Comment: Lake 227 shows clearly that controlling inputs of nitrogen will not reduce or prevent eutrophication of lakes

The conclusion from our earlier paper, that decreasing nitrogen (N) inputs to lakes does not substantially reduce symptoms of eutrophication (Schindler et al. 2008), was recently challenged by Scott and McCarthy (2010). They reanalyzed graphs from our paper showing data from Lake 227 at the Experimental Lakes Area (ELA) in Ontario, which was artificially eutrophied with N and phosphorus (P) from 1969 through 1989 and with P alone from 1990 until the present. They interpreted declining trends in total nitrogen (TN), chlorophyll a (Chl a), and phytoplankton biomass to mean that the lake became less eutrophic after N inputs were terminated in 1990. They then concluded that ‘‘the degree of eutrophication can be controlled by managing N inputs concurrently with P.’’ We now have four more years of data for Lake 227, for a total of 41 yr at a constant P loading rate. Schindler et al. (2008) review the loading history of Lake 227, while Findlay et al. (1994) give full N and P budgets from 1970 to 1992. All samples were collected and analyzed using consistent methods (Stainton et al. 1977; Findlay et al. 1994; Schindler et al. 2008). Statistics reported below were calculated using SYSTAT 12 (SYSTAT 2007). We have used a threshold for significance of 0.05; and, to remain consistent with Scott and McCarthy (2010), we have not corrected probabilities for multiple comparisons. Reanalysis of our results including the more recent data clearly refutes the conclusion of Scott and McCarthy that the lake has become less eutrophic as the result of decreased inputs of N. There has been no significant declining trend in either Chl a (r 5 20.26, p 5 0.27, n 5 20) or phytoplankton biomass (r 5 20.22, p 5 0.36, n 5 20) since we ceased adding N to the lake in 1990 (Fig. 1a,b). Separate analyses of trends after 1996, which Scott and McCarthy used to make many of their arguments, are not different from those for 1990–2009 (Chl a: r 5 20.16, p 5 0.61; phytoplankton biomass: r 5 20.35, p 5 0.14; n 5 13). In fact, there is no significant trend in the annual average Chl a or biomass over the 41 yr since fertilization with P began, regardless of the N : P ratio used in loading (Chl a: r 5 20.05, p 5 0.77; biomass: r 5 20.23, p 5 0.14). We note that in fig. 1 of Schindler et al. (2008), the 1975 data point for Chl a was repeated for 1976 and all subsequent points were offset in error by 1 yr. This error does not affect the major conclusions of Schindler et al. (2008) and only marginally affects the correlations undertaken by Scott and McCarthy (2010). The biomass of Cyanobacteria (not shown) also showed no significant trend over time (r 5 20.38, p 5 0.10, n 5 20 after 1990; r 5 0.08, p 5 0.60, n 5 41 for the entire data set). However, the abundance of heterocysts has increased significantly (r 5 0.59, p , 0.0005, n 5 36), particularly after reduction of N loading in 1990 (Fig. 1c). Using the regression equation of Finday et al. (1994), which is based on studies of Lake 227 by Hendzel et al. (1994), this increase in heterocyst abundance indicates that N fixation has increased considerably since 1990 (Fig. 1d; r 5 0.61, p 5 0.005, n 5 20). The rate of increase in heterocysts roughly doubled after 1997, indicating that N fixation is still increasing. This observation reinforces our earlier arguments (Schindler et al. 1977, 1987) that the responses of lakes to changes in nutrient inputs result from slow changes in species and biogeochemical processes, requiring several years to fully play out. Successful strategies to control eutrophication must account for these rates of change. The negative trend in TN concentrations after 1990 identified by Scott and McCarthy continues to be significant with our longer data set (1990–2009: r 5 20.74, p , 0.0005, n 5 20) (Fig. 2a). Over the same time period, there is no significant change in total phosphorus (TP) (r 5 20.18, p 5 0.26, n 5 20) or TN : TP (r 5 20.32, p 5 0.16, n 5 20) (Fig. 2c,d). Most of the decrease in TN is in total dissolved nitrogen (TDN) (r 5 20.70, p , 0.0005, n 5 20; Fig. 2b) and there are also significant declines in the ratios of dissolved inorganic nitrogen : total dissolved phosphorus (DIN : TDP; r 5 20.50, p 5 0.02, n 5 20) and TDN : TDP (r 5 20.45; p 5 0.05, n 5 20) (not shown). Clearly, reduction in artificial N loading has resulted in lower N concentrations in Lake 227. N limitation for many species of phytoplankton may also have increased because of declines in DIN, which includes the forms of N that are most available to many phytoplankton species. However, N-fixing cyanobacteria increased after the cessation of artificial N loading (Schindler et al. 2008), clearly demonstrating their ability to offset declines in DIN, and overall algal abundance has remained proportional to annual P loading, which has been constant since 1969. To be effective, eutrophication management must reduce excessive phytoplankton abundance. The Lake 227 data show that reductions of N in the absence of reductions in P will shift the competitive advantage to N-fixing cyanobac* Corresponding author: michael.paterson@dfo-mpo.gc.ca