Kinnell argues that we misrepresent the USLE in Parsons et al. (2006). Specifically, he objects to our statement that 'what it provides is an estimate of the erosion that would be measured if the entire area were divided up into 22β’1-mlong plots and the output from all of them added together', and that it does not predict erosion as defined by Trimble (1975). Whilst it is true that some of the plots on which the USLE is based were not standard USLE plots, the vast majority were (389 of 439 data points at the Clarinda site, for example: NSERL, 2007), and, as Laflen and Moldenhauer (2003) point out 'The unit plot concept was widely used in establishing factor values for the USLE' (p. 21). Consequently, the equation is strongly biased towards this plot length. Furthermore, as we pointed out in our paper, 'the fact that USLE has a slope-length scaling factor for longer or shorter slopes only compounds this problem, as the factor assumes a constantly increasing polynomial form. This form flatly contradicts the evidence for an inverse relationship between sediment yield and catchment area' (Parsons et al., 2006(Parsons et al., , p. 1326).
Trimble's definition of 'gross erosion ' (1975, p. 142) would appear to be exactly that of Glymph (1954, p. 181) in his definition of the sediment delivery rate. Whilst it is true that Glymph uses the earlier Musgrave equation to calculate sheet erosion rather than the USLE, he could scarcely be expected to use an equation that was developed some years after his paper on sediment delivery! The Musgrave equation, as an antecedent of the USLE, is well documented (see, e.g., Laflen and Moldenhauer, 2003), and subsequent authors have used the USLE in place of the Musgrave equation in calculations of sediment delivery.
Both of these issues, however, are points of detail compared with the fundamental objection that we raise to the notion of sediment delivery, which is that measurements of sediment flux are not measurements of loss per unit area, and that they are certainly not additive. In contrast, the USLE clearly assumes they are, as the procedure for dealing with heterogeneous areas makes clear. For this the 'procedure is to divide the heterogeneous drainage area into subareas for which representative soil type, slope length, gradient, cover, and erosion-control practices can be defined. The USLE is then used to compute the sheet and rill erosion on each subarea' (Wischmeier and Smith, 1978, p. 46). Kinnell also argues that 'the designers of the USLE could have chosen some other value for the length of the unit plot without having any impact on the amount of soil loss predicted by the model'. This statement demonstrates that he fundamentally does not understand that measurements of sediment flux are scale dependent, and that, in fact, had the designers of USLE used a different plot length, they would indeed have produced a different result. In fact, it is also an arithmetic consequence of the structure of the LS factor in the USLE.
We do not wish to denigrate the achievements of the USLE, but the science on which they were based dates from over half a century ago, and our understanding of hillslope processes has, thankfully, advanced. One unintended consequence of the idea of a 'unit erosion length' -whatever its size -is the biasing of subsequent research, which has typically considered the unit length as a standard. For example, Morgan (2005, p. 98) states that to measure erosion 'the standard plot is 22 m long and 1β’8 m wide although other plot sizes are sometimes used', and the idea of the 'Wischmeier plot' is commonly encountered in soil-erosion research. Precisely because of the scale dependence of sediment flux, the use of this 'standard' seriously biases the results produced, and has thus hindered more progressive understanding of erosion processes. As Laflen and Moldenhauer (2003, p. 22) note, 'While the unit plot concept was widely used, there has never been a unit plot . . . The unit plot concept, while very useful, was apparently a myth as far as soil erosion measurements were concerned'.