Olson (2014; in this issue of New Phytologist, pp. 7–11) provides a valid criticism of our recent commentary (Nardini & Jansen, 2013). The observation by Feild & Brodribb (2013) that high venation density (> 12 mm mm−2) in angiosperm leaves is associated with the evolution of simple perforation plates can indeed help to explain the Baileyan evolutionary trends, as pointed out by Olson, even though vessel diameter, a highly important hydraulic trait, was not measured by Feild & Brodribb (2013). We agree with Olson that neither Bailey nor other students from the Baileyan school provided causal explanations for the transition from scalariform perforation plates to simple ones (Bailey & Tupper, 1918; Frost, 1930; Bailey, 1944), which readers could misinterpret from the title of our commentary. Nevertheless, the fact that Bailey correctly described a trend even without providing a valid mechanistic explanation reveals that not only evolution, but science itself, is far from being a linear process. Clever and careful observations of patterns, even if not supported by rigorous hypotheses about the underlying functional processes, may turn out to be equally as important for the long-term progress of our understanding of nature as ambitious process-based models devoid of observational and/or experimental support. Although we see no controversy surrounding Bailey's functional and evolutionary ideas, we would like to pay attention in this letter to two related concerns: (1) the integration of experimental data from the field of plant hydraulics into traditional, comparative wood anatomy; and (2) the reversibility of evolutionary transitions between perforation plate types that form a continuum.
Based on the evolutionary switch from tracheids to vessels, and from a narrow, long vessel element with scalariform perforation plates to a wide, short vessel element with simple perforation plates, it was thought for several decades that a taxon with simple perforation plates would not give rise to a taxon with scalariform perforation plates (Dickison, 1975). However, a functional interpretation of vessel element length and perforation plate morphology remains, even nowadays, poorly tested and understood (Schulte & Castle, 1993; Ellerby & Ennos, 1998; Schulte, 1999; Christman & Sperry, 2010). There is no lack of speculation over the functional processes behind vessel evolution and perforation plate morphology, and the following hypotheses have been reported for species with scalariform perforation plates (Christman & Sperry, 2010): (1) increased flow resistance; (2) the trapping of freezing-induced air bubbles, which may prevent formation of large air bubbles; (3) increased rates of vessel refilling; and (4) increased mechanical support to prevent implosion of the vessel.
Carlquist's Ratchet hypothesis provides a useful framework to understand the ecological and taxonomic distribution of species with scalariform and simple perforation plates, because it is based on the idea that simple perforation plates are consistent with reduced resistance to flow. This hypothesis is in line with the ecological and comparative wood anatomy approach (Carlquist, 2001), which aims to find large-scale global patterns in how environmental conditions may drive wood anatomical variation. More specifically, simple perforation plates are assumed to have evolved in taxa and environments that require high hydraulic xylem efficiency, and this shift from scalariform to simple perforation plates is very likely to have occurred many times independently during evolution. The higher hydraulic resistance provided by oblique perforation plates with scalariform bars, however, may not be disadvantageous in areas with low transpiration rates, such as tropical mountains and cool, temperate environments (Baas, 1976; Carlquist, 2001; Jansen et al., 2004). This means that species with scalariform perforation plates represent a relict in these environments, where narrow vessels provide a useful strategy to avoid frost-induced embolism formation (Ewers, 1985). The ‘ratchet’ idea presented by Olson suggests that species with simple perforation plates are unlikely to show a reversal to scalariform perforation plates, even though the potential gain in hydraulic efficiency may no longer be of any functional advantage after migration towards cooler environments with low evaporative demands. However, Feild et al. (2002) suggested that vessels in Winteraceae were lost as an adaptation to freezing-prone environments, which is supported by their ecological abundance in temperate wet environments and tropical alpine habitats.
Together with the ecological approach of Carlquist's Ratchet hypothesis, we believe that progress in understanding the driving causes behind perforation plate morphology and xylem evolution in general should rely on tight integration of anatomical, ecological, physiological and phylogenetic evidence at both the xylem tissue and whole-plant level. The traditional idea that comparative wood anatomists are able to work on a much larger number of species than are xylem physiologists may no longer be valid when considering the amount of experimental data that has been accumulated over recent decades on temperate and (to a lesser extent) tropical plant species. One example of how progress can be achieved through synthesis would be the construction of large-scale datasets, such as the Xylem Functional Traits (XFT) Database, which was compiled by a large group of xylem physiologists and ecologists in order to bring together xylem physiological parameters with wood anatomy and climate (Choat et al., 2012).
The XFT dataset demonstrates that perforation plate morphology is strongly related to the mean vessel diameter (D, μm; Fig. 1a) and the hydraulically weighted vessel diameter (DH, μm; Fig. 1b). This finding has been known for many years and explains the fact that scalariform perforation plates can be limited to narrow vessels in juvenile wood and narrow latewood vessels (e.g. Fagus, Platanus). Scalariform perforation plates typically occur in species that have D values below 30 μm. This type of perforation plate is almost completely absent when mean vessel diameters exceed 50 μm. Using the InsideWood Database (Wheeler, 2011), the combination of exclusively scalariform perforation plates and very wide (≥ 200 μm) vessels includes no records (from 6648 modern wood descriptions), and only 55 records show the combination of exclusively scalariform perforation plates with a mean tangential vessel diameter of 100–200 μm.
The difference in DH between species with simple and scalariform perforation plates results in a highly significant difference in the xylem specific conductivity (KS, kg m−1 MPa−1 s−1) between the two groups (Fig. 1b): average KS values for species with scalariform and simple perforation plates are 1.21 kg m−1 MPa−1 s−1 (± 1.03 SD, n = 83) and 2.58 kg m−1 MPa−1 s−1 (± 2.66 SD, n = 690), respectively (P < 0.001 based on Welch's t-test). Moreover, there is a significant difference in photosynthetic capacity (AMAX, μmol m−2 s−1) between species with simple and scalariform perforation plates (Fig. 2a), which is not surprising given the reported correlation between KS and AMAX (Choat et al., 2011). However, an ANCOVA test shows that the slope of the DH vs KS regression lines does not differ significantly (P = 0.2) between the two groups (Fig. 1b). This means that species with narrow (< 50 μm) vessels and simple perforation plates do not have significantly higher KS values than species with narrow vessels and scalariform perforation plates. This finding is in line with Carlquist's Ratchet hypothesis and may also explain why Smith et al. (2013) found no difference (P > 0.1) in KS between 17 species with scalariform perforation plates and 31 species with simple perforation plates.
With respect to the climatic parameters taken from the XFT Database, the strongest influences (P < 0.001) on perforation plate morphology are temperature related: average values of the mean annual temperature are 13.3°C (± 6.8, n = 142) and 16.8°C (± 6.2, n = 864), and the average numbers of ground frost days are 83 (± 78, n = 133) and 55 (± 61, n = 771), for species with scalariform and simple perforation plates, respectively. A highly significant difference is also found for potential evapotranspiration (PET, mm per year), with mean values of 1093 (± 299, n = 134) and 1282 (± 302, n = 845) mm per year for both groups (Fig. 2b). Mean annual precipitation (MAP), however, is not associated with perforation plate morphology: average values of MAP are 1091 (± 761, n = 142) and 970 (± 765, n = 872) mm per year for species with scalariform and simple perforation plates, respectively. The XFT Database also confirms the latitudinal and altitudinal trends in perforation plate morphology, which have been illustrated repeatedly (Baas, 1976; Jansen et al., 2004; Lens et al., 2004).
Although we agree with the overall irreversible trend of scalariform to simple perforation plate evolution, there is no available evidence to believe in its exclusively irreversible nature. Although there is rampant parallel evolution from scalariform to simple perforation plates, students should be alert for possible reversals (Baas & Wheeler, 1996). There are many records of species with the mixed occurrence of simple and scalariform perforation plates in mature wood tissue. Oskolski & Jansen (2009) demonstrated that, for three Araliaceae species, scalariform perforation plates are not randomly distributed, but limited to narrow vessels and vessel endings (Fig. 3a). These findings may suggest that differentiating vessel elements are under the control of positional morphogenetic signals, either through genetic control via one or more morphogens and/or hormonal effects, such as an auxin gradient. Scalariform and ‘unusual’ perforation plates are frequently reported in perforated ray cells in wood that is dominated by simple perforation plates. Moreover, there are many records of various (up to eight) ‘simple’ perforations closely grouped together in a single end wall (Jansen et al., 1997; Fig. 3b,c), ‘combination plates’ with mismatching pairs of simple to scalariform perforation plates (Meylan & Butterfield, 1975) and ‘scalariform’ perforation plates with vestigial bars (Oskolski & Jansen, 2009). These observations make the difference between simple and compound (scalariform, reticulate, foraminate) perforation plates fuzzy, raising the question of how we define these types of perforation plates, and may suggest that a species with (mainly) simple vessel perforation plates is likely to retain the genetic and developmental ability to develop scalariform perforation plates, which could make the evolution from simple to scalariform perforation plates possible, as suggested, for instance, in Meryta (Oskolski & Jansen, 2009).
In conclusion, progress should be made in testing functional hypotheses through the synthesis and integration of ecological, anatomical, phylogenetic and physiological data. Preliminary data from the XFT Database confirm Carlquist's Ratchet hypothesis, but we caution students that vessel evolution and the transition from scalariform to simple perforation plates may not be entirely irreversible. Exploring the degree of irreversibility using well-resolved phylogenies that are time calibrated and linked to palaeoenvironmental reconstructions remains the challenge ahead (Zanne et al., 2014). A useful alternative approach would be to compare the neutral genetic variability (FST) and quantitative genetic differentiation (QST) of wood anatomical features in order to identify the evolutionary forces acting on these traits. Finally, as it is feasible to reconcile the traditional, comparative approach of studying xylem evolution with empirical, physiological methods, readers interested in xylem structure–function relationships are invited to contribute in a collaborative spirit to initiatives such as the XFT Database, which will be made available via the TRY Plant Trait Database (http://www.try-db.org/TryWeb/Home.php).