Discussion
This study provides the first unambiguous experimental proof that leaf hairs reduce the boundary layer conductance for heat transfer between leaves and the air (g bh). Our approach was carefully designed to exclude all potentially confounding influences: we inferred g bh in bald patches and in otherwise identical hairy patches from the time constant for cooling following transient radiative heating, accounted for differences in leaf surface conductance to water vapor and leaf heat capacity, and ensured that the radiative environments of each patch were identical and constant over time. Cooling was faster and boundary layer conductance was greater in the bald patch than in the hairy patch in nearly all experiments, by 2.4 to 38.7% (means within species). This finding could not be attributed to differences in stomatal conductance or other components of energy balance between patches. Our results thus validate the hypothesis that leaf hairs slow convective heat exchange between leaves and air.
Several previous studies have examined the effects of leaf hairs on leaf energy balance or gas exchange, though none was able to directly quantify the effect of hairs on boundary layer conductance in intact leaves in isolation from potential confounding factors. Benz and Martin (2006) found no correlation between trichome cover and gas exchange rates across 12 Tillandsia species, and concluded that hairs had negligible influence on boundary layer conductance; yet, other factors that differed across the 12 species may have confounded the observed differences in gas exchange rates. Wuenscher (1970b) observed higher leaf temperatures and lower transpiration rates in shaved versus unshaved leaves of Verbascum thapsus , and attributed these differences to a 90-fold increase of boundary layer resistance by hairs. However, Parkhurst (1976) reassessed Wuenscher’s results and concluded that hair removal would likely have increased boundary layer resistance two-fold at most, and suggested a role for cuticular or epidermal damage caused by hair removal; differences in stomatal conductance may also have played a role. Ripley et al (1999) measured surface conductance to water vapor by gas exchange before and after removal of hairs fromArctotheca populifolia leaves, and inferred that the hairs generated a resistance of around 0.2 m2 s mol-1; however, as in Wuenscher’s (1970) study, those results could have been influenced by other unobserved changes in leaf surface vapor transport caused by hair removal. Woolley (1964) observed a substantial reduction in wind speed near the surface (both above and within the hair layer) of hairy soybean leaflets as compared to shaved leaflets, which is consistent with an increase in the effective boundary layer thickness and a reduction in g bh. Measurements of water loss were more ambiguous: variability obscured any effect of hairs in excised, living leaflets, whereas shaving increased transpiration rate by 21% in leaflets that had been killed by boiling. Meinzer and Goldstein (1985) simulated effects of hairs on energy balance in Espletia timotensis by assuming hairs entrained a layer of still air equal to their depth and that hairs did not affect absorptance; their simulations reasonably matched observations of greater leaf temperatures in hairy as compared to partially-shaved leaves. Amada et al. (2021) observed lower in situ leaf temperatures in shaved versus hairy leaves of Metrosideros polymorpha ; having verified that surface conductances to water vapor did not differ between the two treatments, the authors attributed the result to reduction in g bh, and thus in convective cooling. That result could also have arisen from effects of leaf hairs on radiation absorption, although hairs in M. polymorpha are light in color and thus probably reduced radiation absorption rather than the converse.
In the present study, we directly quantified g bhitself in hairy and bald leaf patches, and eliminated all factors that could have confounded the results of previous studies described above: (1) We eliminated the confounding influence of differences in light absorption by using dynamic IR thermography rather than steady-state temperature to infer differences in g bh, which enabled us to estimate g bh per se despite any differences in light absorption. (2) We eliminated the confounding influence of differences in leaf size and shape between bald and hairy leaves by choosing adjacent bald and hairy patches identical in size and position relative to the leading edge of the leaf, and ensured identical wind speeds by measuring both patches simultaneously in the same air stream. (3) We eliminated the confounding effects of transplanting hair layers or boiling leaves by measuring only intact leaves from which hairs had been either left in place or removed. (4) We eliminated the confounding influence of differences in stomatal conductance by measuring surface vapor conductance directly in each patch with a porometer immediately before measurements and incorporating those data into our inference of g bh. Surface conductance was very low in general, but was marginally lower on average in bald patches than in hairy patches, ruling out evaporative cooling as a general explanation for faster cooling in bald patches.
Our results do not appear to confirm the theoretical prediction that hairs add a resistance, r h, proportional to the depth of the hair layer (d h). While hair depth and theoretical r h varied nearly 6-fold across species in this study (from 0.04 to 0.80 m2 s mol-1 for r h), our experimental estimates of r h – based on observed differences in g bh between bald and hairy patches – were unrelated to theoretical predictions (Fig 7), although they spanned a similar range. The experimental estimates also depended sensitively on the value of an unknown parameter: the ratio (β ) between the boundary layer resistances at the adaxial and abaxial surfaces in the absence of hairs. It may seem reasonable to assume β ≈ 1. However, in three of our study species (MAGR, OLEU and SOLY), the edges of each leaf had a slight downward curl that may have slightly sheltered the abaxial surface from wind, which could have increased boundary layer resistance at that surface independent of hairs, making β less than 1.0. A similar effect was reported by Grace and Wilson (1976), who found lower wind speeds just behind the downward-curled leading edge of the abaxial surface in Populus × euramericana . β< 1 would bring the results for r h in OLEU and MAGR closer to the theoretical predictions: observed and theoretical r h would match at β = 0.47 for MAGR and β = 0.67 for OLEU (Fig S4). In summary, our data suggestr h is at best weakly related to hair layer depth, but is probably on the same order of magnitude as theoretical predictions based on hair layer depth.
Our results also initially appear inconsistent with modeling by Schreuder et al. (2001), which suggested that leaf hairs mayenhance g bh by promoting turbulence near the leaf surface. However, turbulence was predicted to emerge only above a critical windspeed that depended on trichome height; e.g., for 2.5 cm leaves with 500 µm trichomes, the critical windspeed was 2.18 m s-1, and the critical windspeed increased as trichome height decreased and leaf size increased. In our experiments, wind speed was below 2 m s-1 in most experiments, trichomes were shorter than 300 µm in all species except SOLY (Table 2), and leaves were >> 2.5 cm wide except in OLEU. Thus, our data neither contradict nor confirm the predictions of Schreuder et al. (2001). It is interesting to note, however, that the largest deviation between theoretical and experimental estimates ofr h in this study was in SOLY, which also had the longest trichomes by far. The fact that experimentalr h was less than half the theoretical prediction for SOLY, regardless of what value of β we assumed, may hint at promotion of turbulence by tall hairs as proposed by Schreuder et al. (2001).
We could not rigorously scale up our estimates ofg bh from the patch scale to whole leaves – the estimates reproted here refer only to the individual patches of leaf measured within each experiment – for three reasons. First, we could not locate patches at a constant distance from the leaf edge across experiments, due to the difficulty of finding leaf patches that we could successfully shave without damaging the leaf. Second, it was impossible to remove hairs from entire leaves in most cases. Third, our results did not validate the estimate of hair-layer resistance from hair-layer depth; had they done so, then we could have concluded that the effect of hairs is independent of leaf size, in which case we could merely add a calculated hair-layer resistance to predictions from existing models ofg bh in relation to leaf size and wind speed. For example, prior theory predicts that g bh is proportional to the square root of the ratio of wind speed (v w, m s-1) to leaf size (d l, m), thus: g bh ≈ 0.267(v w/d l)0.5, where g bh is a whole-leaf (two-sided) value in mol m-2 s-1 (Gates, 1968; Nobel, 1975). In the context of that theory, the range of % decreases ing bh that we observed (means of 2.4 to 39% across species) correspond to the effect, predicted by the theory, of a 4.9 – 93% increase in leaf size or a 4.6 – 48% decrease in wind speed. An important caveat is that our data apply only to leaf patches located 2 – 6 mm from the leading edge of a leaf. Given that the wind speed at a given position above the leaf surface declines progressively with distance from the leaf edge (e.g., Grace and Wilson, 1976), the boundary layer resistance in the absence of hairs likewise increases with distance from the leaf edge, which would make hair-layer resistance a smaller proportion of the total resistance in locations farther from the leaf edge. Thus, had we located patches farther from the leaf edge, we may have found smaller percent increases in boundary layer resistance due to hairs.
Authors have speculated for decades about how leaf hairs affect photosynthesis (A ) and transpiration (E ) via boundary layer conductance. Hairs have a wide range of reported effects on gas exchange (e.g., Ehleringer and Mooney, 1978; Baldocchi et al., 1983; Ripley et al., 1999), but it is typically difficult to disentangle the effect mediated by g bh from other effects of hairs, such as on radiation balance. For example, a seminal study by Ehleringer and Mooney (1978) reported large suppression of both Aand E by hairs in Encelia farinosa as compared to its close hairless relative E. californica , though these effects were largely driven by increased reflectance due to hairs; photosynthesis rates were similar with or without hairs after correcting for differences in light absorption, indicating little or no effect of hairs on boundary layer resistance. Amada et al. (2017) found small differences in gas exchange rates in M. polymorpha , consistent with the small (< 10%) increases in total CO2or H2O transport resistance due to trichomes predicted from the thickness of the trichome layer. Our modeling suggests theg bh-mediated effects of hairs on gas exchange should depend sensitively on conditions. A hair-layer resistance of 0.1 m2 s mol-1 – roughly the value we estimated for OLEU – would change instantaneous WUE by -1.7% to +2.9% depending on conditions, according to our model. For example, hairs would increase WUE by nearly 2% at a stomatal conductance of 0.5 mol m-2 s-1 or by 1.5% at a PPFD of 250 µmol m-2 s-1, or decrease WUE by 1.5% in a 1 mm leaf at 10oC (Fig 8). Generally, hairs should tend to improve WUE when stomatal conductance, air temperature or VPD is high or when PPFD is subsaturating. Hairs should influence WUE to a greater degree in small leaves, though the direction of the effect depends on air temperature.