The physiological implications of primary xylem organization in two ferns



    Corresponding author
    1. Department of Ecology & Evolutionary Biology, University of California, 1156 High Street, Santa Cruz, CA 95064, USA
      C. Brodersen. E-mail:
    Search for more papers by this author

    1. Department of Ecology & Evolutionary Biology, University of California, 1156 High Street, Santa Cruz, CA 95064, USA
    Search for more papers by this author

    1. Department of Ecology & Evolutionary Biology, University of California, 1156 High Street, Santa Cruz, CA 95064, USA
    Search for more papers by this author

C. Brodersen. E-mail:


Xylem structure and function are well described in woody plants, but the implications of xylem organization in less-derived plants such as ferns are poorly understood. Here, two ferns with contrasting phenology and xylem organization were selected to investigate how xylem dysfunction affects hydraulic conductivity and stomatal conductance (gs). The drought-deciduous pioneer species, Pteridium aquilinum, exhibits fronds composed of 25 to 37 highly integrated vascular bundles with many connections, high gs and moderate cavitation resistance (P50 = −2.23 MPa). By contrast, the evergreen Woodwardia fimbriata exhibits sectored fronds with 3 to 5 vascular bundles and infrequent connections, low gs and high resistance to cavitation (P50 = −5.21 MPa). Xylem-specific conductivity was significantly higher in P. aqulinium in part due to its wide, efficient conduits that supply its rapidly transpiring pinnae. These trade-offs imply that the contrasting xylem organization of these ferns mirrors their divergent life history strategies. Greater hydraulic connectivity and gs promote rapid seasonal growth, but come with the risk of increased vulnerability to cavitation in P. aquilinum, while the conservative xylem organization of W. fimbriata leads to slower growth but greater drought tolerance and frond longevity.


Xylem organization represents the culmination of over 400 million years of selection for hydraulic systems that balance transport efficiency with resistance to failure via cavitation (Tyree & Zimmermann 2002; Sperry 2003; Sperry, Meinzer & Mcculloh 2008; Pittermann 2010). Coordination between the supply and demand for water is paramount, as stomatal conductance (gs) and photosynthesis are directly linked to hydraulic conductivity (k) (Brodribb & Feild 2000; Brodribb & Holbrook 2004). Our knowledge of the relationships between plant hydraulics and photosynthetic performance is based on an accumulation of studies focused on woody plants, but much less is known about primary xylem structure and function.

Ferns have a seemingly primitive vascular system composed solely of primary xylem, an assemblage of unicellular tracheids, and occasionally vessels (Carlquist & Schneider 2001), which act as a markedly different solution to water transport than the xylem of woody plants, where secondary growth and multicellular vessels offer both structural support and wide, efficient conduits. Fern tracheids are tightly packed, comparatively few in number and have heavily pitted lateral walls bearing pit membranes that are often highly porous (White 1963a; Carlquist & Schneider 1997; Pittermann et al. 2011). These combined features should make ferns extremely vulnerable to cavitation and at a significant physiological disadvantage because of the inherent risk of embolism spread (Hargrave et al. 1994; Wheeler et al. 2005; Hacke et al. 2007). Despite these limitations, ferns have been persistent through their evolutionary history and represent highly successful forms in both the past and present (Taylor, Taylor & Krings 2009; Pittermann 2010). Recent research even suggests that many ferns have surprisingly efficient and cavitation-resistant xylem (Watkins, Holbrook & Zwieniecki 2010; Pittermann et al. 2011), but to what extent selection has acted to modify the spatial organization of primary xylem is not well understood.

Recent work in woody plants has shown that xylem organization and conduit diameter play critical roles in cavitation resistance (Loepfe et al. 2007; Mencuccini et al. 2010; Savage et al. 2010; Lens et al. 2011), but few studies have addressed these issues in plants with primary xylem. Hydraulic integration (i.e. xylem connectivity) in woody plants is often coupled with small conduit diameter, lower k and increased vulnerability to the spread of drought-induced embolism (Zanne et al. 2006; Espino & Schenk 2009). Physical separation of the xylem into discrete sectors, most often associated with xeric species with wide, efficient conduits and high k, compartmentalizes water transport and prevents the systemic spread of embolism from compromised conduits to functional adjacent ones (Orians, Babst & Zanne 2005; Salguero-Gómez & Casper 2011). As shown by Schenk et al. (2008), woody species across aridity gradients display a continuum of xylem sectoriality patterns in stem and trunk architecture that help plants adapt to a wide range of habitats with variable moisture availability. Similar redundancy and hydraulic integration in leaves aid in the recovery of gas exchange and k following xylem damage or drought-induced embolism (Sack et al. 2008; Scoffoni et al. 2011). The question remains of whether ferns composed solely of primary xylem operate under the same set of selective pressures as woody plants.

Ferns may have overcome the constraints of primary xylem through the diversification of vascular bundle organization (i.e. stelar patterns). Stelar patterns within fern fronds range from the simple, integrated haplostele, to the more derived and complex eustele, siphonostele or dictyostele with many discrete vascular bundles, making them functionally similar to the subdivided stems of xeric woody plants (Fig. 1; Stein 1993; Schneider et al. 2002; Zanne et al. 2006; Schenk et al. 2008), thereby mirroring the sectoriality and integration found in secondary xylem. Embolism spread within a frond should be determined by the spatial separation of bundles and the number and frequency of connections within the xylem network. While xylem organization has been studied in a variety of fern species (Gibson, Calkin & Nobel 1985), and with specific focus on hydraulic conductance and photosynthesis in leaves (e.g. Schulte, Gibson & Nobel 1987; Franks & Farquhar 1999; Brodribb et al. 2005), the physiological implications of xylem organization and connectivity have yet to be fully explored (Woodhouse & Nobel 1982; Lo Gullo et al. 2010).

Figure 1.

Frond architecture and xylem anatomy of P. aquilinum fronds. The triangular shaped frond of P. aquilinum is 3-pinnate pinnatifid (a). The number of vascular bundles (b, squares) decreases from the base of the frond (0 cm) to the apex (∼120 cm) in three representative fronds. Representative transverse sections (c) spaced approximately 20 cm apart (see Supporting Information Fig. S1 for larger images) and the location of connections (d) in three representative fronds show the organization of the vascular bundles. Grey bars in (a) indicate the three sampling regions of the stipe or rachis used for hydraulics and gas exchange measurements.

The striking variation in stelar patterns and the ease of manipulating vascular bundles in ferns (e.g. severing individual xylem bundles) present a good opportunity to study the fundamental relationships between xylem organization and physiology that would otherwise be difficult or impossible in woody plants. The goal of this study was to determine what structure-function trade-offs might exist in ferns as the result of adaptations towards different life history strategies, where vascular arrangement should have significant consequences for gas exchange and cavitation resistance. We chose two ferns, Pteridium aquilinum (L.) Kuhn (Dennstaedtiaceae) and Woodwardia fimbriata (Sm.) (Blechnaceae), which have differences in phenology, stelar patterns, conduit morphology and resistance to cavitation, yet frequently occupy similar habitats. P. aquilinum is a drought deciduous pioneer species known for its invasiveness (Robinson, Sheffield & Sharpe 2010) and has moderate resistance to cavitation (Pittermann et al. 2011). The vascular bundles of P. aquilinum are tightly packed and appear to fuse readily throughout the frond; an organization that we hypothesize should favour both efficient k and the spread of drought-induced embolism (Fig. 1). In addition, P. aquilinum is one of the few known fern species to have true vessels in addition to tracheids (Bliss 1939; White 1963b; Carlquist & Schneider 1997, 2007; Pittermann et al. 2011). In contrast, W. fimbriata is a slow-growing evergreen species with fronds that typically last at least 2 years, is highly resistant to cavitation, and fronds that have few vascular bundles which appear to be isolated throughout their axial length (Figs 2 and 3). The focus of this study was to characterize the relationships between xylem organization, cavitation resistance and gs based on the known differences in anatomy and phenology.

Figure 2.

Frond architecture and xylem anatomy in W. fimbriata fronds. The lanceolate frond of W. fimbriata is 1-pinnate pinnatifid (a). The number of vascular bundles (b, squares) decreases from the base of the frond (0 cm) to the apex (∼100 cm) in three representative fronds. Representative transverse sections (c) spaced approximately 20 cm apart (see Supporting Information Fig. S2 for larger images) and the location of connections (d) in three representative fronds show the organization of the vascular bundles. Grey bars in (a) indicate the three sampling regions of the stipe or rachis used in hydraulics and gas exchange measurements.

Figure 3.

Representative xylem network maps for (a) P. aquilinum and (b) W. fimbriata fronds. Each vertical line represents a vascular bundle and each white dot represents the location of a connection. Connections were clustered around the axis junctions in P. aquilinum while connections were less frequent in W. fimbriata. The number of bundles decreases acropetally in both species, with many bundles diverted to the axis junctions in P. aquilinum, while the pinnae in W. fimbriata were supplied by short bundles (not shown) bifurcating from the two large bundles [dark lines in (b)].


Plant material and collection

For this study, we chose two fern species commonly found in the seasonally moist Sequoia sempervirens understorey. The deciduous triangular-shaped fronds of P. aquilinum are 3-pinnate pinnatifid and joined together by a long-creeping, branched rhizome (Fig. 1a) (Calkin, Gibson & Nobel 1985). The evergreen lanceolate fronds of W. fimbriata are 1-pinnate pinnatifid, growing from a short, stout rhizome, typically around sources of fresh water (Fig. 2a). The organization of the primary xylem in P. aquilinum was described by Wardlaw (1950) as a perforated polycyclic solenostele, with multiple vascular bundles arranged in concentric circles, while W. fimbriata has a dictyostele consisting of two large vascular bundles and one to four smaller abaxial bundles (Beck, Schmid & Rothwell 1982; Schneider et al. 2002) (Figs 1 and 2, Supporting Information Figs S1 and S2). In both species, each vascular bundle consists of a central grouping of xylem surrounded by phloem, all of which are encapsulated by an endodermis. Fully developed current-year fronds from P. aquilinum and W. fimbriata were cut from plants growing on the UC Santa Cruz campus and used for anatomical and hydraulics measurements from May to June, 2011. Fronds from the same plants were used for in situ gas exchange measurements in the field during the same time period.

Anatomical measurements

We characterized xylem anatomy and organization by measuring conduit diameters and the spatial distribution of bundles from transverse cross sections at three sampling points on the fronds (Figs 1 and 2, Supporting Information Figs S1 and S2, Table 1). Transverse sections were made from three representative fronds from three separate individual plants from both species, with each frond subdivided into three segments from the base of the frond (stipe) at the soil level, within the rachis above the axis junction (lower rachis), and from the rachis near the apex (upper rachis, approximately 90 cm from the soil surface). Freehand sections were stained with toludine blue, mounted in glycerol and then photographed at 100× magnification with a digital camera mounted to a Motic BA400 compound microscope (Motic Inc., Richmond, British Columbia, Canada). All measurements were taken from digital images analysed using ImageJ software (freeware available from the site Conduit diameters, the ratio of xylem lumen area to cross-sectional area of the stipe or rachis, were measured. Conduit lumen area was measured first and then converted to the equivalent circle diameter. The degree of sectoriality (S) of the xylem bundles was characterized by calculating the relationship between the total perimeter of the xylem bundles to the combined cross-sectional area of the xylem using the following formula modified from Schenk et al. (2008):


where p = xylem perimeter, A = combined cross-sectional area of the xylem.

Table 1.  Anatomy and sectoriality of P. aquilinum and W. fimbriata fronds at the stipe, lower rachis and upper rachis positions used in gas exchange and hydraulic conductivity measurements
  P. aquilinum W. fimbriata
  1. All values are means ± SD.

  2. n = 4 for each sampling position. Values with different superscript letters denote statistical significance following anovas with post hoc t-tests using Bonferroni corrections (P < 0.05).

  3. D, diameter of the stipe or rachis (mm); NVB, number of vascular bundles; S, Sectoriality Index, where increasing S represents increasing xylem integration; XSA, xylem specific area (xylem lumen area/cross-sectional area of the stem) (%).

Stipe9.11 ± 0.81a33.25 ± 5.56a12.11 ± 0.47a0.06 ± 0.01a10.50 ± 1.64a5.00 ± 0.00a9.63 ± 2.86a0.01 ± 0.01a
Lower rachis5.52 ± 0.29b14.66 ± 6.02b10.62 ± 1.49a0.09 ± 0.01b8.25 ± 1.83a3.50 ± 0.57b8.84 ± 2.17a0.02 ± 0.01a
Upper rachis3.27 ± 0.09c8.00 ± 2.16b6.75 ± 1.85b0.09 ± 0.01b4.77 ± 1.53b2.50 ± 0.57c7.78 ± 3.64a0.03 ± 0.03a

Xylem connectivity

To visualize the entire xylem network in the fronds, including connections between bundles, we extracted all vascular bundles in three mature fronds from each species. Fronds were cut from the field and immediately brought back to the lab for dissection. Firstly, all of the pinnae were cut from the fronds so that approximately 1 cm of the midrib remained attached to the rachis. Next, the stipe and rachis were gently compressed with a pair of pliers to dislodge the xylem bundles from the surrounding epidermis and cortex tissue. The frond was then pinned to a rigid board and the xylem bundles were carefully extracted with tweezers by removing all of the tissue not containing xylem such that the xylem connections were preserved and the xylem network was displayed as a complete unit. The location of every connection was measured as a distance from the base of the frond. The total number of xylem bundles was then measured at 10 cm intervals along the frond. Finally, the xylem network was photographed with a digital camera and the images were used to create a xylem network map (Fig. 3).

High-resolution computed tomography (HRCT)

To visualize the three dimensional (3D) organization of xylem bundles, well-hydrated plants were cut from the UC Santa Cruz campus and the UC Berkeley Botanical Garden, placed in plastic bags and transported to the Lawrence Berkeley National Laboratory Advanced Light Source, Beamline 8.3.2 microtomography facility. Frond sections from the lower and upper rachis were cut and allowed to dehydrate under ambient conditions for approximately 3 h to eliminate the majority of the water in the xylem but not significantly change the shape of the samples. Next, stems were wrapped in Parafilm (Peichiney Plastic Packaging, Inc., Chicago, IL, USA) to prevent further dehydration, mounted in the microtomography instrument and imaged at 15 keV at 0.125° increments over 180° following the methods of Brodersen et al. (2011). The resulting two-dimensional images were reconstructed into a 3D dataset approximately 1.5 cm in length using Octopus software (University of Ghent). The datasets were then visualized with Avizo 6.2 software (Visualization Sciences Group, Burlington, MA, USA).

Hydraulics measurements

We performed a variety of experiments aimed at studying cavitation resistance in the xylem network of both species with and without mechanical damage to the xylem bundles at different points on the fronds. Fronds were cut from well-hydrated plants in the field and brought back to the lab. Sections were recut under water from the stipe, lower rachis and upper rachis (Figs 1a and 2a). The sections were then cut to 142 mm in length and the ends were shaved with a razor blade to expose a smooth transverse surface. If any axis junctions fell within the excised section, the pinnae were cut close to the rachis and then sealed with cyanoacrylic glue (Loctite 409, Henkel Corporation, Rocky Hill, CT, USA). Sections were then degassed overnight under vacuum in a 20 mm KCl solution to remove any native embolism in the xylem.

The k was measured following the methods of Sperry (1993) and calculated as a function of the flow rate through the section at a given pressure. Segments (n = 6) were mounted in a tubing apparatus to direct the flow of a 20 mm KCl solution through the xylem at a pressure of 6–8 kPa to measure maximum k (kmax) after degassing. The flow rate through the sections without a pressure head was measured before and after each gravimetric measurement to control for equilibrium drift in the system. These background measurements were averaged and subtracted from the pressure-induced flow rates.

The centrifuge method (Pockman, Sperry & O'Leary 1995; Alder et al. 1997) was then used to measure the vulnerability to cavitation of the xylem at the three sampling points on the frond in response to a range of xylem pressures. Following the kmax measurement, the sections were mounted in a custom rotor for a Sorvall RC-5C centrifuge (Thermo Fisher Scientific, Waltham, MA, USA) and spun for 3 min at speeds known to induce xylem pressures (Px) of −0.5, −1.0, −3.0, −5.0, −7.0 and −9.0 MPa. After each spinning, the k of the sections were measured. The percent loss of conductivity (PLC) resulting from progressively negative Px was then calculated as a function of kmax at Px = 0 MPa:


A Weibull function was then used to fit the vulnerability curves from each section to estimate the xylem pressure at which 50% loss of conductivity (P50) was reached (Neufeld et al. 1992).

A second group of W. fimbriata stipe segments (n = 6) was then used to estimate the relative hydraulic contribution of large and small bundles. Segments were degassed as above and the kmax of each section was measured. Then, k was measured after individual xylem bundles were sealed on both ends of the segment with cyanoacrylic glue (Loctite 409). The large number, small size, convoluted shape and non-symmetric arrangement of P. aquilinum xylem bundles (Fig. 1) made adequate sealing exceedingly difficult and prevented these measurements from being made with consistent treatment between sections. The W. fimbriata sections, however, had large, discrete bundles that were easily sealed.

A third group of W. fimbriata stipe segments (n = 12) was used to test the cavitation resistance of the large and small bundles. Segments were prepared as above for centrifugation but half of the segments had large bundles sealed with glue while the other half had the small bundles sealed. We then proceeded to measure kmax and k at the same negative xylem pressures as above.

Another group of stipe segments (n = 6) from both species was notched at the midpoint of the segments to sever the xylem bundles and then compared with un-notched control sections (n = 6) to determine the hydraulic consequences of air entry. The xylem was notched with a razor blade perpendicular to the direction of flow through the segment, and a thin piece of polystyrene plastic was inserted into the notched area to prevent flow through the severed xylem. The notched area was then wrapped in Parafilm to prevent the polystyrene from falling out while spinning in the centrifuge. P. aquilinum segments were notched so approximately 25% of the cross-sectional area was rendered non-conductive, while W. fimbriata segments were notched so that one large adaxial bundle was rendered non-conductive. A second group of segments (n = 6) was cut in the same manner but only to a depth of <1 mm to control for any wound response effects.

Gas exchange measurements

To evaluate the role of redundancy in xylem organization in situ we measured gs in plants with and without mechanical damage to the xylem network. Thirty minutes prior to gas exchange measurements, half of the stipes were notched with a razor blade 10 cm above the soil surface and perpendicular to transpiration stream. For P. aquilinum, approximately 25% of the cross-sectional area of the stipe was notched, while W. fimbriata stipes were notched such that either one or two large adaxial vascular bundles were severed. Twelve total P. aquilinum plants, six control and six notched, and 18 W. fimbriata plants, six control, six with a single large vascular bundle notched and six with both large vascular bundles notched, were used for the gas exchange measurements. Next, a piece of polystyrene plastic was cut and inserted into the notched area to block k within the bundle, and then the notched area was wrapped in Parafilm. Stipes of control plants were cut such that the epidermis and part of the cortex were notched to a depth of <1 mm, but without damaging the vascular bundles to minimize potential wounding response differences between control and notched plants. The gs was measured with a Decagon SC-1 Leaf Porometer (Decagon Devices Inc., Pullman, WA, USA) on consecutive days in June and July 2011 with cloudless skies. Pinnae from the lower and upper rachis were measured to determine the effect of vascular bundle disruption. To capture diurnal changes in gas exchange, gs was measured on two pinnae, one on each side of the frond, at both sampling positions at five points during the day (0900, 1100, 1300, 1500 and 1700 h). Concurrent with gas exchange measurements, frond water potential was measured at 0900, 1300 and 1500 h on six plants at the middle rachis position using a Scholander style pressure chamber (PMS Instruments, Corvallis, OR, USA). In addition, water potential measurements were made at the lower, middle and upper rachis positions at 1300 h on six fronds from both species to test the water potential gradient within the fronds.

Statistical analysis

The P50, gs, anatomical measurements and water potential values were compared using analyses of variance (anovas) and post hoc t-tests with Bonferroni corrections to test for statistical significance between multiple treatments (P < 0.05) using SPSS software (v. 20.0.0; SPSS Inc., Chicago, IL, USA).


Frond anatomy and architecture

We observed two extremes of xylem organization after dissecting the fronds of P. aquilinum (complex and convoluted) and W. fimbriata (simple and discrete) (Figs 1–3, Table 1). The basal end of P. aquilinum stipes had an average of 34 discrete vascular bundles that anastomose along the frond to approximately eight bundles at the apex (Figs 1b,c, Supporting Information Fig. S1; Table 1). Connections in P. aquilinum were concentrated at the axis junctions where pinnae emerge from the rachis, and decreased in frequency towards the apex (Figs 1d & 3a). Such connections typically consisted of the fusion of two large bundles or the bifurcation of an axial bundle, with the secondary bundle leading to pinnae (Fig. 3a). The basal end of W. fimbriata stipes had an average of five vascular bundles consisting of two large adaxial bundles and one to four smaller abaxial bundles (Fig. 2b,c). Large bundles spanned the entire frond, eventually merging in the apex, and solely supplied the pinnae through bifurcations from the main axial bundles (Figs 2b,c & Supporting Information Fig. S2). The small bundles made infrequent connections to other small bundles, and occasionally to the large bundles. We observed increasing vascular integration from the base to the apex in P. aquilinum fronds, while W. fimbriata showed nominal changes in integration (Table 1). The ratio of xylem lumen area to cross-sectional area increased along the frond in both species, with the greatest axial increase in W. fimbriata (Table 1).

HRCT imaging

The 3D HRCT images confirmed the results from the manual dissection of vascular bundles, with the benefit of preserving the spatial orientation of the bundles in situ (Fig. 4 & Supporting Information Fig. S3, Supporting Information Movies S1 & S2). In the upper rachis of the P. aquilinum segment (Fig. 4a,b), no axial bundles merged; however, several secondary bundles merged after bifurcating into axis junctions leading to the pinnae. In the upper rachis of W. fimbriata, there were no connections between axial bundles, yet both bundles did bifurcate with the secondary bundles leading to the pinnae (Fig. 4e,f). In the lower rachis segment of P. aquilinum, there were four connections between the axial bundles (Fig. 4c,d), while there was only one connection in W. fimbriata between two small bundles (Fig. 4g,h). The large axial bundles of W. fimbriata remained isolated throughout the segment.

Figure 4.

High-resolution computed tomography volume renderings of P. aquilinum (a–d) and W. fimbriata (e–h). Frond segments 1.5 cm in length from the upper (a, b, e, f) and middle rachis (d, c, h, g) are viewed in longitudinal plane and then rotated in 45° towards the reader, revealing primary xylem bundles (blue) and the surrounding tissue (green). The upper and middle rachis of P. aquilinum consists of many xylem bundles with multiple connections located in the axis junctions. The upper and middle rachis of W. fimbriata consists of two and five xylem bundles, respectively, with few connections. Scale varies with perspective. Bar = 5 mm.

Hydraulic measurements

Xylem integration varies along the frond; so to characterize the susceptibility of different frond segments to cavitation, we generated vulnerability curves from the stipe, lower rachis and upper rachis positions from both species. Vulnerability curves for P. aquilinum were limited to centrifugation speeds that generated up to −7.0 MPa, as segments spun at greater speeds had a tendency to disintegrate in the centrifuge rotor. By −7.0 MPa, most P. aquilinum segments had reached at least 90% loss of conductivity. Vulnerability curves from P. aquilinum fronds revealed that segments from all three frond positions were equally resistant to cavitation (no significant difference between mean P50 values, P > 0.05) (Fig. 5a, Table 2). Vulnerability curves in W. fimbriata revealed that this species was much more resistant to cavitation, yet became significantly more vulnerable towards the apex (Fig. 5b, Table 2). We then compared the vulnerability curves of large and small vascular bundles in W. fimbriata stipes and found no statistically significant differences in P50 (−5.94 MPa ± 3.52 SD and −6.19 MPa ± 2.32 SD for small and large bundles, respectively; Fig. 5a). We found a direct trade-off between safe and efficient water transport, with xylem-specific conductivity of P. aquilinum stipes being significantly higher than W. fimbriata (67.1 ± 13.2 SD and 41.6 ± 6.0 SD mg mm−1 kPa−1 s−1, respectively, P < 0.05).

Figure 5.

Vulnerability to cavitation at three sampling points on fronds of (a) P. aquilinum and (b) W. fimbriata, measured at the base of the stipe (black symbols), the lower rachis (light grey symbols) and the upper rachis (white symbols). Bars –±SE, n = 6. Mean P50 values are reported in Table 2.

Table 2.  Vulnerability to cavitation as measured by the percent loss of conductivity in fronds of P. aquilinum and W. fimbriata
  P. aquilinum W. fimbriata
P 50 P 50
  1. All values are means ± SD.

  2. n = 6 for all treatments. Values are means ± SD. Values with different superscript letters denote statistical significance following anovas with post hoc t-tests using Bonferroni corrections (P < 0.05).

  3. P 50, 50% loss of conductivity; 1/4 cut, one quarter of the xylem bundles severed; 1 LB cut, one large bundle cut.

Stipe−2.23 ± 1.64a−5.21 ± 2.11a
Middle rachis−1.47 ± 0.64a−3.71 ± 3.09a
Upper rachis−1.23 ± 0.59a−1.11 ± 0.72b
Stipe 1/4 cut−2.11 ± 0.62aNA
Stipe 1 LB cutNA−2.00 ± 1.21b

To assess how quickly air can spread through an integrated versus sectored xylem network, we severed either one quarter of the cross-sectional area of P. aquilinum segments and one large xylem bundle in W. fimbriata segments, and then generated vulnerability curves. When the xylem was cut in the integrated P. aquilinum fronds, vulnerability curves became steeper; however, the P50 values were not significantly different from uncut segments (Fig. 6b, Table 2). In W. fimbriata, however, segments with one large bundle cut became significantly more resistant to cavitation compared with uncut segments (P50 = −5.21 MPa ± 2.11 SD and −2.00 ± 1.21 MPa SD for cut and uncut stipes, respectively; Fig. 6c, Table 2). When analysing the relative contributions of large and small vascular bundles in W. fimbriata, we found that large bundles were responsible for the bulk of water transport (45.0% ± 0.25 SD of kmax per large bundle), with minimal contribution from the small bundles (4.98% ± 0.02 SD of kmax per small bundle).

Figure 6.

Vulnerability to cavitation of xylem in stipes of P. aquilinum and W. fimbriata. In (a), W. fimbriata xylem in large bundles (black circles) was more vulnerable to cavitation than xylem in small bundles (white circles). Vulnerability to cavitation increased in (b) P. aquilinum stipes with one quarter of the xylem bundles severed (white circles) compared with intact stipes (black circles). Vulnerability to cavitation decreased in (c) W. fimbriata, with one large xylem bundle severed (black circles) compared with intact stipes (white circles). Vulnerability curves were normalized to data after the first centrifugation at −0.5 MPa. Bars –±SE, n = 6 in all figures. Mean P50 values are reported in Table 2.

Tracheid diameter distribution

We then measured conduit lumen diameters in the stipe, lower rachis and upper rachis to test whether the differences we observed in higher resistance to cavitation along the frond could be explained by variations in conduit diameter distributions. P. aquilinum xylem had a broader range in conduit diameters than W. fimbriata, and also showed a shift towards smaller mean tracheid diameter towards the apex while W. fimbriata tracheids maintained a constant mean tracheid diameter at all sampling positions (Fig. 7 & Supporting Information Fig. S4). An analysis of tracheid diameters in large and small xylem bundles of W. fimbriata showed a slight shift towards larger diameter tracheids in the large bundles (Fig. 7c).

Figure 7.

Tracheid diameter distributions in (a) P. aquilinum stipes, (b) W. fimbriata stipes, and (c) large (grey bars) and small (white bars) xylem bundles in W. fimbriata stipes. Bars –±SE, n = 4. Arrows show bins with low frequency for large and small bundles.

Gas exchange measurements

To test the functional relationship between xylem integration and gas exchange, we measured diurnal changes in gs in plants with and without mechanical damage to the xylem network. By severing the xylem bundles, our goal was to disrupt the transpiration stream and introduce air into the xylem network. We observed gs rates up to threefold higher in P. aquilinum compared with W. fimbriata, but P. aquilinum suffered greater decreases in gs following xylem damage (Fig. 8).

Figure 8.

Diurnal trends in mean stomatal conductance and mean frond water potential in P. aquilinum (a,b) and W. fimbriata (c,d) fronds measured at upper (a,c) and lower (b,d) rachis positions. Stomatal conductance of control plants (black circles) was typically higher than plants with one quarter of the stipe severed (a,b; grey circles), one large vascular bundle severed (c,d; grey circles) or fronds with two large vascular bundles severed (c,d; white circles). Frond water potential measured at the lower rachis (b,d; grey triangles) declined at midday and recovered by 1700 h in P. aquilinum (b), while W. fimbriata values declined throughout the day. Bars –±SE, n = 9. Superscript letters denote statistical significance between control and experimental pinnae within a sampling period following anovas with post hoc t-tests using Bonferroni corrections (P < 0.05).

In W. fimbriata the effects of severing a single large xylem bundle, representing approximately 45% of the total k, were not significant at any point during the day in either the lower or upper rachis (Fig. 8c,d). Severing both large xylem bundles in W. fimbriata, which represented approximately 90% of the total k, resulted in significant declines in gs during the 0900 and 1100 h sampling points, yet the maximum loss in gs was only 49%. In P. aquilinum, where approximately 25% of the k was severed, losses in gs were more substantial, up to 56% in the lower rachis during the 1100 h sampling point. We then observed diminishing effects of xylem damage as the day progressed beyond 1300 h. The greatest decreases in gs for both species occurred in pinnae from the lower rachis, while the highest gs rates were observed in the upper rachis (Fig. 8). Lower rachis W. fimbriata pinnae showed little diurnal variation in gs with two large bundles severed (Fig. 8d).

Frond water potential

The water status of P. aquilinum fronds measured on pinnae at the middle rachis decreased from 0900 to 1300 h, followed by a recovery by 1700 h (Fig. 8b). Frond water status declined modestly throughout the day in W. fimbriata, with no recovery at the 1700 h sampling point. The lowest water potential measurement at 1300 h, −1.4 MPa, fell very close to the P50 value for P. aquilinum. Midday (1300 ) water potentials in P. aquilinum were significantly lower in the middle and upper rachis compared with the lower rachis (−1.4 ± 0.21a, −2.05 ± 0.19b, −2.07 ± 0.21b MPa for the lower, middle and upper rachis, respectively, P < 0.05). The midday frond water potential was significantly lower at the upper rachis compared with the lower and middle rachis in W. fimbriata (−0.5 ± 0.14a, −0.67 ± 0.31a, −0.75 ± 0.12b MPa for the lower, middle and upper rachis, respectively, P < 0.05).


This study illustrates how the balance between safe and efficient water transport directly influences gas exchange in two ferns with markedly different life history strategies. P. aquilinum has adopted a phenology optimized for rapid early season growth prior to the onset of late season drought. To support high rates of gs, P. aquilinum has wide, efficient conduits within a highly integrated xylem network. While the xylem of P. aquilinum may be more efficient in delivering water to the transpiring pinnae, the interconnections within the xylem ultimately make it more vulnerable to embolism spread. The opposite and more conservative strategy was seen in W. fimbriata, which favours perennial growth, lower gs, drought-resistant xylem and a habitat preference for consistently moist areas. This strategy allows W. fimbriata to tolerate the seasonal fluctuations in temperature and water availability across its broad habitat range, including freezing temperatures.

While P. aquilinum may appear to be highly sectored in transverse cross sections with many discrete vascular bundles (Fig. 1 & Supporting Information Fig. S1), the connections between bundles increase the total integration of the frond (Figs 3 & 4, Table 1). The degree of integration is not immediately apparent from transverse sections (Fig. 1 & Supporting Information Fig. S1), but when viewing the xylem network as a whole or in 3D (Figs 3 & 4, Supporting Information Fig. S3, Supporting Information Movies S1 & S2), the connections between xylem bundles are revealed, and highlight the benefits of both of the techniques used to study xylem organization. While the physical dimensions of the HRCT instrument and our limited data processing capabilities prevented the 3D analysis of longer segments, HRCT analysis is clearly superior for visualizing xylem networks in 3D and preserving the spatial orientation of the xylem in situ (Fig. 4, Supporting Information Fig. S3, Supporting Information Movies S1 & S2). However, manual dissection of the xylem bundles was easy to perform, did not require any specialized equipment and yielded network connectivity data on a much larger scale (whole fronds versus short segments) (Fig. 3). While manual dissection of the xylem network was only performed on the two species described in this study, we have subsequently found that the technique works well for most ferns and is a good alternative to HRCT imaging.

In P. aquilinum fronds, comparable vulnerability curves and P50 values when coupled with negligible differences in conduit diameters between frond positions all strongly suggest that vascular integration influenced embolism spread in this species (Fig. 5a). We suspect that increasing xylem integration along W. fimbriata fronds contributed to decreases in cavitation resistance, specifically the fusion of the large bundles near the apex and the loss of the more resistant small bundles (Figs 2c, 5b, 6a, Supporting Information Fig. S2, Tables 1 & 2). Fern tracheids have at least one side of their lateral walls completely pitted (Bierhorst 1960; White 1963b; Evert 2006), and there is a higher likelihood of inter-tracheid contact in large bundles than in small bundles. Given the propensity in this species for greater integration and higher vulnerability to cavitation in more distal parts of the frond, it may be that the proportionality between pit area and tracheid volume also increases. Consequently, the higher probability of the presence of a weak, ‘rare pit’ may contribute to cavitation resistance in some ferns (Christman, Sperry & Adler 2009; Christman, Sperry & Smith 2012) though membrane thickness and porosity (Lens et al. 2011) may also play an important role.

Gas exchange in P. aquilinum was more vulnerable to mechanical damage than in W. fimbriata with its sectored xylem network. The gs measurements in situ were consistent with frond hydraulics, with both experiments showing a higher propensity for embolism spread in the integrated xylem of P. aquilinum. However, xylem connectivity may provide some benefit when sections of the network become dysfunctional. Xylem bundles within P. aquilinum frequently merge in the upper rachis (Fig. 3) and those connections appear to facilitate hydraulic redistribution where water is rerouted around the severed bundles through alternate pathways, similar to what has been reported in the leaves of woody plants (Sack et al. 2008; Scoffoni et al. 2011). Support for this hypothesis can be seen in the declines in gs between the upper and lower rachis following xylem damage (Fig. 8a,b). The greatest differences in gs between cut and uncut fronds were at 1100 h in both the upper and lower rachis, but the pinnae closest to the cut (lower rachis) showed the greatest decline. As shown in other ferns (e.g. Calkin et al. 1985), we observed increasing xylem tension towards the frond apex, a driving force that would facilitate the distal movement of water around dysfunctional bundles through functional secondary pathways.

Because the two large W. fimbriata bundles each contribute approximately 45% of the k, we expected that severing one large bundle would lead to a comparable decrease in gs. Instead, severing a single bundle resulted in no significant decline in gs, and only when two large bundles were severed (estimated 90% loss in conductivity) was there a significant decrease in gs, causing gs to fall by up to 49% compared with control plants (Fig. 8c,d). Hydraulic redistribution is likely more difficult in W. fimbriata due to the limited number of connections compared with P. aquilinium. For redistribution to work in W. fimbriata, water would have to travel the entire length of the frond in the functional bundle to the union of the two large bundles near the apex and then flow basipetally (against the water potential gradient) to the transpiring pinnae through the severed bundle, a highly unlikely scenario. Alternatively, water could flow through the few connections between small bundles, but their infrequency would limit the effectiveness of this pathway (Fig. 3). Instead, there may be internal capacitance provided by water stored in the pinnae, cortex or rhizome that compensates for short-term xylem dysfunction (Meinzer et al. 2009), a hypothesis consistent with the isohydric pattern in diurnal frond water potential (Fig. 8d). Having all of the pinnae directly connected to the two large vascular bundles appears to be a risky strategy, but the narrow, cavitation-resistant tracheids and internal capacitance appear to buffer the fronds from hydraulic dysfunction.

In comparison with xylem integration trends observed in woody plants (e.g. Zanne et al. 2006; Schenk et al. 2008), these two ferns present an interesting contrast. Following the woody plant model, for these mesic fern species one would expect integrated xylem with low conductivity, small diameter tracheids and low resistance to cavitation. Xylem within the individual bundles of both species was highly integrated (Supporting Information Figs S1 & S2), but in ferns integration must also be assessed at the scale of the frond. W. fimbriata utilized the beneficial safety features of both integrated and sectored xylem. Within the bundles conduits are small and redundant, while the bundles were spatially discrete to limit embolism spread. As a result, gs was low, but the xylem network was significantly more resistant to cavitation and less susceptible to mechanical damage. In contrast, P. aquilinum relied on a highly integrated xylem network within and between bundles; with wide, efficient conduits to maintain high gs but at the cost of low cavitation resistance. Indeed, there was a clear trade-off between safety and efficiency, as xylem-specific conductivity was significantly higher in P. aqulinum, which also suffered from lower cavitation resistance, similar to trade-offs present in woody plants (Wheeler et al. 2005; Loepfe et al. 2007). To overcome the effects of drought, W. fimbriata relies on safe xylem networks and hydraulic capacitance of the surrounding tissue rather than hydraulic redistribution and a phenology targeted at early season growth prior to the seasonal drought common in this Mediterranean climate as seen in P. aquilinum.

Recent research into the evolution of stomatal control (Brodribb & McAdam 2011) raises a number of questions related to fern physiology. Evidence for stomatal control in ferns is mixed, with reports of active stomatal control in P. aquilinum to avoid excessive water loss in open high-light areas, but not in ferns in occupying moist, shaded areas (Nobel, Calkin & Gibson 1984). Conversely, others report minimal stomatal response to changes in vapour pressure deficit in ferns (including P. aquilinum) (Roberts, Wallace & Pitman 1984; Franks & Farquhar 1999). The most recent research suggests a passive, hydraulic response in ferns, likely the result of the lack of an ABA signalling mechanism in fern stomata common among angiosperms (Brodribb & McAdam 2011). Our gas exchange experiments showed dramatic decreases in gs in response to mechanical damage to the xylem in the drought-deciduous P. aquilinum fronds, and suggest a passive, hydraulic response in the control of stomata closure that is consistent with the findings of Brodribb & McAdam (2011). An ABA-regulated stomatal response may not be as critical for ferns because many are understorey species that prefer moist habitats.

This study illustrates the functional consequences of xylem organization on cavitation resistance and gas exchange and highlights two extremes in phenology in two coexisting ferns. Our results raise a number of questions to be addressed in future research. Because pit membranes play such a critical role in cavitation resistance, research is needed to determine what pit membrane characteristics are important to cavitation resistance in ferns (Wheeler et al. 2005; Choat & Pittermann 2009). The weak or highly porous pit membranes reported for a number of ferns (Carlquist & Schneider 2007) suggest that xylem organization may be even more important for fern cavitation resistance compared with angiosperms, particularly in light of the higher safety margins and conservative drought-resistance strategies employed by ferns when faced with desiccation (Brodribb & Holbrook 2004).


The authors would like to thank the NSF (IOS-1027410 to J.P.) for supporting this research. We would also like to thank E. Limm, C. Rico and S. Bastami for their assistance in the collection of hydraulics and anatomical pilot data. Holly Forbes at the UC Berkeley Botanical Garden kindly provided access to plant material for 3D imaging. We also thank Alastair MacDowell and Dula Parkinson of the Lawrence Berkeley National Laboratory Advanced Light Source microtomography beamline 8.3.2 for their assistance in 3D imaging. The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the US Department of Energy under contract no. DE-AC02-05CH11231.