Response to Plos one query:

Thank you for your close reading of, and interesting comments on our paper. Below are numbered responses to your questions.

1. Weed seedbanks: Thank you for stressing the importance of weed seedbanks to overall weed population dynamics, a point that we readily agree with. While it is true that there is a slight uptick in the mean number of weed seeds in the 3 and 4 year rotations in the final year of the study, for several reasons, we do not believe that this signals a shift in the long-term behavior of the seedbank, or that it constitutes an immediate weed management problem. First, to put things in perspective, the number of weed seeds measured in 2011 in the 3-yr rotation, with the greatest seedbank population densities, was equivalent to about 1000 seeds m-2. While this may sound like a lot, it is less than one-tenth of the commonly observed population densities for weed seeds in grain production systems of the north central region. And these low seedbank population densities were linked to the very low weed biomass seen in all plots in all years. Second, our mixed-effects models of seedbank decline, which accounted for temporal autocorrelation of the seedbank measurements, indicated a strong negative slope for seedbank population densities in all treatments. Moreover, there was no statistical support for models in which slopes of the seedbank trajectories were allowed to vary (for interaction term, P > 0.25; Akaike weight for the best model without interaction term = 0.82, indicating an 82% probability that this was the more appropriate model for the data), supporting our assertion that the important weed management goal of reducing weed seedbanks was being achieved in all systems. Figure 2a is somewhat problematic, in that it shows the means, across reps, for the seedbank population densities, rather than the BLUPs (best linear unbiased predictors), which are the true way to represent the results of a mixed-effects model analysis. We chose to present a scatterplot of the means because we felt it would be easier for a general readership to understand than the BLUPs. The BLUPs show strong negative linear trends in all reps of all systems measured. Finally, you’ll notice that weed seedbank densities oscillated in a very similar way between 2007 and 2008, dropping again after that. The small increase in seedbank population density in 2007, roughly the same size as that observed in 2011, was not large enough to prevent timely weed management from bringing numbers back down in 2008. The seedbank data for 2012 are not yet available, but we are confident that we will be able to maintain our management objectives of depleting the seedbank to some low, population density at which weeds may be readily managed with integrated tactics.
Regarding your other weed seed questions: the species represented in the seedbank were dominated by velvetleaf (Abutilon theophrasti) and giant foxtail (Setaria faberi), with half-lives in the seedbank of 8.5 and 0.8 years, respectively (Lueschen et al. 1993; Buhler and Hartzler 1999). Only seeds determined to be viable through tetrazolium testing were included in our analyses of seedbank population densities, so the numbers reported above are for viable seeds. The seeds were shed with strong primary dormancy that quickly declined after the first overwinter period.

2. In the PLOS One study, we assumed that different weed management strategies would be used in the different rotation systems. The 2-year system used the dominant approach presently used in the Corn Belt: broadcast application of pre-emergence or post-emergence herbicides, without cultivation between crop rows. The 3-year and 4-year rotations employed banded applications of post-emergence herbicides over corn and soybean rows and cultivation between rows. No herbicides were applied to the small grain and forage legume phases of the 3-year and 4-year rotations, though grain stubble was clipped for weed suppression. The diversified, multi-tactic weed management strategies used in the 3-year and 4-year systems are not widely used at present in the Corn Belt, but are employed on some commercial farms, and have been identified as examples of what is necessary to address the growing problem of herbicide-resistant weeds (Mortensen et al. [2012], Navigating a critical juncture for sustainable weed management, BioScience 62:75-84). Thus, while the ecotoxicity results might well differ somewhat among individual farms using different herbicide chemistries, should diversified systems become common enough to compare over long periods of time to the 2-year system that is currently dominant, we believe that our projections represent a reasonable comparison among systems based on present-day practices.
A companion study conducted within the Marsden Farm plots in 2008-2010 by Gomez et al. ([2012], Comparison of crop management strategies involving crop genotype and weed management practices in conventional and more diverse cropping systems, Renew. Ag. Food Sys. doi: 10.1017/S1742170512000142) examined the use of transgenic and non-transgenic corn hybrids and soybean varieties paired with high-herbicide-input and low-herbicide-input weed management strategies. For the transgenic crop / high-herbicide-input regime, mean herbicide use for the 2-year, 3-year, and 4-year rotations was 1.6, 1.1, and 0.8 kg active ingredients per ha per year, respectively. For the non-transgenic crop / low-herbicide-input regime, mean herbicide use for the 2-year, 3-year, and 4-year rotations was 0.1, 0.7, and 0.5 kg active ingredients per ha per year, respectively. Weed dry matter production in corn and soybean was very low (<62 kg/ha), regardless of crop rotation system, crop genotype, and herbicide use.

3. As stated in the article, we assume that the 3- and 4-year rotation systems are linked to livestock operations on the same farm or on neighboring farms. Given the long history of integrated crop-livestock systems in the US Midwest and northern Europe, and the potential financial and environmental benefits of using this strategy in the 21st century (Russelle et al. [2007], Reconsidering integrated crop-livestock systems in North America, Agron. J. 99: 325-334; Sulc and Tracy [2007], Integrated crop-livestock systems in the US Corn Belt, Agron. J. 99: 335-345), we feel this is a reasonable assumption of what could be done in the near future.
As stated in Appendix S1, we assigned labor and machinery costs to spreading manure, but assigned no cost to material itself, since we assumed that recycling nutrients and organic materials is an intrinsic component of integrated crop-livestock systems, and that failure to do so would create a waste problem, which a livestock producer would be compelled to deal with. It is possible, however, to envision a scenario in which manure has a price determined by its nutrient content. An analysis of the financial consequences of doing this in the Marsden Farm experiment is presented by Johanns et al. [2012], http://www.extension.iast.... For the period of 2006-2011, Johanns et al. found that including the costs of manure priced by its nutrient content decreased returns to land and management of the 3-year and 4-year rotations by 7% and 5%, respectively.

4. Labor inputs for each rotation system are shown in Fig. 1H and were determined using data from Hanna [2001], http://www.extension.iast.... We assumed all farm operations were done with machines, not by hand.