importantly, the evolution of a bifacial vascular cambium and the resultant secondary xylem in conifers and woody angiosperms marked a radical departure in the evolution of xylem function as well as overall plant structure because it allowed plants to have the architectural flexibility to vary the height and horizontal display of their foliage, an otherwise impossible endeavor in plants limited by a unifacial cambium or simple strands of primary xylem (Rowe & Speck, 2005; Spicer & Groover, 2010). In conifers and pro-gymnosperms, the xylem acquired the capacity not only to transport water to the leaf canopy, but also to structurally support it (Meyer-Berthaud et al., 1999; Tyree & Zimmermann, 2002; Pittermann et al., 2006a,b; Sperry et al., 2006; Pittermann, 2010).

Recent work has shown that, in north-temperate conifers, the combined requirements for canopy support along with reinforced, implosion-resistant tracheids constrain the maximum hydraulic efficiency of conifer xylem because of the necessity to build a strong, secondary cell wall coupled with a narrower lumen diameter (Pittermann et al., 2006a; Sperry et al., 2006). Both the cell size and the volume of the conduit wall are limited by the metabolic output of the developing xylem cell over the growing season, so it is impossible for conifer tracheids to be long and wide and also sufficiently fortified to offer structural support (Pittermann et al., 2006a; Sperry et al., 2006). By contrast, fern xylem is released from the structural support requirement by virtue of the hypodermal sterome, a ring of schlerenchyma fibers that surrounds the main axis of the frond as well as the thicker secondary axes in some species (Rowe & Speck, 2004; Rowe et al., 2004). This tissue supports the frond and provides it with a high degree of flexural stiffness (Niklas, 1992; Rowe et al., 2004). Consequently, without a support function, fern tracheids can occupy a much broader morphospace in the form of longer and wider tracheids that have little need for exceptionally reinforced cell walls. However, fern tracheids lack the torus-margo pitting particular to conifer tracheids and instead possess the ancestral, homogenous pit membrane that confers much greater pit area resistance than the torus-margo arrangement, especially in derived angiosperms (Sperry et al., 2005; Pittermann et al., 2005; but see Hacke et al., 2007).

Without torus-margo pit membranes, secondary xylem and a canopy support function, just how does the transport efficiency of the tracheid-based xylem of ferns compare to that of conifers? The goal of this study was to examine the structure and function of fern xylem across a broad phylogenetic sampling scheme that includes self-supporting ferns, climbers, and perennial and seasonally deciduous ferns as well as the desert-dwelling ferns in order to capture the broadest possible variation in tracheid structure. Because water transport occurs under negative pressure and xylem conduits are vulnerable to the entry air (cavitation), we examined how the cavitation resistance of fern xylem compares to that of conifers and whether ferns exhibit any of the trade-offs associated with cavitation resistance such as reduced hydraulic efficiency and increased fortification of tracheids against implosion, as observed in north-temperate conifers (Pittermann et al., 2006a,b; Pittermann et al., 2010; Hacke et al., 2001; Sperry et al., 2006).

Materials and Methods

Plant material was gathered locally from campus forests at the University of California in Santa Cruz (UCSC) or from glasshouse collections at UCSC and UC Berkeley, USA. Fronds were collected from a minimum of four individuals (but only three individuals for the Lygodium species) from 16 species belonging to several families and exhibiting a variety of life history strategies, statures, and native habitats (Table 1). Species were sampled in May–August of 2009 and 2010.

Table 1.   A list of species used in this study and their collection sites, growth habit and native habitat features
SpeciesFamilySymbolCollection siteHabitEndemic habitatClimate
Adiantum capillus-veneris L.PteridaceaeACGlasshouse collectionErectNorth American warm-temperate forestsMesic
Asplenium bulbiferum G. ForstAspleniaceaeABGlasshouse collectionErectSouthern-hemisphere temperate forestsMesic
Cheilanthes bonariensis (Willd)PteridaceaeCBBotanical gardenErectCentral American desertsXeric
Dicksonia antarctica (Labill)DiscksoniaceaeDAGlasshouse collectionErectSouthern-hemisphere temperate forestsMesic
Dryopteris arguta (Kaulf.) WattDrypteridaceaeDRCoastal Redwood forestErectWestern North American temperate forestsMesic
Lygodium flexuosum (L.) Sw.LygodiaceaeLFGlasshouse collectionClimbingSoutheast Asia tropical forestsMesic
Lygodium japonica (Thunb.) Sw.LygodiaceaeLJGlasshouse collectionClimbingSoutheast Asia tropical forestsMesic
Microlepia strigosa (Thunb.) C. PreslDennstaedtiaceaeMSGlasshouse collectionErectHawaiian Island tropical and subtropical forestsMesic
Osmunda regalis L.OsmundaceaeORBotanical gardenErectEuropean North African, and Eastern North American bogs and swampsMesic
Polypodium aureum (L.) J.Sm.PolypodiaceaePAGlasshouse collectionErectNorth and South American tropical and subtropical forestsMesic
Polystichum munitum (Kaulf.) C. PreslDryopteridaceaePMCoastal Redwood forestErectWestern North American temperate forestsMesic
Psilotum nudum (L.) BeauvoisPsilotaceaePNGlasshouse collectionErectHawaiian Island tropical and subtropical regionsMesic
Pteridium aquilinum (L.) KuhnPteridaceaePQUCSC CampusErectTemperate and subtropical regionsMesic
Pteris cretica (L.)PteridaceaePCGlasshouse collectionErectSouthern and Northern Hemisphere temperate regionsMesic
Pyrrosia lingua (Thunb.) FarwellPolypodiaceaePLGlasshouse collectionErectAsian temperate regionsMesic
Woodwardia fimbriata Sm.BlechnaceaeWFGlasshouse collectionErectWestern North American temperate forestsMesic

Gas-exchange measurements

Gas-exchange measurements were performed on the same fronds used for the hydraulics and anatomical studies described below using the Li-Cor LI-6400XT portable gas-exchange system (Li-Cor Biosciences Inc., Lincoln, NE, USA) fitted with a standard opaque 2 × 3 cm LED chamber. All plants were watered to saturation the day before the gas-exchange measurements were collected such that the lowest midday water potentials were only −0.75 MPa, as measured using a standard pressure chamber (PMS Instruments, Corvallis, OR, USA). Each measurement was obtained on mid-rachis pinnae, requiring between 2 and 5 min for a stable reading, with leaf temperature at 21°C, the flow rate at 300 ml min−1 and the CO2 mixer set to 400 μmol m−2 s−1. The chamber humidity ranged from 70 to 80%, only slightly higher than ambient glasshouse humidity. For glasshouse-grown species, the minimum saturating light intensity was 500 μmol m−2 s−1, but it was increased to 700 μmol m−2 s−1 to ensure that the pinnae were indeed light saturated. Outdoor-growing, sun-exposed species such as Cheilanthes bonariensis and P. aquilinum required the equivalent of full sun (2000 μmol m−2 s−1) to achieve saturation, while the understory fern Woodwardia fimbriata saturated at 700 μmol m−2 s−1. An average of 10 measurements were made on glasshouse species (one to two measurements per frond), and that was doubled for outdoor-grown ferns.

Conifer gas-exchange data were collected in May to late June in 2009 and 2010 from 20 individuals from the Pinaceae, Cupressaceae, Podocarpaceae and Araucariaceae in the context of another project, and supplemented by data from the literature (Pittermann et al., 2006a,b). The gas-exchange measurements were performed by J.P. and E.L. on well-hydrated branches (the average midday water potential was −0.4 MPa) sampled from trees growing at the UC Santa Cruz Arboretum, San Francisco Botanical Garden at Golden Gate Park and the UC Botanical Garden in Berkeley, all locations in coastal, central California. The measurements were made as described in the previous paragraph, with a leaf temperature of 21°C, a flow rate of 300 ml min−1, 400 μmol m−2 s−1 ambient CO2, chamber humidity between 40 and 60% and a light intensity of 2000 μmol m−2 s−1. Six measurements were made on four to six stems collected from at least two trees, depending on availability at the local arboreta.

Hydraulic measurements

The stipes represent the leafless portion of the fern frond, and unlike the frond rachis, the total xylem area is generally invariable along this segment. Stipes were collected from the bottom 18–20 cm of the frond for a sample size of = 6–10. Stipe diameters ranged from 2 to 8 mm, with Lygodium sp. and Adiantum capillus-veneris possessing the narrowest stipes and P. aquilinum and W. fimbriata the widest. Stipe samples were re-cut under water to a length of 142 mm, and the distal ends shaved smooth with a razor blade. Although all plants were well hydrated before collection, any remaining embolism was removed by submersing the segments in distilled and filtered 20 mM KCl solution (0.22 μm; E-Pure filtration system; Barnstead International, Dubuque, Iowa, USA) and degassing overnight under vacuum. Stipes were degassed rather than flushed in order to minimize handling damage and wound effects arising from compressed cortex tissue.

Hydraulic conductivity (k) was measured according to the method of Sperry (1993) and calculated as the flow rate for a given pressure gradient standardized per unit of stem length. The stipes were mounted on a tubing apparatus where k was measured gravimetrically under a pressure of 6–8 kPa using filtered 20 mM KCl solution. The flow rate through the segments was determined without a pressure head before and after each gravimetric flow measurement. These background flows were averaged and subtracted from the pressure-induced flow in order to improve accuracy. Functional xylem area was measured on samples perfused for several hours with basic fuchsin whereby hand-cut cross-sections were photographed and analyzed using ImagePro software (Media Cybernetics, Carlsbad, CA, USA). Xylem specific conductivity (ks) and leaf specific conductivity (kL) represent k standardized for functional xylem area and distal leaf area, respectively.

Several workers have previously reported the presence of mucilage that often inconvenienced hydraulic measurements on ferns, but with the exception of P. aquilinum and D. antarctica we did not experience this problem. In the aforementioned species, the stipe ends were carefully shaved to expose the xylem, and following a mild −0.5 MPa spin in the centrifuge (see below), the mucilage was eliminated from the cut ends. All hydraulic measurements were made promptly to minimize artifacts caused by wounding effects.

We used the centrifuge method to determine species’ vulnerability to cavitation in response to a range of xylem pressures (Pockman et al., 1995; Alder et al., 1997). Stems were secured in a custom rotor designed to fit a Sorvall RC-5C centrifuge (Thermo Fisher Scientific, Waltham, MA, USA) and spun for 3 min at speeds that induce a known xylem pressure (Px). The per cent loss of conductivity (PLC) caused by centrifugation at each Px was calculated from the k measured after spinning, relative to the maximum conductivity after degassing (kmax) at Px = 0 Mpa, such that

image(Eqn 1)

where kmax was determined at Px = 0 MPa following degassing. The segments were spun to progressively more negative Px at −1 MPa increments until the PLC exceeded 90%, or alternatively until Px = −10 MPa, which is the most negative Px that can be achieved using the centrifuge. A Weibull function was used to fit the vulnerability curves from each stipe (Neufeld et al., 1992), and the xylem pressure at which segments exhibited a 50% loss of conductivity (P50) was computed as an average ± 1 SD per species.

Ferns such P. nudum and A. capillus-veneris had fronds that were either too short or too weak to endure centrifugation, so these species were subjected to the bench-dry method of obtaining vulnerability curves on at least 15 fronds. Intact fronds were cut and dehydrated down to a range of desired water potentials measured on mid-rachis pinnae, at which point the stipes were removed and the hydraulic conductivity and PLC calculations were made as described above. The data were fitted with a Weibull function that was used to compute a single P50..

Similarly, we used the bench-dry method to verify the validity of the centrifuge-generated PLC data on P. aquilinum, a species known to possess vessels longer than 142 mm in both the rhizome and the stipes (Carlquist & Schneider, 2007). Recent work on vines and ring-porous species suggests that the centrifuge method may overestimate vulnerability to cavitation (Choat et al., 2010; Cochard et al., 2010), but we found that, in P. aquilinum, the two methods yielded vulnerability curves that were indistinguishable from one another (data not shown). Similarly, we observed no differences in kmax on P. aquilinum stipes ranging in length from 14 to 22 cm, 22 cm approximating the length of the longest infrequent vessel. We would expect kmax in the 14.2-cm segment to be much higher if the removal of conduit end walls reduced a significant proportion of end-wall resistance.

Anatomical measurements

In north-temperate woody plants, the conduit double-wall-to-lumen-span ratio, (t/b)h2, is positively correlated with cavitation resistance and reflects the potential of the xylem conduits to withstand implosion caused by negative water potentials (Hacke et al., 2001). This metric has since been applied to woody plants across a variety of habitats (Pittermann et al., 2006a; Jacobsen et al., 2007). Conduit diameters and double-wall thickness:lumen span measurements were determined on xylem located in the center of the stipe previously used for hydraulic measurements according to Pittermann et al. (2006a,b) and Hacke et al. (2001). The sections were treated with phloroglucinol to highlight lignified tissue such as xylem, rinsed, and mounted in glycerin. All of the xylem was photographed under 200–400× magnification with a digital camera mounted on a Motic BA400 compound microscope (JH Technologies, San Jose, CA, USA). Because of the limited amount of xylem present in the stipes, it was possible to measure all conduits located in each cross-section. Conduit features were measured using ImagePro analysis software.

Tracheid lumen areas were converted to equivalent circle diameters and the hydraulic mean diameter was calculated from tracheid diameter distributions as Dh = ∑D5/∑D4 according to Kolb & Sperry (1999). All conduits were measured, with the sample size ranging from from 70 to 300 conduits depending on species xylem area. The thickness:span ratio was measured on adjacent conduits where at least one cell (although usually both) was within 10% of Dh. The double-wall thickness was determined from the shared walls of at least 40 and up to 100 of these cells.

Conduit length and diameter were measured on individual tracheids obtained from macerations according to the methods of Mauseth & Fujii (1994). Individual vascular bundles at least 5 cm in length were excised from the stipe and submersed in a 50 : 50 solution of 30% hydrogen peroxide and 80% glacial acetic acid (Sigma-Aldrich, St. Louis, MO, USA), and heated to at least 110°C for c. 2–3 d. ImagePro was used to measure at least 50 tracheids per species at ×200–400.

Anatomical data from conifer root and shoot xylem were collected from previously published work by Pittermann et al. (2006a,b), as well as unpublished data collected in 2008–2010 by J. P.

Conduit length measurements

We used the silicone injection method of Hacke et al. (2007) and Christman et al. (2009) to obtain conduit length distributions in species of ferns suspected to possess long conduits such as Lygodium flexuosum and P. aquilinum (Carlquist & Schneider, 2007). Five fronds of each species were injected basipetally from the base of the stipe with a silicone-fluorescent dye-hardener mix at a pressure of 50 kPa, and left overnight (Christman et al., 2009). The silicone was hardened over 3 d, after which the stipes were sectioned at regular intervals starting at 5 mm from the base of the injection site. In P. aquilinum, which is known to possess vessels, the increments were progressively increased from 5 mm to 5 cm. The fraction of silicone-filled conduits at each length increment was counted and the data analyzed according to Christman et al. (2009).


Conduit diameter frequency distributions show that with the exception of P. nudum and P. lingua, all surveyed stipes possess conduits that are wider than 40 μm, and average conduit diameters can vastly exceed even the largest maximum tracheid diameters (c. 60 μm) found in riparian conifer roots (Fig. 1; Pittermann et al., 2006a,b). Tracheids in excess of 100 μm in diameter were frequently present in the viney, indeterminately growing stipes of Lygodium flexuosum, a finding that is consistent with the need to maximize water transport to distal leaves by packing large conduits in narrow, 2.5-mm-diameter stipes with low xylem areas. The observed wide range in conduit diameter frequencies explains the significant standard deviations associated with mean and hydraulic tracheid diameter values reported in figures below.

Figure 1.

Examples of conduit diameter distributions in stipes belonging to six species of north-temperate and tropical terrestrial ferns with some of the narrowest and broadest ranges in conduit sizes (= 5–6 stipes ± SD). Additional data can be found in Supporting Information, Fig. S1 and Table S1.

We generated vulnerability curves on stipes from all species examined to determine how susceptible ferns are to drought-induced cavitation given the tremendous variation observed in conduit dimensions and xylem stele arrangements (Fig. 2). Fig. 3 shows vulnerability curves in six species that span a broad range of cavitation resistances, from a P50 of –0.68 MPa in P. nudum to a surprising −11.8 ± 7.6 MPa (mean ± SD) in Dryopteris arguta. In several fronds of D. arguta, the P50 was extrapolated from the Weibull fits assigned to each vulnerability curve. Surprisingly, several D. arguta stipes showed < 60% loss of hydraulic conductivity at −9 MPa but had a tendency to shred when subjected to −10 MPa xylem pressures in the centrifuge, so data at this pressure are not reported.

Figure 2.

Mid-segment cross-sections of six fern stipes, stained with toluidine blue, mounted in glycerol and photographed under a compound microscope. Bars, 1 mm. These species were chosen on the basis of their cavitation resistance and the arrangement of their primary xylem, beginning with species that exhibit the least cavitation resistant and integrated xylem and ending with those with the most resistant and sectored xylem: (a) Psilotum nudum, P50 = −0.69 MPa; (b) Pteridium aquilinum, P50 = −1.05 ± 0.53 MPa; (c) Dicksonia antarctica, P50 = −0.96 ± 0.45 MPa; (d) Cheilanthes bonariensis, P50 = −3.18 MPa; (e) Lygodium flexuosum, P50 = −3.97 ± 1.88 MPa; and (f) Dryopteris arguta, P50 = −11.8 ± 7.6 MPa. The arrangement of L. flexuosum tracheids within the stele is unusual because parenchyma cells separate the conduits from one another. A thick ring of schlerenchyma fibers known as the hypodermal sterome (Rowe & Speck, 2005) is located just below the epidermis in all cross-sections. In addition to its mechanical role, the sterome may also aid in preventing water loss.

Figure 3.

The hydraulic response to increasingly negative water potential in six ferns representing the most vulnerable and the most resistant species to drought-induced cavitation. Because of its short stature, the vulnerability curve in Psilotum nudum stipes was obtained using the bench-dry method while other species were evaluated using the centrifugal method of Alder et al. (1997).

We observed no hydraulic or structural xylem trade-offs associated with species’ cavitation resistance, in contrast to trends evident across a broad sampling of the conifer xylem response (Fig. 4). Indeed, cavitation resistance in conifers comes at the cost of reduced conduit diameters and thicker conduit walls, both of which tend to lower hydraulic conductivity, especially within the Pinaceae and north-temperate Cupressaceae (Pittermann et al., 2006a,b), but no such costs were apparent in fern xylem. We can, however, make the generalization that cavitation resistance in ferns is on par with cavitation resistance in conifers.

Figure 4.

The relationship between the xylem pressure causing a 50% loss of hydraulic conductivity as a result of cavitation (P50) and xylem specific conductivity (a), conduit hydraulic diameter (b) and conduit double-wall thickness (c) in conifers (circles) and ferns (letters; see Table 1 for species abbreviations). In contrast to conifers, fern xylem showed no hydraulic or structural trade-offs associated with P50. ns, not significant.

Despite showing a broad range of P50 values, fern tracheid allometry is surprisingly different from conifer tracheids (Fig. 5). In conifers, the tracheid double-wall thickness to tracheid lumen diameter ratio (t/b)h2 is a proxy of tracheid resistance to implosion, and this trait strongly correlates with cavitation resistance in the north-temperate Pinaceae and Cupressaceae. In these two families, higher (t/b)h2 ratios are found in increasingly drought-resistant species, indicating greater conduit strength (Fig. 5; Hacke et al., 2001; Pittermann et al., 2006a). However, conifer tracheids also tend to exhibit more reinforcement than strictly necessary, a trait interpreted as a safety factor protecting against implosion, which is shown by the relationship between the (t/b)h2 ratio and species’P50 (Fig. 5). By contrast, (t/b)h2 ratios in fern conduits are decoupled from cavitation resistance and exhibit no safety factor from implosion. Thus, fern conduits may show some degree of mechanical flexibility. The outlier in Fig. 5 is P. nudum, a basal filicopsid fern, which, despite showing modestly reinforced conduits, is the least cavitation-resistant species sampled.

Figure 5.

Conduit double-wall thickness plotted as a function of conduit hydraulic diameter in conifers (circles) and ferns (letters; see Table 1 for species abbreviations) (a). In contrast to conifers, we observed in fern conduits a strong, positive correlation between conduit hydraulic diameter and double-wall thickness. Surprisingly, this scaling results in a low (t/b)h2 ratio (b) that exhibits no safety factor from implosion, even under very negative xylem pressures (ns, not significant).

Several studies indicate that photosynthesis in pteridophytes is inherently constrained by the low hydraulic efficiency characteristic of this group (Brodribb et al., 2005; Watkins et al., 2010). Although our photosynthesis and stomatal conductance rates agree with previous studies (see also Watkins et al., 2010), it is important to point out that, across a broad sampling of ferns including climbing as well as north-temperate species, photosynthesis and stomatal conductance rates can equal or exceed those of evergreen conifers (Fig. 6). We attribute this result in part to the ferns’ much wider and longer conduits which account for the surprisingly high hydraulic conductivity of fern xylem, and which can, in turn, support a broad range of stomatal conductance rates. Indeed, it is likely that the ferns’ large conduit volumes compensate for the ferns’ small functional xylem areas that are two orders of magnitude smaller than those of conifers, despite serving a similar amount of distal foliage (Figs 7, 8). If evolution acted as expected to increase hydraulic efficiency in the xylem-limited ferns, then P. aquilinum represents the pinnacle of hydraulic selection in fronds as it is the one species in our pool known to contain vessels in both fronds and rhizomes (not shown), with some vessels exceeding 20 cm in length (Fig. 7).

Figure 6.

Photosynthesis (a) and stomatal conductance (b) as a function of hydraulic conduit diameter in conifers (circles) and ferns (letters; see Table 1 for species abbreviations). Cheilanthes bonariensis (CB), an outdoor-grown desert fern, exhibited the highest rates of gas exchange, consistent with its thicker pinnae which presumably provide some degree of capacitance. The highest photosynthesis and conductance rates in the conifers belong to three deciduous species of Cupressaceae. ns, not significant.

Figure 7.

Average conduit length plotted as a function of conduit diameter in conifers (circles) and ferns (letters; see Table 1 for species abbreviations). Assuming a cylindrical geometry, fern tracheid volumes are 8 to 60× greater than those of conifers. This factor would be considerably higher in Pteridium aquilinum, where vessel length exceeds 20 cm.

Figure 8.

Fern xylem supports similar leaf areas as conifer xylem despite relying on primary xylem with a total area that is two orders of magnitude smaller than that of conifers. The tight scaling between xylem and leaf area in ferns is likely to reflect the limited venation and low capacitance of fern leaves (Brodribb et al., 2005, 2007), a result consistent with the constant scaling between stipe length and leaf area observed in Polystichum munitum (Limm & Dawson, 2010). In the absence of the indeterminately growing fronds of Lygodium japonica, the r2 of the regression increases to 0.61 and < 0.001. Conifers are indicated by circles and ferns by letters (see Table 1 for species abbreviations).

The hydraulic consequences of the high-volume tracheids of ferns are readily apparent when xylem conductivity is expressed per leaf and functional xylem area, kL and kx, respectively, and compared with equivalent data in conifers (Fig. 9). The similar kL values in conifers and ferns support the finding that leaf-level function in both plant groups can exhibit a similar range of assimilation and conductance rates (Fig. 6).

Figure 9.

Leaf (a) and xylem (b) specific conductivities plotted as a function of hydraulic conduit diameter in conifers (circles) and ferns (letters; see Table 1 for species abbreviations). Species’ transport efficiency is a function of xylem conduit diameter in both conifers and ferns. The large discrepancy in kL values between Lygodium flexuosum and Lygodium japonica can be attributed to the indeterminately growing habit of the fronds, which varied between 0.5 and 1.5 m in length.

Lastly, our data indicate that fern xylem can exhibit lower transport resistivity (rx, the inverse of kx) for a given conduit diameter than the xylem of the more derived conifer and angiosperm lineages (Fig. 10; additional data from Pittermann et al., 2005). Put another way, when we compared the area-specific xylem resistivities of fern, conifer, vine, and ring- and diffuse-porous species with equivalent mean conduit sizes, fern xylem exhibited higher transport efficiencies than the xylems of the remaining plant groups with the exception of vines, which had a mean conduit diameter of 150 μm and an rx of 13 MPa s m−2 (Fig. 10).

Figure 10.

Xylem specific resistivity plotted as a function of hydraulic conduit diameter in conifers (closed circles), angiosperms (open circles) and ferns (letters; see Table 1 for species abbreviations). Some conifer and angiosperm data were re-plotted from the original comparison of tracheid vs vessel transport efficiency in Pittermann et al. (2005).


The greater xylem specific conductivity observed in the xylem of fern stipes relative to conifer xylem can be attributed in part to the ferns’ longer and wider tracheids. Large lumen diameters reduce the frictional resistance to water transport that arises from conduit walls, while increasing conduit length reduces the frictional resistance attributable to pit membrane crossings from one conduit to the next (Sperry et al., 2006). In ferns, high-volume conduits were presumably free to arise because the xylem serves solely in a hydraulic capacity, in contrast to the double-duty xylem of conifers where tracheids not only deliver water to the canopy but also support it (Pittermann et al., 2006a,b). This support function reduces conifer tracheid diameter and length, and frequently calls for extra wall thickening, as observed in compression wood (Spicer & Gartner, 2002). In ferns, however, the presence of the hypodermal sterome (Fig. 2) relaxes the support constraints imposed upon the tracheids, such that high-volume conduits function solely in the efficient transport of water. An analogous situation occurs in the xylem of woody angisperms, where thick-walled fibers constitute the structural matrix that supports the canopy, leaving the hydraulic function to the much longer vessels (Hacke et al., 2001; Sperry et al., 2006), as well as in leaves, which are supported by turgor pressure and collenchyma. In conifers, root xylem is also released from the support function and characteristically exhibits tracheids that are longer, wider and more conductive (Pittermann et al., 2006a). The primary xylem of conifer seedlings and juvenile shoots is developmentally most analogous to fern xylem, but whether or not the structure–function patterns of primary tracheids resemble those of conifer roots or ferns is unknown.

Fern xylem may be hydraulically efficient, but it is primary xylem nonetheless and is subject to limits imposed by a basal developmental program. First, the overall structure of ferns is constrained by the lack of a vascular cambium, which gives rise to lateral branching and secondary xylem (Rowe & Speck, 2004; Spicer & Groover, 2010). The absence of wood imposes a fundamental limit on fern size and canopy development even in tree ferns, which raise their relatively modest canopy by virture of anomalous secondary growth courtesy of roots and fibers. Secondly, ferns possess much smaller xylem areas in the absence of a cambium than conifers, despite supporting an equivalent amount of foliage and similar rates of gas exchange (Figs 8,9). Hence, the observed frond allometry in Fig. 8 is the result of a limited vascular system that supplies water in sufficient quantities for transpiration, but probably lacks the buffer against fluctuations in water availability that capacitance provides in woody plants. The reliance on primary xylem for water transport may explain why ferns exhibit more rapid stomatal closure than angiosperm leaves (Brodribb & Holbrook, 2004).

The extent to which pit membrane characteristics contribute to fern hydraulic efficiency is largely undetermined. However, the few studies that addressed the relative contributions of pit and lumen resistance in fern xylem indicate that pit membranes in fern tracheids account for 36 to 47% of total tracheid resistance (Calkin et al., 1985; Schulte et al., 1987). By contrast, conifer and angiosperm pit membranes confer considerably greater resistance to flow, accounting for c. 64 and 56% of conduit resistance, respectively (Wheeler et al., 2005; Pittermann et al., 2006b; Sperry et al., 2006). Whether or not reduced pit membrane resistance in ferns is attributable to greater porosity and/or a higher fraction of conduit pit area remains to be seen, but some evidence suggests that fern pit membranes not only may be quite porous, but may extend along the full length of the tracheid wall (Carlquist & Schneider, 2007). Selection has been shown to act on pit membrane area as well as porosity, and both traits are thought to have a profound effect on hydraulic efficiency and cavitation by the entry of air, otherwise known as ‘air-seeding’ (Choat et al., 2008; Jansen et al., 2009; Lens et al., 2011). Certainly, greater pit area and membrane porosity should compensate for xylem comprised of single-celled conduits lacking the low-resistance torus-margo pit membrane (Pittermann et al., 2005; Sperry et al., 2006). Interestingly, the homogeous pit membranes found in the tracheids of vesselless angiosperms including the Amborellaceae, Winteraceae and Trochodendrales exhibit considerably higher porosity than those of eudicots, a finding that is consistent with the low, conifer-like area-specific pit resistance in this basal lineage (Hacke et al., 2007). Similar to these basal dicots, fern xylem exhibits hydraulic conductivities that are at least 38× greater than predicted for tracheids lacking conifer-type torus-margo pits (Pittermann et al., 2005; Hacke et al., 2007). Whether Botrichium dissectum, an Ophioglossid fern with torus-margo pit membranes (Morrow & Dute, 1998), exhibits comparable hydraulic efficiencies that are closer to conifers or ferns remains to be seen.

Recent work indicates that pit membrane features control cavitation resistance in both conifers and angiosperms (Christman et al., 2009; Pittermann et al., 2010; Lens et al., 2011; Delzon et al., 2010), but what controls cavitation resistance in ferns? Currently, we can entertain one or a combination of two possibilities: the fern pit membrane may exhibit variable porosity that corresponds to species’ resistance to air entry air-seeding, as previously mentioned; and the integration of the vascular bundles, specifically the frequency of conduit contact between two or more vascular bundles, affects the rate at which air spreads through the xylem. Empirical and theoretical work on woody plants has shown that hydraulically integrated vascular networks are associated with increased hydraulic conductivity, but at the cost of more rapid spread of air-seeding-induced embolism (Zanne et al., 2006; Loepfe et al., 2007; Schenk et al., 2008). Looking across the stelar patterns of the sampled fern species (Fig. 2), we suspect that cavitation resistance in these plants is controlled by a similar, spatially determined vascular arrangement that may inhibit or facilitate the spread of air through the conduit network depending on conduit-to-conduit or bundle-to-bundle contact. For example, hydraulically integrated xylem belonging to P. aquilinum and P. nudum was the most vulnerable to cavitation, while xylem belonging to the cavitation-resistant D. arguta and P. cretica was packaged in a limited number of spatially separated (sectored) vascular bundles (Fig. 2). Variation in P50 values was greater in species with sectored xylem, presumably because of irregular and as of yet uncharacterized bundle-to-bundle connectivity over the length of the sampled stipe segments. A combination of methods including dye perfusions and high-resolution computed tomography (HRCT) could be used to construct three-dimensional renderings of variable xylem patterns along the length of the frond and as well as in different species (Schulte et al., 1987; Brodersen et al., 2010).

Perhaps a more fundamental question than what controls air-seeding in ferns is what selects for cavitation resistance in these overwhelmingly mesophylic plants? Watkins et al. (2010) show that tropical epiphytic ferns are more cavitation resistant than terrestrial species, a result consistent with the epiphytes’ lower midday water potentials. In our assortment of terrestrial ferns, the issue may also be complicated by ecology and life history strategy. For example, P. aquilinum is a drought-deciduous fern, and exhibits xylem with high hydraulic conductivity and low cavitation resistance. This hydraulic ‘boom and bust’ strategy is similar to the phenology and physiology of deciduous ring-porous or riparian trees. By comparison, D. arguta is a semi-perennial species that retains fronds for at least 2 yr, during which it endures the dry, mediterranean summer climate of coastal California. In addition to drought stress, cavitation resistance in ferns is probably associated with frond longevity. An extreme example of this is L. flexuosum, the fronds of which grow indeterminately to lengths well beyond 2 m. Although this species is native to mesic, subtropical habitats, its stipe is surprisingly cavitation resistant, presumably to ensure water transport to distal pinnae. Whether or not reproductive, spore-bearing fronds show alternative hydraulic strategies is an open question.

Despite the high cavitation resistance exhibited by a number of fern species, the tracheid allometry of ferns is considerably different from conduits in conifers and woody angiosperms. In north-temperate conifers, the trend toward increased implosion resistance comes courtesy of wall thickness:lumen diameter ratios that increase with P50, but do so by reducing tracheid diameter rather than fortifying the tracheid walls (Pittermann et al., 2006a; Fig. 5). Consequently, reduced transport efficiency constitutes the fundamental hydraulic trade-off associated with cavitation resistance in north-temperate conifers (Pittermann et al., 2006a,b; Fig. 4). These safety–efficiency trade-offs are also apparent at the level of the pit, whereby increased cavitation resistance is associated with reduced pit conductivity across a range of conifer species and within a single tree (Pittermann et al., 2006b, 2010; Domec et al., 2008). By contrast, we found no hydraulic or structural ‘investment’ trade-offs in vulnerable vs cavitation-resistant ferns, similar to the results of Watkins et al. (2010). This is consistent with the weaker relationship between vessel diameter and P50 observed in angiosperms, and the probabilistic role that pit porosity plays in controlling cavitation resistance in dicots (Wheeler et al., 2005; Hacke et al., 2007; Christman et al., 2009; Lens et al., 2011). Across the pteridophytes, the costs of drought resistance may be reflected in the increased production of osmolytes and proteins that protect living tissues from low water potentials rather than xylem structure per se (Proctor & Tuba, 2002).

In ferns, larger conduit diameters scale directly with greater wall thickness (Fig. 5). Interestingly, despite the unusually thick tracheid walls found in several ferns, the relationship between species’P50 and the conduit (t/b)h2 ratio falls directly on the predicted conduit implosion line – a dramatic departure from the pattern observed in conifer tracheids. Conduit implosion thresholds are computed without k, the safety factor from cell collapse, so our data indicate two things: first, ferns do not allocate any more resources to their xylem than is necessary to maintain a circular geometry, and second, the conduit (t/b)h2 ratio cannot be used to infer cavitation resistance in these plants. Fern tracheids may exhibit some degree of flexibility to tolerate bending or torsion stress associated with life in the understory or to facilitate the climbing habit. To wit, the largest tracheids in the climber L. flexuosum are located in the stele center, that is, the neutral axis region, which experiences minimal tensile and compressive stress (Niklas, 1992). Also, flexible tracheids may accommodate shape change in response to the tension imposed on them by the water column. A good example of this phenomenon is transfusion tracheids in conifer leaves and needles. These unlignified tracheids flex and collapse under negative xylem pressure, presumably to avoid cavitation and to facilitate rapid refilling (Cochard et al., 2004; Brodribb & Holbrook, 2005). Similarly distorted tracheids have been observed in the xylem of variably dehydrated fronds (J. Pittermann & E. Limm, unpublished), and it is not unreasonable to suspect that, in the very cavitation-resistant species, the reductions in hydraulic conductivity with progressively more negative water potentials may be attributable to conduit distortion, rather than embolism. The aforementioned HRCT method is nondestructive, and could provide a window into the intricacies of fern vascular function in response to drought and rewatering in a live plant (Brodersen et al., 2010).

The extent to which we understand variation in xylem structure and function across tracheophytes will affect the inferences we draw about the evolution of transport tissue across both extinct and extant lineages, including the Lycopodiales, Sellaginales and perhaps even some bryophytes. For example, empirically based biophysical models of xylem performance have provided important insights into the xylem function of the earliest land plants as well as the more derived seed ferns, pro-gymnosperms and other lineages with tracheid-based xylem (Wilson et al., 2008; Wilson and Fischer, 2010; Wilson & Knoll, 2010). These models suggest that, far ahead of the appearance of heteroxylous angiosperm xylem, xylem evolution in ancient plants was consistently guided by selection for hydraulic efficiency, cavitation resistance and, eventually, mechanical support. Our data show that extant ferns have succeeded in mastering these three essentials, although not in the manner one would predict based on the structure–function model derived from woody plants. The physiological resilience of ferns may go a long way toward explaining why they have persisted across nearly all parts of the globe despite several mass extinctions and the intense competitive pressure exerted by angiosperms (Schneider et al., 2004; Schuettpelz & Pryer, 2009; Watkins et al., 2010).


We are grateful to the NSF (IOS-1027410; J.P.) and Save the Redwoods League (E.L. and J.P.) for supporting this research. C.R. received funding through an NSF REU grant and from the Save the Redwoods league. M.A.C. was supported by NSF-IBN-0743148 to John Sperry (University of Utah). The conifer data were collected with support from the Miller Institute for Basic Research (UC Berkeley; J.P.) and assistance from Todd Dawson. We wish to thank Lucy Lynn, Stephanie Ko and Shervin Bastami for contributing to the xylem anatomical measurements, and Duncan Smith and John Sperry for kindly helping with the silicone injection method. Holly Forbes, Mona Bourrell and the staff at the University of California Botanical Garden (Berkeley, CA) and the San Francisco Botanical Garden kindly provided access to plant material, as did Jim Velzy and Denise Polk (UC Santa Cruz), who also assisted with propagation and glasshouse space. We also thank three anonymous reviewers whose insightful comments helped improve the manuscript.