Temporal variations in leaf traits, chemical composition and in vitro true digestibility of four temperate fodder tree species

Additional keywords: fatty acids, feed quality, goats, in vitro digestibility, tannins.

Tree and shrub foliage is an important component of small ruminant diet in many parts of the world. It plays an essential role for browsing animals, especially where livestock systems are based on rangeland and grazable forestland exploitation (Leng 1997). Due to their deep root system, trees can maintain the high nutrient quality of their foliage also in dry periods, when herbage quality decreases (Papachristou and Papanastasis 1994). For this reason, the interest for fodder tree species is rising also in European temperate regions where drought periods are increasing in length and frequency over large areas due to global warming (Vandermeulen et al. 2018a). In this context, the so called ‘silvopastoral systems’, integrating fodder trees and grasslands, are spreading as a possible solution to face climate changes and to support a sustainable and resilient agriculture, especially in marginal areas (Novak 2014; Seidavi et al. 2018).

The knowledge of the importance of tree foliage for ruminant nutrition has led to several studies aimed at evaluating leaf forage productivity, chemical composition and digestibility (e.g. Boufennara et al. 2012). Proximate and fatty acid (FA) composition, phenolic concentration and digestibility have been largely assessed for various fodder tree species in several environments such as savanna (Tefera et al. 2008), tropics (Arigbede et al. 2012), subtropics (Prakash et al. 2009), American temperate region (Ricklefs and Matthew 1982), Mediterranean region (Kökten et al. 2012) and non-European alpine areas (Khanal and Subba 2001).

According to these and other studies, tree and shrub foliage is increasingly recognised as a high-quality feedstuff for small ruminants, either offered as sole feed or as dietary supplement (Leng 1997). Tree foliage has a superior mineral composition compared with that of some grasses (e.g. dry grasses), making foliage a significant source of minerals for the animals and rumen microbiota (Leng 1997). Some foliage has high protein contents and is, therefore, used as protein source to supplement other feeds, particularly when ruminants are offered roughages of a low nutritive value (Simbaya 2002; Kongmanila 2012). Tree and shrub foliage contains endogenous secondary compounds able to bind with proteins and protect them from microbial attack in the rumen afterwards. For this reason, foliage can represent a source of by-pass protein for the ruminant that, escaping rumen degradation, can be digested in the intestine, enhancing the animal protein status (Leng 1997). Plant secondary compounds are also able to alter the balance of microorganisms in the rumen (Vasta and Bessa 2012). Therefore, besides variations in FA, differences in concentrations of plant secondary compounds may also confer specific intrinsic sensory and chemical attributes to ruminant-derived dairy and meat products (Iussig et al. 2015a). Some plant secondary compounds contained in a large number of fodder tree species, such as phenolic compounds, are also characterised by potential anti-nutritional effects (Waghorn 2008). Among phenolic compounds, tannins can cause either adverse or beneficial effects on nutrient utilisation, health and animal production, in relation to their molecular characteristics and concentrations (Min et al. 2003). Particularly, tannins (depending on the type, chemical characteristics and amount ingested) can (1) increase intestinal escape protein availability, (2) reduce methane production, (3) defend against bloat and (4) wane gastrointestinal parasites (Mueller-Harvey 2006; Vandermeulen et al. 2018a). Francisco et al. (2015) reported also that inclusion of low concentrations (up to 16.3 g/kg DM) of condensed tannins (CT) in lamb diet could wax antioxidant properties in meat, improving its stability after storage. Similarly, milk from goats fed with low concentrations of hydrolysable tannins (HT) showed an increasing trend of conjugated linoleic acid isomers and oleic acid, as a result of an enhancement of Δ9-desaturase activity (Abo-Donia et al. 2017).

Temporal variations have also been analysed, since tree foliage productivity, chemical composition and digestibility can significantly change due to phenological advancement and climatic variations (Papachristou and Papanastasis 1994; Ndondo et al. 2004; Kökten et al. 2012).

In the last decades, extensive goat rearing has spread in European alpine areas, particularly in marginal areas (Battaglini et al. 2014; Álvarez‐Martínez et al. 2016). Here, due to their increasing presence and resistance to browsing and drought, extensive shrublands and small trees can provide abundant leaf fodder, which represents the basic component of goat diet. Particularly, a general increase of tree leaf consumption has been observed in late summer, when the availability and nutritive value of herbage decrease (Castro and Fernández Núñez 2016). Moreover, goats are well adapted to exploit low-quality forages due to their ruminal bacterial population, suitable for degradation of highly lignified material and resistance to tannin toxicity, resulting in an increased tree foliage digestibility (Manousidis et al. 2016).

Various research has been conducted on European temperate tree species (see, among others, Luske and van Eekeren 2015; Emile et al. 2016; Vandermeulen et al. 2018b). These plants have been a widespread feeding resource since Neolithic (Gómez García and Fillat 1984; Kühn et al. 2013; Hejcman et al. 2014), especially for goat nutrition (Leng 1997; Castro and Fernández Núñez 2016). However, the leaf biomass production of fodder tree species in these environments is poorly documented. Moreover, only a limited number of authors experienced the assessment of foliage production using allometric equations based on leaf traits (Forrester et al. 2017). In addition, there is a lack of knowledge on the FA profile of such tree foliage, even if this feature is of increasing interest for the possible effects on the nutraceutical properties of the derived dairy and meat products (Glasser et al. 2013).

The present study aimed to fill the gap of knowledge about the above-mentioned topics. For this purpose, we selected, in the western Italian Alps, the following four temperate tree species that are widespread on European uplands (San-Miguel-Ayanz et al. 2016): Acer pseudoplatanus L. (sycamore maple), Fraxinus excelsior L. (ash), Salix caprea L. (goat willow) and Sorbus aucuparia L. (rowan). These trees are among the browses most preferred and used by goats, either directly by browsing or by being fed after cutting (Guinier 1950; Thiebault 2005; Iussig et al. 2015a, 2015b). The objective of the work was to characterise their leaves across the whole vegetative season in terms of leaf traits (i.e. leaf length and biomass), proximate composition (i.e. dry matter (DM), ash, crude protein (CP), ether extract (EE), and fibre fractions), FA profile, phenolic compounds (i.e. total extractable phenols (TEP), non-tannin phenols (NTP), total tannins (TT; CT and HT) and in vitro true DM digestibility (IVTD).

The study was conducted at Oasi Zegna, located in the Piedmont Region, north-western Italy (45°40′N, 8°09′E), within the boundaries of Valle Sessera Site of Community Interest (SCI IT11300002). The study area is characterised by a suboceanic climate (Köppen’s classification: Cfb), with annual mean temperature of 7.3°C and precipitation of 1700 mm (mean value for the years 2002–2015 of the pluviometric station of Bielmonte). Dominant soils have originated from siliceous parent rock. In the study area, tree stands are mixed broadleaved populations, dominated by Fagus sylvatica L. (European beech) and Betula pendula Roth (silver birch), currently managed by occasional selective cutting.

For each investigated species (i.e. A. pseudoplatanus, F. excelsior, S. caprea and S. aucuparia), a group of four trees was selected in December 2014, during the vegetative dormancy, in a 500 × 150 m² area with homogeneous exposure (north–north-west) and elevation (1270–1320 m asl). Within each species group, the trees were located at a maximum distance of 125 m from each other and had a similar height, stem diameter, age, and no disease evidence. The selected trees were protected from animal browsing by using deer fences during the whole trial.

Leaf traits (i.e. leaf length and biomass) were recorded 12 times, while leaf sampling for laboratory analyses (i.e. chemical composition and IVTD) occurred four times during the trial, as detailed in Table 1.

Table 1.  Survey schedule
All surveys were conducted in 2015. DOY, day of the year; GDD, growing degree-days, expressed as the mean ± s.e. of the four tree species

Four sprouts at bud stage per tree were selected at a maximum height of 1.80 m (i.e. as far as goats can browse buds and leaves, assuming a bipedal stance; Iussig et al. 2015b). Each sprout was located in a different cardinal direction to avoid differences due to light exposure. The elongation of emerging leaves was monitored on each sprout, from the budburst of the earliest sprout until all trees reached the maximum vegetative phenological stage, according to the extended BBCH scale (Hack et al. 1992). At each survey date, following the survey schedule provided in Table 1, the phenological stage (vegetative and, when present, reproductive) of each individual was recorded. In addition, one sprout, comparable to the monitored one in terms of leaf number and development, was harvested and transported to the laboratory. The emitted leaves were then separated from the sprout and weighed. Leaf traits (i.e. leaf length and biomass) were computed by cumulating the values of all unfolded leaves per sprout in each survey date, accounting for a total of 16 replicates per species (four trees × four sprouts).

Four HOBO data loggers (Onset Corporation, Pocasset, MA, USA), placed near the centroid of each species group, recorded air temperature every 30 min for the whole trial. Heat units from 1 January 2015 to the end of the trial, expressed as cumulative growing degree-days (GDD), were calculated for each species group from daily air temperature, by cumulating all mean daily temperatures of >5°C (Prislan et al. 2013). For each sprout, the GDD corresponding to the 25%, 50%, 75% and 100% of the final leaf length and biomass were assessed by data interpolation.

Since all the selected sprouts ended leaf emission (i.e. 4 June 2015), one leaf sample per tree (i.e. a total of four replicates per species) was collected for four dates from June to August, as detailed in Table 1. Such time interval approximately corresponds to the goat browsing season in these grazable forestlands. A quantity of ~400 g of fresh leaves (including petioles and rachises; Hejcmanová et al. 2014) was harvested all around the canopy, bottom-up to 1.80-m height, so as to simulate goat browsing (Iussig et al. 2015b). All leaves damaged by pathogens (insects, fungi) were avoided. The samples were placed in sealed polyethylene bags, immediately stored at 4°C in a portable refrigerator and transported to the laboratory. In the laboratory, each sample was divided into two homogeneous aliquots of ~200 g each. The samples were freeze-dried (Freeze Drying Equipments, Criofarma, Torino, Italy), grounded with a cutting mill to pass through a 1-mm-screen sieve (Pulverisette 15, Fritsch GmbH, Idar-Oberstein, Germany) and used for further laboratory analyses.

AOAC (2000) procedures were used to determine DM (Method no. 930.15), ash (Method no. 942.05), CP (Method no. 984.13), acid detergent fibre (ADF) and acid detergent lignin (ADL; Method no. 973.18). The EE was determined following Method no. 920.39 of AOAC (2003). Neutral detergent fibre (NDF) was analysed according to Van Soest et al. (1991); α-amylase (Sigma Aldrich, Saint Louis, MO, USA), but no sodium sulfite was added and results were corrected for residual ash content. The proximate composition was expressed as g/kg DM.

The FA composition was assessed using a combined direct transesterification and solid-phase extraction method, as described by Alves et al. (2008). The FA methyl esters were separated, identified and quantified as detailed by Renna et al. (2014). The FA composition was expressed as g/kg DM.

The TEP and phenol fractions (NTP and CT) were determined using standard protocols, as detailed in Iussig et al. (2015a). The absorbance was recorded at 725 nm (TEP and NTP, expressed as gallic acid equivalents) and 550 nm (CT, expressed as leucocyanidin equivalents) using a UV–VIS spectrophotometer (Shimadzu UVmini-1240, Shimadzu Corporation, Kyoto, Japan). The TT were computed as the difference between TEP and NTP. The HT were estimated as the difference between TT and CT. The phenolic composition was expressed as g/kg DM.

Four slaughtered adult male Alpine goats were used as rumen fluid donors. The male goats belonged to an extensive dairy goat farm located in Rueglio (Torino, Italy). They were fed on alpine grazable forestlands rich in fodder tree and shrub species, including those under investigation in the present trial. Slaughtering procedures followed Council Regulation (EC) No. 1099/2009 of the European Union. They occurred four times (once per each male goat), every three weeks between October and December 2015. A representative sample of the rumen contents was collected from each animal into a thermos container, which was kept at 39°C using a warm water-containing cooler, and immediately transported to the laboratory, where the sample was filtered through four layers of cheesecloth under continuous flush of CO₂ at 39°C. The IVTD was determined according to the ANKOM Daisy procedure (ANKOM 2017). Following the slaughtering time schedule, four runs were performed. For each run, all 64 leaf samples were digested using the rumen fluid of one goat, accounting for a total of eight replicates per plant species and sampling date (two trees × four rumen fluids).

Leaf samples (0.25 g) were weighed into filter bags (ANKOM® #F57; pore size 25 μm, ANKOM Technology Corporation, Fairport, NY, USA), which were then heat-sealed. Rumen fluid was diluted into the buffer medium in a proportion of 1 : 4 (v/v). Thereafter, a volume of 2 L of buffered rumen fluid was transferred into four 5-L jars at 39°C under anaerobic conditions. Each jar, containing leaf samples and one blank, was placed in a revolving incubator (ANKOM Daisy^II digestion system, ANKOM Technology Corporation) at 39°C for 48 h, under continuous rotation. After incubation, the samples were rinsed with cold water and subjected to an extraction with NDF solution at 100°C for 1 h, so as to remove microbial debris and any remaining endogenous products.

To evaluate species precocity in leaf development at different phenological stages, one-way ANOVA was used to analyse for differences among the GDD corresponding to budburst and to 25%, 50%, 75% and 100% of the final leaf trait values. Moreover, to test for differences of final leaf length and biomass values among species at the end of the growing season, a one-way ANOVA was performed. For both analyses, the sprout was considered as the statistical unit and tree species as a fixed factor.

The relationship between leaf length and biomass was then explored with linear regressions, using a separate dataset for each species. The regression equations were based on the best-fitting model and the sprout was considered as the statistical unit.

General linear models accounting for repeated measures were performed on proximate composition, FA profile (namely, total FA (TFA), groups of FA, and main represented individual FA), phenolic composition, and IVTD of each tree species. Tree was considered as the statistical unit, species as a fixed factor, and sampling date as a repeated measure.

To test for differences among species during the vegetative season, a one-way ANOVA was performed on the same variables (i.e. proximate composition, FA profile, phenolic composition and IVTD) for each of the four sampling dates, using tree as the statistical unit and tree species as the fixed factor. Additionally, the temporal variations in terms of proximate composition, FA, phenolic composition and IVTD within the same species were tested performing a one-way ANOVA, with tree as the statistical unit and sampling date as a fixed factor.

Assumptions of normality and homogeneity of variance were checked with Shapiro–Wilk and Levene’s tests respectively. Variables that were not normally distributed were log-transformed before further statistical analysis; results are presented as non-transformed data. When normal distribution or homogeneity of variances were not met, even after log-transformation, the non-parametric Kruskal–Wallis test or Welch one-way ANOVA were used, respectively. When significant differences were found, Tukey’s, Steel’s and Tamhane’s post hoc tests were performed for ANOVA, Kruskal–Wallis and Welch one-way ANOVA, respectively.

All statistical analyses were performed using SPSS (Version 24.0, IBM SPSS Statistics, Armonk, NY, USA). Significance was set at P < 0.05.

A total of 479 leaves was monitored during the trial. Mean number of emitted leaves per sprout was eight for A. pseudoplatanus and S. caprea, nine for F. excelsior and five for S. aucuparia.

The vegetative season started at 135 GDD (105th day of the year), with S. caprea unfolding of the first leaves (Table S1, available as Supplementary material to this paper). Budburst occurred first (at a lower GDD) for S. caprea and S. aucuparia, followed by F. excelsior and then by A. pseudoplatanus (Table 2). Salix caprea and S. aucuparia developed earlier in terms of leaf length and biomass for most of the considered stages, followed by F. excelsior and, then, by A. pseudoplatanus.

Table 2.  Growing degree-days (GDD, mean ± s.e.) for each development stage
Data are provided for budburst and cumulative leaf length and biomass per sprout at 25%, 50%, 75% and 100% of the final recorded value of each species. Different letters within a row indicate significant differences (P < 0.001) among tree species

Differences among tree species in terms of final leaf traits were similar between leaf length and leaf biomass, with F. excelsior showing the greatest values, followed by A. pseudoplatanus, S. aucuparia and S. caprea (Fig. 1).

Fig. 1.  Cumulative leaf development. Increase of cumulative leaf length (a) and leaf biomass (b) per sprout in relation to growing degree-days (GDD) and day of the year (DOY, expressed as a mean for the four tree species). Grey highlights indicate leaf sampling dates for laboratory analyses. Error bars represent the s.e. of the means, while different letters indicate significant differences among the tree species. ***, P < 0.001.

For each of the four species, a relationship between leaf length and leaf biomass was demonstrated. A remarkable amount of variance was explained by the regressions leaf length versus leaf biomass (regression parameters and equations are shown in Fig. 2).

Fig. 2.  Allometric equations for leaf development. Changes in cumulative leaf biomass per sprout in relation to changes in cumulative leaf length per sprout.

The proximate composition was significantly different among the considered fodder tree species (Fig. 3). The average DM contents of S. caprea and S. aucuparia leaves were higher than those of A. pseudoplatanus and F. excelsior leaves. The ash content in S. aucuparia leaves was always lower than in the leaves of the other species. In F. excelsior, ash increased across time, displaying the highest values at the end of the season (Table 3). The highest EE values were observed in A. pseudoplatanus, followed by S. caprea, F. excelsior and S. aucuparia. Salix caprea was found to be the most fibrous species, with the highest concentrations of NDF, ADF and ADL. Fraxinus excelsior showed the lowest concentrations of ADF and ADL, while intermediate amounts of these fractions were found in A. pseudoplatanus and S. aucuparia leaves. Concerning NDF, F. excelsior showed values similar to those observed in S. caprea, whereas the lowest concentrations were found in A. pseudoplatanus and S. aucuparia.

Fig. 3.  Proximate composition of the four tree species: overall ranges. Provided data are expressed as g/kg dry matter (DM), except for DM, which is expressed in g/kg fresh matter. CP, crude protein; EE, ether extract; NDF, neutral detergent fibre; ADF, acid detergent fibre; ADL, acid detergent lignin. Error bars represent the s.e. of the means, while different letters indicate significant differences among the tree species. ***, P < 0.001.

Table 3.  Differences among tree species and sampling dates (expressed in day of the year, DOY) in terms of proximate composition
ADF, acid detergent fibre; ADL, acid detergent lignin; CP, crude protein; DM, dry matter; EE, ether extract; NDF, neutral detergent fibre. Uppercase letters indicate significant differences among tree species within a sampling date according to post hoc tests; significance levels are provided on the right side of each line. Lowercase letters indicate significant differences among sampling dates within a tree species according to post hoc tests; significance levels are provided on the bottom of each column. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., P ≥ 0.05

The statistical analyses performed on the temporal variations in the proximate composition highlighted a general variability related to phenology advancement, with an overall decrease in CP and an increase in fibre content (Table 3).

Regarding the FA profile of the leaves, the following 15 FA were detected in all samples: C12:0, C14:0, C15:0, C16:0, C16:1 c9, C18:0, C18:1 c9 (n9), C18:1 c11, C18:2 c9c12 (n6), C18:3 c6c9c12 (n6), C18:3 c9c12c15 (n3), C20:0, C20:1 c11, C22:0 and C20:4 c5c8c11c14 (n6). Among them, six FA, namely, palmitic acid (C16:0), stearic acid (C18:0), oleic acid (C18:1 c9), linoleic acid (C18:2 n6), α-linolenic acid (C18:3 n3) and γ-linolenic acid (C18:3 n6), comprised 92–97% of TFA and were then considered for further statistical analyses. The remaining FA were cumulated in the ‘Other FA’ group. In all tree species, C18:3 n3, C18:2 n6, and C16:0 were the main detected FA (Fig. 4). In S. aucuparia, the lowest total monounsaturated FA (MUFA, together with S. caprea), polyunsaturated FA (PUFA) and TFA concentrations were detected (Fig. 4). Concerning individual FA, S. aucuparia showed the highest C18:0 and the lowest C18:3 n6 and C18:3 n3 concentrations. Acer pseudoplatanus leaves had the highest C16:0 concentration, while F. excelsior showed the highest MUFA and C18:1 c9 concentrations. No differences were found for the ‘Other FA’ group among the species. The majority of individual FA and FA groups showed differences among the considered species during the whole trial. However, for total saturated fatty acids (SFA), C16:0, C18:3 n6, C18:3 n3, and ‘Other FA’, the differences among species tended to wane as DOY advanced.

Fig. 4.  Fatty acid profile of the four tree species: overall ranges. Provided data are expressed in g/kg dry matter. SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; TFA, total fatty acids; Other FA, C12:0 + C14:0 + C15:0 + C16:1 n9 + C18:1 c11 + C20:0 + C20:1 c11 + C22:0 + C20:4 n6. Error bars represent the s.e. of the means, while different letters indicate significant differences among the tree species. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., P ≥ 0.05.

Among sampling dates, FA profiles (Table 4) showed few significant variations. The concentration of C18:2 n6 was significantly (P < 0.001) higher at the first sampling date for all the species. Concerning F. excelsior and S. caprea, the sampling date influenced the concentration of total PUFA, which was significantly (P < 0.05 for both) higher at the first sampling date than at the third (F. excelsior) or the second and third sampling dates (S. caprea).

Table 4.  Differences among tree species and sampling dates (expressed in day of the year, DOY) in terms of fatty acid profile
DM, dry matter; MUFA, monounsaturated fatty acids; Other FA, C12:0 + C14:0 + C15:0 + C16:1 n9 + C18:1 c11 + C20:0 + C20:1 c11 + C22:0 + C20:4 n6; PUFA, polyunsaturated fatty acids; SFA, saturated fatty acids; TFA, total fatty acids. Uppercase letters indicate significant differences among tree species within a sampling date according to post hoc tests; significance levels are provided on the right side of each line. Lowercase letters indicate significant differences among sampling dates within a tree species according to post hoc tests; significance levels are provided on the bottom of each column. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., P ≥ 0.05

On the basis of their mean phenolic compound values, the fodder tree leaves were significantly different (box plots in Fig. 5). Acer pseudoplatanus and S. caprea showed higher concentrations of TEP, TT and HT than did F. excelsior and S. aucuparia. The concentration of CT was lower in A. pseudoplatanus than in S. caprea and S. aucuparia. The CT were not detected in F. excelsior. No significant differences in NTP concentration were detected among the species. Variations in the phenolic composition among the sampling dates were not significant or negligible (Table 5).

Fig. 5.  Phenolic compound concentration of the four tree species: overall ranges. Provided data are expressed in g/kg dry matter, as gallic acid equivalents, except for condensed tannins (CT), which are expressed as leucocyanidin equivalents. TEP, total extractable phenols; NTP, non-tannin phenols; TT, total tannins; HT, hydrolysable tannins. Fraxinus excelsior values for CT were null and are not presented. Error bars represent the s.e. of the means, while different letters indicate significant differences among the tree species. ***, P < 0.001; **, P < 0.01; n.s., P ≥ 0.05.

Table 5.  Differences among tree species and sampling dates (expressed in day of the year, DOY) in terms of phenolic compound content
CT, condensed tannins; DM, dry matter; HT, hydrolysable tannins; NTP, non-tannin phenols; TEP, total extractable phenols; TT, total tannins. Fraxinus excelsior values for CT were null and are not presented. Uppercase letters indicate significant differences among tree species within a sampling date according to post hoc tests; significance levels are provided on the right side of each line. Lowercase letters indicate significant differences among sampling dates within a tree species according to post hoc tests; significance levels are provided on the bottom of each column. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., P ≥ 0.05; n.d., not determined

The IVTD of leaves differed significantly among the fodder tree species (Fig. 6). On the basis of leaf mean digestibility values, F. excelsior and S. aucuparia were significantly more digestible than were A. pseudoplatanus and S. caprea.

Fig. 6.  In vitro true dry matter digestibility (IVTD) of the four tree species: overall ranges. Provided data are expressed in g/kg of dry matter. Error bars represent the s.e. of the means, while different letters indicate significant differences among the tree species. ***, P < 0.001.

Sampling date affected A. pseudoplatanus (highest value: third sampling date; lowest value: final sampling date) and S. aucuparia (highest values: first and final sampling dates; lowest value: third sampling date) digestibility. Fraxinus excelsior and S. caprea leaves were characterised by a more stable digestibility throughout the vegetative season (Table 6).

Table 6.  Differences among tree species and sampling dates (expressed in day of the year, DOY) in terms of in vitro true dry matter digestibility (IVTD)
DM, dry matter. Uppercase letters indicate significant differences among tree species within a sampling date according to post hoc tests; significance levels are provided on the right side of each line. Lowercase letters indicate significant differences among sampling dates within a tree species according to post hoc tests; significance levels are provided on the bottom of each column. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., P ≥ 0.05

The results obtained in terms of the relationships among leaf traits confirmed the leaf length as an easy recordable in-field plant trait useful to estimate leaf biomass of a tree or shrub species via allometric equations (Itô and Sumida 2017; Škɹma et al. 2018). The use of allometric equations can, thus, represent a simple and effective tool to estimate foliage production, as it is a non-destructive and time-affordable method (Elzein et al. 2011; Konôpka et al. 2015; Pajtík and Konôpka 2015).

Concerning leaf development across the vegetative season, budburst dates for F. excelsior and S. aucuparia were consistent with those obtained in Swiss environments by Basler and Körner (2012). The same authors reported a later budburst for A. pseudoplatanus. As outlined in Table 2, precocity of leaf development appeared inversely proportional to final values of both leaf traits; leaves started to develop earlier (i.e. at a lower GDD) in the species with shorter and lighter leaves at the end of the vegetative season (i.e. S. caprea and S. aucuparia).

These considerations concerning variations in leaf length and biomass across the vegetative season can be used as proficient tools for fodder tree management, since productivity is one of the basic criteria for using a particular tree species as feeding resource (Papanastasis et al. 1998). Indeed, in European alpine regions, F. excelsior trees growing close to farm households were often used to forage ruminants (Gómez García and Fillat 1984). Trees were managed with pollard practice and yearling sprouts were given to animals either as fresh (in summer) or conserved (in winter) forage (Thiebault 2005).

Concerning the proximate composition, the results obtained for DM and EE were consistent with those already reported by other authors (Ricklefs and Matthew 1982; Kökten et al. 2012; Emile et al. 2016). The CP contents of the four species were higher than the minimum level of 70–80 g/kg DM required for optimum rumen function and feed intake in goats (Van Soest 1994), with A. pseudoplatanus showing the highest values. The range of CP content in tree foliage was comparable to the one of herbage growing in similar alpine environments across the vegetative season (Revello-Chion et al. 2011; Luske and van Eekeren 2015; Ravetto Enri et al. 2017). Conversely, in dry to arid environments, other authors reported higher CP contents in the foliage of deciduous fodder tree species than in local grass or hay (Yayneshet et al. 2009; Gardiner et al. 2013). Lu et al. (2005) showed that ADF concentrations of 180–200 g/kg DM or NDF concentrations of 410 g/kg DM meet the nutritional requirements for dairy goats. Other research highlighted that increasing the ADF or NDF contents over 180–200 g/kg DM and 600 g/kg DM respectively, may limit feed intake due to rumen fill (Santini et al. 1992; Mertens 1994; Riaz et al. 2014). Therefore, all the studied species showed adequate average NDF amounts according to these authors. However, if the leaves of the four species were given as sole feed, A. pseudoplatanus, S. aucuparia, and (with a stronger effect) S. caprea, may reduce the feed intake, due their high average ADF contents (Santini et al. 1992; Lu et al. 2005). In this respect, F. excelsior may exert the mildest effect, being slightly over the ADF limit. Other studies provided results sometimes different for the proximate composition, even if leaf samples were collected from the same tree species, in similar environments, and at comparable dates. Regarding F. excelsior, Emile et al. (2016) found lower DM and NDF and, together with Luske and van Eekeren (2015), comparable CP contents, while Masson et al. (1980) reported lower DM and higher ash and CP contents. Hejcman et al. (2016), instead, reported similar ADF but lower NDF, ADL and ash concentrations for S. aucuparia leaves in Iceland.

The temporal fluctuations in the proximate composition can be commonly observed in many fodder-tree species with similar trends, also in other than European environments (Papachristou and Papanastasis 1994; Yayneshet et al. 2009; Kökten et al. 2012), and in fresh grass from semi-natural pastures (Bovolenta et al. 2008).

The concentrations of the main FA were consistent with the findings for other fodder tree species and with grassland fodder values, also from environments other than the alpine ones (Ndondo et al. 2004; Uysal et al. 2015; Ravetto Enri et al. 2017). However, the concentration of C18:3 n3 in S. aucuparia leaves was lower than that observed in herbage collected in the same environment (Iussig et al. 2015a). Rosenqvist and Laakso (1991) reported a FA profile and main FA percentages comparable to those of the present study for S. caprea in northern Finland. Concerning A. pseudoplatanus, F. excelsior and S. aucuparia, no data on FA profile of leaves are currently available in literature. Other studies conducted in dryer environments have found amounts of some individual FA and FA groups similar to those of the fodder-tree species studied, while higher SFA and lower PUFA contents have been reported (Ndondo et al. 2004; Uysal et al. 2015). According to the obtained results on FA profile, A. pseudoplatanus, F. excelsior and S. caprea leaves could be used as feedstuff for goat nutrition, also in late summer, when quality (and particularly C18:3 n3 concentration) of grassland species decreases (Revello-Chion et al. 2011; Uysal et al. 2015; Ravetto Enri et al. 2017). Indeed, the lipid metabolism in the rumen and in the mammary gland can be affected by such decrease in the quality of grassland species, usually resulting in higher concentrations of hypercholesterolaemic SFA and lower concentrations of beneficial FA (i.e. vaccenic acid, rumenic acid and n3 FA) in the derived dairy and meat products (Ferlay et al. 2017).

The TEP, NTP, TT, CT and HT concentrations reported for other fodder tree species from drier regions were similar or even higher than those for the species here studied (Ammar et al. 2008; Kökten et al. 2012; Gemeda and Hassen 2015). Overall, CT concentrations were lower than the minimum threshold of 20 g/kg identified by Min et al. (2003) for ruminant nutrition. According to these and other authors, such CT values should have a limited negative effect on ruminal digestion, especially for goats. In fact, goats prefer tannin-rich foliage, minimising CT consumption through modification of the feed-intake pattern (Mkhise et al. 2018), and produce tannin-binding proteins in saliva, overcoming the negative impacts on digestibility in the rumen (Lamy et al. 2011).

Interesting and innovative results were shown by IVTD of the four species, which was similar to that of some Mediterranean tree and shrub species, either deciduous or evergreen (Ammar et al. 2005, 2008). Digestibility in rumen environment is a complex process, partially still unknown, in which several factors interact with each other (Aguerre et al. 2016; Seidavi et al. 2018). Chemical composition (i.e. mainly fibre fractions (NDF, ADF, ADL), FA profile (particularly PUFA), and phenolic compounds (above all, CT and HT) exert a crucial multi-layered influence on microbial activity (Waghorn 2008; Boufennara et al. 2012; Ferlay et al. 2017). The NDF constituents, being mainly represented by hemicellulose and cellulose, are the dominant feed fraction for grazing ruminants, supplying energy for maintenance and production. However, fibre has a slower ruminal passage rate than do other dietary elements because their chemical conformation is difficult to cleave or is even indigestible, resulting in a filling effect over time (Mertens 1994). The inverse relationship between the amount of fibre and digestibility rate reported for other fodder tree species in the Mediterranean region (Ammar et al. 2004; Boufennara et al. 2012; Gemeda and Hassen 2015) can be observed also in the present trial. This means high NDF, ADF and ADL and low IVTD in S. caprea, low NDF and high IVTD in S. aucuparia, and low ADF and ADL and high IVTD in F. excelsior. Interestingly, these relationships were not displayed in other cases, such as in A. pseudoplatanus (low NDF and low IVTD), F. excelsior (high NDF and high IVTD) and S. aucuparia (high ADF and ADL and high IVTD), as observed also in previous studies (Van Soest 1994; Riaz et al. 2014; Gemeda and Hassen 2015). This could be attributable to the additional influence of the other chemicals on the microbiome. The interaction of fibre with different PUFA concentrations could have also played an important role in digestibility, since PUFA can inhibit and/or alter the microbial activity and the biohydrogenation pathways in the rumen (Ferlay et al. 2017). For this reason, the high PUFA concentrations (especially concentrations of C18:3 n6, C18:3 n3 and C18:2 n6) in A. pseudoplatanus and S. caprea leaves, together with their high ADL contents, could have, consequently, contributed to the reduction of the degradation capacity of goat microbiome and IVTD. Indeed, ADL represents a physical barrier between digestible cell content and bacteria (Waghorn and McNabb 2003). In contrast, IVTD was promoted in F. excelsior because of the low amount of ADL (despite the high PUFA concentration) and in S. aucuparia because of the low PUFA concentration (despite the high ADL concentration). Concerning phenolic compounds, CT may have affected digestion mechanisms (i.e. by binding to dietary protein, fibre and enzymes and by reducing microbiome degradation activity; Makkar 2003; Mueller-Harvey 2006; Waghorn 2008). The chemical structure of CT can differ among tree species and can exert different binding activities, regardless of their concentration, resulting in different IVTD (Hove et al. 2001; McSweeney et al. 2001). In the present case, the lower IVTD associated with lower CT concentrations observed in A. pseudoplatanus leaves may be due to a higher binding activity of the CT molecules. The higher IVTD associated with the higher CT concentration in S. aucuparia leaves is probably due to a lower CT activity. The highest HT concentrations found in A. pseudoplatanus and S. caprea leaves (approximately 6-fold those of F. excelsior and S. aucuparia) may have negatively influenced the digestibility of these leaves, because of the potential release of compounds toxic for bacteria in the rumen (Makkar 2003; Waghorn 2008). Therefore, the high CT (with a probable high binding activity) and HT concentrations in S. caprea leaves could have combined and lowered the digestibility. Conversely, F. excelsior leaves showed a phenolic profile more suitable for ruminal digestion, since they were totally free of CT and had low HT concentrations.

The four species showed a consistent stability of IVTD throughout the vegetative season. This is in contrast with the results previously obtained for Quercus coccifera L. (an evergreen Mediterranean tree species), which showed a progressive decline of leaf digestibility while dry season advanced. However, it should be pointed out that evergreen species generally show lower quality at late phenological stages (Roukos 2016).

Additional research on productivity and chemical and digestibility features of other tree portions (such as sprouts) or other temperate tree species selected by goats in extensive European farming systems (e.g. Alnus viridis (Chaix) DC., B. pendula, Corylus avellana L., F. sylvatica, Quercus sp., Tilia cordata Mill., Ulmus spp.) would be advisable (Hejcman et al. 2014; Castro and Fernández Núñez 2016; Vandermeulen et al. 2018a). Moreover, the preferences by goats for the investigated tree species should be taken into consideration. Future trials should address topics such as the palatability of tree and shrub leaves and sprouts, the related goat diet selection and the effects of tree and shrub leaves and sprouts, used as sole feed or as supplement, on voluntary DM intake by goats, on digestion mechanisms (especially on protein and organic-matter digestibility and on total tract digestion), even through in vivo studies, and on goat milk yield, gross composition and FA profile.

The present study has provided original data with a comprehensive approach on leaf traits, and chemical features of four tree species widespread in Europe.

In situ, chemical, and in vitro measurements can be considered useful tools in initial screening studies to rank forages according to their nutritive quality. The present results have highlighted that the investigated species, due to their foliage production, proximate composition, FA profile, phenolic composition and digestibility, can represent good-quality feedstuff for goat nutrition. This is especially true in late summer when herbage quality decreases, particularly in terms of CP and the FA profile.

These species can supply goat feedstuff either for direct browsing or as fresh or dried fodder (Bestman et al. 2014), also depending on foliage production. Trees could be managed through copping and maintained at a maximum height of 1.80 m to facilitate direct browsing by goats. Alternatively, they could be pollarded to prevent damages from wild ungulates and their branches could be periodically cut for fresh- or dry-fodder delivery. Otherwise, new specific plantations for goat browsing could be established. In this case, the species of the stand and the relative abundance should be selected, taking into account goat feeding preferences as well as leaf nutritive values of the tree species.

The differences among the four selected species were remarkable, even if weakly related to phenological stage advancement, especially when considering the FA and phenolic composition. The high digestibility of F. excelsior leaves throughout the vegetative stage could be due to the positive influence of the low concentrations of ADL and phenolic compounds, irrespective of the low CP concentrations. This species could be regarded as potentially highly-nutritive feedstuff, also improving the quality of goat dairy products for human nutrition, as confirmed by its traditional use in European mountain areas (Gómez García and Fillat 1984; Thiebault 2005). Also S. aucuparia leaves showed similar digestibility values; however, the low CP and PUFA concentrations (especially C18:3 n3 concentration) can partially reduce the interest for this species as a feeding resource for goat dairy and meat products with healthy properties. Conversely, a lower digestibility was found in S. caprea samples, which was also the species with the less productive sprouts, especially at the beginning of the vegetative season, as a consequence of its high phenolic and ADL concentrations. Nevertheless, the FA profile of this species highlighted high C18:2 n6, C18:3 n6 and C18:3 n3 concentrations, generally being recognised as main lipid precursors in ruminant metabolism for the synthesis of FA considered beneficial to human health (e.g. vaccenic acid, rumenic acid and n3 FA). A modest digestibility was recorded for A. pseudoplatanus leaves, despite their high CP and medium–low ADL concentrations, which were probably insufficient to contrast the high concentrations of phenolics (TEP, TT and HT) and the high binding activity (CT), even if they displayed a good level of the same ‘healthy precursors’ as reported for S. caprea foliage.

The authors declare no conflicts of interest.