In-field measurement of starch content of cassava tubers using handheld vis-near infrared spectroscopy implemented for breeding programmes

https://doi.org/10.1016/j.compag.2020.105607Get rights and content

Highlights

  • NIR spectroscopy has potential for measuring starch content of on-field fresh tuber.

  • Effect of measurement position (head, middle and tail) was examined.

  • Model of mixed sample was recommended in real situation measurement.

  • Model ability for future sample was validated by unknown in-field spectra, providing RMSEP < 1.87%.

Abstract

This paper involves the prediction of cassava tuber starch content (SC) using near-infrared (NIR) spectroscopy, aiming to follow the change of SC in individual tubers utilised for a breeding programme. This study applies a portable NIR spectrometer at wavelengths of 570–1031 nm in the evaluation of SC in fresh cassava tubers. The prediction models are established using partial least squares (PLS) regression with NIR spectra obtained in the interactance mode. The effective model was developed from the wavelength region of 600–1000 nm with spectral pre-processing of the second derivative, giving the coefficient of determination of prediction set (r2) and root mean square error of prediction (RMSEP) of 0.62 and 2.21%, respectively. The effect of tuber section (including head, middle and tail) on the performance of the SC model was investigated. The individual head, middle and tail models were acceptable for screening. However, the performances of the combined model (which is the model developed a mix of all individual section samples) and the individual section model were not significantly different. Therefore, the combined model was suitable in real application because of the ease of in-field scanning. The result demonstrates that the SCs of cassava tubers can be measured by a NIR spectroscopy method. Furthermore, it can be used as an alternative tool which is appropriate for breeders to use to follow the behaviour of SC during breeding.

Introduction

Cassava (Manihot esculenta) is an important economic crop. It has been utilised as a raw material for producing food and bioenergy. Thailand is the second-largest producer of cassava in the world and has the highest export value coming from cassava products (OAE, 2018). All parts of cassava are valuable. Cassava root is used to produce food, animal feed and biofuel (ethanol). Other parts, such as leaves, stalks and rhizomes, are used as raw materials in various industries, such as power plants, pelletisation factories and so on (Kasetsart university research and development institute (KURDI)., 2015a, Koopmans, 2005). In the commercial market, the price of cassava root is set by its weight, moisture content (MC) and starch content (SC). The SC is an especially key index of the quality and value of cassava root (Buddhakulsomsiri et al., 2015, Mulualem and Bekeko, 2015). Cassava with higher SC is sold at a high price. Therefore, the variety that can provide the highest SC, store SC the fastest and keep SC constant as long as possible is required for breeding and farming (Buddhakulsomsiri et al., 2015, Chuasuwan, 2018). The SC and yield of cassava root can vary according to the planting conditions, such as soil moisture content, temperature, soil fertiliser and environmental surroundings. Therefore, if a good variety is planted but in a different season and different area, then it may not provide high quality and productivity (Janket et al., 2018). Even with the same variety, if the planting conditions are changed, SC and productivity can be different. To maintain and increase quality and productivity, it is necessary to develop a specific variety always suited to the culture. Moreover, because of climate change, a new variety has to be improved more often because the current variety will not have good quality in the future.

Presently, the lead time of cassava breeding is long and has a rather low accuracy, waiting for more than 15 years before release to farmers (KURDI, 2015b). For the breeding protocol, after pollination to produce hybrids and waiting to get cloned hybrid seeds, each seed is an individual variety. Each seed (each variety) is planted to select the best variety among hybrid populations. When the cassava age reaches 12 months, the breeder must compare the yield of tubers and other important agronomic traits of the cloned hybrids. In this step, each cassava variety has only one stalk, and so SC cannot be measured to follow the change of SC during planting. This leads to the breeder using information related to the appearance of external features to evaluate SC (Hahn et al., 1973). There is a high possibility of mistakes in variety selection. When the cassava age is 12 months, the SC of each variety is measured using the traditional method based on the specific gravity (SG) of cassava roots. In addition, the traditional method for determining SC of cassava can follow the manual of “ISO 10,520 Native starch - Determination of starch content - Ewers polarimetric method” (ISO, 1997). The cassava peel is removed and then dried using a hot-air oven at 60 °C until a constant weight is observed, and about 100 g of dried cassava is ground and sieved through 180-mesh. The ground sample is utilised to determine %polarimetric. Determining the SC of cassava by the polarimetric method is inconvenient, expensive and requires skilled personnel. Thus, breeders use the specific gravity (SG) method for estimating SC instead of a polarimetric method, because the difference of SC measurement between the polarimetric and SG methods was not over 6.2% (Bantajan and Rittiron, 2016).

For the SG measurement method, the cassava root is pulled out of the ground, and then the stem is removed using a chopping knife. The requirement is at least approximately 5 kg to determine the SG of cassava roots. The SC can be estimated if the SG of the sample is known, according to a strong relationship among SG, SC and dry matter reported by Wholey and Booth (1979). This method was not suggested for young cassava due to some stalks not providing enough samples of tubers for measuring SC. Therefore, the breeder cannot follow the change of SC until harvesting (age from 4 to 12 months). For example, some varieties such as ‘RY9′ and ‘RY11′ cannot provide enough samples for monthly measurements. Therefore, non-destructive evaluation of cassava tuber quality is necessary, which can increase the number of times of measuring the SC of living cassava stalks and reduce operating costs and time consumption as well. The selection process for searching for the best variety, which is very important, also takes a long time, around 3–6 years, depending on the variety and weather conditions, for screening to be confident before sending to the next process (Ceballos et al., 2016), which is trial outplanting. A non-destructive test is needed for the selection process because it reduces selection mistakes and decreases the loss of time.

Near-infrared (NIR) spectroscopy has been used for measuring the internal quality of food and agriculture products. Due to having the advantages of being a rapid, accurate, reliable and, importantly, non-destructive analytical method, NIR spectroscopy is widely used in several fields, such as soil science (Steffens et al., 2014), medical technology (Lu & Fei, 2014) and especially agricultural sciences. The rapid measurement of apples (Fan et al., 2016, McGlone et al., 2003), tomatoes (Hahn et al., 2004, Khuriyati et al., 2004) and mangoes (Saranwong et al., 2001, Rungpichayapichet et al., 2017) was carried out by NIR technique to guarantee the product quality. NIR spectroscopy has been utilised in plant breeding programmes such as forest trees, agronomy crops and Kraft pulping (Meder, 2015, Meder, 2018). For example, Maraphum et al., 2018, Phuphaphud et al., 2019 applied the NIR spectroscopic technique to evaluate parameters to index sugarcane quality, such as the Pol value, Brix value and fibre content, of the sugarcane stalk. They obtained a coefficient of determination of prediction set (r2) of around 0.70–0.80. Pomares-Viciana et al. (2017) developed NIR models to predict the SC of freeze-dried zucchini samples, and the model could be used for screening purposes (r2 of 0.66 and SEP of 21.29 mg g−1 DW). NIR spectroscopy could become an especially useful selection tool in zucchini breeding programmes and quality control as well.

Although the NIR spectroscopy provides a good prediction for fruit, for in-field measurement, it should be used with caution due to the impact of sample temperature. The model performance can be accepted if the sample temperature is not too high. From a literature review, the sample temperatures < 45 °C had no impact on NIR spectra and model accuracy, which was shown from measuring tomato dry matter (Acharya et al., 2013), latex dry rubber content (Arsaipanich and Udompetaikul, 2014), the Brix values of intact peaches, pears, persimmons and apples (Jannok et al., 2017) and molasses sugar content prediction (Chapanya et al., 2019). The measurement of the starch contents of cassava tuber using NIR spectroscopy was reported by Bantajan and Rittiron (2016); they used a portable NIR instrument with a wavelength range of 700–1100 nm to evaluate the starch content of cassava tubers. The measurement was done at a trading station for setting the price bought from the farmer. The cassava samples were collected from the trading station at almost full maturity stage (10–12 months old, approximately). The models were created using spectra at the surface (scanning through the peel) and flesh of cassava root (at the cross-section of cassava root that makes a transverse cutting at the middle). The coefficients of determination (r2) were approximately 0.52 and 0.85, and the standard errors of prediction (SEP) were 3.79% and 1.74%, respectively.

Therefore, the purpose of this study is: 1) to investigate the ability of NIR spectroscopy to evaluate the cassava SC of living fresh tubers for the breeding programme; 2) to search for the optimal measurement conditions to introduce to the breeder for the measurement of in-field tubers and 3) to prove whether NIR spectroscopy can be used instead of the traditional method. If the prediction model is satisfied, it benefits the breeding programme. A cassava breeder can track the same tubers to study behaviours of SC accumulation in cassava tubers. This method could be suggested to the breeder to be used as an alternative method of SC measurement in a breeding programme. It can help breeders increase the possibility of success in discovering a good variety and reduce the cost and time consumption of the breeding programme.

Section snippets

Sample collection

Cassava variety ‘Kasetsart 50’ was planted on the breeding field of the Faculty of Agriculture, Khon Kaen University, Thailand was used. The cassava tubers used were grown under normal fertilisation and local weather conditions. The first dose of NPK fertilisers was applied at the age of 4–8 weeks after planting. A second dose of NPK fertiliser was applied again at the age of 16 weeks after planting (https://agro4africa.com/cassava-farming-process/). Cassava tubers were pulled out of the

Reference analysis

After scanning, the SCs of the cassava samples were determined using the specific gravity (SG) method. SC can be calculated as follows (Wholey and Booth, 1979):SC%=159.1×SG-147.0where SC is starch content when SG is known. SG is specific gravity and can be determined by weighing the same sample in air and water, respectively. Then, SG is calculated as follows:SG=WaWa-Wwwhere Wa is the weight of the sample in the air (mg), and Ww is the weight of the sample in water (mg). Wa and Ww were measured

Characteristics of the near-infrared spectra of cassava tubers

Fig. 2 illustrates the raw spectra of cassava tuber ranged from 570 to 1031 nm. Three obvious peaks can be found at the wavelength of 628 nm, related to the vibration of C–H stretching and at 741 nm, involving the functional group of O–H alkyl alcohols with no hydrogen bonding (R–C–OH) in CCl4 (Workman and Weyer, 2008) as well as at 970 nm, corresponding to O–H stretching with the third overtone of water content (Osborne et al., 1993). The NIR spectra of cassava tubers showed high variation in

Conclusion

This study tried to use a non-destructive method as an alternative method in in-field measurement of starch content of living cassava tubers and to examine the most effective measurement method. Therefore, the statistical data of SC in cassava tubers were investigated. The result found that the SC of the head section was slightly higher than that of the middle and tail sections, while SC values of the middle and tail sections were not different significantly. The average root mean square (RMS)

CRediT authorship contribution statement

Kanvisit Maraphum: Conceptualization, Methodology, Software, Formal analysis, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization. Khwantri Saengprachatanarug: Validation. Seree Wongpichet: Validation. Arthit Phuphaphud: Validation. Jetsada Posom: Conceptualization, Formal analysis, Data curation, Writing - review & editing, Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to thank the National Science and Technology Development Agency (NSTDA), RD&E FUNDING CONTRACT (FDA-CO-2562-10130-TH), Thailand, the Research EN KKU and Applied Engineering for Important Crops of the North East research group, Khon Kaen University, for providing research fund of this experiment. We also thank the Crop Science Laboratory, Faculty of Agriculture, Khon Kaen University for providing the cassava samples used in this experiment.

References (51)

  • P. Rungpichayapichet et al.

    Prediction mapping of physicochemical properties in mango by hyperspectral imaging

    Biosyst. Eng

    (2017)
  • P. Sirisomboon et al.

    Nondestructive estimation of maturity and textural properties on tomato ‘Momotaro’ by near infrared spectroscopy

    J. Food Eng.

    (2012)
  • J. Wang et al.

    Development of multi-cultivar models for predicting the soluble solid

    Postharvest Biol. Technol. J.

    (2017)
  • Acharya, U.K., Walsh, K.B., Subedi, P.P., 2013. Effect of temperature on SWNIRS based models of fruit DM and colour....
  • S.K. Angra et al.

    Non-destructive evaluation of apples of different origin using near infrared filter based reflectance spectroscopy

    Instrum. Sci. Technol.

    (2009)
  • Arsaipanich, P., Udompetaikul, V., 2014. Determination of Dry Rubber Content in Field Latex Using Near-Infrared...
  • Bantajan, Y., Rittiron, R., 2016. A Portable Starch Analyzer in Fresh Cassava Root using NIR technique. The 5th Asian...
  • J. Buddhakulsomsiri et al.

    Determining Appropriate Production and Inbound Logistics Practices for a Cassava Supply Chain in Thailand

    Kasetsart J.

    (2015)
  • H. Ceballos et al.

    Cassava Breeding I: The Value of Breeding Value

    Front. Plant Sci.

    (2016)
  • C.W. Chang et al.

    Near-infrared reflectance spectroscopy – Principal components regression analysis of soil properties

    Soil Sci. Soc. Am. J.

    (2001)
  • P. Chapanya et al.

    Temperature Compensation on Sugar Content Prediction of Molasses by Near-Infrared Spectroscopy (NIR)

    Sugar Tech.

    (2019)
  • C. Chuasuwan

    Cassava industry. Thailand industry outlook 2017–2020

    Krungsri Res.

    (2018)
  • A.A. Alves

    Cassava botany and physiology. Cassava: Biology

    Prod. Utilization. Chaper

    (2002)
  • P. Dardenne

    Some considerations about NIR spectroscopy: Closing speech at NIR-2009

    NIR News.

    (2010)
  • J.W. Ellis et al.

    The infrared absorption spectrum of water containing deuterium

    J. Chem. Phys.

    (1934)
  • Cited by (19)

    • Achieving robustness across different ages and cultivars for an NIRS-PLSR model of fresh cassava root starch and dry matter content

      2022, Computers and Electronics in Agriculture
      Citation Excerpt :

      The wavelength range of 600–1000 nm was used for developing a calibration model. In a previous report, Maraphum et al. (2020a,b) used NIR spectroscopy to estimate the SC in cassava tubers. They looked for the optimal wavelength region and the optimal pre-processing method.

    View all citing articles on Scopus
    View full text