I have now gone over the revised manuscript of Pelletier et al. The authors have paid careful attention the reviewer comments and I am happy with their responses. In particular, the authors have adequately answered the concern about transience in the drainage density calculations (they show that drainage density can be adequately modelled by an equilibrium model based on the timescales of adjustment) and I also feel that they have adequately dealt with the question about applying a long term model to sites with erosion and sediment transport data that spans decades.
The paper is a very nice example of how to integrate careful field measurements with a mathematical model of landscape evolution to say something meaningful about how landscapes respond to changes in vegetation.
I do have a few concerns, and thus suggest minor revisions before we accept the paper at ESURF.

S. Mudd

Lined comments:
p.6 Line 14: Refer to the figure here. Also the “63” designation is not in the figure so this could be slightly confusing. Perhaps mention it in the figure caption.
p.6 Lines 15-22: The citations for the soil data, vegetation types, etc. are missing in this paragraph, which is odd since they are included in the following paragraph. Add the appropriate citations.
p. 8, Line 21: “Multiple-flow-direction algorithm” is vague since this could refer to an entire family of flow algorithms (this could refer to any algorithm that routes flow to more than one pixel, so could include, for instance, the d-inf algorithm). Add a citation so readers know which one it is.
p. 8, Line 22: State how were the flow lines were calculated.
p. 8, Line 23: This is a bit confusing since it states an averaging routine was used for “all pixels draining into the channels” but previously it was stated that the hillslope length was calculated on the basis of flow lines emanating from topographic divides. If every pixel is used, rather than those emanating from divides, the hillslope length will be underestimated significantly so it is important to clarify this point.
p. 9 line 20- p 10 line 2: Culling was a pioneer and should be cited here, but he didn’t show that “sediment transport by colluvial processes leads to a diffusion equation for topography if slopes are uniformly soil mantled and topographic gradients are modest”. He proposed a linear flux law because it was convenient. Culling himself says that the linear flux law is “the fundamental assumption…analogous to the assumption made in the theory of heat conduction”. As far as I know, there isn’t much evidence for a linear flux law beyond McKean’s 1993 geology paper (based on 2 data points and a line passing through the origin) and the rainsplash papers of Furbish et al 2007 and Dunne et al 2010, both in JGR. There are a number of papers that suggest that topography is consistent with a nonlinear flux law (Roering’s 1999 and 2008 papers, papers from Martin Hurst and Stuart Grieve from my own group) which approximates to an erosion statement like equation (1) for low topographic gradients. So while I am perfectly happy with equation (1) and think it is justified in this setting, the text, as it is currently written, gives insufficient justification and should be modified.
Line 11, line 14: I like Figure 5 but it doesn’t show the square approximation of sediment flux that is used to arrive at equation (6). I think this would be clearer if, in addition to what is already there, a small figure showing the gridded abstraction of channel head geometry is used to illustrate equation (6). Basically this is in response to looking at Figure 5 and thinking “what 3 adjacent hillslopes?” and then interpreting the 3 adjacent hillslopes as the two side slopes and the one upslope pixel if you used a gridded approximation to Figure 5.
Page 12, Line 1: Refer to the actual slopes here.
Page 12, Line 5: I suggest adding a very brief note here that explains you will demonstrate this base on Nearing et al.’s field data.
Page 14, Line 5: How many of these measurements were made? I ask because I wonder how noisy this data is. Also, the slope of low order channels isn’t really relevant since the main purpose of presenting channel slope is to support the assumption that there is no colluvial transport out of the cell at the channel head.
Page 15: Lines 1-2: I very much enjoyed the way a range of careful field measurements is balanced in this paper to support the mathematical model. However, there is a component of the model that raises some questions. That is that the p value is greater than 1. This implies that erosion is a function of drainage area (according to equation (4)). The erosion goes as nearly the square root of the area. So if you have the junction between two small basins of equal size, the erosion rate below the junction is ~sqrt(2) larger than the erosion rate just upstream of the junction. The erosion rate, according to this equation, suggests that as you move downstream erosion rates increase and you end up with something that looks very much like a wave of incision moving through the landscape. I do not think that the authors believe there is a wave of incision in the landscape. What do the authors think is behind this trend? What does a linear fit through figure 9 look like? I am not going to hold the paper up on this point but perhaps the authors would like to comment on the apparent dissonance between the model fit suggesting erosion is increasing downstream (with the consequence that slope angles will need to be transient) and the equilibrium assumption in equation (11) and (12), which rely on slope angles remaining constant.
Page 15, lines 3-5: A single value of D is used, on the basis that there is no way to otherwise measure it. But curvature on ridgetops reflects D. According to figure 8 the ridgetop curvature of both sites is the same, yet there are large differences in erosion rates, suggesting that D is quite different in the two sites. The fact that the shrub landscape has the same ridgetop curvature but a much higher erosion rate than the grassland suggests that the D value is higher in the shrub landscape. This is consistent with the findings of Dunne et al (2010-JGR-ES) who showed that decreasing cover can increase sediment transport rates by rain splash by orders of magnitude. The authors later mention that this assumption leads to uncertainty in the results but I feel a few more sentences on the issue would be useful, if only to refer the reader to the discussion in section 4.2 asserting that there has not been enough time to modify the ridgetops.
Page 18, line 21-25: Earlier the authors state that curvature values on ridgetops are not affected because the erosion signal has not had time to propagate to the ridgetops. But if this is true then increased erosion in the channels and downslope would increase relief. The fact that relief is lower in the shrubs suggests that D could have increased in the shrubland. I suggest adding a sentence or two on this point.
Page 20, line 12-14: This passage needs citations. I am not aware of a paper that shows increasing vegetation cover monotonically increases D values but the Dunne et al 2010 paper suggests transport rates increase with decreasing cover. The Gabet et al. (2003) paper would suggest a positive relationship between biomass and D, but these were the results from various mathematical models and not data.
Figure 8: This plot needs some information on uncertainties since hoe else is the reader to know if these differences are real or just noise?

While I am still perplexed by the issue of p > 1 in this manuscript, I feel that the authors have justified their (mostly lack of) changes and am prepared to accept this manuscript in its current form. Again, this is a very nice combination of careful field measurements and modelling, and although I have some reservations about the model I do not think any model is "prefect" and this does a better job than the vast majority of papers out there. I'm pleased to see such high-quality work submitted to ESURF and hope to see more such contributions in the future.

Dear Jon,

I have now had the chance to go through the AE's recommendations. Based on these, and the high quality of your paper, I am really pleased to accept your paper for publication in ESURF.

Thank you for submitting your work to us, and contributing to the success of the journal.

Best regards,
Frederic