Application of near-infrared spectroscopy to predict the cooking times of aged common beans (Phaseolus vulgaris L.)

Highlights

• Cooking times of beans were predicted from NIR spectra of milled raw beans.

• De-hulling beans prior to NIR measurements improved prediction accuracy.

• NIR spectra of aged beans clustered into groups according to variety.

• Variety-specific calibrations more accurately predict cooking times of aged beans.

Abstract

The cooking time of beans is an important quality indicator which can change considerably during ageing. Therefore, this study investigated the potential of near-infrared spectroscopy to rapidly predict cooking times of aged common beans. Four bean varieties were aged under different storage conditions, resulting in a range of samples for each of these varieties. The cooking kinetics of the aged beans were determined by finger pressing and modelled using logistic regression to obtain the times it took to cook 95% of the beans. The cooking times obtained were predicted from the NIR spectra of milled raw bean samples. This was done using partial least squares regression, after carrying out wavelength selection. Model performance was improved up to an average prediction error of 8 min by de-hulling the beans and reducing the number of varieties included. In conclusion, NIR spectroscopy has high potential to predict the cooking times of aged beans.

Introduction

Legumes are widely considered important foods as part of diverse, nutritious, healthy and environmentally sustainable diets (Willett et al., 2019; Castro-Guerrero et al., 2016). This is true for the majority of lower-income populations in developing regions that highly depend on plant foods due to the high cost of foods from animal sources (Chávez-Mendoza and Sánchez, 2017; Castro-Guerrero et al., 2016; Câmara et al., 2013) as well as for high-income populations (Willett et al., 2019). Among the legumes, common beans (Phaseolus vulgaris L.) are the most economically significant species and are widely accepted (Broughton et al., 2003). For all these reasons, common beans have great potential to contribute to food and nutrition security now and in the future.

The cooking process (cooking time) is one of the major factors affecting the consumption convenience and general acceptability of common beans (Katungi et al., 2011; Bernal-Lugo et al., 1997). Conventional cooking is usually a two-step procedure of texture degradation which starts with soaking followed by thermal processing, mostly boiling at household or industrial level. During cooking, the rate limiting step of bean softening has been reported to be the solubilization of pectin present in the middle lamella and the cell walls of the cotyledon (Chigwedere et al., 2018). The cooking process of beans has been previously measured subjectively and/or objectively as the percentage of beans that have attained a stipulated endpoint of texture degradation or chemical change (e.g. starch gelatinization), then called cooked. The time it takes to reach the desired endpoint is an important quality indicator for breeders, seed distributors, food industries and consumers (Wood, 2017).

The cooking time of beans depends on variety (Bernal-Lugo et al., 1997; Kinyanjui et al., 2015) and even within the same variety they can differ depending on postharvest storage conditions (Kinyanjui et al., 2017). Cooking times can range from 30 to 120 min when beans are freshly harvested (Garcia et al., 2012) up to 8 h due to inappropriate storage (Garcia et al., 1998). This has been attributed to the development of the Hard-To-Cook (HTC) textural defect. This defect causes the bean cotyledon to harden and although the bean properly hydrates, bean softening is hindered, resulting in a prolonged cooking time (Shiga and Lajolo, 2006; Shiga et al., 2004).

The mechanism of HTC development in beans has yet to be fully elucidated, nonetheless, several mechanisms have been suggested. The mechanisms depend on enzyme-catalyzed reactions and transport of small molecules largely controlled by the plasticizing effect of water. Therefore, temperature and relative humidity (typically above 25 °C and 65%) during storage (Stanley, 1992) and probably soaking are key factors of HTC development (Chigwedere et al., 2019a). The current overall hypothesis suggests that during adverse storage, membrane disruptions and (lipoxygenase catalyzed) lipid oxidation occur (Hussain et al., 1989). The pectin-cation-phytate theory (Kilmer et al., 1994; Hentges et al., 1991; Jones and Boulter, 1983), involves divalent cations released when phytase hydrolyzes phytates. The cations diffuse from protein bodies into the cell walls where they cross-link with pectin demethoxylated by pectin methylesterase. The lignification (Garcia et al., 1998; Hohlberg and Stanley, 1987) and multiple reactions theories (Aguilera and Rivera, 1992) involve the transport of phenolics to the cell walls resulting in lignification and/or phenolic-pectin cross-links. Besides the abovementioned theories, it has also been suggested that protein denaturation during storage and cooking hinders starch denaturation and thus bean softening (Garcia-Vela and Stanley, 1989).

To date, the approaches used to determine the cooking times of beans and eventually the HTC defect at the user level include finger pressing and texture analysis among others after cooking (Wood, 2017). Both these methods are time-consuming. Additionally, the finger pressing method requires trained personnel to obtain reliable results as it is subjective (Kinyanjui et al., 2015). This calls for further exploration of alternative techniques to address these challenges.

Near infra-red (NIR) spectroscopy is a high throughput analytical tool (Manley, 2014; Osborne, 2000b) that is widely applied in the quality check of raw materials and end-products in food research and industry (Nicolai et al., 2007; Williams, 2004; Osborne, 2000a). The NIR region of the electromagnetic spectrum ranges from 780 nm to 2800 nm. Samples are scanned to obtain spectra containing absorption information of functional groups (O–H, C–H, CO, N–H) and scattering information induced by microstructural heterogeneities. Subsequently, chemometrics is used to correlate the spectra to sample quality attributes obtained using conventional methods. This results in calibration models used to predict quality attributes of new samples (Saeys et al., 2019; Pasquini, 2018; Williams, 2004). NIR spectroscopy has been successfully used to predict the chemical composition, physical characteristics and sensory properties of common beans (Wang et al., 2014; Plans et al., 2014, 2013; 2012; Pojić et al., 2010).

NIR spectroscopy and hyperspectral imaging have also been successfully used to predict cooking times of Andean Diversity Panel accession covering a wide range of fresh common bean seed types (Mendoza et al., 2018b; Cichy et al., 2015). However, these studies excluded the development of storage-induced HTC. Also, the spectra were acquired for whole beans, which to the best of our knowledge likely put limitations on the vis/NIR approach when using the reflectance mode to evaluate cooking times associated with an interior structure, the cotyledon. Therefore, the aim of this study was to evaluate to what extent NIR spectroscopy can be used to determine the cooking time for aged beans. Hence, the experimental design involved using a limited number of varieties stored under different conditions to generate aged samples with a broad range of cooking times within each variety. As it has been shown that HTC development is cotyledon related, the aged samples were separated into two groups, the first with seed coats and the second as cotyledons only. The samples were milled before the acquisition of spectra. By doing this, we hypothesized that the molecules described in one or more of the mechanisms leading to HTC development could be identified during model development.