Research Article

Classification of Legume Quality Based on Chemical Composition and Digestibility Values using Multivariate Analysis

Dedy Nanda Kurniawan1, Nahrowi Nahrowi2, Yulianri Rizki Yanza3,4, Vincent Niderkorn5, Anuraga Jayanegara2,4*

1Graduate Study Program of Nutrition and Feed Science, IPB University, Bogor, Indonesia; 2Department of Nutrition and Feed Technology, Faculty of Animal Science, IPB University, Bogor, Indonesia; 3Department of Animal Nutrition and Feed Technology, Faculty of Animal Science, Universitas Padjadjaran, Jatinangor, Indonesia; 4Animal Feed and Nutrition Modelling (AFENUE) Research Group, Faculty Animal Science, IPB University, Bogor, Indonesia; 5INRAE, VetAgro Sup, UMR Herbivores, Université Clermont Auvergne, Saint-Genès-Champanelle, France.

Received | September 01, 2024; Accepted | November 28, 2024; Published | February 11, 2025

*Correspondence | Anuraga Jayanegara, Department of Nutrition and Feed Technology, Faculty of Animal Science, IPB University, Bogor, Indonesia; Email: [email protected]

Citation | Kurniawan DN, Nahrowi N, Yanza YR, Niderkorn V, Jayanegara A (2025). Classification of legume quality based on chemical composition and digestibility values using multivariate analysis. Adv. Anim. Vet. Sci. 13(3): 573-583.

DOI | https://dx.doi.org/10.17582/journal.aavs/2025/13.3.573.583

ISSN (Online) | 2307-8316; ISSN (Print) | 2309-3331

Copyright: 2025 by the authors. Licensee ResearchersLinks Ltd, England, UK.

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

INTRODUCTION

Leguminous forages are widely recognized for their exceptional quality, characterized by high protein content and low fiber content. The protein content of legumes typically ranges from 18% to 36%, whereas the fiber content ranges from 3.3% to 18.3%, with varying levels of lipids, such as linoleic and linolenic acids, as well as bioactive compounds, including isoflavones, phenolic acids, saponins, and polyphenolic compounds (Bahadoran and Mirmiran, 2015). Recently, the use of legumes as animal feed has presented several advantages, particularly for ruminant livestock. These benefits include improved ruminant performance (Castro-Montoya and Dickhoefer, 2018; Simbaya, 2002), enhanced nutrient digestibility (Ratnawaty et al., 2018), increased fat content in milk (Brito and Broderick, 2006), and reduced methane emissions (Zeleke et al., 2006). Moreover, the high quality effects of feeding legumes to ruminants can be attributed to their protein, which can demolish animal growth and development, which is then converted into energy. Energy derived from proteins, lipids, and carbohydrates, which are constituents of the chemical composition of feed ingredients, is crucial for the survival of livestock (Guadayo et al., 2019; Jayanegara et al., 2019). Hence, digestibility is a measure of the quality of feed ingredients is digestibility (Darma et al., 2023).

Legumes can be classified based on various criteria such as the type of stem, climate in which they are grown, and edible parts. According to stem type, legumes are divided into trees, shrubs, and herb legumes (Castro-Montoya and Dickhoefer, 2020). Herb legumes have mild stem structures and grow by trailing on the ground, climbing, and attaching to other trees or fences. Shrub legumes, on the other hand, have woody stems and grow to form shrubs, whereas tree legumes possess strong and upright stem structures and grow upward. Based on their growing region, legumes can be classified into tropical, subtropical, and temperate categories. Tropical and subtropical legumes have optimal growth temperatures slightly above 30°C, whereas temperate legumes have optimal growth temperatures below 30°C (Pitman dan Vendramini, 2020). The differentiation process of the morphological development and physiological responses of plants is also affected by climatic conditions (Myers et al., 2014). Therefore, climate can be one of the factors that influence the chemical content of legumes. All leguminous plants morphological structures can not be used as animal feed, whereas ruminants can only digest the edible parts of plants, such as branches, leaves, pods, flowers, and seeds (Myers et al., 2014; Soedarmanto et al., 2022). Moreover, the chemical contents of the edible parts of legumes also vary. For example, the protein content in leaves is higher than that in stems (Shih et al., 2011).

Numerous studies have been conducted on leguminous plants, with particular focus on their chemical composition and use in ruminants (Castro-Montoya and Albarrán-Portillo, 2023; Kelln et al., 2023; Marsetyo et al., 2020). In the fact, high forage digestibility can also have significant environmental impacts. According to Restirinani et al. (2016), feed with high digestibility can reduce methane production. This effect occurs because highly digestible feed accelerates the fermentation process and brief existence of feed in the rumen to reduce production methane from methanogen (Chaokaur et al., 2015). To address this gap, it is crucial to identify the effects of growth type, growth climate, and edible parts of legumes on their chemical composition and digestibility. To discern the relationship between legume types and their chemical composition, a substantial dataset is required to identify patterns. Consequently, this study utilized a database from Feedipedia, employing a multivariate analysis approach. Multivariate analysis enables the simultaneous examination of multiple variables to identify patterns and characteristics within the data (Everitt, 1975). Furthermore, this analysis is widely employed to explore data patterns and classifications and has also found application in animal husbandry (Jayanegara et al., 2011; Miedico et al., 2022). Therefore, the objective of the present study was to classify leguminous plants according to the type of stem, climate region, edible parts, chemical composition, and digestibility rates using a multivariate analysis.

 

Table 1: Categories and number of legumes included in the present study.

Legume Categories	Legume Class	Number of Samples
Legume type	Shrub	89
	Tree	148
	Herb	78
Climatic region of plant origins	Tropical	143
	Subtropical	62
	Temperate	32
Edible part	Aerial part	131
	Leaves	31
	Pods	41
	Seeds	35

 

MATERIALS AND METHODS

Legume Dataset

This study obtained information on legume plants from Feedipedia (www.feedipedia.org), a joint project of INRAE, CIRAD, AFZ, and the FAO. Feedipedia is an open-access information system that provides data on the nature, presence, chemical composition, and nutritional value of nearly 1400 animal feeds worldwide. Moreover, Feedipedia provided a comprehensive overview of research findings on feed ingredients, particularly on leguminous

 

Table 2: Descriptive statistics of the chemical composition and digestibility values of legumes.

Vari-able	Legume type									Climatic region
	Shrub			Tree			Herbs			Tropical			Subtropical
	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV
CP	17.66	6.38	36.13	16.42	6.48	39.47	18.51	4.92	26.58	17.63	5.22	29.63	16.42	7.12	43.37
CF	26.37	10.37	39.34	22.22	8.15	36.68	26.40	9.58	36.30	25.51	9.18	36.00	24.55	9.82	39.98
NDF	42.63	12.59	29.52	40.31	9.85	24.44	43.82	12.09	27.59	42.92	11.15	25.98	41.89	11.26	26.89
ADF	31.07	10.28	33.10	29.67	8.39	28.28	31.29	10.83	34.62	31.02	9.81	31.61	30.88	8.80	28.50
Lignin	7.66	3.10	40.45	9.32	4.84	51.91	7.19	3.12	43.39	8.30	4.08	49.13	7.94	3.55	44.74
EE	2.78	1.35	48.44	3.27	1.70	52.19	2.76	1.21	43.84	2.95	1.42	48.17	2.97	1.59	53.65
Ash	8.40	2.87	34.15	6.72	2.87	42.79	8.94	3.47	38.80	8.02	3.22	40.17	7.55	3.12	41.30
GE	18.57	0.66	3.57	18.67	0.71	3.80	18.48	0.64	3.49	18.58	0.70	3.79	18.61	0.64	3.43
OMD	68.77	13.36	19.43	77.47	10.77	13.90	68.24	11.21	16.42	70.95	12.73	17.94	73.35	11.50	15.68
ED	64.89	12.57	19.37	73.44	10.38	14.13	65.18	11.44	17.55	67.25	12.07	17.95	69.37	11.12	16.03
DE	11.91	2.55	2.01	13.52	2.24	16.57	12.03	2.19	18.17	12.35	2.37	19.16	12.82	2.30	17.96
ME	9.54	2.01	21.10	11.04	1.71	15.48	9.59	1.78	18.59	9.97	1.92	19.27	10.33	1.84	17.78

 

CP: Crude Protein; CF Crude Fiber; NDF: Neutral Detergent Fiber; ADF: Acid Detergent Fiber; EE: Ether Extract; GE: Gross Energy; OMD: Organic Matter Digestibility; Energy Matter Digestibility; DE: Digestible Energy; ME: Metabolize Energy; SD: Standard Deviation; CV: Coefficient of Variance; SD: Standard Deviation; CV: Coefficient of Variance.

 

Table 3: Descriptive statistics of the chemical composition and digestibility values of legumes (Cont.).

Vari-able	Climatic Region			Edible Part
	Temperate			Aerial Part			Leaves			Pods			Seeds
	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV	Mean	SD	CV
CP	19.50	6.25	31.05	18.20	5.03	27.66	17.19	6.46	37.59	17.63	5.22	29.63	16.42	7.12	43.37
CF	22.40	10.93	48.80	26.66	9.72	36.44	23.74	9.46	39.84	25.51	9.18	36.00	24.55	9.82	39.98
NDF	38.39	14.46	37.67	43.41	11.83	27.25	41.21	11.67	28.33	42.92	11.15	25.98	41.89	11.26	26.89
ADF	27.38	12.31	44.98	31.28	10.46	33.45	30.02	9.71	32.35	31.02	9.81	31.61	30.88	8.80	28.50
Lignin	6.70	3.47	51.74	7.36	3.36	45.71	8.37	4.15	49.60	8.30	4.08	49.13	7.94	3.55	44.74
EE	2.96	1.38	46.55	2.88	1.24	43.01	3.01	1.58	52.58	2.95	1.42	48.17	2.97	1.59	53.65
Ash	8.61	3.29	38.20	8.93	2.78	31.16	7.41	3.33	44.98	8.02	3.22	40.17	7.55	3.12	41.30
GE	18.56	0.62	3.33	18.48	0.54	2.91	18.65	0.74	3.98	18.58	0.70	3.79	18.61	0.64	3.43
OMD	71.49	13.33	18.65	67.43	12.08	17.91	74.20	12.13	16.35	70.95	12.73	17.94	73.35	11.50	15.68
ED	68.74	14.10	20.52	64.65	12.45	19.26	70.02	11.53	16.47	67.25	12.07	17.95	69.37	11.12	16.03
DE	12.74	2.99	23.49	11.94	2.39	20.02	12.88	2.43	18.83	12.35	2.37	19.16	12.82	2.30	17.96
ME	10.15	2.40	23.68	9.48	1.93	20.33	10.46	1.92	18.34	9.97	1.92	19.27	10.33	1.84	17.78

 

CP: Crude Protein; CF Crude Fiber; NDF: Neutral Detergent Fiber; ADF: Acid Detergent Fiber; EE: Ether Extract; GE: Gross Energy; OMD: Organic Matter Digestibility; Energy Matter Digestibility; DE: Digestible Energy; ME: Metabolize Energy; SD: Standard Deviation; CV: Coefficient of Variance; SD: Standard Deviation; CV: Coefficient of Variance.

 

plant chemical content and digestibility rates in ruminants. However, there is currently a lack of widely available information regarding the classification of leguminous plants based on their chemical composition and digestibility It is database sourced from research whole the world published in journal. The database was then built on three subcategories based on the type of stem, climate in which they are grown, and edible parts, including legumes from families faboideae, caesalpiniodeae, and mimosoideae in the database, which consisted of information regarding their chemical contents and digestibility rates in ruminants (Table 2 and 3). It was recorded that approximately 387 legume species were included in the datasets. The z-score analysis was introduced to statistically verified that those legumes quantitative values are within reliable ranges. The z-score analysis was conducted using the following formula (Warner, 2016):

where z is the z-score, xi the observed data value, μ is the mean of the data, and s is the standard deviation. Through normality ranges data with a z-score greater than 2 or smaller than -2 will be identified before values that are still within the normal according to biolgically value not to considered as outlier data and were removed from the database. Values falling outside the established z-score range and deemed biologically implausible were excluded from further analysis. As a result, 236 leguminous plant data points were obtained for further multivariate analysis. The legume categories and number of legume samples used are presented in Table 1.

The parameters utilized in this study were nutrient content of legumes, such as crude protein (CP, % of DM), crude fiber (CF, % of DM), neutral detergent fiber (NDF, % of DM), acid detergent fiber (ADF, % of DM) lignin (% of DM), ether extract (EE,% of DM), Ash (% of DM), and gross energy (GE, MJ/kg of DM). Digestibility values include: organic matter digestibility (OMD, %), energy matter digestibility (EMD, %), digestible energy (DE, MJ/Kg of DM), and metabolize energy (ME, MJ/Kg of DM). The classification of edible parts of legume forages and climate region were determined based on information from the Feedipedia website. Meanwhile for the type of stems, they were classified using the USDA plant database (https://plants.usda.gov) and divided into “herbs” (including forbs, vines, and herbs), “shrubs” (including sub-shrubs and shrubs), and “trees” (including shrub-trees and trees).

Statistical Analysis

The multivariate analyses employed in this study included correlation coefficient analysis, principal component analysis (PCA), hierarchical cluster analysis, and k-means cluster analysis. All statistical procedures were performed using R software version 16. Meanwhile for the multivariate analyses were conducted using the R vegan package version 2.5.6 (Oksanen et al., 2014). For the correlation analysis was used to measure the linear relationship between two random variables (Schober dan Schwarte, 2018). The correlation analysis was used the “corr” program code with the “pearson” method was executed in R Studio. The analysis results are presented in the form of a matrix table to simplify data interpretation.

The PCA analysis was performed using the “prcomp” code in R Studio whereas the algorithm includes data scaling, eigenvector value determination, and principal component formation. Briefly, the analysis began with data standardization to equalize the units. The software then the datasets were reduced and generated eigenvalues, and sequentially explained the variation in the reduced data. Furthermore, factor values are generated by forming new variables called principal components (Jollife dan Cadima, 2016). At the last, factor values were then visualized as biplots directed towards facilitating the data interpretation.

The cluster analysis used in the present study was K-means non-hierarchical cluster analysis, utilizing the program code “kmeans.” The cluster analysis algorithm consisted three stages: dimension reduction, cluster identification, and result evaluation (Dalmaijer et al., 2022). The cluster analysis began by determining the number of clusters using the “silhouette” method. Data scaling and K-means cluster analysis were performed. The results of the analysis are visualized in the form of plots. Finally, the results of the analysis are evaluated and identified.

RESULTS AND DISCUSSION

Correlation between Nutrient Contents and Digestibility Values

The results of the correlation matrix are presented in Table 4. According to the Pearson correlation approach, some chemical composition variables were positively correlated with digestibility rates, whereas others were negatively correlated. All chemical composition variables were significantly correlated (p<0.01) with digestibility values, except for EE, which was significantly correlated (p<0.05). Protein content was positively correlated with the digestibility variables OMD (0.45), ED (0.51), DE (0.59), and ME (0.51). Proteins in ruminants are categorized into two types: rumen degradable protein (RDP) and rumen undegradable protein (RUP) (Savari et al., 2018). Both have different roles in ruminants. RDP will be degraded in rumen into NH3 and amino acids. NH3 will be used in the formation of microbial protein synthesis for carbohydrate degradation.

Meanwhile, RUP can passed through rumen and degrades into amino acids and mostly absorbed in the small intestine and delivered to the entire animal tissues that affects for growth, production, and reproduction (Putri et al., 2021). Therefore, as an essential compound, high protein content in legume forages may positively influence ruminant health and performance. Jayanegara et al. (2016) and Widodo et al., (2023) also showed that forage protein content had a strong positive relationship with digestibility rates. However, CP was negatively correlated with fiber components CF (-0.57), NDF (-0.55), ADF (-0.62), and lignin (-0.41). Some of fiber molecules in feed can bond with the protein, making it difficult for the livestock body to be degraded, and can adversely influence ruminant digestibility on feed intake (Hervik and Svihus, 2019). It is also confirmed that the fiber fraction molecules such as CF, NDF, ADF, and lignin were significantly negatively correlated (p<0.01) with

 

Table 4: Pearson’s correlation matrix of chemical composition and digestibility values of legumes.

Variable

CP

CF

NDF

ADF

Lignin

EE

Ash

GE

OMD

EMD

DE

ME

CP

1.00

CF

-0.57**

1.00

NDF

-0.55**

0.68**

1.00

ADF

-0.62**

0.77**

0.87**

1.00

Lignin

-0.41**

0.35**

0.48**

0.57**

1.00

EE

0.29**

-0.26**

-0.15*

-0.17**

0.06ns

1.00

Ash

-0.10ns

0.26**

0.28**

0.29**

0.20**

0.10ns

1.00

GE

0.39**

-0.05ns

-0.15*

-0.17**

-0.08ns

0.34**

-0.63**

1.00

OMD

0.45**

-0.72**

-0.59**

-0.64**

-0.46**

0.14*

-0.46**

0.22**

1.00

EMD

0.51**

-0.76**

-0.62**

-0.69**

-0.48**

0.16*

-0.43**

0.20**

0.96**

1.00

DE

0.59**

-0.72**

-0.62**

-0.67**

-0.47**

0.22**

-0.43**

0.33**

0.85**

0.94**

1.00

ME

0.51**

-0.73**

-0.61**

-0.67**

-0.47**

0.21**

-0.53**

0.34**

0.94**

0.98**

0.96**

1.00

 

CP: Crude Protein; CF: Crude Fiber; NDF: Neutral Detergent Fiber; ADF: Acid Detergent Fiber; EE: Ether Extract; GE: Gross Energy; OMD: Organic Matter Digestibility; Energy Matter Digestibility; DE: Digestible Energy; ME: Metabolize Energy, * Significant at p<0.05, ** Significant at p<0.01, ns: not significant p>0.05.

 

the digestibility variables OMD (-0.72, -0.59, -0.64, -0.46), ED (-0.76; -0.62; -0.67; -0.47), DE (-0.72; -0.62; -0.67; -0.47), and ME (-0.73; -0.61; -0.67; -0.47). Because some of leguminous forages that contain high fiber fractions consist of high hemicellulose and cellulose compounds that are difficult to digest by rumen microbes, and adversely affects the digestibility rates. Those evidence are in accordance with the Jayanegara et al., (2016) findings, showed that NDF and ADF consisted in feed particles commonly correlated with the negative organic matter digestibility.

EE is an organic material that is broken down into energy and has a significant positive correlation (p<0.05) with OMD (0.14) and EMD (0.16) as well as a highly significant positive correlation (p<0.01) with DE (0.22) and ME (0.21). Fat is one of energy sources for ruminants, especially dairy cattle, because it produces more ATP than proteins (Hervik and Svihus, 2019; Palmquist, 1994). In contrast, ash content was found to be significantly negatively correlated (p<0.01) with the digestibility variables OMD (-0.46), ED (-0.43), DE (-0.43), and ME (-0.53). Ash is the inorganic residue that the removal of water and organic matter or describes total mineral content of the feed (Thiex et al., 2012). In energy estimation, a higher ash content will result in lower feed energy. This is because ash is not an organic material that can be converted into energy. An overview of the energy content in the feed can be found in the GE content. GE represents the energy derived from feed organic matter such as carbohydrates, proteins, and fat (Jayanegara et al., 2019). The analysis showed that GE content was significantly positively correlated with the digestibility variables OMD (0.22), ED (0.20), DE (0.30), and ME (0.34). This positive correlation is consistent with the relationship between protein and fat, which also correlates positively with digestibility. Therefore, an increase of protein and fat will enhance the GE value of the feed.

 

Table 5: Eigenvalue of principal component (PC).

PC	Eigan value
	Total	% Total variance	%cumulative
1	6.66	55.50	55.50
2	1.48	12.37	67.87
3	1.26	10.54	78.40
4	0.92	7.67	86.08
5	0.61	5.10	91.18
6	0.48	3.97	95.15
7	0.25	2.11	97.26
8	0.13	1.08	98.34
9	0.11	0.91	99.24
10	0.07	0.61	99.86
11	0.01	0.09	99.95
12	0.01	0.05	100.00

 

Principal Component Analysis (PCA)

The results of the PCA analysis in the form of eigenvalues and factor loadings will be visualized as graphs and plots for the reader to understand easily. This eigenvalue explains the data variation that can be explained based on the principal component (PC) formed; the higher the eigenvalue, the higher the data variation that the PC can explain. According to Kaiser (1960), only PCs with eigenvalues> 1.0, are used for data exploration. The PCA analysis showed that PC1, PC2, and PC3 had eigenvalues>1.0, and were able to explain 78.40% of the data (Table 5). PC1, PC2, and PC3 were further analyzed to determine the factor loading values of the variables formed in the three main components. The factor loading value describes the correlation between the original variables and the variables formed by PCA. Stevens (Stevens, 2009) recommends a factor loading value of >0.60 can describe a strong correlation between variables. The factor-loading values are presented in Table 6.

 

Table 6: Factor loading of principal component (PC) whose eigenvalue>0.1.

Variable	PC1	PC2	PC3
CP	0.68	-0.02	0.43
CF	-0.83	0.25	-0.14
NDF	-0.78	0.25	-0.09
ADF	-0.84	0.25	-0.10
Lignin	-0.58	0.27	0.19
EE	0.25	0.25	0.82
Ash	-0.49	-0.64	0.47
GE	0.36	0.86	0.17
OMD	0.90	-0.01	-0.20
EMD	0.94	-0.05	-0.15
DE	0.93	0.05	-0.05
ME	0.94	0.10	-0.14

 

CP: Crude Protein; CF: Crude Fiber; NDF: Neutral Detergent Fiber; ADF: Acid Detergent.

 

In this study, PC1 showed a variance value of 55.50%. The variables that were formed and correlated with PC1 were CP (0.68), CF (-0.83), NDF (-0.78), ADF (-0.84), OMD (0.90), ED (0.94), DE (0.93), and ME (0.94). Positive and negative values indicate the relationship between variables; positive values indicate an increase, while negative values indicate a decrease. Fiber; EE: Ether Extract; GE: Gross Energy; OMD: Organic Matter Digestibility; Energy Matter Digestibility; DE: Digestible Energy; ME: Metabolize Energy.

In PC1, the digestibility variable had a positive connected with protein but a negative connected with the fiber component. This result was similar to that of the correlation analysis, which showed a connected between protein, fiber, and digestibility rates. It can be suggests that the protein and fiber content of the feed significantly associated to the digestibility rates. Digestibility rate is an essential aspect of feed quality; the higher the digestibility rate, the better the feed quality (Darma et al., 2023). Feeding high-protein feed, such as legumes, can increase the digestibility of livestock feed (Araújo et al., 2020; Truong and Trung, 2023). However, legumes have issues owing to the hemicellulose and cellulose content in their cell walls, making it difficult for rumen microbes to digest (Truong et al, 2022). However, some studies have found that rumen microbes can digest hemicellulose. According to Zhou et al. (2022), NDF that still contains hemicellulose and has a quadratic relationship with ruminant digestibility. This indicated that hemicellulose content of NDF can still be degraded by rumen microbes to the concentration level. Meanwhile, PC2 and PC3 were difficult to interpreted due to the limited variable components followed with factors higher than 0.6. In PC2 only ash and GE are correlated, while PC3 only EE. Unequivocally, the associated variables between PC1 and PC2 are shown in Figure 1, whereas the horizontal line represents PC1, and vertical line represents PC2.

 

Biplot visualization illustrated the relationship between the variables and leguminous data in this study. As shown in Figure 1, the leguminous plots in region I are legumes with high fiber components (CF, NDF, ADF, and lignin). Conversely, region II contained legumes with high crude protein content and digestibility. The results of the PCA score classification are shown in Figure 2. In the stem type category, legume plots were scattered across regions I and II. This shows that legumes with tree, shrub, and herb stem types do not show significant differences in chemical content; therefore, they do not affect the digestibility rate. According to Upadhyay et al. (2022), the factors that significantly affect the productivity and chemical content of plants are the soil, water, and fertilizer. The climate category showed the same results as the stem type (Figure 2). The classification of legume based on their typical region such as in the tropical, subtropical, and temperate regions were not significantly associated with their chemical contents (Figure 1). Legumes can grow in warm and cold environments and are able to adapt to maintaining their productivity (Krner, 2016; Lee et al., 2017). This was assumed to be reason for no difference in the chemical content of the plant growth with the climate region category. Moreover, the edible part category also showed no significant difference in chemical composition. However, it was found that seed overall plot in region II, which was high in protein and digestibility rates (Figure 1). This indicated that leguminous seed parts contained high protein and digestibility rates.

 

Cluster Analysis

Cluster analysis is combines several objects based on their similarities. The formation of cluster analysis groups was based on the distance between the objects. The cluster analysis used is a non-hierarchical k-means cluster analysis, whose algorithm must first determine the number of clusters to be formed. One of the cluster optimization methods to be developed is the silhouette method (Figure 3). Determining the optimal point with this method by looking at the peak point of the graph; therefore, in this study, the number of clusters to be formed was two.

The results of the cluster analysis are shown in Figure 3. The red and blue plots represent clusters 1 and 2, respectively. The cluster analysis results are presented in Table 7. Cluster 1 was characterized by CF, NDF, ADF, Lignin, and Ash characteristics, and Cluster 2 by CP, EE, GE, OMD, EMD, DE, and ME characteristics. Result of this analysis was similar with correlation analysis and PCA which showed that protein was positively related to digestibility rates and fibre was negatively related, vice versa. Hence, legumes with high crude protein content potentially improve ruminant digestibility, while high fiber content, in contrary might lowered digestibility rates.

 

 

This can also be clearly seen in Table 8. It can be assumed that high protein legumes in Cluster 2 can be categorized as favourable legumes, while high fiber legumes in Cluster 1 can be categorized as unfavourable legumes. Protein has an essential role in livestock growth, productivity, and reproduction. Therefore, it is beneficial for livestock if the feed has a high protein content. In addition, protein can enhance feed digestibility, subsequently impacting environmental sustainability. It is crucial to prioritize feed with a high digestibility rate when selecting feed, particularly forage for animal feed. Numerous studies (Álvarez et al., 2022; Chaokaur et al., 2015; Kannan et al., 2017) have demonstrated that feeding highly digestible feed can reduce methane gas production, making it a potential strategy to mitigate methane gas emissions in livestock feed formulations. Legumes in cluster 2 can be a suitable option for providing forage to livestock. In contrast, high fiber can decrease digestibility, that negative impacts for livestock productivity (Castro-Montoya and Dickhoefer, 2018;

 

Table 7: Profiling and number of cluster members.

Cluster	CP	CF	NDF	ADF	Lignin	EE	Ash	GE	OMD	EMD	DE	ME
1	16	28.2	45.7	34	9.03	2.89	8.85	18.5	66.7	62.7	11.4	9.22
2	22.7	13.8	30	19	4.56	4.56	5.12	18.9	87.8	85	16.1	12.9
Identification on the number of cluster members per category
Category				Number of cluster members
				1	Percentage in cluster (%)			2		Percentage in cluster (%)
All data in study				181	23.63			23.63		23.63
Edible part
	Aerial part			129	98.47			2		1.53
	Leaves			28	90.32			3		9.68
	Pods			25	60.98			16		39.02
	Seed			0	0			35		100
Plant origin climate
	Tropical			113	79.02			30		20.98
	Subtropical			44	70.97			18		29.03
	Temperate			24	75.00			8		25.00
Type of stem
	Shrub			66	84.62			12		15.38
	Tree			38	48.72			30		38.46
	Herb			67	83.75			13		16.25

 

CP: Crude Protein; CF: Crude Fiber; NDF: Neutral Detergent Fiber; ADF: Acid Detergent Fiber; EE: Ether Extract; GE: Gross Energy; OMD: Organic Matter Digestibility; Energy Matter Digestibility; DE: Digestible Energy; ME: Metabolize Energy.

 

Table 8: The highest and lowest quality legumes determined from the analysed database.

Highest quality legumes
Species	Common names	Edible part	CP	CF	NDF	ADF	Lignin	EE	Ash	GE	OMD	EMD	DE	ME
Trigonella foenum-graecum	Fenugreek	Seed	26.2	1.0	15.5	2.9	3.6	6.4	4.0	19.7	96.0	95.0	18.9	15.2
Vicia sativa	vetch	Seed	28.4	4.7	15.7	7.5	0.9	1.5	4.0	18.9	92.5	91.1	17.2	13.8
Senna tora	Wild senna	Seed	18.2	4.6	21.3	13.0	3.4	7.4	9.1	18.6	92.5	88.0	16.2	13.3
Lens culinaris	Lentil	Seed	26.9	4.9	13.0	6.3	1.6	1.6	3.8	18.5	92.4	90.9	16.8	13.5
Phaseolus vulgaris	bean	Seed	24.8	5.2	20.0	7.6	0.2	1.7	4.6	18.6	92.3	90.6	16.8	13.6
Lowest quality legumes
Crotalaria brevidens	Slenderleaf	Aerial part	12.7	42.5	57.8	40.0	11.5	2.9	5.8	19.5	32.1	30.6	6.0	4.7
Prosopis tamarugo	Tamarugo	Leaves	12.0	13.8	27.3	39.3	17.6	2.0	17.2	16.1	38.7	41.7	10.2	5.1
Vicia benghalensis	Purple vetch	Aerial part	11.0	39.8	65.8	47.7	13.3	3.7	8.5	18.4	44.0	42.0	7.8	6.2
Vicia faba	Faba bean	Aerial part	10.9	45.2	26.3	12.5	13.0	1.8	3.1	19.8	44.8	42.8	8.5	6.7
Cicer arietinum	Chickpea	Aerial part	5.4	41.0	65.6	46.9	11.9	1.0	7.4	18.1	45.0	41.4	7.5	6.1

 

Ratnawaty et al., 2018). However, fiber also essential as a main source of energy in ruminants. Fermentation of forage fiber will produce VFA that is used by cattle as their main energy (Guadayo et al., 2019). Most of legumes in this study were located in Cluster 1, approximately 180 plants species. Which means, further identification was performed based on previous legume categories.

Cluster 1 had the highest percentage of aerial parts in the edible category, whereas Cluster 2 had the highest percentage of seed parts. Aerial parts refer to the above ground portions of plants that ruminants can consume. The chemical composition of these parts varies considerably depending on the mixture of the edible parts used. Similarly, Hao et al. (2021) found that the whole edible part had a lower nutrient content than the leaves. Aerial parts, one of which is the stem, contain high lignin content that can decrease digestibility. In contrast, seed parts have a high protein content because of their role in developing new individuals, which require high protein levels for growth (El-Maarouf, 2022). The regional climate category and subtropical legumes dominated Cluster 2, whereas the tropical legumes dominated Cluster 1. According to Lee et al. (2017), plants in tropical areas have higher NDF and lower protein contents than those in temperate areas. Legumes in tropical regions receive more sunlight, prompting them to thicken their cell walls and reduce evaporation (Kering et al., 2011). Climate significantly affects the structure and polysaccharides of plants, such as lignin, cellulose, and hemicellulose, resulting in variable cell wall compositions and fluctuations in the nutritional value of forage (Gang et al., 2015; Lee et al., 2017; Upadhyay et al., 2022). In the stem type category, tree legumes dominated Cluster 2, while shrub legumes dominated Cluster 2. This study similar (Castro-Montoya and Dickhoefer, (2018) and Castro-Montoya and Albarrán-Portillo, (2023), Tree legumes have higher protein and lower NDF content than shrub legumes or herbs. High protein content correlated with high digestibility, making tree legumes more prevalent in Cluster 2. Conversely, shrub legumes had a lower protein content, mostly in cluster 1.

CONCLUSIONS AND RECOMMENDATIONS

Multivariate analysis showed that legume characteristics in the categories of stem type, regional climate, and edible parts had similar associations with chemical content and digestibility rates. This indicates that there were no specific differences among the three categories. However, the seeds were characterized by high protein content and digestibility rates. Furthermore, protein content was positively associated with ruminant digestibility, whereas the fiber content was vice versa showed negative association. Therefore, it can be concluded that the high quality legumes indicated by high protein content, whereas low quality legumes are high fiber content. In addition, legumes in cluster 2 can be used as forage for ruminants.

ACKNOWLEDGEMENTS

The authors are thankful to the Ministry of Education, Culture, Research and Technology, Republic of Indonesia, for the PHC Nusantara research grant, year 2024, grant number 15819/IT3.D10/PT.01.02/P/T/2024. The first author is grateful to Adelia, Martin, and Giovani for their technical assistance.

NOVELTY STATEMENTS

This study provides a novel classification of leguminous plants based on chemical composition and digestibility using advanced multivariate analysis techniques, such as PCA and k-means cluster analysis. By integrating data from Feedipedia and categorizing legumes by stem type, climatic origin, and edible parts, this research highlights the distinct association between protein content and digestibility, and the inverse relationship with fiber components. These findings not only provide a comprehensive understanding of legume feed quality but also propose a practical framework for selecting high-quality forage to enhance ruminant performance and reduce environmental impacts.

AUTHOR’S CONTRIBUTIONS

Dedy Nanda Kurniawan: Investigation, formal analysis, methodology, software, visualization, writing - original draft.

Nahrowi Nahrowi: Validation, data curation, supervision, writing - review and editing.

Yulianri Rizki Yanza: Validation, data curation, supervision, writing - review and editing.

Vincent Niderkorn: Validation, supervision, writing - review and editing.

Anuraga Jayanegara: Conceptualization, validation, resources, data curation, supervision, writing - review and editing.

Availability of Data and Materials

The data will be made available from the corresponding author upon reasonable request.

Conflict of Interest

The authors declare that there is no conflict of interest.

REFERENCES

Álvarez C, Nielsen NI, Weisbjerg MR, Volden H, Eknæs M, Prestløkken E (2022). High-digestible silages allow low concentrate supply without affecting milk production or methane emissions. J. Dairy Sci., 105(4): 3633–3647. https://doi.org/10.3168/jds.2021-21479

Araújo LC, Valente EEL, da Trindade Filipini V, Carvalho ST, Stahlhofer M, Cabral CHA, Damasceno ML, Barbizan M (2020). Effects of dietary protein level on intake, digestibility, and energy expenditure in dairy heifers. Semin. Cienc. Agrar., 41(6): 3211–3223. https://doi.org/10.5433/1679-0359.2020v41n6Supl2p3211

Bahadoran Z, Mirmiran P (2015). Potential Properties of Legumes as Important Functional Foods for Management of Type 2 Diabetes: A Short Review. Int. J. Food. Sci. Nutr., 4(2): 6. https://doi.org/10.11648/j.ijnfs.s.2015040201.12

Brito AF, Broderick GA (2006). Effect of varying dietary ratios of alfalfa silage to corn silage on production and nitrogen utilization in lactating dairy cows. J. Dairy Sci., 89(10): 3924–3938. https://doi.org/10.3168/jds.S0022-0302(06)72435-3

Castro-Montoya J, Dickhoefer U (2018). Effects of tropical legume silages on intake, digestibility and performance in large and small ruminants: A review. Grass Forage Sci., 73(1): 26–39. https://doi.org/10.1111/gfs.12324

Castro-Montoya JM, Albarrán-Portillo B (2023). Digestibility of Additional Crude Protein From Tropical Legumes in Mixed Grass-Legume Rations for Ruminants. Chil. J. Agric. Anim. Sci., 39(1): 65–74. https://doi.org/10.29393/CHJAA39-6DAJB20006

Castro-Montoya JM, Dickhoefer U (2020). The nutritional value of tropical legume forages fed to ruminants as affected by their growth habit and fed form: A systematic review. Anim. Feed Sci. Technol., 269: 114641. https://doi.org/10.1016/j.anifeedsci.2020.114641

Chaokaur A, Nishida T, Phaowphaisal I, Sommart K (2015). Effects of feeding level on methane emissions and energy utilization of Brahman cattle in the tropics. Agriculture, Ecosystems and Environment, 199: 225–230. https://doi.org/10.1016/j.agee.2014.09.014

Dalmaijer ES, Nord CL, Astle DE (2022). Statistical power for cluster analysis. BMC Bioinformatics, 23(1): 1–28. https://doi.org/10.1186/s12859-022-04675-1

Darma ING, Jayanegara A, Sofyan A, Budiartilaconi E, Ridla M, Herdian H (2023). Evaluation of nutritional values of tree-forage legume leaves from Gunungkidul District, Indonesia. Biodiversitas, 24(5): 2733–2745. https://doi.org/10.13057/biodiv/d240527

El-Maarouf-bouteau H (2022). The Seed and the Metabolism Regulation. Biology, 11(2): 1-16. https://doi.org/10.3390/biology11020168

Everitt BS (1975). Multivariate analysis: the need for data, and other problems. BJPsych., 126(3): 237–240. https://doi.org/10.1192/bjp.126.3.237

Gang C, Suping W, Xiang H, Juan H, Lei D, Lihong Z, Lixia Y (2015). Environmental factors affecting growth and development of Banlangen (Radix Isatidis) in China. Afr. J. Plant Sci., 9(11): 421–426. https://doi.org/10.5897/ajps2015.1266

Guadayo GF, Rayos AA, Merca FE., Tandang AG, Loresco MM, Angeles AA (2019). Prediction of in situ ruminal degradability of forages in Buffaloes using the in vitro gas production technique. Trop. Anim. Sci. J., 42(2): 128–136. https://doi.org/10.5398/tasj.2019.42.2.128

Hao Y, Huang S, Liu G, Zhang J, Liu G, Cao Z, Wang Y, Wang W, Li S (2021). Effects of different parts on the chemical composition, silage fermentation profile, in vitro and in situ digestibility of paper mulberry. Animals, 11(2): 1–12. https://doi.org/10.3390/ani11020413

Hervik AK, Svihus B (2019). The role of fiber in energy balance. J. Nutr. Metab., 2019: 1-11. https://doi.org/10.1155/2019/4983657

James P. Stevens (2009). Applied Multivariate Statistics For The Social Sciences. 1–23.

Jayanegara A, Dewi SP, Ridla M (2016). Nutrient content, protein fractionation, and utilization of some beans as potential alternatives to soybean for ruminant feeding. Media Peternakan, 39(3): 195–202. https://doi.org/10.5398/medpet.2016.39.3.195

Jayanegara A, Ridla M, Nahrowi, Laconi EB (2019). Estimation and validation of total digestible nutrient values of forage and concentrate feedstuffs. IOP Conference Series: Mater. Sci. Eng., 546(4): 1-5. https://doi.org/10.1088/1757-899X/546/4/042016

Jayanegara A, Wina E, Soliva CR, Marquardt S, Kreuzer M, Leiber F (2011). Dependence of forage quality and methanogenic potential of tropical plants on their phenolic fractions as determined by principal component analysis. Anim. Feed Sci. Technol., 163(24): 231–243. https://doi.org/10.1016/j.anifeedsci.2010.11.009

Jollife IT, Cadima J (2016). Principal component analysis: A review and recent developments. Philosophical Transactions of the Royal Society A: Math., Phys. Eng. Sci., 374(2065). https://doi.org/10.1098/rsta.2015.0202

Kaiser HF (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement. Educ. Psychol. Meas., 20(1): 141–151.

Kannan A, Bhar R, Radotra S, Mal G, Singh B, Jadhav VR, Kumar M, Gaurav A (2017). Nutrient composition, in vitro true digestibility and methane production potential of feed resources of North Western Himalayan region. Indian J. Anim. Sci., 87(10): 1243–1250. https://doi.org/10.56093/ijans.v87i10.75292

Kelln BM, Penner GB, Acharya SN, McAllister TA, McKinnon JJ, Saleem AM, Biligetu B, Lardner HA (2023). Effect of mixtures of legume species on ruminal fermentation, methane, and microbial nitrogen production in batch and continuous culture (RUSITEC) systems. Can. J. Anim. Sci., 103(4): 326–337. https://doi.org/10.1139/cjas-2022-0095

Kering MK, Guretzky J, Funderburg E, Mosali J (2011). Effect of nitrogen fertilizer rate and harvest season on forage yield, quality, and macronutrient concentrations in Midland Bermuda grass. Commun. Soil Sci. Plant. Anal., 42(16): 1958–1971. https://doi.org/10.1080/00103624.2011.591470

Krner C (2016). Plant adaptation to cold climates. F1000Research, 5: 1–5. https://doi.org/10.12688/f1000research.9107.1

Lee MA, Davis AP, Chagunda MGG, Manning P (2017). Forage quality declines with rising temperatures, with implications for livestock production and methane emissions. Biogeosciences, 14(6): 1403–1417. https://doi.org/10.5194/bg-14-1403-2017

Marsetyo M, Mumu MI, Basri M (2020). The Effect of Various Legume Herbs Supplementation on Intake, Digestibility and Growth of Fat-tailed Lambs Given Mulato Grass. Anim. Prod., 21(3): 129-135. https://doi.org/10.20884/1.jap.2019.21.3.784

Miedico O, Nardelli V, D’Amore T, Casale M, Oliveri P, Malegori C, Paglia G, Iammarino M (2022). Identification of mechanically separated meat using multivariate analysis of 43 trace elements detected by inductively coupled mass spectrometry: A validated approach. Food Chem., 397(July), 133842. https://doi.org/10.1016/j.foodchem.2022.133842

Myers SS, Zanobetti A, Kloog I, Huybers P, Leakey ADB, Bloom AJ, Carlisle E, Dietterich LH, Fitzgerald G, Hasegawa T, Holbrook NM, Nelson RL, Ottman MJ, Raboy V, Sakai H, Sartor KA, Schwartz J, Seneweera S, Tausz M, Usui Y (2014). Increasing CO2 threatens human nutrition. Nature, 510(7503): 139–142. https://doi.org/10.1038/nature13179

Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Wagner H (2014). Vegan: Community Ecology Package. R Package Version 2.2-0. http://cran.rproject.org/package=vegan

Palmquist DL (1994). Conference: regulating lipid metabolism to increase productive efficiency. J. Nutr., 124(8): 1371-1381. https://doi.org/10.1093/jn/124.suppl

Pitman WD, Vendramini JMB (2020). Legumes for Tropical and Subtropical Areas. Forages, II: 277–296. https://doi.org/10.1002/9781119436669.ch15

Putri EM, Zain M, Warly L, Hermon H (2021). Effects of rumen-degradable-to-undegradable protein ratio in ruminant diet on in vitro digestibility, rumen fermentation, and microbial protein synthesis. Vet. World, 14(3): 640–648. https://doi.org/10.14202/VETWORLD.2021.640-648

Ratnawaty S, Hartutik H, Chuzaemi S (2018). The Effect of Herbaceous Legume of Feed in In-Vitro Digestibility. IOP Conf. Ser. Earth Environ. Sci., 119(1). https://doi.org/10.1088/1755-1315/119/1/012012

Restitrisnani V, Nugroho TA, Rianto E, Purnomoadi A (2016). Methane Emission Factor at Different Total Digestible Nutrients and Feeding Level in Ram. Pro. Intsem., 352–356. https://doi.org/10.14334/proc.intsem.lpvt-2016-p.352-356

Savari M, Khorvash M, Amanlou H, Ghorbani GR., Ghasemi E, Mirzaei M (2018). Effects of rumen-degradable protein:rumen-undegradable protein ratio and corn processing on production performance, nitrogen efficiency, and feeding behavior of Holstein dairy cows. J. Dairy Sci., 101(2), 1111–1122. https://doi.org/10.3168/jds.2017-12776

Schober P, Schwarte LA (2018). Correlation coefficients: Appropriate use and interpretation. Anesthesia and Analgesia, 126(5): 1763–1768. https://doi.org/10.1213/ANE.0000000000002864

Shih MC, Chang CM, Kang SM, Tsai ML (2011). Effect of different parts (leaf, stem and stalk) and seasons (summer and winter) on the chemical compositions and antioxidant activity of Moringa oleifera. Int. J. Mol. Sci., 12(9): 6077–6088. https://doi.org/10.3390/ijms12096077

Simbaya J (2002). Potential of fodder tree/shrub legumes as a feed resource for dry season supplementation of small holder ruminant animals. Intl Atomic Energy Agency, 69–76.

Soedarmanto I, Nururrozi A, Ramandani D (2022). Benefits and risk of giving oleifera moringa as ruminant animal feed : brief. Ind. J. Vet. Scie., 3(1): 20-32. https://doi.org/10.22146/ijvs.v3i1.82655

Stevens JP (2009). Applied Multivariate Statistics for the Social Sciences (5th ed.). Routledge, New York, USA.

Thiex N, Novotny L, Crawford A (2012). Determination of ash in animal feed: AOAC Official Method 942.05 revisited. J. AOAC Int., 95(5): 1392–1397. https://doi.org/10.5740/jaoacint.12-129

Truong NB, Thu N, Van (2022). Effect of Dietary Levels of Neutral Detergent Fiber on Feed Intake, Nutrient Digestibility, Rumen Parameters and Nitrogen Retention of Beef Cattle. Adv. Anim. Vet. Sci., 10(2): 316–322. https://doi.org/10.17582/JOURNAL.AAVS/2022/10.2.316.322

Truong NB, Trung TT (2023). Effect of Dietary Crude Protein Levels on Feed Intake and Nutrient Digestibility of Wagyu Crossbred Cattle. Online Journal of Animal and Feed Research, 13(4): 295–301. https://doi.org/10.51227/ojafr.2023.44

Upadhyay H, Juneja A, Turabieh H, Malik S, Gupta A, Bitsue ZK, Upadhyay C (2022). Exploration of Crucial Factors Involved in Plants Development Using the Fuzzy AHP Method. Math. Probl. Eng., https://doi.org/10.1155/2022/4279694

Warner RA (2016). Using Z Scores for the Display and Analysis of Data. Optimizing the Display and Interpretation of Data: 7–51. https://doi.org/10.1016/b978-0-12-804513-8.00002-x

Widodo S, Shiddieqy MI, Wahyono T, Widiawati Y, Muttaqin Z (2023). Analysis of Correlation between Nutrient Content, Digestibility, and Gas Production of Forages in Indonesia. Adv. Anim. Vet. Sci., 11(11): 1770-1778. https://doi.org/10.17582/journal.aavs/2023/11.11.1770.1778

Zeleke AB, Clément C, Hess HD, Kreuzer M, Soliva CR (2006). Effect of foliage from multi-purpose trees and a leguminous crop residue on in vitro methanogenesis and ruminal N use. Int. Congr. Ser., 1293: 168–171. https://doi.org/10.1016/j.ics.2006.01.017

Zhou J, Xue B, Hu A, Yue S, Wu M, Hong Q, Wu Y, Wang Z, Wang L, Peng Q, Xue B (2022). Effect of dietary peNDF levels on digestibility and rumen fermentation, and microbial community in growing goats. Front. Microbiol., 13: 1-20. https://doi.org/10.3389/fmicb.2022.950587