Commentary: What We Know About Stemflow's Infiltration Area

Department of Geography and Environmental Studies, Thompson Rivers University, Kamloops, BC, Canada, Department of Disaster Prevention, Meteorology and Hydrology, Forestry and Forest Products Research Institute (FFPRI), Ibaraki, Japan, Department of Technology Assessment and Substance Cycles, Leibniz Institute for Agricultural Engineering and Bioeconomy, Potsdam, Germany, 4 Institute of Environmental Assessment and Water Research, Spanish National Research Council (IDAEA-CSIC), Barcelona, Spain, 5 Soil Science, Institute of Geography, Friedrich Schiller University Jena, Jena, Germany, University of Tsukuba, Ibaraki, Japan, Department of Geography & Spatial Sciences, University of Delaware, Newark, DE, United States, Department of Plant & Soil Sciences, University of Delaware, Newark, DE, United States


INTRODUCTION
Stemflow represents the portion of precipitation routed by vegetation to the base of tree boles or plants stems. Van Stan and Allen (2020) (herein referred to as VS&A) is a mini review of studies that have attempted to quantify the infiltration area of stemflow once it has reached the soil surface, I T . More specifically, VS&A provide an overview of: (i) the ability of vegetation canopies to funnel rainfall; (ii) the various approaches used to estimate or measure the size of I T ; (iii) the different soil properties that may influence the magnitude of I T , and (iv) the potential for and limitations to using dye and stable isotope tracers in I T research. The objectives of this commentary are to: (i) highlight and expand upon important points raised by VS&A in order to advance the understanding of the controls regulating the size of I T , and (ii) provide corrections to and clarification of prior I T results presented in VS&A.

ADVANCEMENT OF THE SCIENTIFIC UNDERSTANDING OF STEMFLOW INFILTRATION AREA, I T
VS&A state the importance of stemflow in the hydrology and biogeochemistry of vegetated environments is dependent upon I T size. These authors rightfully note that there is a need for further research, especially in natural forest systems, to characterize the size of I T . Previous studies (e.g., Iida et al., 2005;Chinen, 2007) have estimated the magnitude of I T using litter marks (the displacement of leaf litter) or soil scour marks caused by the excess overland flow of stemflow. As VS&A state, litter and scour marks are difficult to interpret quantitatively as they neither represent mean nor maximum I T for a given storm. As such, litter and scour marks have little utility estimating I T .
VS&A correctly state that factors, such as soil hydrophobicity, could influence stemflow infiltrability in certain environments. Nonetheless, the methodology of Herwitz (1986), in which I T values are derived by dividing the stemflow volumetric input rate by the infiltration capacity of the surface soil (i.e., the saturated hydraulic conductivity, K sat ), remains a theoretically sound approach. What is important to highlight is that in situ measurements of K sat , as a surrogate for stemflow infiltrability in the proximal bole/stem area that include the effect of macropore flow (i.e., K sat measured with no tension; hydraulic head = 0 cm) are likely to be more representative of the actual infiltrability of stemflow than K sat measured using tension or K sat values derived from pedotransfer functions [e.g., ROSETTA model- Schaap et al. (2001)], which estimate soil matrix K sat .

CRITIQUE OF REPORTED FORMULA AND FINDINGS OF PREVIOUS RESEARCH
VS&A (page 1) suggest that the following equation (Equation 1 in VS&A) is the funneling ratio derived by Herwitz (1986): where F is the funneling ratio (dimensionless), S T represents stemflow volume (L tree −1 ), P is precipitation depth (mm), and I T is the stemflow infiltration area (m 2 tree −1 ). The funneling ratio proposed by Herwitz (1986), however, differs from that of Equation (1) in that the basal area of the tree bole, B (m 2 ), rather than I T , is multiplied by P in the denominator of the equation: VS&A (page 2) also suggest that ". . . Herwitz's (1986) equation for F employs the concept of I T . . . "; however, Herwitz (1986) never advocated that B was a surrogate for I T or that B played any role in I T size. Instead, and as aforementioned, Herwitz (1986) derived I T by taking the stemflow input rate and the infiltration capacity of the surface soil into account, and the derived values of I T were markedly different than B. VS&A (page 3) cite various studies supporting their claim that "there are pieces of evidence that suggest that I T is larger, 10 −1 to 10 1 m 2 , than the areas assumed elsewhere, e.g., 10 −4 -10 −1 m 2 (Iida et al., 2016;McKee and Carlyle-Moses, 2017;Carlyle-Moses et al., 2018)". Iida et al. (2016) make no mention of I T (or stemflow) and it is unclear why this study was cited. Furthermore, the range of I T provided by Carlyle-Moses et al. (2018) is for conditions of average rainfall / stemflow input rates within mature, natural forests. They are not representative of extreme precipitation events (e.g., Herwitz, 1986) nor orchards or agricultural fields (e.g., Keen et al., 2010) where soil compaction may reduce stemflow infiltrability. The majority of prior studies report the maximum extent of I T (e.g., Voigt, 1960;Pressland, 1973) or use "litter marks" or erosional soil scouring for estimating I T (e.g., Iida et al., 2005;Chinen, 2007) which simply do not provide reliable quantitative evidence of average I T . Litter marks may be seasonal and are at least episodic phenomena persisting across events (e.g., Iida et al., 2005). Litter marks are not created during low intensity events (as stated by VS&A) but rather during peak periods of heavier rain with high stemflow funneling. What does emerge from Table 1 is that studies conducted thus far using in situ dye experiments and direct observations of stemflow infiltration or studies utilizing physically-based approaches such as dividing the stemflow input rate by the soil K sat suggest that I T associated with average rainfall and stemflow rates are limited < 1 m 2 tree −1 in environments (e.g., mature, natural forests) where the soil infiltrability can be expected to have a magnitude of order of 1 x 10 2 or 1 x 10 3 mm h −1 . Additionally, the findings presented in Table 1 suggest that I T ≥ 1 m 2 tree −1 may sometimes arise during large / extreme rainfalls and stemflow rates in these forest environments and under relatively smaller rainfall and stemflow rates in environments (e.g., agricultural plantations, orchards, agroforestry areas, and urban environments) where infiltrability is likely < 1 x 10 2 mm h −1 .  Tanaka et al. (1991;1996). Mean basal areas of the 6 species ranged from 1.5 × 10 −2 to 9.1 × 10 −2 m 2 .

Dye Experiment
Blue dye was applied to the downslope sides of two sycamore trees in Aberdeenshire, UK using a 20-L backpack sprayer for 35 min resulting in an equivalent rainfall intensity of 45.7 mm h −1 to identify areas of double-funneling. It should be noted that the authors describe the precipitation at the site as being characterized by frequent, low-intensity rain events. The two trees were part of a stand of trees found on a 20.3 ± 11.6 • slope. K sat of the soil was 256 mm h −1 .
No data Correction: VS&A state that, based on correspondence with the corresponding author of the article, the dye extended 1.27 and 0.63 m downslope of the two study trees. VS&A use the distance the dye extended downhill as the radius of the I T areas; however, the dye stained I T areas are clearly not circular and occupies only a fraction of the areas suggested by VS&A [see Figure 2B. of Gonzalez-Ollauri et al. (2020)].

SAVANNA AND SHRUBLAND Chinen (2007) Erosional Scour Marks and Rills
The extent of scour marks, including rills, were measured and assumed to be associated with stemflow produced during an intense rainfall from three tree species occupying an immobile sand dune in the Republic of Niger. The rainfall depth was 20.7 mm rainfall in which the bulk of the rain fell within 20 min (intensity c. 60 mm h −1 ).
No data Single extreme event Clarification: I T was not measured, but the extent of traces of surface runoff and rills extended c. 4 to 7 m in the downslope direction of the trees. There is no mention of I T varying from 1.12 to 4.75 m 2 tree −1 as indicated in Table 1 of VS&A, although it seems VS&A based their values on Figure 4 of Chinen (2007), which provides a sketch and scale of the traces of erosional scours and rills that developed during an extreme rainfall event.
0.10-1.14 Range of maximum I T extents for individual trees over 18 months Clarification: Observed infiltration was constrained to within 0.15 m of the boles of small trees and 0.45 m of large tree boles. This suggests, taken the basal area of the trees into account, the given maximum I T range for rain events up to 120 mm.

Pressland (1976) Direct Observations
Arid woodland in proximity to where the Pressland (1973) study took place. Stemflow was not measured, but stemflow infiltration was observed during rainfall events.
No data Clarification: Stemflow infiltrated to within 50 cm of large trees (circumference > 40 cm, basal area > 1.27 x 10 −2 m 2 ) and to within 30 cm of small trees (circumference < 20 cm, basal area < 3.18 x 10 −3 m 2 ). It is not possible to derive I T with available information, but likely on the order of that for Pressland (1973).  Charlier et al. (2009), stemflow is shown to not be the main contributor to overland flow from the plot. Extreme example due to special morphology of banana plants (funnel like shape). Gómez et al. (2002) Stemflow Rate divided by K sat Stemflow measured from three mature olive trees (mean basal area of 5.3 x 10 −2 m 2 ) in an orchard situated in Spain. I T estimated as mean stemflow rate divided by measured K sat of 81 mm h −1 .

0.108
Average I T value for three trees over 12 rain events Clarification: I T average presented in this table was calculated for the three trees for the study period (12 rain events). For the largest rain event (77.1 mm) I T for the three trees averaged 0.762 m 2 tree −1 (range = 0.53 to 1.12 m 2 tree −1 ).

(Continued)
Frontiers in Forests and Global Change | www.frontiersin.org

Keen et al. (2010) Erosional Scour Marks
Macadamia orchard, Australia. Seven 9-year-old trees were sampled for stemflow and the erosion associated with stemflow was monitored. Total study-period rainfall depth was not provided but did include a 217 mm event with a mean intensity of 6.8 mm h −1 . No estimate of K sat is provided; however, the authors state that the exposed soil is inherently erodible and is also subjected to erosion during harvesting.

2.1
Maximum I T extent for any tree over 16 months Correction: Study period was 16 months, not 18 months as reported by VS&A. Clarification: I T was not derived, however, the authors state that it was "confined to small areas of the entire orchard." It is not entirely clear from the article, but the area of erosion from the base of the trees appears to have been 2.1 m 2 . If so, this may represent the maximum I T per tree in the orchard.
Litter Marks 18 to 19-year-old oil palm plantation in Malaysia. I T determined using the litter mark method for 30 trees in which the extent of bare dark areas around the base of trees was assumed to be created by stemflow.
6.8-11.8 Range of maximum I T values. No time scale provided.
Clarification: From Figure 1 of Rashid and Askari (2014) no leaf litter can be seen. Since bare areas around the base of trees may be caused by a variety of factors (allelopathy, competition, herbicide use) and because stemflow was not measured nor were direct observations of stemflow induced overland flow made during this study, there is no definitive proof that these dark, bare areas were caused by stemflow or represent I T . Addition: Rashid et al. (2015) include the same I T data as Rashid and Askari (2014). VS&A did not reference Rashid et al. (2015).  Table 1 of VS&A are not correct (those are the diameters at the tree base) and ranged from 18.1 to 39.2 cm with an average of 28.6 cm. Also see Table 1 footnote (*) .

Tanaka et al. (1991) ** Infiltration Area, Erosional Scour, and Vegetation Marks
The extent of infiltration area marks (i.e., wetted infiltration areas on the soil surface), erosional scour marks and vegetation marks were measured for 12 trees within and outside of the University of Tsukuba campus. It should be noted that intensities of stemflow and K sat were not measured. DBH values provided in Table 1 of VS&A were likely derived from the diameter at tree base indicated in Figure 4 of Tanaka et al. (1991).
0.17-1.03 (Average = 0.60) Range of maximum I T extents derived from infiltration area mark for 7 trees for a single 2.0 mm rain event; erosional scour marks and vegetation marks based on some earlier rainfall events Clarification: All marks except for one indicated that I T < 1 m 2 . The infiltration area marks generated by stemflow associated with the 2.0 mm rainfall are not to be confused with litter marks or erosional scour marks that may be formed during high stemflow funneling episodes. Also see Table 1 footnote (**) .
Summary In all but a few extreme rainfall events, I T is < 1 m 2 under average conditions for forested ecosystems. There is no compelling evidence to indicate otherwise. For agricultural and urban settings with soil compaction average I T could be larger than 1 m 2 in some cases but convincing evidence is lacking at this juncture. More work is necessary to quantify I T for a range of ecosystems, especially different forest types.
*The statements in VS&A "I T > 1 m 2 tree −1 has been reported under low rainfall intensity, 1-2 mm h −1 " and "photographs of litter marks showing I T = 0.4 to 1.3 m 2 tree −1 under non-extreme precipitation conditions" cannot be derived from or substantiated by Iida et al. (2005) as the litter marks were formed during an earlier 88.5 mm rain event when the maximum intensity of 9.5 mm h −1 was reached, not during portions of that event with lower rain intensity. In addition, the 22-23 March, 2005 event only created limited ponding close to the tree trunk (Figure 4, Iida et al., 2005) when rain intensity was 1.5 mm h −1 and no litter was displaced during the entire storm, despite a maximum intensity of 6.5 mm h −1 . As a comparison, I T values of 0.34 and 0.30 m 2 were calculated based on a maximum stemflow intensity of 1,100 cm 3 (30 s) −1 and average infiltration capacities of 383 and 441 mm h −1 for two Formosa sweet gum trees (Iida et al., unpublished data). ** The description "post-storm litter marks caused by infiltration excess (Tanaka et al., 1991.)" in VS&A (page 3) is not correct since Tanaka et al. (1991) did not observe any litter marks. Instead, the extent of I T was inferred as the extent of the infiltration area marks (i.e., the area of wetted surface soil). It should be noted that I T using this method may be overestimated due to capillarity of the surface soil increasing the wetted area in the absence of infiltration.

AUTHOR CONTRIBUTIONS
DC-M was the primary author of the manuscript and a cooriginator of the commentary. SI contributed to the text and was a major contributor to the table. SG and PL contributed to the text of the paper, making several editorial changes and suggestions. SG also played a major role in the revision, reconfiguring the