Dear editor, dear authors,
Thank you for considering my previous comments and expanding on the missing explanations, which enabled me to understand the contents of the paper and results much better. I found a few more contradicting statements (see detailed comments below), which may suggest that I have still misunderstood something. However, based on my now more complete understanding, I cannot recommend publication of the paper in its current form in HESS, for the following reasons.
The main conclusions of the paper are:
C1) Soil moisture exerts a strong, dominant influence on root density growth and vertical root distribution. (L200, L212)
C2) Root profiles can be predicted realistically from information on soil moisture profiles only. (L201)
C3) The root system is "insensitive to above ground processes and overall biomass growth". (L204)
C4) Implementation of the proposed root growth model in regional Land-Atmosphere models would result in "more correct representation of soil-plant water fluxes, and a more realistic representation of root biomass". (L216-219)
These conclusions are based on a single experiment where water was applied to a soil profile at progressively deeper depths as the root system of a single maize plant developed over 50 days in a rhizobox. Since there was no control experiment with a different irrigation scheme, it is not possible to tell from the results presented here whether the root system development was indeed governed by soil moisture availability or whether the pattern found here is just a result of the seedling exploring an empty soil volume. The irrigation scheme was designed "in an attempt to follow the water demand of the plant" (L76) and to "control the vertical soil moisture gradients and stimulate vertical root growth" (L78), such that water was always applied towards the bottom of the expanding root system (compare Fig. 2C with Fig. 6A-D), resulting in a vertical soil moisture distribution that largely reflected the irrigation scheme (Fig. 2). Although C1) is plausible, due to the lack of a control experiment, I cannot confirm that the results support this conclusion. Consequently, C2) is not supported by the presented results, as it is not clear whether root growth responded to soil moisture at all in this experiment, or whether the implied correlation between root growth and soil moisture was due to selective irrigation in soil layers where root growth was expected to be highest (at the bottom of the advancing root system). The conclusion that the root system "is insensitive to above-ground processes" (C3) is entirely unsupported, as above-ground processes were neither controlled nor monitored. Similarly, there is little support for Conclusion C4) in the results presented here, as the dynamics of the root system development of a single seedling growing in an initially root-free soil is of limited relevance for the simulation of the root system of a plant community in a regional Land-Atmosphere model. There are other papers in the literature that point out the importance of representing root system dynamics in such models, some of which have been cited here, but it is really far fetched to conclude that the model presented here would improve land surface models based on the results presented in this study.
The different model versions presented here provide limited potential for new insights. The finding that the exponential root distribution led to very different simulations of the vertical water uptake profile than simulations based on measured or dynamically simulated root distributions is not surprising given that root water uptake was simulated as a linear function of root length density and soil moisture. In fact, the representation of root water uptake adopted here ignores the non-linearity of the water retention and hydraulic conductivity curves, so it is not clear what can be learned from these simulations.
The main insights I drew from the paper are:
1) Maximum root growth was observed predominantly at the bottom of the advancing root system (Fig. 2c). At the same time, this is where the watering was predominantly taking place. Therefore, it is not clear if this growth pattern was due to the watering or a natural root system development of this variety of maize, whereas the watering followed the root development. I assume that the decision to design the irrigation scheme in a way to "stimulate vertical root growth" was based on prior experiments where vertical root growth did not occur to the same extent. It might help to include these experiments in the paper as controls or different treatments, so that the effect of the watering scheme becomes more obvious.
2) Vertical root distribution deviated substantially from an exponential distribution, with an almost inverted exponential root density profile, having the highest root density at the bottom of the root system at the end of the experiment (Fig. 6D). Believing that root growth does respond positively to soil moisture, or at least that it is hampered when the soil dries out, the vertical root profiles found in this study are likely due to the specific irrigation scheme at the bottom of the root system while the top of the root system is left to dry out after 20 days. This does show that root systems can deviate from the exponential distribution under certain conditions, but it does not automatically mean that exponential root distributions are wrong representations in land surface models where water replenishment happens predominantly by infiltration at the soil surface.
My co-workers and I are in the process of preparing a manuscript ourselves where we documented the dynamic responses of maize root systems to water pulses, so I do believe that your Conclusion C1) is correct, but unfortunately, I cannot see clear support by the data presented here. Since the other conclusions are in my view even less supported by the results presented here, I cannot recommend publication of the paper in its current form. However, I hope that my detailed comments below help re-structure and expand the paper in a way that it can be published in a suitable journal in the future.
DETAILED COMMENTS:
Unfortunately, the manuscript structure is still confusing:
1 Introduction
2 Experiments
2.1 Experimental Set-up
2.2 Diagnostic model of root growth: root follows moisture
Time series and correlations
Model formulation
Model parameter evaluation and results
2.3 Soil moisture and water uptake model
Model formulation
Model parameter evaluation
Results and discussion
2.4 A prognostic model for coupled soil moisture and root growth
Model formulation
Model calibration
Results and discussion
3 Conclusions
As shown above, there are numbered and un-numbered sections, where the un-numbered section titles are repeated multiple times. This makes the reading confusing, so I would recommend numbering all sections, and ideally consolidate them into one methods section where all three model versions are described, followed by one results and discussion section.
L76: How was water demand in each layer measured? According to Fig. 2A, soil moisture was highest at the depths where water was added, which suggests that water was added beyond the amount of water taken up by roots.
Eq. 1: As L is increasing over time (Eq. 4), local root growth would decline at constant theta_n over time. Is this realistic?
Eqs. 1-4: Please clarify in the equations that it is r(t), not r, and L(t), not L. So r(t) is the observed overall rate of increase in root length, whereas Eq. 1 provides the rate of increase in root length in a given soil layer. So in essence, Eq. 1 is:
dR(z, t)/dt = integral(d(R(z,t)/dt) dz) * theta_n/integral(theta_n dz)
The hypothesis formulated in Eq. 1 is hence that the vertical root growth distribution follows the vertical soil moisture distribution in a linear way. It would be good to make this clearer.
L119: This is not actually root growth, but extension rate of rooting depth, see Eq. 4.
L119: Based on what observations? Based on observed rate of extension of the rooting depth or calibrated to reproduce measured root distribution?
L124: Why is this an extra condition? According to Eq. 2, theta_n<0 if theta<0.075, and according to Eq. 1, root growth should be negative if theta_n<0, so this should be already satisfied.
L135: I would put the fraction in brackets to make very clear that the exponent applies to the whole fraction.
Eq. 9: This ignores the non-linearity of the water retention and hydraulic conductivity curve. A linear relation between normalised soil moisture and root water uptake rate seems very unrealistic.
L148: Driven by the measured root density data?
L161: Why "except for the first period"? I see 0.75 for the exponential profile, compared to 0.77 and 0.81 for the others.
L162-169: This suggests that the fact that the vertical soil moisture distribution was relatively similar between the exponential and modelled root profile simulations was due to a negative feedback loop between water depletion by root water uptake and reduced root water uptake by reduced soil moisture. However, this argument is not supported by the experimental evidence, as the experiment was designed in a way to avoid depletion of soil moisture by root water uptake. In fact, root water uptake was over-compensated, leading to a soil moisture profile with the highest soil moisture where the root density was highest (Fig. 6, except for 40-50 days). An alternative explanation is that the vertical soil moisture distribution is determined by the location of water input to the system. Fig. 2C illustrates that the soil moisture was generally highest where the water inputs took place. Since the top soil was left to dry out after 20 days (same figure), it is not surprising that root growth subsided thereafter in this part, eventually leading to the inverted vertical root distribution shown in Fig. 6D.
L175: I don't think normalization is the right term, as this would suggested that the maximal "local root growth tendency" would be 1.
L176: So now Eq. 1 is replaced by Eq. 10, i.e. root growth rate is a linear function of normalized local water content rather than the local fraction of total water content? I understand that r(t) from Eq. 1 had to be removed for prognostic simulations, but why was the division by integrated water content removed, too?
L179: "with identical overall root length": If the exponential root profile is set to have the identical overall root length as the dynamically simulated root profile, how come there is up to 54% difference in total modelled root length between the two in Table 3?
L193-: I do not understand this explanation, as according to L179, the exponential profiles were set to have the same overall root length as the dynamic root profiles, so I expect the profile to be always exponential and never become "off-shape".
L200-: This is not clear, as irrigation was applied locally where the largest root growth was expected, so it is not clear if root growth followed irrigation and soil moisture or if the dynamics would have been similar under homogeneous soil moisture.
L201: What do you mean by vertical rooting depth? Do you mean vertical root distributions?
L202: Not infiltration, but vertical water transfer model.
L206-: Not necessarily, as all the results show is that differences in vertical root distributions did not have much effect on the simulated soil moisture profiles, which could be due to the strong irrigation signal in the soil moisture profile, overwhelming the more distributed root water uptake profiles. This is underligned by the fact that soil moisture was highest where root abundance was highest in most cases, so there was no obvious effect of root water uptake on the vertical soil moisture distribution.
L212: Not necessarily, see above.
L215: Why would it prevent water demand?
L216: I would strongly advice against implementation of an empirical root growth model in LSMs, which was based on an experiment with maize seedlings.
L217: There is no evidence for such a benefit in LSMs in this paper.
Table 1: To avoid confusion, I would write in the top row: "measured RP", "modeled RP", and "exponential RP", and in the caption: "...root profiles (RP)."
Figure 2C: What do the units of root growth given in cm/cm mean? Should this be cm/cm2, or is it the root length added divided by the initial root length? The line colours in the legends and in the plots are not the same. Perhaps it would be enough to have only one legend, as it is always the same between the solid, dashed and dotted lines.
Figure 6: Why not use squares for soil moisture in the middle column to avoid confusion? What does "driver" and "results" mean?
Figure 8: The lines are wrong in the legend, as there are only solid and dashed lines in the plot. |
