INVESTIGADORES
GOICOECHEA hector Casimiro
congresos y reuniones científicas
Título:
Response Surface Methodology for Optimization of Biopesticides Production using Industrial Wastewater as Substrate
Autor/es:
GOICOECHEA, HÉCTOR C; AUGUSTO MOREIRA, GUILHERME; ALEJANDRO BECCARIA,
Lugar:
Campinas
Reunión:
Congreso; 10th Internacional Conference on Chemometrics in Analytical Chemistry (CAC 2006); 2006
Institución organizadora:
CAC
Resumen:
Insecticides based on entomopathogens are
generally specific and present low or no toxicity to vertebrates. One of the
leads to successful production of biopesticides is the development of the
medium composition. The objective of this work was to optimize the parasporal
crystals production of Bacillus thuringiensis var. kurstaki using
an experimental desing of the medium components (substrates) coupled with
Response Surface Methodology (RSM) followed by multiple variable response
optimization through a desirability function.
The
strain Bt var. kurstaki HD-1 was provided by Dra. Graciela
Benintende, IMYZA, Instituto Nacional de Tecnología Agropecuaria, Argentina, was grown on Tryptic Soy Agar slants
(Britania, Argentina) and stored at 4°C. The substrates used were sugar
cane molasses (30º Brix in DI water), reduced-fat milk (2.624g.L-1 in DI water with Biochemical Oxygen Demand
equivalent to 1915.5mg.L-1) and brewery wastewater (with Chemical
Oxygen Demand equivalent to 3473.0mg.L-1 and pH 10.2). Effluents
proportion was analysed by using a simplex lattice {3,2} augmented
with the overall
centroid and axial points. This design has 10 points, with four of these points
in the interior of the simplex. Additional replicates and a random point were
added in order to increment the number of experiments for modelling purposes. A
culture medium TSB was employed as control culture. The responses were analysed
by counting, in a calibrated microscope, vegetative cells, spores and
parasporal crystals (fixed by flame and coloured by gentian violet) in a 0.28
cm2 slide. Design Expert version 6.0.10 trial (Stat-Ease,
Inc., Minneapolis, USA) was used for performing
experimental design and data analysis.
We
found that the F values (F tests performed for the responses prediction models)
are lower than expected, causing not significance of the models (cubic models).
We also found that the F for LOF (lack-of-fit) is also low (p>0.05)
which means that instead of the low reliability of the models they still can be
fit in order to get an approximation of the optimum medium composition. For this
purpose, we used the desirability function (D) for crystals and vegetative
cells. The value for D was found to be 0.979 (the maximum value for D is equal
to 1) which gives a prediction for the best composition (76.5 % of reduced-fat
milk, 22.0 % of brewery wastewater and 1.5 % of sugar cane molasses). The
results are the following: 4.5E+8 crystals/mL, 3.3E+8 vegetative cells/mL,
6.4E+6 vegetative cells/mL. Further analysis with Artificial Neural Networks
(ANNs) resulted in an optimal mixture as follows: 74% reduced-fat milk, 26%
brewery wastewater, 0% sugar cane molasses (D = 0.896). The predicted response
for this mixture is: 4.1E+8 crystals/mL, 2.9E+8 vegetative cells/mL, 9.0E+7
vegetative cells/mL.
We
found an optimal mixture of effluents defined by RSM, although the fitted model
presented low reliability. Further analysis made with ANNs showed an optimal
composition consistent with the one obtained by RSM but analytically more
reliable. This probably was due to the higher complexity of the bioprocesses,
the strong interaction between the components and high experimental noise
level. We achieved good Bt var. kurstaki growth and good
parasporal crystals production, according to our objectives. However,
complementary studies should be made to evaluate the efficiency and potency of
the products in target larvae. The study could be enhanced with a new screening
of combinations of other effluents by using this methodology.
Acknowledgment.
Universidad
Nacional del Litoral and CONICET (Consejo Nacional de Investigaciones
Científicas y Técnicas) are acknowledged for financial support. G.A. Moreira. thanks AUGM for a
fellowship. G.A. Micheloud thanks CONICET for a fellowship.