This paper analyses some field data on interception, throughfall, and stemflow in two shrub taxa from the Chinese drylands. The data were collected in 2014-2015. The authors explore some of the behaviour of the interception parameters both among separate rainfalls and at the intra-event scale.
On the whole, the paper is very systematically set out, and generally clear. The work reflects a considerable and commendable effort in instrumenting dryland shrubs to record interception, stemflow, and throughfall. It is good to see such work carried out under natural rainfall, and not rainfall simulation, which generally fails to reproduce key characteristics of natural rainfall events, notably including their durations. Nevertheless, there are some more serious issues with the data and the interpretation of the results, which I explore further below.
A fundamental issue that the authors do not dicuss is whether it is actually important to investigate water partitioning for individual shrubs (as they do in this paper) or whether area-wide water balance is more significant, given that shrub root systems can extend widely, well beyond the limits of the canopy. In other words, for a given shrub, is it not possible that an important component of the water accessed via the roots is actually from open-field rainfall, and not from stemflow or throughfall? A critical issue here is whether, and to what extent, shrubs can generate metabolically useful stemflow, say, focussed on the root system. Otherwise, shrub canopies do nothing but reduce the depth of rainfall arriving at the root system (owing to interception losses on the above-ground plant parts). This is why extensive root networks become so important. It is also generally considered that small showers of rain are not of great use to shrubs, many of which have deep root systems that access water accumulated in the deeper soil over months or years. Small rainfall events can nevertheless benefit microphytic plants and microbial communities.
The authors need not resolve the issue of individual shrub versus landscape-level analysis, but in their paper, they certainly need to discuss this, and show that they have considered what their results really tell us that is of ecological use, in helping to understand the true water balance of shrublands. For example, what is the actual ground cover fraction covered by shrub canopies, in their study area, and how much soil is exposed directly to rainfall with no interception?
There are also technical issues that I think require further classification and justification.
The authors derive an estimate of I10max. But as far as I can see, their rainfall data are locked to the time-step of the logger. To find short-term maximum intensities correctly, much higher temporal resolution of the rainfall data is needed. For instance, suppose the logger records the rainfall in the periods 10:00-10:10, 10:10-10:20, etc. Then, if the most intense 10 minutes actually occurred from 10:05-10:15, it will be missed by the logger data, and the maximum intensity recorded will only be about half of the true value.
The authors express canopy water holding capacity (C) by weight gain of branch specimens dipped in water. They then proceed to estimate the canopy storage capacity of entire shrubs by multiplying by the total biomass of the shrub, estimated from an allometric growth model. Why they did not actually weigh some entire shrubs is unclear. In any case, the problem with this procedure is that it is far more likely to be the surface area of the plant parts that governs C, and not weight. The authors need to discuss this, and to defend or justify the procedure that they adopted, based on mass not surface area.
Likewise, the authors estimate total shrub stemflow by measurements on a few branches, then multiply the result by the total number of branches on the shrub. This again seems unlikely to be reliable, given that some branches form the outer perimeter of a shrub, and some are more sheltered, internal branches. There are also upper, more exposed branches, and lower, more sheltered branches within the canopy. The presumption that all of these generate the same stemflow flux warrants justification.
The maximum throughfall fractions reported by the authors - around 80% - seem very low in light of the depth of some of the rainfall events (more than 40 mm). I would expect that in such an event, the throughfall fraction would approach 100%.
Consider some rough calculations:
For both shrubs, the value of Cm is about 5 litre.
For C Korshinskii, the canopy area averages ~ 5 m2, so the depth of rain over the canopy required to fill Cm is about 1 mm.
For S psammophila, the canopy area averages ~ 24 m2, so the depth of rain over the canopy required to fill Cm is about 0.2 mm.
Both depths required to fill the canopy stores are negligible in relation to a rainfall of 40 mm, and should have resulted in an interception loss (even allowing for ongoing evaporation during rainfall) of at least 97%. Where did the rainwater go to yield throughfall fractions no larger than 80%? Here I begin to have serious doubts about the reliability of the field data. Perhaps the few throughfall gauges deployed underestimated that parameter? Perhaps intra-event evaporation is actually very significant for these shrubs? The authors need to comment on these possibilities and explain why the maximum throughfall fraction that they report is so low in events in which the loss to canopy storage should be no more than 1% or so of the incident rainfall.
These queries lead me once again to recommend that the authors provide more detail on the rainfall events that they recorded. How were separate events delineated? How were intensity and duration related, for these events? Further, given the air temperature and humidity data that the authors had available, did they estimate ongoing wet canopy evaporation losses during rainfall, as a component of the shrub canopy water balance?
Finally, of course, some further data on the shrublands themselves would be of use. What was the areal cover provide by shrubs (as distinct from shrub interspaces)? What were the shrub root systems actually like, and can the shrubs actually make use of small amounts of throughfall or stemflow? Though arguing that these aspects of shrub hydrology are important for us to understand, the authors have not really shown that they are significant at all, at least in the many small rainfall events that the authors describe (34% of rainfall events were smaller than 5 mm depth - see line 271).
Minor errors:
line 49: interception loss is not comprised of transpiration
lines 328-329: English expression needs to be improved
line 338: 'slighter' should I think be 'slightly'
David Dunkerley
Monash University |