Production, physiology, and plant/seed production

0800. Methyl Bromide fumigation alternatives for sweetpotato hotbeds in California: a preliminary report (back)

C. S. Stoddard^1*, M. Davis², A. Ploeg³, A. Shrestha⁴, J. Stapleton⁵

¹UC Cooperative Extension, Merced, CA;²UCCE Plant Pathology, UC Davis; ³UCCE Nematology Specialist, UC Riverside; ⁴UCCE IPM Weed Ecologist, Kearney Agriculture Center;⁵UCCE IPM Plant Pathologist, Kearney Agriculture Center.

In California, sweetpotato hotbeds are commonly fumigated in the late fall with a MeBr + Pic combination, tarped with standard plastic. Currently, MeBr is allowed under a Critical Use Exemption (CUE) with the U.S. Environmental Protection Agency, however, this is unlikely to continue indefinitely. Products containing Telone and chloropicrin are a strategic part of the methyl bromide alternatives strategy for this industry. The most common replacement would be Telone (1,3-D) + Pic. Unfortunately, Telone has numerous regulatory restrictions because it cannot be applied in December or January and is subject to maximum use caps that renew at the beginning of each year. Despite its widespread use, there has been little research in California investigating fumigant effects in sweetpotato hotbeds on plant production or the subsequent effects in the field (most research has taken place in production fields, not hotbeds). A small trial performed in 2006 in Merced County, CA, showed that chloropicrin did not provide satisfactory weed control unless combined with Telone. Further fumigation research is needed to investigate the impacts of multiple years without chemical fumigation.Beginning in the summer of 2007, a project has been approved by the USDA ARS with the objective of evaluating alternatives to MeBr for sweetpotato hotbeds. The project includes Telone, Pic, and Vapam combinations, as well as solarization, compared to MeBr. Additionally, main plots will be split at bedding to compare chemical and variety combinations that may be viable alternatives to straight chemical fumigation. Two fungicides (Botran, Mertect), herbicides (Devrinol, Valor), and varieties (Beauregard, Golden Sweet) will be compared. Nematodes, weed control, plant stand production, disease incidence if present, and effects on field production will be measured.

0815. Empirical modeling of sweetpotato yield using heat units and climatic variables: modeling frames and model performance (back)

A. Villordon^1*, C. Clark², D. Ferrin², and D. LaBonte³.¹LSU AgCenter Sweet Potato Research Station, ²LSU AgCenter School of Plant, Environmental, and Soil Sciences; ³LSU AgCenter Department of Plant Pathology and Crop Phsyiology.

U.S.#1 yield data from 107 planting dates (2002-2007) and from representative sweetpotato growing locations in Louisiana were used in this study. A combination of minimum coefficient of variation (CV) and linear regression (LR) was used to identify candidate growing degree day models (cGDDs) using various base (B) and ceiling (C) temperatures. LR (stepwise selection) was conducted for U.S.#1 sweetpotato production [total data (TD) set=107; dependent variable, DV=U.S.#1]. This approach identified the following candidate models and combinations of B and C: M2 (Tmax-B, Tmax=maximum air temperature, where if M2<0, then M2=0; 60-80F, 60-80F, 60-85F, and 60-95F), maximum-limited M2 (M4, 65-100F; M6, 70-95F, 60-100F), triangle (M7, 60-85F), and sine (M8, 60-90F). LR (stepwise selection, DV=U.S.#1) analysis that included cGDDs and several climatic variables identified the following predictor variables (PVs): C2 (60-80F, 60-85F, 6090F, 6095F), mean relative humidity five days after transplanting (RH5), mean minimum RH5 (MNRH5), and mean minimum RH 30 days to harvest (MNRHT30). LR (DV=U.S.#1, TD) was compared to the following adaptive techniques: generalized linear model (GLM), regression tree (RT), and neural network (NN). GLM analysis on TD (link function=identity; selection method=stepwise, criteria=AIC; optimization method=conjugate gradient) yielded models with the following PVs: M2 (60-85F, 60-90F, 60-95F, or 60-100F), RH5, MNRH5, and MNRHT30. GLM, RT (splitting criterion=F test, significance level=0.20), and NN (model selection criteria=average error; network architecture=multilayer perceptron; training technique=standard back propagation) modeling was subsequently conducted on partitioned TD: 50% training (TRAIN), 25% testing (TEST), and 25% validation (VALID). RT models had the lowest TRAIN, TEST, and VALID average square errors (ASEs). All RT models identified M1 ([(Tmax-Tmin)/2)-B], where if [(Tmax-Tmin)/2]<B, then [(Tmax-Tmin)/2]=B), RH5, and MNRHT30 as PVs. GLM analysis yielded a model with a low Schwarz Bayesian Criterion score and the following PVs: M2 (60-90F), RH5, MNRH5, MNRHT30, and mean rainfall 30 days to harvest (RAINT30). NN-based models had relatively higher TRAIN, TEST, and VALID ASEs compared to GLM and RT-based models. M2 with B=60F and C=90F appears to be the best method for calculating GDD for sweetpotatoes grown in Louisiana. The best performing LR and GLM models for predicting U.S.#1 yield shared common PVs: M2, RH5, MNRH5, and MNRHT30. A fifth PV (RAINT30) was unique to the GLM model calculated with partitioned data.

0830. The "rootcam" and other methods for observing and quantifying sweetpotato adventitious root initiation, growth, and development (back)

A. Villordon¹* and D. LaBonte².¹LSU AgCenter Sweet Potato Research Station, ²LSU AgCenter School for Plant, Environmental, and Soil Sciences.

Relatively few studies have documented how the interaction of biological and environmental variables influences the early initiation, growth, development, and morpho-anatomical characteristics of sweetpotato adventitious roots (ARs). Yet, these studies appear to underscore the significance of AR initiation and development on the timing of storage initiation events and final storage root yield. Although destructive sampling methods have provided information on some quantifiable morphological and anatomical characteristics, not all aspects of AR initiation and early development can be captured by these conventional sampling approaches. For example, the phenomenon of diurnal growth can be more effectively quantified by time-course measurements of in-situ samples. We used consumer-grade webcams in tracking the early development of roots from transplants grown in liquid growing medium. Timed images of initiating and developing ARs were captured every 2-4 hours by off-the-shelf software. Such images provided data on time of initiation as well as AR root number, diameter, and length. We describe the basic components of this real-time AR observation system as well as its potential applications and limitations. We will also describe the use of sand-based growing medium in conducting limited tracking of root development as well as quantifying the effects of external variables on morpho-anatomical characteristics and other quantifiable traits of newly-initiated and developing ARs.