Statistical Modeling of Yield and Variance Instability in Conventional and Organic Cropping Systems

Published in Agron. J. 103:673–684 (2011) Published online 9 Mar 2011 doi:10.2134/agronj2010.0420 Copyright © 2011 by the American Society of Agronomy, 5585 Guilford Road, Madison, WI 53711. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. T diversity and intensity of cropping systems in the Corn Belt have been changing over time in response to several interacting biophysical and social factors (Posner et al., 2008). During the second half of the 20th century, cropping systems in large parts of the Corn Belt became more specialized at fi eld, farm, and landscape levels, where reduction in number of crops and variability within fi elds led to the development of monocultures that potentially increase environmental risks because they reduce biodiversity, ecosystem functions, and ecological resilience (Rozenzwieg and Tubiello, 2007). Th e longstanding debate over the trajectory of extensive cropping systems based on the corn–soybean rotation in the Corn Belt deserves a new look because it focuses on a single ecosystem service (i.e., production), overconsumes environmental resources, and releases chemicals to the environment (Turinek et al., 2009). Th e sustainability of cropping systems is most eff ectively evaluated by long-term experiments that simulate management practices and conditions encountered in farmers’ fi elds (Singh and Pala, 2004). Moreover, long-term experiments are necessary if the sustainability of a production system is to be determined, such that long-term yield trends, dynamics of the availability and balance of nutrients and capacity of the soil to maintain productivity over time can be measured (Stanger and Lauer, 2008). Long-term experiments, which provide a measure of sustainability, can be used to detect problems that may aff ect productivity (Berzsenyi et al., 2000). Due to their complex nature, cropping systems cannot be fully studied using reductionist (e.g., factorial) experiments (Drinkwater, 2002); whereas, a systems approach, where treatments represent intact management strategies, is more powerful in elucidating how functions of a cropping system (e.g., yield) are determined by the interrelationships among treatments and biophysical processes (Drinkwater, 2002; Brandt et al., 2010) An ideal study site for cropping systems experiment is a land area representative of a large ecogeographical region, in which the plot size and soil variability have been optimized for a specifi c crop, crop rotation, or treatment (Legendre et al., 2004). Finding such an experimental fi eld is a challenge in large parts of the upper Midwestern United States where research on fi eld stations was originally developed for and continues to be devoted to small, randomized plot experiments (Cambardella et al., 1994; Porter et al., 2003). A combination of reductionist and holistic systems approaches in the design and analysis of long-term experiments is essential for understanding the complex interactions of plants, soils, climate, and management (Fagroud and van Meirvenne, 2002). Th is approach is also critical in estimating the covariance or correlation structure of experimental errors over time (Singh and Jones, 2002) if clear insights into the eff ects of crop rotation and the management or input factors associated with them are to be obtained. Spatial (between-plots) and temporal (within-plot) variation are the major components of ABSTRACT