Nutrient Cycling in Agroecosystems 70: 323–324, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands. 323 Response to the commentary by Hardy, Reich and Tucker Lawrence B. Cahoon 1,∗ and Scott Ensign 2 1 Department of Biological Sciences, UNC Wilmington, Wilmington, NC 28403, USA; 2 UNC Institute of Marine Science, 3437 Arendell St., Morehead City, NC 28557, USA; ∗ Author for correspondence (e-mail: cahoon@uncw.edu) We are pleased to respond to the comments ad- dressed to you by Drs. Hardy, Reich, and Tucker regarding our article, ‘Spatial and temporal variability in excessive soil phosphorus levels in eastern North Carolina’, published in Nutrient Cycling in Agroe- cosystems (69: 111–125). The intent of our article was to examine large-scale patterns in excessive soil P-I values in a region facing substantial impacts from many potential sources. We relied almost exclusively on data reported by various North Carolina state agen- cies in our analysis. Hardy et al. raise several issues, which we address here. Hardy et al. correctly point out that the North Carolina Division of Agronomy’s (NCDA) soil test- ing procedure has changed over the period we ex- amined, from Mehlich II extraction to Mehlich III extraction and colorimetry, then Mehlich III extrac- tion and ICP-AES. Our statement in the article that the soil test laboratory ‘employs the Mehlich III ex- traction method’ was strictly correct in the present tense, but we knew and could have stated in the art- icle that the laboratory’s methods have changed over the years. However, NCDA reports soil P levels us- ing a scale (soil P-I) indexing soil test results to plant needs. P-I values > 50 are considered to indicate no need for additional fertilizer application. NCDA defines ‘excessive’ soil P-I values as those greater than 100, i.e., twice the level at which additional fertilization would normally be expected to yield no response, and the distributions of index values >100 where the data we examined. Moreover, in a conversa- tion between the article’s senior author (Cahoon) and Dr. Tucker in late 2000, assurances were given that the soil phosphorus index (soil P-I) values in the NCDA data reports were indeed normalized across methods. Comparability must surely have been a major con- sideration in the NCDA’s use of the soil index and reporting policy. Given the statement by Hardy et al. that the Mehlich soil test extractants were developed in the NCDA’s laboratory, it seems curious that they question our presentation of soil P-I index data. They offer no specific explanation of how changes in their soil P analytical methodology might have biased their data or our analysis of them. Finally, we explicitly stated in our description of methods that “...limitations and assumptions involved in the soil P-I data summar- ized here must be considered before evaluating their full importance...” and that “...categorization of soil P-I data sets into classes, particularly the soil class considered here, ‘P-I>100’, likely also understates the magnitude of excessive soil P-I levels”. Our intention was to be conservative in our assessment. Hardy et al. offer no evidence that the incidence of soil P-I values > 100 has been overstated in our analysis. Hardy et al. point out that soil sampling depths can be up to 15–20 cm, depending on crop, whereas we stated that sampling depths were “supposed to be col- lected from the top 10 cm of soil and so represent only surface concentrations of phosphorus”. Though we thank Hardy et al. for this explanation of their meth- ods, we remind readers that our spatial analysis of soil P was not an investigation of the vertical distribution of P in the soil. As we discussed in our Methods, soil P concentration varies greatly with depth, and therefore we relied upon the NCDA’s soil P-I data to be repres- entative only to actual sample depth. We must clarify that our analysis did not imply that there were changes in soil P concentration across time, space or land use types (as Hardy et al. insinuate), nor could it, as quan- tifying the proportion of the class of soil P-I values >100 allowed no calculation of mean values of soil P. Instead, we emphasized that increased proportions of soil P-I values >100 between FY90-91 and FY00-01 represent an increase in excess P relative to cover crop needs. Unless the NCDA published erroneous soil P-I data for FY00-01, our analysis could not have ‘overes- timated’ the amount of soil P available for plant needs relative to the cover crop. Hardy et al. argue that changes since the mid- 1990s in the proportions of ‘forage’ crops and con- servation tillage may bias our overall assessment of excessive soil P-I frequencies. However, that is why we broke down our analysis by land use type. For example, our Table 3 shows that the frequencies of