J. Agric. Food Chem. 2008, 56, 10433–10438
10433
Logistic Regression Modeling of Cropping Systems To Predict Fumonisin Contamination in Maize PAOLA BATTILANI,*,† AMEDEO PIETRI,§ CARLO BARBANO,† ANDREA SCANDOLARA,† TERENZIO BERTUZZI,§ AND ADRIANO MAROCCO# Institute of Entomology and Plant Pathology, Institute of Food Science and Nutrition, and Institute of Agronomy, Faculty of Agriculture, Universita` Cattolica del Sacro Cuore, via Emilia Parmense 84, 29100 Piacenza, Italy
The aims of this research were to monitor the presence of fumonisins in maize crops in northern Italy over a 6 year period, to study the role of the cropping system on fumonisin levels, and to contribute to the development of a predictive system for fumonisin contamination. In the 6 year period from 2002 to 2007, 438 maize samples were collected in five regions, supported by agronomic data, and analyzed for fumonisin content. Fumonisin was detected in almost all of the grain samples, but 2007 was less and 2005 more contaminated compared to the other years. Preceding crop, maturity class of hybrids, nitrogen fertilization, sowing and harvest week, and grain moisture significantly affected the level of contamination. The logistic regression developed explained around 60% of variability with major roles for longitude, maturity class, and growing weeks. The function can be used to quantify the effect of these factors in a predictive system. KEYWORDS: Cropping system; maize; mycotoxins; fumonisins; modeling; logistic regression
INTRODUCTION
Fusarium Verticillioides Sacc. (Niremberg) is the most common toxigenic fungus in maize; it causes ear rot disease, typically occurring on random kernels, groups of kernels, or physically injured kernels, and consists of a white or light pink mold (1). In addition to causing ear rot symptoms, F. Verticillioides is also frequently found in symptomless kernels (2). F. Verticillioides produces primarily fumonisin toxins; these are a group of at least 15 compounds, the most prevalent being fumonisin B1. Mycotoxins represent a serious, multifaceted economic problem, and maize crops are the most commonly affected, economic losses in maize being due to yield loss caused by diseases induced by the fungi, direct loss of grain which is unfit for sale due to mycotoxin contamination, losses in animal productivity because of mycotoxin-related health problems, and human health costs (3). Approximately 10 million tons of maize per year are produced in Italy, primarily in the Po Valley (northern Italy); a high percentage of this production (88%) is directly consumed as animal feed. In Italy, fusarium ear rot is the most common disease associated with maize ears, and late in the season it can be found at low levels in nearly all maize fields; consequently, fumonisins are the main mycotoxins that affect maize in this * Corresponding author (telephone +39 0523 599254; fax +39 0523 599256; e-mail
[email protected]). † Institute of Entomology and Plant Pathology. § Institute of Food Science and Nutrition. # Institute of Agronomy.
geographic area (4, 5). Very high levels of fumonisin contamination were recorded in northern Italy over the period from 2004 to 2006 (6). The usual route of mycotoxin exposure is consumption of contaminated food or feed by humans or animals; however, dermal exposure and inhalation may also be important. The primary concern with regard to fumonisin exposure is its possible connection with human esophageal cancer and with a higher incidence of neural tube defects; furthermore, it shows acute toxicity for reared animals (7, 8). Worldwide food legislation safeguards the health of consumers and the economic interests of producers and traders, imposing limits on the concentrations of specific mycotoxins in foods. A European Union regulation sets maximum levels for Fusarium toxins in foodstuffs (9). The maximum tolerated level for total fumonisins (FB1 + FB2) in raw maize is 4000 µg/kg, and lower levels are fixed for maize-derived foods for direct human consumption. Recommendations have also been made at the European level regarding fumonisin (FB1 + FB2) content in animal feeds (10); the lowest concentration is 5 mg/kg, tabulated for complete feeds destined for pigs and horses. Fusarium ear rot is more common in warm and dry areas, and it is generally favored by warm, dry weather during the grain-filling period (1, 11). In several studies cited by Munkvold (12), fumonisin levels were negatively correlated with seasonlong rainfall or with rainfall in June. Any management practice able to maximize plant performance and decrease plant stress reduces mycotoxin contamination (13). Most of the fungi causing ear rot survive in crop residues; good management of residues is suggested as a control
10.1021/jf801809d CCC: $40.75 2008 American Chemical Society Published on Web 10/09/2008
10434
J. Agric. Food Chem., Vol. 56, No. 21, 2008
Battilani et al.
Figure 1. Sampling points in the 2002-2007 maize survey carried out in northern Italy.
measure for this disease (14), but there is little direct evidence of the success of this approach. Studies on the survival of Fusarium species causing fusarium ear rot suggest that tillage and crop rotation are unlikely to affect this fungus and its mycotoxins (15), and tillage practices did not affect the incidence of ear rot caused by F. Verticillioides or Fusarium graminearum in South Africa (16). Previous cropping history has been shown to influence soil populations of toxigenic fungi (17), but the importance of initial inoculum has not been established. Sowing dates can influence the risk of ear rot; earlier sowing results in a lower risk, greatly depending on annual meteorological conditions. With regard to toxin levels, crop nutrition stress and weed-related stress have been associated with high contamination (18). Jones and Duncan (19) reported that a higher rate of nitrogen fertilizer consistently reduced aflatoxin levels. The aims of this research were to monitor the presence of fumonisins in maize crops of northern Italy over a 6 year period, to study the role of the cropping system on fumonisin levels, and to contribute to the development of a predictive system for fumonisin contamination at harvest. MATERIALS AND METHODS Sampling and Data Collection. In the 6 year period from 2002 to 2007, samples were collected in five regions where maize is an important crop destined for food or feed: Piedmont, Lombardy, Veneto, Friuli Venezia Giulia, and Emilia Romagna. The monitored area was between 43.8589 and 46.1667 latitude north and between 7.4989 and 13.3327 longitude east. A total of 98, 98, 84, 77, 37, and 44 maize samples, grown in 2002, 2003, 2004, 2005, 2006, and 2007, respectively, were collected during harvesting from maize crops managed using the ordinary cropping system for each area (Figure 1). Sampling of maize was performed according to EC Directive 76/371, Commission of the European Communities (20). Samples were dried at 45 °C and analyzed for fumonisins; fumonisin B1 (FB1) was determined in all of the samples and FB2 starting from 2005. A questionnaire was prepared and completed for each sample to collect the relevant information on the cropping system: soil texture, previous crop, debris management, tillage and other field operations, hybrid seeded, sowing period and investment, mineral nutrition, weeds control, irrigations, flowering period, crop injuries (borers, hail, and wind), chemical control of ECB, harvesting period, and moisture of kernels at harvesting.
A database was built up, including data from the questionnaires and on fumonisin content in all of the samples. All data were georeferenced using GIS Arc View 8.2 [Environmental System Research Institute (ESRI), Redlands, CA]. Fumonisin Analysis. Fumonisins were determined according to the method of Visconti et al. (21). Fumonisins were extracted from 10 g of sample in a plastic centrifuge bottle with 50 mL of acetonitrile/ methanol/water (25:25:50, v/v/v). After extraction for 45 min using a rotary-shaking stirrer and centrifugation at 4500g for 6 min, the supernatant was poured into a flask; another 50 mL of the same solution was added to the residue in the centrifuge bottle, and a second extraction was performed for 30 min. The combined extracts were filtered through a folded filter paper. An aliquot of 2 mL was diluted with 20 mL of 0.1 M phosphate-buffered saline (PBS, pH 7.4) and purified through an immunoaffinity column (R-Biopharm Rhoˆne Ltd., Glasgow, Scotland); after the column had been washed with PBS (2 mL), the fumonisins were slowly eluted (0.5 mL/min) with methanol (6 mL) into a graduated glass vial; subsequently, the eluate was concentrated to 2 mL under a gentle stream of nitrogen. Analysis was carried out using a LC-MS/MS system, consisting of an LC 1.4 Surveyor pump (Thermo-Fisher Scientific, San Jose, CA), a PAL 1.3.1 sampling system (CTC Analytics AG, Zwingen, Switzerland), and a Quantum Discovery Max triple-quadrupole mass spectrometer; the system was controlled by Excalibur 1.4 software (Thermo-Fisher). After dilution of the extract (0.1 mL brought to 1 mL) with acetonitrile/water (30:70, v/v, acidified with 0.4% acetic acid), the fumonisins were separated on a 150 mm × 2.1 mm i.d., 5 µm, Betasil RP-18 column (Thermo-Fisher) with a mobile phase gradient of acetonitrile/water (both acidified with 0.4% acetic acid) from 25:75 to 55:45 in 9 min, then isocratic for 3 min; the flow rate was 0.2 mL/min. Ionization was carried out with an ESI interface (Thermo-Fisher) in positive mode as follows: spray capillary voltage, 4.0 kV; sheath and auxiliary gas, 35 and 14 psi, respectively; temperature of the heated capillary, 270 °C. The mass spectrometer analysis was operated in selected reaction monitoring (SRM). For fragmentation of [M + H]+ ions (m/z 722 for FB1, m/z 706 for FB2), the argon collision pressure was set to 1.5 mTorr and the collision energy to 36 V. The detected fragment ions were m/z 704, 352, and 334 for FB1 and m/z 688, 336, and 318 for FB2. Quantitative determination was performed using LC-Quan 2.0 software. Fumonisin standards were obtained from Sigma-Aldrich (St. Louis, MO). FB1 and FB2 (1 mg) were separately dissolved in 10 mL of acetonitrile/water (1:1, v/v); the concentration was calculated using the weight indicated by the manufacturer. These solutions were diluted to obtain HPLC calibrant solutions in acetonitrile/water (30:70, v/v, acidified with 0.4% acetic acid) at individual concentrations of FB1 and FB2 between 2.5 and 50 µg/kg.
Prediction of Fumonisin in Maize
J. Agric. Food Chem., Vol. 56, No. 21, 2008
Table 1. Index Groups Defined for Agronomic Traits of Maize Samples soil index sand preceding maturity sowing group content crop classa week 1 2 3 4 5 a
20 40 60
maize wheat others
110 118 125 130 135
nitrogen
Table 2. Mean Yearly Content and Minimum/Maximum Levels of Fumonisin B1 and B2 in Samples Collected in the 6 Year Survey in Northern Italy
harvest grain week moisture
e13 e200 e35 14-16 200-324 36-37 g17 g325 38-39 g40
e20 20-23 g24
Maturity class was expressed as mean number of days from emergence to
ripe.
10435
FB1 (µg/kg)
FB2 (µg/kg)
FB2/FB1
year
mean
min/max
mean
min/max
mean
min/max
2002 2003 2004 2005 2006 2007
5132 (55/98)a 5415 (49/98) 6303 (44/84) 6910 (53/77) 4018 (10/37) 662 (0/44)
