Fourier transform infrared spectral features of plant biomass components during cotton organ development and their biological implications

The majority of attenuated total reflection Fourier transform infrared (ATR FT-IR) investigations of cotton are focused on the fiber tissue for biological mechanisms and understanding of fiber development and maturity, but rarely on other cotton biomass components. This work examined in detail the ATR FT-IR spectral features of various cotton tissues/organs at reproductive and maturation stages, analyzed and discussed their biological implications. The ATR FT-IR spectra of these tissues/organs were analyzed and compared with the focus on the lower wavenumber fingerprinting range. Six outstanding FT-IR bands at 1 730, 1 620, 1 525, 1 235, 1 050 and 895 cm−1 represented the major C=O stretching, protein Amide I, Amide II, the O–H/N–H deformation, the total C–O–C stretching and the β-glycosidic linkage in celluloses, respectively, and impacted differently between these organs with the two growth stages. Furthermore, the band intensity at 1 620, 1 525, 1 235, and 1 050 cm−1 were exclusively and significantly correlated to the levels of protein (Amide I bond), protein (Amide II bond), cellulose, and hemicellulose, respectively, whereas the band at 1 730 cm−1 was negatively correlated with ash content. The resulting observations indicated the capability of ATR FT-IR spectroscopy for monitoring changes, transportation, and accumulation of the major chemical components in these tissues over the cotton growth period. In other words, this spectral technology could be an effective tool for physiological, biochemical, and morphological research related to cotton biology and development.


Introduction
Cotton (Gossypium spp) is grown in over 100 countries/ regions covering an area of 33 million hectare and is fulfilling around 31% of fiber needs for textile purposes (Mollaee et al. 2020). It also produces edible oil from cottonseed for human consumption, and cottonseed meal as animal feed (Cheng et al. 2020;Kumar et al. 2021).
Other cotton organs or biomass byproducts could also be used as animal feed supplements, bioenergy sources, soil amendments, and industrial raw materials (Al Afif et al. 2020;He et al. 2016He et al. , 2020bKirkan et al. 2018;Kurtulbaş et al. 2018;Vancov et al. 2018;Yue et al. 2020). Furthermore, biological importance of different tissues of cotton plants (e.g., leaves, bolls, stalks, and stems), components of cotton leaves and their functional role/biological activities, and effect of cotton leaf traits (e.g., leaf color phenotypes and curly leaf morphology) on cottonseed nutritional qualities have been investigated (Bellaloui et al. 2019(Bellaloui et al. , 2021Egbuta et al. 2017;Shakhidoyatov et al. HE et al. Journal of Cotton Research (2022) 5:11 1997. However, their correlation with Fourier transform infrared (FT-IR) features have not been well characterized (He and Liu 2021). FT-IR spectroscopy has been widely used to characterize various agricultural biomass materials (Bekiaris et al. 2020;Goydaragh et al. 2021;He et al. 2009He et al. , 2011Hssaini et al. 2021;Lazzari et al. 2018;Waldrip et al. 2014). Especially, combined with attenuated total reflection (ATR) sampling device, FT-IR method has facilitated the research of different cotton plant related materials (Cheng et al. 2019;He et al. 2021a;Kim et al. 2017;Nam et al. 2017). However, the majority of the FT-IR investigations are focused on the fiber tissue for biological mechanisms and understanding of fiber development and maturity, but rarely on other cotton biomass components and their biological implications (Abidi et al. 2014;He and Liu 2021;Liyanage and Abidi 2019;Natalio and Maria 2018). One representative study was to monitor changes in cellulose within developing cotton fibers at 10 to 56 days postanthesis (DPA) between two cultivars (TM-1 and TX55) by Abidi et al. (2014). Their results revealed the usefulness of the integrated intensities of the bands at 3 286, 1 738, 1 639, 1 543, 1 161, 897, 710, and 667 cm −1 in evaluating the secondary cell wall cellulose formation, especially, the potential of the integrated intensities of two bands at 897 and 667 cm −1 in assessing cotton fiber cellulose content indirectly. For the same two cultivars, Abidi and Manike (2018) reported the linear response of the percentage crystallinity from wide-angel X-ray diffraction to either the two-band ratio at 1 372 and 2 900 cm −1 or the single-band integrated intensities of two bands at 667 or 897 cm −1 . In order to evaluate cotton fiber development directly and quantitatively, Liu and Kim (2017) and Liu et al. (2011) identified a number of unique IR bands, and then proposed simple algorithms for fiber maturity with the use of three IR band intensities at 1 500, 1 032, and 956 cm −1 , for fiber crystallinity with the use of three IR band intensities at 800, 730, and 708 cm −1 , and also for developmental index with the use of three IR band intensities at 1 800, 1 315, and 1 236 cm −1 . In addition, there are a few studies on FT-IR features of cotton plant materials (e.g., leaf, hull, stem, bract, and seed coat) (Allen et al. 2007;Fortier et al. 2011Fortier et al. , 2017Himmelsbach et al. 2006). These FT-IR studies of cotton plant derivatives are mainly for their identification and detection due to their presence as contaminants mingled into commercial cotton fiber, not for biological characterization of these biomass fractions during different growth phases.
In exploring and optimizing crop management practices, He et al. (2017He et al. ( , 2020a collected different cotton plant biomass fractions from field-grown cotton plants at the reproductive stage (i.e., mid-season) and the maturation stage (i.e., late season just before defoliation for harvesting readiness), and reported their elemental composition and biomolecule (e.g., carbohydrate and amino acid) profiles. In the meantime, the ATR FT-IR spectra of these biomass samples were also briefly examined, demonstrating the potential of ATR FT-IR as a diagnostic tool for monitoring plant biomass biosynthesis for cotton physiology and biology research (Liu et al. 2016). It is also reported that the FT-IR data of different stem parts of Arabidopsis thaliana have been applied for metabolic fingerprinting of the wild type and mutant lines (Brown et al. 2005). Thus, in this work, we examined in detail the ATR FT-IR spectral features of these biomass samples, analyzed and discussed their biological implications. As ATR FT-IR spectroscopy could identify exclusively and consistently the major chemical components (e.g., protein, hemicellulose) in different cotton tissue materials, it was hypothesized that such a capability would make this spectral technology an effective monitoring tool for physiological, biochemical, and morphological research related to cotton biology and development (Brown et al. 2005). Our overall goal is to provide more knowledge about cotton biology and development for researchers and growers to optimize field management practices and processing strategies for higher yield and quality of cotton fiber and biomass materials as valuable natural resources.

Cotton materials and sample preparation
Two sets of cotton plant biomass materials were collected and used in the work. The details of collection and treatment of the first set of samples were reported previously (He et al. , 2020a. Briefly, they were prepared from a cotton variety Deltapine (DP) 1321 B2RF grown at the Mississippi Agricultural and Forest Experiment Station near Pontotoc, MS (34° 8′ 30″ N, 88° 59′ 36″ W) (Tewolde et al. 2015). The soil type was Atwood silt loam. Four to eight whole plants from each plot of four replications were collected in 80 and 112 days after planting (DAP) that were classified into vegetative, reproductive, and maturation stages, respectively (Fig. 1). The cotton plants in the reproductive stage were separated into six different organs including main stem, branch stem, leaf blade, petioles, root, and reproductive tissues (Figs. 1B,2A). In addition to the six organs, fiber and seeds were also classified from the cotton plants in the maturation stage (Figs. 1C, 2B). The seed samples were delinted with concentrated H 2 SO 4 , rinsed with tap water, and dried in a forced-air oven at 80 °C (He et al. 2020c). All samples except for cotton fibers were ground to < 1 mm and kept at a relative humidity of (65 ± 2)% and temperature of (21 ± 2) °C prior to characterization. Cotton fibers Fig. 1 Seasonal development of cotton. Cotton developmental stages were classified according to the three different seasons described in Oosterhuis and Jernstedt (1999). Green garden stakes (90 cm) were used as a scale marker Fig. 2 Cotton plant biomass fractions used in this study. A and B collection of plant organs of variety DP 1321 B2RF, at reproductive stage (mid-season) and maturation stage (late season at pre-defoliation for harvesting), respectively. Bur, bract, and peduncle are not separated separated from the seeds were subjected to the FT-IR analysis without further treatment.

ATR FT-IR spectral collection and analysis
All samples were conditioned at a constant relative humidity of (65 ± 2)% and temperature of (21 ± 2) °C for at least 24 h, prior to ATR FT-IR spectral acquisition. An FTS 3000MX FT-IR spectrometer (Varian Instruments, Randolph, MA, USA), aligned with ATR attachment, was employed to collect ATR FT-IR spectra. It was equipped with a ceramic source, KBr beam splitter, and deuterated triglycine sulfate (DTGS) detector (Liu et al. 2015(Liu et al. , 2016Liu and Kim 2017). The ATR sampling device utilized a DuraSamplIR single-pass diamond-coated internal reflection accessory (Smiths Detection, Danbury, CT, USA), and a consistent contact pressure was applied by way of a stainless steel rod and an electronic load display. Least three measurements for individual sample were collected over the range of 4 000-600 cm −1 at 4 cm −1 and 16 co-added scans. The spectra were present in absorbance and no ATR correction was applied. Spectral normalization was performed based on the mean of peak intensities from the 1 800-600 cm −1 . The mean intensities (i.e., band heights) of interesting bands centered at 1 730, 1 620, 1 525, 1 235, 1 050, and 895 cm −1 were determined by a multi-point average of the intensities at respective range of 1 760-1 700, 1 700-1 570, 1 570-1 480, 1 290-1 200, 1 200-910, and 910-880 cm −1 .

Chemical composition measurements
Moisture content was determined as the loss in weight upon drying a sample in a forced draft oven at 105 °C for 5 h (McCall and Jurgens 1951). All the measured data in this section were converted to a dry basis via the correction of the moisture content. Ash content was determined by measuring the residual mass of a sample (1.0 g) after heating in a muffle furnace at 550 °C for 4 h. Selected mineral contents were analyzed following acidic digestion. Specifically, 0.50 g of ground sample was mixed in 10.0 mL of concentrated trace metal grade HNO 3 for 1 h in a HotBlock ™ Environmental Express Block Digester. The sample was then heated to 115 °C for 2 h and 15 min. The concentrations of six macro elements (i.e. P, Ca, K, Mg, Na, and S) and seven trace elements (i.e. Fe, Zn, Cu, Mn, B, Ni, and Al) in these digests were determined by a Spectro CirOs ICP Spectrometer (Mahwah, NJ, USA).
Content of cellulose was calculated by the difference between acid detergent fiber (ADF) and acid detergent lignin (ADL). Hemicellulose content was determined by the difference between neutral detergent fiber (NDF) and ADF. The values of ADF, NDF, and ADL were determined using the filter bag methods with a Fiber Analyzer (Ankom Technology, Macedon, NY, USA). The total N and C contents of the ground samples were directly determined by a LECO Truspec Dry Combustion C/N Analyzer (St. Joseph, MI, USA) without pretreatments. Crude protein content was calculated by multiplying the N content value by the factor of 6.25 ).

Data treatment and statistical analysis
The averaged FT-IR spectra of the same type of biomass fractions were reported previously (Liu et al. 2016). However, those data were analyzed as groups without elaboration of their detailed features. In this work, ATR FT-IR spectral features of each sample were re-analyzed with quantitative data of band height. Furthermore, the correlation analyses were performed. The data of contents of selected elements, carbohydrate, and amino acids of various cotton organs collected from the Mississippi field were reported in He et al. (2017He et al. ( , 2020a. In this work, these data were used to calculate the correlation coefficients between the chemical properties and the FT-IR band features. These coefficient values were then discussed to explore the biological implications of these FT-IR features. The quantitative spectral data (band height) were loaded into Microsoft Excel 2016 to execute simple algorithmic analysis. Separately, P values were calculated using one-way analysis of variance (ANOVA) under Data Analysis in Microsoft Excel 2016. The Descriptive Statistics Tool Data were used to calculate means and standard errors of tetraplicate field samples. The Correlation Analysis Tool was used to analyze correlation coefficients between the data sets of different parameters.

Results and discussion
Classification of cotton tissues and organs at two different developmental stages Fig. 1 shows three different developmental seasons of a cotton variety DP 1321 B2RF used in this study. In the early season (Fig. 1A), vegetative development occurred for establishing roots and expanding leaf area (Oosterhuis and Jernstedt 1999; Robertson and Roberts 2010). In the mid-season (Fig. 1B), reproductive growth began with the emergences of floral buds or squares around 40 DAP. Flowers and developing bolls appeared during the reproductive stage (Fig. 1B). In the late season ( Fig. 1C), mature cotton bolls started to open approximately at 100 DAP, and fibers were exposed to air.
The first set of cotton plants was collected in 80 DAP that was in the midst of the reproductive stage ( Fig. 1B). They were classified into 6 different organs as shown in Fig. 2A. Unlike the five different vegetative organs (main stems, branch stems, leaf blades, petioles, and roots), the reproductive tissues were composed of multiple organs including squares, flowers (yellowish petals, 0 DPA), and young bolls (reddish petals, < 3 DPA) with bracts and peduncles as shown in Fig. 2A. The cotton ovules inside the young cotton bolls contained fiber initials composed exclusively of the primary cell walls since the secondary cell walls were produced after 23 DPA (Abidi et al. 2014;Kim 2015).
The second set of cotton plants was harvested in 112 DAP that was at the maturation stage of the late season before defoliation for harvesting readiness (Fig. 1C). In addition to the five vegetative organs (main stems, branch stems, leaf blades, petioles, and roots), the mature cotton bolls were separated into three types including seeds, fibers, and others (burs, bracts, and peduncles) as shown in Fig. 2B. Fully developed fibers are composed of almost pure cellulose (Kim 2015).
On the dry basis, the average biomass of main stem at the maturation stage (17.1 ± 5.0 g) shows no significant difference (P = 0.869) with that at the reproductive stage (17.5 ± 2.1 g) as shown in Fig. 3. Similarly, the biomass of branch stem also shows no significant variation (P = 0.643) between the reproductive (7.1 ± 1.1 g) and maturation (7.8 ± 2.5 g) stages. The similar biomass patterns of both stems between the two stages are expected based on the almost identical cotton heights between the reproductive (89.1 ± 5.9 cm) and maturation (90.3 ± 14.6 cm) stages. In contrast, the average biomass of leaf blades (11.2 ± 1.1 g) and petioles (1.5 ± 0.2 g) at the maturation stage are significantly (P < 0.001) reduced compared with the leaf blades (16.3 ± 0.6 g) and petioles (2.9 ± 0.2 g) at the reproductive stage (Fig. 3). These patterns are consistent with the leaf wilting and abscission process at the maturation stage as shown in Fig. 1C. The average biomass of the reproductive organs strikingly and significantly (P < 0.001) increases at the transition from the reproductive stage (8.0 ± 2.7 g) to the maturation stage (61.2 ± 10.3 g) because of the seed and fiber development. We excluded cotton root biomass comparison in Fig. 3 due to practical difficulties of digging up entire cotton roots of individual plants. Under regular field conditions, cotton roots can grow three times more than the aboveground portion of the cotton plant (Oosterhuis and Jernstedt 1999). Fig. 4 shows the representative ATR FT-IR spectra in the 3 800-600 cm −1 region of six different cotton biomass samples collected from the reproductive stage (or mid-season). Although these cotton plant fractions are inherently different, their FT-IR spectra in Fig. 4 are quite similar. This is due to the fact that these natural materials areprimarily composed of varying types of carbohydrates in the forms of cellulose, hemicellulose, pectins, lignins, and sugars, and also of small amounts of proteins, waxes, and other inorganics. FT-IR spectral features of these cotton crop related components, if not all, have been investigated in detail for a variety of interests, and characteristic bands have been assigned to different functional groups (Himmelsbach et al. 2003(Himmelsbach et al. , 2006Liu et al. 2012). Typically, the broad and strong bands from 3 700 to 3 000 cm −1 region are due to the O-H or N-H stretching modes of carbohydrates, adsorbed water, and proteins, while relatively weak bands at 2 924 and 2 850 cm −1 are owing to hydrophobic CH 2 asymmetrical and symmetrical stretching vibrations, respectively (He et al. 2021b;Zhang et al. 2021). One band close to 1 730 cm −1 contributed by the C=O stretching modes of carbonyl groups (Himmelsbach et al. 2003) should be due to the minor organic acids (e.g., malic and citric acid) (He et al. 2009;Wakelyn et al. 2007). The broad band centered at 1 620 cm −1 and the minor band at 1 520 cm −1 are mainly from the O-H bending, Amide I and II (N-C=O) of protein and pectic acid esters as well as H-bonded C=O of conjugated ketones (Abidi et al. 2014;He et al. 2009). Numerous absorptions in the region of 1 500-1 200 cm −1 consist of a combination of CH 2 deformations and C-O-H bending vibrations, and those in the 1 200-800 cm −1 region are ascribed to the C-O and C-C stretching vibrations. Especially, the outstanding band around 1 020 cm −1 is mainly due to C-O stretching of carbohydrates in these plant materials although some inorganic components (e.g., P-O, Si-O bonds) may also Fig. 3 Biomass allocation of field-grown cotton. Average biomass of main stem, branch stem, leaf blade, petiole, and reproductive tissues at reproductive stage (80 DAP) was compared with the corresponding organs at maturation stage (112 DAP). Data are present on the dry basis with average with standard deviation bars. Symbol *** above the bar indicates statistical significance (P < 0.001) between the two stages (n = 4). No significant difference at P = 0.05 was found for the data pairs of main stem (P = 0.869) and branch stem (P = 0.643) have some contribution (Bekiaris et al. 2015;He et al. 2006;Himmelsbach et al. 2006). A weak absorption near 895 cm −1 is attributable to β-glycosidic linkage in linear cellulose portion of cotton fibers, as its integrated intensity or area was proportional positively to the percentage of cotton cellulose content (Abidi et al. 2014;Bekiaris et al. 2015;Liu and Kim 2017).

ATR FT-IR features of six biomass samples collected at the reproductive stage
Chemically, stem-like substances, represented by root, main stems, and branch stem, are mainly composed of carbohydrates at varying amounts and derivatives, such as cellulose, hemicellulose, pectins and lignins, and also other organic/inorganic components (Renuka et al. 2005). Whereas leaf-like samples, such as petioles, leaf blade, and reproductive tissues, are comprised of hydrocarbons, organic/amino acids, phenolic compounds, carotenoids, sugars, pectins, and others (Shakhidoyatov et al. 1997). Visually, these cotton plant biomasses exhibit similar FT-IR features, but with differences in relative band intensities and positions. For example, the intensity of the band around 1 620 cm −1 increased from the root material, along with the biomass order of the plant topology, to the reproductive material. Thus, it was estimated that this band and other bands in the lower wavenumber range (i.e., the fingerprinting region) may be applicable, while the two band groups (i.e., around 3 320 cm −1 covering the 3 700-3 000 cm −1 and 2 900 cm −1 covering the 3 000-2 800 cm −1 ) in the high wavenumber wing are too broad or weak for meaningfully quantitative applications. To this regard, the fingerprinting region was divided into six spectral intervals subjectively, with the 1 760-1 700 cm −1 region noted as I 1 730 representing C=O stretching modes typically from esters, ketones, and acids, the 1 700-1 570 cm −1 region noted as I 1 620 presenting the overlapped contributions from adsorbed water at 1 640 cm −1 and proteins (Amide I) at 1 620 cm −1 , the 1 570-1 480 cm −1 region noted as I 1 525 originating from proteins (Amide II), the 1 290-1 200 cm −1 region noted as I 1 235 focusing on O-H or N-H deformation, the 1 200-910 cm −1 region noted as I 1 050 showing a total C-O-C stretching contributed from all C-O-C containing substances regardless of cellulosic or noncellulosic molecules, and the 910-880 cm −1 region noted as I 895 considering the sole origin of β-glycosidic linkage in cellulose (Table 1). It is worth noting that the subjective division of the fingerprinting region into six spectral intervals is more complicated than those relevant studies which focused more on fewer band regions for fiber quality, maturity and crystallinity indices (He and Liu 2021).
As given in Table 1 from the reproductive stage, the intensities of three bands at 1 620 cm −1 (I 1 620 ), 1 525 cm −1 (I 1 525 ), and 1 235 cm −1 (I 1 235 ) increase from root material to main stem, branch stem, petiole, and leaf blade, but the reproductive specimen has reduced intensities compared with the leaf blade sample. The intensities of two bands at 1 050 cm −1 (I 1 050 ) and 895 cm −1 (I 895 ) decrease in general from root material, to main stem, branch stem, petiole, and leaf blade specimen, while the reproductive material increases in the two intensities compared with the leaf blade sample. The intensity of the 1 730 cm −1 (I 1 730 ) band increases from root samples, to main stem, branch stem, and petiole, and then decreases from petiole to leaf blade and increases from leaf blade to reproductive sample. This is expected, as the three bands at 1 620, 1 525, and 1 235 cm −1 are related to protein and amine compositions and other bands at 1 730, 1 050, and 895 cm −1 are due to carbohydrate derivatives. Relative standard deviation (RSD), calculated from the ratio of standard deviation (SD) to mean value, shows the variation of six biomasses for a specific IR intensity index. In general, the I 1 525 value possesses the greatest RSD of 0.35, followed by the I 1 620 index (RSD = 0.26), the I 1 050 index (RSD = 0.12), the I 1 730 and I 895 values (RSD = 0.08), as well as the I 1 235 values (RSD = 0.06). Hence, changes in intensities of six FT-IR indices reveal differing biosynthesis mechanisms of Amide/amine and C-C/C-O Table 1 The normalized intensities of six ATR FT-IR bands of the plant biomass from the reproductive stage. Data are presented as average ± standard deviation (n = 4) a Relative standard deviation (RSD), defined as the ratio of standard deviation to mean value for each column I  functional groups between these cotton plant parts, respectively.

ATR FT-IR features of nine biomass samples collected at the maturation stage
Unique and characteristic FT-IR bands for individual cotton plant parts in Fig. 5 represent striking and reasonable differences of biomass origins during the fiber maturation stage. Cotton fibers and seeds are two extreme portions, and their spectral features differ from those of other cotton plant parts significantly. Commonly, cellulose is a major chemical component in mature cotton fibers (Hsieh 2007), while oils, proteins, fats, highly digestible fibers, and inorganic minerals are abundant in cotton seeds (He and Liu 2021;Rohman and Man 2010;Talpur et al. 2014). Obvious FT-IR spectral difference between cotton fiber and other cotton plant portions is the absence of two unique bands at 1 725 and 1 235 cm −1 in cotton fiber, which come from the C=O groups in carboxylic acids of cellulosic derivatives and N-H deformation modes in proteins, respectively. Characteristic FT-IR bands of cotton seeds are two strong and separated absorption bands at 1 745 and 1 535 cm −1 from the respective ester C=O group and Amide II mode in two major components of cotton seeds (i.e., oil and protein). Compared with the spectra of both cotton fiber and seed, FT-IR spectra of some cotton plant biomass have been investigated before (Himmelsbach et al. 2006), primarily because these cotton plant parts are the undesired contaminants (or foreign matters) mingled into commercial cotton bales. There were expected chemical, compositional, and structural differences of plant biomass between reproductive stage and maturation stage, as revealed by principal component analysis (PCA) of their FT-IR spectra (Liu et al. 2016). The results showed a tendency of the first principal component (PC1) scores increasing with plant samples in the order of main stems < root = branches < petioles < burs < reproductive = bracts < leaf blades, and also of the second principal component (PC2) score separating the same type of plant parts collected at reproductive stage and maturation stage. Similar to the pattern in Table 1, Table 2 shows the intensity for the two bands at 1 620 cm −1 (I 1 620 ) and 1 525 cm −1 (I 1 525 ) increasing initially from root sample to main stem, branch stem, petiole and leaf blade, then the intensity of these two bands decreasing from leaf blade to burr and bract samples at the maturation stage. It also indicates the intensity for the two bands at 1 050 cm −1 (I 1050 ) and 895 cm −1 (I 895 ) decreasing from root to main stem, branch stem, petiole, and leaf blade, and the intensity increasing from leaf blade to bur and bract substance. Because of the fact that fibers consist of cellulose component dominantly, while cotton seeds include oils and proteins as major compositions, it is anticipated a distinctive pattern between the two in Table 2. Compared with the same indices among the nine materials, the fiber sample is with the largest I 1 050 and I 895 indices and also the smallest I 1 730 , I 1 620 , I 1 525 , and I 1 235 indices. On the other hand, the seed material has the least I 895 indices, but the greatest I 1 730 , I 1 620 , and I 1 235 indices among the nine materials. In addition, RSD of nine biomasses from the maturation stage (Table 2) decreases in the order of the I 1 525 value (RSD = 0.36), the I 1 620 value (RSD = 0.27), the I 1 730 value (RSD = 0.16), the I 895 value (RSD = 0.14), the I 1 050 index (RSD = 0.13), and the I 1 235 value (RSD = 0.12). The tendency of RSD in Table 2 differs slightly from that in Table 1, with the I 1 730 and I 895 values overpassing the I 1 050 index. Therefore, variations of six FT-IR indices address the different biosynthesis mechanisms of Amide and C-C/C-O functional groups between these cotton plant biomasses at the maturation stage. Table 2 The normalized intensities of six ATR FT-IR bands of the plant biomass from the fiber maturity stage. Data are present as average ±standard deviation (n = 4) a Relative standard deviation (RSD), defined as the ratio of standard deviation to mean value for each column I It is noticeable that there are some spectral intensity differences between the reproductive stage (Table 1) and the maturation stage (Table 2). For example, intensities of the bands at 1 730 and 1 235 cm −1 increase among root, main stem, and branch stem , whereas that at 1 050 cm −1 decreases among root and petiole from the reproductive stage to the maturation stage. While for leaf blade from the reproductive stage to the maturation stage, apparent spectral intensity reduction occurs in the 1 650-1 480 cm −1 region due to noncellulosic components, and intensity increase happens at 870 cm −1 due to β-D-fructose (Türker-Kaya and Huck 2017). In order to differentiate the same type of cotton plant biomasses picked at the reproductive stage and the maturation stage, Table 3 compares the changes in averaged values of four field replicates for each sample collected at two periods. Root samples show significant difference in biomass composition by four of six IR indices, followed by main stem and petiole sample with three IR indices, and also branch stem sample and leaf blade sample with two IR indices. This observation is much in line with expectation, as individual cotton plant organ undergoes unique biosynthesis. On the other hand, the I 1 235 index is found to be effective in separating all samples (from root to main stem, branch stem, petiole, and leaf blade) at the maturation stage from those at the reproductive stage, then the I 1 730 and I 895 value, and next the I 1 620 , I 1 525 , and I 1 050 index. Table 4 lists only the selected chemical components of cotton fiber sample as such composition data of other cotton biomass samples (Additional file 1: Table S1) have been reported previously (He et al. , 2020a. The fiber sample, as it was, contained 5.44% of water (moisture), converted to 57.5 g·kg −1 on a dry basis, comparable to about 8% reported before (McCall and Jurgens 1951). As a major element in organic materials, carbon   accounted for > 1/3 of the biomass of the cotton fiber sample. It appeared mainly in the form of cellulose, with minor forms of hemicellulose, lignin, protein, and other unmeasured components (e.g., wax, small organic acid) (Wakelyn et al. 2007). The main form was acid detergent fiber (ADF, 934.5 g·kg −1 ). The difference between ADF and acid detergent lignin (ADL, 19.3 g·kg −1 ) was the content of cellulose (915.2 g·kg −1 ) of the tested fiber sample. This cellulose value was quite consistent with the historic data of the cellulose content ranging from 880 to 960 g·kg −1 (McCall and Jurgens 1951). Combined our result with other cellulose measurements in last couple of decades (Corradini et al. 2009;de Morais Teixeira et al. 2010), methodology evolution from that one in 1950's (McCall andJurgens 1951) has not greatly impacted the measurement of fiber cellulose much. In contrast to the cellulose content, contents of protein (22.1 g·kg −1 ) and ash (19.95 g·kg −1 ) in our fiber sample were higher than the historic ranges of 11∼19 g·kg −1 and 7∼16 g·kg −1 , respectively (McCall and Jurgens 1951;Wakelyn et al. 2007). Our data indicated that the hemicellulose contents in the sample were at the same level of protein and ash around 20 g·kg −1 . These values were higher than the content of hemicellulose (5 g·kg −1 ) in white cotton fiber reported by de Morais Teixeira et al. (2010), but lower than the corresponding value (80.0 g·kg −1 ) reported by Corradini et al. (2009). The ash content represented the inorganic salts (phosphates, carbonates, and oxides) and salts of organic acids present in the raw fiber (Wakelyn et al. 2007). Thus, the specific contents of six macro and seven trace elements (minerals) in the fiber samples were measured (Table 4). Those minerals in intact raw cotton enhance fiber's resistance to heat and flame functionally (Nam et al. 2014). As reported by Wakelyn et al. (2007), K was the most abundant macro element (about tenfold higher than other macro elements) in the fiber. While these elements are biologically necessary for the development of the cotton plant and fiber (Tewolde et al. 2018;Wakelyn et al. 2007), uptake of K under Ca-deficient conditions is favorable to the fiber elongation and maturation (Gamble 2009;Guo et al. 2017). It seems that a high level of Ca results in the formation of rigid fiber morphology to inhibit the fiber elongation (He et al. 2021b). Future investigation of the Ca uptake and distribution from cotton roots to the leaf blade and reproductive organs over the fiber development phase would be helpful to investigate if higher Ca expression activity or Ca signaling genes are involved in the inhibitory mechanisms of fiber elongation.

Correlation analysis of the ATR FT-IR parameters and chemical composition measurements
Correlation analysis is useful to elucidate the relationships between chemophysical properties and FT-IR features in agricultural studies (He et al. , 2021aWaldrip et al. 2014). Thus, the correlation coefficients of the six major chemical composition measurements (i.e., the content of cellulose, hemicellulose, protein, ADL, ADF, and ash) and the six fingerprinting FT-IR bands with all biomass samples collected at both the reproductive stage and the maturation stage were calculated (Table 5). There were 13 of 36 pairs of data between the two measurements showing the correlation coefficients significant at P ≤ 0.05. The FT-IR intensities of I 1 620 and I 1 525 are only positively correlated to the protein content, indicating that the two FT-IR bands could be used to monitor the protein synthesis and abundance during the cotton growth. Previously, Waldrip et al. (2014) reported that the FT-IR intensity at 1 650 cm −1 was positively correlated (P ≤ 0.05) to N forms in cattle manure samples. Similarly, the increase of I 1 235 is correlated exclusively to the increase of cellulose content. The intensity of I 1 050 was an indicator of hemicellulose content apparently. On the other hand, the change of I 895 is positively correlated to cellulose, hemicellulose, and ADF contents, indicating I 895 as a general carbohydrate parameter, but not serving well as an exclusive measurement to any of the three. As a comparison, two separate studies reported strong and linear relationships between the integrated intensity Table 5 Correlation coefficients between the six ATR FT-IR indices and six chemical properties of all plant biomass samples collected at both the reproductive stage and the maturation stage (n = 15) Symbol *, ** and *** indicate the coefficient value significant at P = 0.05, 0.01, and 0.001, respectively I of the 895 cm −1 band and cellulose content in developmental cotton fibers (Abidi et al. 2014;Liu and Kim 2017). Those statistically significantly negative correlation coefficients are the indicators of the competitively reversing change trends of the functional groups assigned to the FT-IR bands with the chemical component measured. For example, the negative correlation between I 1 730 and ash content suggests that the change of minerals (ash) content was in a reverse order with the organic acid level (I 1 730 ) in these cotton tissues. This observation may imply that most minerals (if not all) in these cotton plant materials do not form biominerals (such as Ca oxalate) as reported in some species of Cacteae grown under desert environment (la Rosa-Tilapa et al. 2020). No statistically significant correlation between ADL and any FT-IR bands implied that rapid FT-IR evaluation was not very useful for ADL estimates from the current ADL testing procedure. Although two FT-IR bands are significantly correlated to ADF, application of the whole spectral range with a regression method of modified minimum partial least squares could be an alternative in ADF quantification, as in an analysis of ADF of turnip greens and tops (Obregón-Cano et al. 2019).

Conclusions
ATR FT-IR spectroscopy in combination with statistical analysis was applied to evaluate a series of cotton plant tissues (biomass parts), including the root, stems, leaves, and reproductive parts, at reproductive and maturation stages. The FT-IR spectra displayed distinct spectral patterns, especially in the lower wavenumber range with recognizable functional group characteristics of the biomass parts studied. Correlation coefficient analysis of the six FT-IR bands in the fingerprinting region with six major chemical components revealed that the FT-IR band intensities at 1 620, 1 525, 1 235, and 1 050 cm −1 were exclusively and positively correlated to the levels of protein (amide I), protein (amide II), cellulose, and hemicellulose of the 15 cotton biomass samples tested, respectively, at P ≤ 0.05. The negative correlations (P ≤ 0.05) between the organic acid-related I 1 730 and ash content implied that the presence of minerals and small organic acid molecules was in a complementary mode in these cotton tissues. These resulting observations of ATR FT-IR analysis were sufficiently unique to be used as fingerprinting to monitor the synthesis, transport, accumulation, and other biological implications of these major chemical components in various cotton tissues over the cotton growth period.
Additional file 1. Table S1. Chemical composition of non-fiber biomass material.