Discussion
Experiment 1 and 2 show that spreading of drought-induced embolism and thus embolism resistance can be strongly affected by the proximity of the xylem area studied to pre-existing embolism such as cut xylem tissue. The observation that embolism initiation (PEP12) occurs over a > 1MPa range of xylem water potential for four out of six species studied challenges the assumption that embolism spreads once a certain threshold difference has been reached. While cut-open conduits may facilitate embolism spreading, hydraulic segmentation may limit this potential artefact. As predicted, embolism spreading in minor veins was also affected by the proximity to cut conduits, but showed a rather limited, local distribution without affecting the entire leaf. Also, we found no difference between the optical and pneumatic method for five species, despite considerable variation in the P12 values of both methods. The consequences and broader significance of these findings present some old questions and assumptions in a new light, and provides a novel gas diffusion hypothesis as a possible mechanism for embolism spreading.
Embolism spreading depends on pre-existing embolism
Since embolism spreading happens largely from one embolised conduit to a neighbouring one, as predicted by the air-seeding hypothesis (Zimmermann, 1983; Sperry and Tyree, 1988), the observation that embolism formation may depend on proximity to an existing gas source such as cut-open conduits or pre-existing embolised conduit is not surprising. Embolism formation appears to be unlikely if a conduit is not connected to a pre-existing embolism. Novel, de novo embolism formation has been observed in very few conduits that are not connected to embolised ones based on microCT (Brodersen et al. , 2013, Choatet al. , 2015, 2016), and embolism formation in seemingly isolated conduits could occasionally be observed in our experiments. However, the rather two-dimensional view associated with the optical method, its limited resolution to accurately detect narrow vessel ends (Oskolski and Jansen, 2009), and its shortcoming to detect pre-existing embolism, did not allow us to confirm that these conduits were completely disconnected from neighbouring gas sources.
If availability of a pre-existing gas source or embolised conduit is important, then where does the gas come from to induce embolism in intact xylem? It is possible that there is almost always an embolised conduit available, perhaps in primary xylem or in older xylem from an older growth ring. This would be an obvious gas source if functional, sap-filled conduits show any direct connection with these embolised conduits via bordered pits. However, these pre-existing gas sources may be limited due to poor connectivity or compartmentalisation of the hydraulic network (Kitin et al. , 2004; Morris et al. , 2016). Since vessels and tracheids do not share pits with non-conductive fibres (Sano et al. , 2011), it is unlikely that air entry from these cells or intercellular spaces will contribute to embolism formation in conduits.
Hydraulic segmentation reduces embolism spreading in xylem tissue
While the proximity of a studied xylem area to cut conduits seems to be important, the speed of embolism spreading over a certain distance also depends on the vessel dimensions. If each individual vessel would embolise separately (Johnson et al. , 2020), wide and long vessels would show a faster propagation of embolism over a given distance than narrow, short vessels. Spreading of embolism would especially be reduced in xylem with a high degree of hydraulic constrictions, making xylem patches at the distal side of embolised conduits seemingly more resistant to embolism. The four species that showed a reduced embolism resistance in detached leaves as compared to leaves attached to branches, have open vessels running directly from the base of the petiole into the midrib (Table 1, Figure 2). Since the maximum vessel length in petioles of L. tulipifera and B. pendula were shorter than the petiole length (Table 1), both species showed relatively small differences in embolism resistance between detached leaves and leaves attached to a stem segment (Fig. 1c, e).
Wide and long vessels in the midrib and secondary veins were found to embolise before the high 3rd to 5thvein orders. This pattern confirms various studies based on the optical method and microCT observations (Brodribb et al. , 2016a, b; Scoffoni et al. , 2017; Klepsch et al. , 2018). However, the observation of local spreading of embolism in minor veins near cut vessels in experiment 2 supports the hypothesis that embolism can spread from pre-existing gas sources, and in minor veins prior to embolism formation in the large vessels of major veins (Fig. 3). This observation suggests that proximity to a gas source is the main driver for embolism spreading, and not the conduit diameter per se. Nevertheless, the limited and short dimensions of minor veins (Chatelet et al. , 2006; Lechthaler et al. , 2019; Hua et al. , 2020) are a plausible explanation for why artificially induced embolism near cuts in minor veins propagate locally, rather than spread across the lamina. Short conduits in minor vein orders may have been selected for in order to maximise leaf hydraulic function over the life span of a leaf. Minor orders of leaf veins are believed to be prone to xylem embolism formation over the life of a leaf due to localized damage caused by insects, pathogens or mechanical damage to the lamina, and because they sustain the most negative pressures (Brodribb et al. , 2010). Considerable restriction to the extent of embolism spread through minor vein orders would ensure a limited impact on the hydraulic capacity of a leaf that sustains localised damage to the lamina.
It is also possible that wide and long vessels are more likely connected to a pre-existing embolism than narrow, short conduits, and that large vessels show a higher amount of intervessel pit membrane area than narrow, short vessels. In other words, the reason why large and wide vessels are likely to embolise first, may reflect a difference in the rate of air entry, which is caused by their connectivity to a gas source, and not any inherent difference in embolism resistance per se. We are not aware of an alternative mechanism that would explain why wide conduits are more vulnerable to embolism than narrow ones. Pit membrane thickness, which is strongly associated with embolism resistance (Liet al. , 2016; Kaack et al. , 2019), was found not to be related to conduit diameter (Kotowska et al. , 2020; Wu et al. , 2020).
Vulnerability segmentation may reflect hydraulic segmentation, but not intrinsic differences in embolism resistance per se
The observation that embolism formation in leaves of an individual tree may occur under highly variable (> 1 MPa) xylem water potentials, suggests that embolism resistance may represent a relative trait that does not capture the absolute, intrinsic embolism resistance of its xylem. It is likely that hydraulic segmentation includes highly reduced conduit dimensions, especially with respect to conduit length and width, with a high number of interconduit end walls over a short stretch of xylem tissue. Conduit end walls have been suggested to hold up embolism spreading at least temporarily, with pit membranes functioning as safety valves and preventing further spreading of embolism due to their tiny pores (Zhang et al. , 2017, 2020; Kaacket al. , 2019; Johnson et al. , 2020). Moreover, narrow and short tracheids or fibriform vessels may be more confined than long and wide vessels, with a small interconduit pit membrane area for air entry in narrow tracheids.
Quantifying embolism resistance across the entire xylem pathway could be complicated by measuring artefacts or the proximity to pre-existing embolism, such as cut xylem. In our study, comparison of xylem embolism resistance of leaves attached to long stem segments with embolism resistance of stem xylem based on previous papers (Klepsch et al. , 2018; Zhang et al. , 2018), indicates that leaf xylem of all six species was between 0.5 and > 1 MPa more embolism resistant than stem xylem. Our result for B. pendula was consistent with Klepsch et al. (2018), with leaf xylem being more resistant than stem xylem. Most angiosperms species, however, showed that stem xylem was either more embolism resistant, or equally resistant than leaf xylem (Zhu et al. , 2016; Skelton et al. , 2018; Losso et al. , 2019). Therefore, caution is needed to directly compare absolute values of embolism resistance between organs, since measured values of embolism resistance could be relative estimations only, especially if destructive methods are used and cut-open xylem accelerates embolism spreading.