The plants were grown at the facilities of a local government experimental farm located in the southwest of Fuerteventura Island (Canary Islands, Spain). This farm features a 1,200 m² mesh nursery dedicated exclusively to the reproduction of native flora ( Supplementary Figure S1 ). Twelve native and endemic species from the island of Fuerteventura, regarded in oral tradition as goat feed, were selected for the present study. These included four legumes (Lotus lancerottensis, Bituminaria bituminosa, Coronilla viminalis, and Retama rhodorhizoides), three grasses (Cenchrus ciliaris, Phalaris coerulescens, and Tricholaena teneriffae), and five other herbaceous and shrub species from different families denominated as mixed group (Lavatera acerifolia, Periploca laevigata, Campylanthus salsoloides, Echium decaisnei, and Crambe sventenii) ( Figure 1 ).

All plants were germinated under controlled conditions and grown in a substrate made from a mix of Fuerteventura soil (Typic Torrifluvents; Soil Survey Staff, 2014) with a sandy-loam texture (clay 198.1 g/kg, silt 594.0 g/kg, and sand 207.9 g/kg), peat, and gravel in a 1:1:1 ratio, and irrigated with desalinated brackish water (EC_iw ~ 0.4 dS/m). The soil had a very low organic matter and nitrogen content, typical of arid region soils, requiring the use of basal fertilization (~ 6 g of NPK 17:6:14 fertilizer per 3 L volume container) to prevent nutritional deficiencies.

Plant tissue sampling took place in early November 2021 when the plants were between 10-18 months old. Samples (mainly leaves) from five individuals of each of the twelve selected species were collected. Once in the laboratory, the samples were pre-washed with distilled water, and divided into two subsamples. One of the subsamples was dried at 60°C for 72 hours and ground (1 mm) for fiber, protein, and minerals analysis, while the other subsample was stored at -80°C until their lipid analysis.

The nutritive composition of the leaf tissues was determined using standard procedures (AOAC, 1990, 2000). The nutritional potential of the forages was assessed based on the levels of CP, ash, neutral detergent fiber (NDF), acid detergent fiber (ADF), acid detergent lignin (ADL), enzymatic digestibility of organic matter (EOMD), relative feed value (RFV), ME, lipid content, lipid classes and fatty acid profiles, and minerals. CP was determined by dried combustion and measurement in the resulting gases with a thermoconductivity cell from CN828 equipment (LECO Corporation, St. Joseph, MI, USA). The ash content after incineration at 550°C, NDF, and ADF were analyzed using the Van Soest fractionation analysis (Van Soest et al., 1991). ADL was extracted using 72% sulfuric acid after the ADF procedure described by Robertson and Van Soest (1981), and EOMD following de la Roza and Argamentería (1992). The RFV was calculated from the ADF and NDF as follows: RFV = [88.9 - (0.779 * ADF)] * [(120/NDF)/1.29] (Putnam et al., 2008). The ME was calculated from the ash content and the in vitro digestibility of organic matter (IVOMD) as follows: ME = K * (100 - ash) * (IVOMD/100), where K = 0.16 for forages and the IVOMD was predicted from EOMD values (ARC, 1984; de la Roza and Argamentería, 1992).

Lipid characterization was based on the content of total lipids (TL), lipid classes and FAs profiles. TL were extracted by homogenization in chloroform/methanol (2:1 v/v) containing 0.01% butylated hydroxytoluene (BHT) as an antioxidant according to the method described by Folch et al. (1957) with slight modifications (Christie and Han, 2010). The organic solvent was evaporated under a stream of nitrogen, and the lipid content was determined gravimetrically. The lipid residue was then resuspended in chloroform/methanol (2:1 v/v) containing 0.05% BHT and stored under an inert nitrogen atmosphere at -20°C until further analysis.

The lipid classes were determined by high-performance thin-layer chromatography (HPTLC) with dual one-dimensional development (Olsen and Henderson, 1989), using 1-propanol/chloroform/methyl acetate/methanol/potassium chloride at 0.25% (5:5:5:2:1.8, v/v) to separate the polar lipids (PL), and hexane/diethyl ether/acetic acid (20:5:0.5, v/v) for the neutral lipids (NL). Lipid classes were quantified by calibrated densitometry by means of a CAMAG TLC Visualizer dual-wavelength flying-spot scanner (Camag, Muttenz, Switzerland), as described by Reis et al. (2019). Lipid class identification was performed by comparison with external lipid standards (cod roe lipid extract and a mix of digalactosyldiacylglycerol (DGDG), monogalactosyldiacylglycerol (MGDG), and sulfoquinovosyldiacylglycerol (SQDG) (Avanti Polar Lipids, Inc., Alabaster, Alabama, USA) placed on the same HPTLC plate ( Supplementary Figure S2 ). The proportion of each lipid class was expressed as a percentage of the TL in the analyzed sample.

Another aliquot of TL was subjected to acid-catalyzed transesterification according to Christie and Han (2010). The resulting fatty acid methyl esters (FAMEs) were purified by thin-layer chromatography (TLC) and separated and quantified using a TRACE-GC Ultra gas chromatograph (Thermo Fisher Scientific Inc., Milan, Italy) under the chromatographic conditions described in Galindo et al. (2022). Individual FAMEs were identified by reference to a well-characterized commercial standard mixture (Mix C4-C24 and PUFA No. 3 from menhaden oil (Supelco Inc., Merck KGaA, Darmstadt, Germany)), and their relative abundance was expressed as a percentage of the total FAs in the analyzed sample ( Supplementary Figure S3 ). Before transesterification, nonadecanoic acid (19:0) was added to the lipid fraction as an internal standard. When necessary, the identity of FAME was confirmed by GC-MS (DSQII, Thermo Scientific), under the same chromatographic conditions.

Nutritional quality of plant FA composition was assessed by calculating the index of atherogenicity (IA), the index of thrombogenicity (IT) following Ulbricht and Southgate (1991), and the hypocholesterolemic (h) to hypercholesterolemic (H) FAs ratio (hH) as described by Santos-Silva et al. (2002):

The elements analyzed for the study of mineral composition were N, P, K, Ca, Mg, Na, Fe, Mn, Cu, Zn, and B. All elements were determined by digestion combined with inductively coupled plasma mass spectrometry (ICP-MS), except N, which was measured using LECO CN828 equipment. The concentrations of these elements were compared with the established maximum tolerable levels (MTLs), defined as the highest level of a particular element in an animal’s diet over a specified period that does not lead to health deterioration or a decrease in animal performance (NRC, 2005).

Statistical methods were implemented using R version 4.3.0 (R Core Team, 2023) and IBM SPSS Statistics V26.0. The significance level for all tests was set at p< 0.05. Prior to each analysis, the assumptions of normality (Shapiro-Wilk test) and homogeneity of variance (Levene’s test) were verified within groups, and, where necessary, appropriate variance stabilizing transformations were performed. For all parameters, differences between the plant species were evaluated using a one-way ANOVA followed by the Tukey HSD post-hoc test. When transformations failed, Welch and Brown-Forsythe test followed by the Dunnett T3 test, or non-parametric Kruskal Wallis and Mann-Whitney U tests with Bonferroni adjustment were performed. Principal component analysis (PCA) for the main nutritional parameters of plant species was carried out. Two hierarchical cluster heatmaps were created using Euclidean distance, one for the lipid classes and the other for the main FAs groups, to identify patterns among species.

Table 1 presents the main nutritional parameters related to fiber, protein and ash contents of the analyzed plant species. Levels of NDF, ADF, and ADL varied strongly between families and species ( Table 1 ). Poaceae species exhibited higher average values of NDF and ADF compared to the other groups (e.g., 64.8, 41.4 and 28.0% for NDF in Poaceae, Fabaceae and mixed groups, respectively). T. teneriffae was the grass species with the highest concentrations of NDF and ADF among all analyzed plants. In contrast, ADL levels were up to two-three times higher in R. rhodorhizoides and E. decaisnei than in the other species. Content of CP varied significantly from ~ 6% (T. teneriffae) to ~ 32% (P. coerulescens), both from the Poaceae family. Generally, legumes surpassed grasses in terms of CP, except for P. coerulescens which presented the highest proportions of all analyzed species. Species within the mixed group showed CP levels close to those of legumes (average values 14.2 vs. 17.5%), with C. salsoloides reaching an average concentration of 18.9% ( Table 1 ). Ash content ranged from ~ 5% (R. rhodorhizoides) to 19% (C. sventenii). The average EOMD for all species in the mixed group was 80.0%, which was significantly higher compared to legumes (64.1%) and grasses (40.1%). C. sventenii, B. bituminosa, and C. ciliaris notably stood out within their respective groups. The energy potential of forage, calculated as ME, ranged from 8.1 to 8.6 MJ/kg DM on average per group. Among all studied species, B. bituminosa recorded the highest significant values (~ 10.9 MJ/kg DM), unlike specimens of R. rhodorhizoides which averaged 5.9 MJ/kg DM. Levels of RFV ranged from 75 in T. teneriffae to 318 in C. salsoloides. The mixed group had the highest average RFV indices (260), followed by Fabaceae (165) and Poaceae (108).

The mean values of macronutrients showed significant variations among species ( Table 2 ). Thus, Ca concentrations were higher in legumes than in grasses (8.8 and 4.7 g/kg, respectively), but the highest concentrations were reached in E. decaisnei and P. laevigata (up to 22.7 and 34.2 g/kg, respectively). Regardless of the plant group, Mg levels remained low with average values around 3 g/kg. Potassium dominated in E. decaisnei and C. viminalis, not exceeding 35 g/kg in any specimen. The lowest K values were found in C. salsoloides, C. sventenii, R. rhodorhizoides, and T. teneriffae (ranging from 3.6 to 8 g/kg). Sodium concentrations were high for E. decaisnei and P. laevigata (~ 26 g/kg), followed by two legumes species C. viminalis and L. lancerottensis. not exceeding 5 g/kg in the remaining species. Content of P was similar in all species, ranging between 0.3 and 2.3 g/kg, except in E. decaisnei which presented the highest average value of 3.4 g/kg. Nitrogen followed the same pattern as CP, peaking in P. coerulescens ( Table 2 ).

The mean micronutrient concentrations in the collected plant species are presented in Table 3 . Similarly to macroelements, significant differences were detected between groups and species. Iron levels showed high variability, with differences of up to approximately two orders of magnitude, led by L. lancerottensis and at the opposite end C. salsoloides (803.6 vs. 9.3 mg/kg). In most species, average Mn values fluctuated between ~ 46 and 81 mg/kg, except for R. rhodorhizoides, P. coerulescens, C. salsoloides and C. sventenii where it accounted for 1-2 mg/kg. Copper and Zn followed similar trends, with the highest concentrations in legumes and mixed group (average ~ 37.0 and 45.3 mg/kg for Cu and Zn, respectively). In both cases, E. decaisnei presented the highest concentrations of these elements. Levels of B also exhibited high variability, ranging from less than 10 mg/kg in grasses such as P. coerulescens and T. teneriffae to levels exceeding 100 mg/kg in the legume C. viminalis ( Table 3 ).

Forage quality generally improves with increases in CP, EOMD, RFV, and ME, and reductions in NDF, ADF, and ash content (Suyama et al., 2007a; Díaz et al., 2018; Pirasteh-Anosheh et al., 2023). The levels of NDF, ADF and ash found in our present work coincide largely with those reported by other authors for species such as B. bituminosa, C. viminales, L. lancerottensis and E. decaisnei under natural conditions (Chinea et al., 2014). However, CP values were slightly higher than those previously reported, possibly due to the more favorable nursery conditions. When compared to a protein-rich forage like alfalfa (15-20% CP; Annicchiarico et al., 2015; Yang et al., 2021), four of the studied species (L. lancerottensis, B. bituminosa, C. viminalis and C. salsoloides) exhibited CP levels within this range ( Table 1 ), demonstrating their potential as high-quality forage. The critical threshold of CP required for maintaining an adult non-pregnant sheep is approximately 7-9% (Pirasteh-Anosheh et al., 2023), and all examined native species, except T. teneriffae, surpassed this minimum, reinforcing their suitability for forage based on nitrogen and protein benchmarks.

The fiber percentages in our study (24.2-64.5% NDF and 9.0-40.5% ADF) also reflect the nutritional diversity within the studied forages. Eight out of twelve native species analyzed had an NDF content below 45%, and nine out of twelve had an ADF content below 35%, positioning them well within the ranges considered suitable for high-quality forage (Pirasteh-Anosheh et al., 2023). This has a positive impact on intake, energy density, milk production, cattle health, and feed cost (Christensen et al., 2015; Anderson et al., 2023). All forages had acceptable levels of ME (i.e., > 7 MJ/kg DM in most cases) for beef cattle and goats fed at a maintenance energy level (NRC, 1996). Considering that a 53 kg sheep needs forage with an ME content of 10-12 MJ/kg (Pirasteh-Anosheh et al., 2023)., only B. bituminosa met this requirement. This matches findings from Pirasteh-Anosheh et al. (2023) who reported that none of the selected halophyte species reached the required ME content.

The most valuable species in terms of forage quality within legumes were B. bituminosa and C. viminalis, both P. coerulescens and C. ciliaris within grasses, and C. salsoloides, E. decaisnei and C. sventenii for the mix group. Another criterion based on the quality parameters analyzed indicates that legumes can be classified as “premium” quality, grasses as “good-utility” and the rest between “premium” and “good” (Putnam et al., 2008; USDA Agricultural Marketing Service, 2024), which makes most of them suitable for dairy cattle and for low production or maintenance level beef cattle (Suyama et al., 2007b), reflecting their high nutritional content.

The TL content of the studied plants was in line with the generally low lipid levels of forages stated in the literature (< 8% DM; Davis et al., 2022). In recent years, there has been growing interest in under-exploited but promising plant species as alternative sources of vegetable oils. Some of these species can contain significant amounts of oil and, in addition, a high proportion of nutritionally desirable fatty acids, such as omega-3 PUFA (Manso et al., 2009; Timilsena et al., 2017). The PUFA levels recorded in our work were slightly higher than those determined by Cerci et al. (2011) in fresh alfalfa forages (30-50%), except in the case of P. laevigata, E. decaisnei, and C. sventenii ( Figure 4 ). Previous studies have provided evidence of the differences in the FA profiles among various legumes, grasses, and herbaceous forage species, in which grasses tend to have higher levels of the C18 omega-3 PUFA, ALA than legumes, and conversely, lower contents of the C18 omega-6 PUFA, LA (Clapham et al., 2005; Elgersma et al., 2013; Whetsell and Rayburn, 2022). In this context, both C18 PUFA have been described to exert several benefits to animal health and well-being, and must be incorporated through the diet as they cannot be synthesized de novo (Morales-Almaráz et al., 2011; Davis et al., 2022). However, the increasing concern on the excessive intake of omega-6 fatty acids in the occidental diet, have raised the interest to search for plants and forage particularly rich in omega-3 fatty acids, with higher n-3/n-6 ratios, both for human and farmed animals feeding (Almeida et al., 2019; Zárate et al., 2017). It should also be stressed that E. decaisnei was the only species containing substantial amounts of two fatty acids of increasing nutraceutical value: GLA (0.8% of total FA) and SDA (2.3 of total FA). In this regard, species of the Echium genus are known to be rich sources of GLA and SDA (Guil-Guerrero et al., 2000a, b), two physiologically valuable FA for mammals’ health (López-Martínez et al., 2004; Wang et al., 2022; Berti et al., 2007), that have shown promising results in the cultivation of aquatic species (Díaz-López et al., 2009, 2010) and terrestrial farmed animals (Navarro et al., 2008).

It is well established that in the feeding of ruminants, the most important lipids are those containing fatty acids bound to glycerol (i.e., phospholipids, glycolipids and triglycerides). Glycerolipids are the most abundant type of lipid in our forages species, represented by major phospholipids including PC, phosphatidylserine (PS), phosphatidylinositol (PI), PG, and PE, and galactolipids such as SQDG, MGDG, and DGDG (Liu et al., 2017). Among the NLs, DAG and TAG were prominent. Like FAs, the relative proportions of these compounds fluctuate among plant species and are strongly influenced by the nutrient levels in the plants (Liu et al., 2017). SQDG, MGDG, and DGDG have been described as compounds with anti-inflammatory and antithrombotic properties (Ibañez and Cifuentes, 2013). Similarly, the levels of PTS in the forage can be very important as it has been demonstrated to reduce cholesterol levels in the form of low-density lipoprotein (LDL) in humans, thereby reducing the risk of cardiovascular diseases and inflammatory processes (Ibañez and Cifuentes, 2013; Kendel et al., 2015; Simonetti et al., 2021).

The mineral composition of forage plant tissue shows significant variations, which could have direct implications on dietary formulations for ruminants. The levels of macro- and micro-minerals generally fall within the range that meets ruminants’ nutritional requirements. Species such as C. viminalis, L. lancerottensis, E. decaisnei, L. acerifolia, and P. laevigata had mineral contents that exceed the normally adequate values, particularly for Ca, Na, Fe and Cu (NRC, 2005). High Ca levels may cause health risks such as milk fever in dairy cows (Lean et al., 2006). The MTLs for dietary Ca in ruminants is 1.5% DM (NRC, 2005), yet E. decaisnei and P. laevigata recorded Ca concentrations of 22.7 and 34.2 g/kg DM, respectively, which can depress feed intake and milk yield. Sodium levels in some specimens reached 28 g/kg DM, potentially reducing appetite and increasing urine output, as well as the risk of kidney damage (ARC, 1980; NRC, 2007). Excessive dietary intake of Na is usually manageable as long as non-saline water is readily available for animals (Meyer et al., 1955; Nelson et al., 1955). Iron toxicity could lead to diarrhea and metabolic acidosis, in our case only L. lancerottensis exceeded the MTLs for this element (~ 500 mg/kg DM; NRC, 2005). Additionally, E. decaisnei exhibited Cu levels slightly above the safe threshold at 40 mg/kg DM (NRC, 2005), which could result in liver necrosis. Chronic Cu toxicity, particularly challenging to diagnose in cattle, involves symptoms such as hemoglobinuria and jaundice, typically more obvious in sheep, and is often indicated by high liver, kidney, and serum Cu concentrations (López-Alonso and Miranda, 2020).

Our results reveal strong interspecific variations in all parameters, indicating the potential benefits of using a diverse mix of species to optimize livestock nutritional intake. This is especially crucial in arid regions where plant species naturally exhibit unique nutritional profiles (Lee, 2018). Selecting appropriate forage species can optimize mineral intake and prevent deficiencies or toxicities that could impair livestock health and productivity.

Globally, most of the evaluated species have an adequate nutritional value, in addition to presenting a FA profile and certain lipid classes, such as PTS, beneficial for animal health and well-being. In particular, the species B. bituminosa, P. coerulescens, E. decaisnei, and C. sventenii could be promising candidates to be introduced as part of the diet of ruminants. The present work is limited to evaluating plant specimens under controlled conditions, so future research should focus on studying these species in field conditions, under different agronomic managements, that allow evaluating the influence of seasonal and environmental factors on the nutritional composition.