Ethical considerations

The experimental procedures followed guidelines by Instituto de Ciencias Agropecuarias y Rurales (ICAR) of Universidad Autónoma del Estado de México (Mexico), and were institutionally approved (DICARN-1319).

The study followed an adaptive participatory rural research approach methodology validated in different countries (^{Kraaijvanger and Veldkamp 2015}; ^{Flor et al. 2017}), where the main goal is to find solutions to problems faced by farmers through collaborative work between farmers and researchers promoting innovation in the utilization of local resources (^{Hauser et al. 2016}; ^{Aare et al. 2021}).

Location of the study

The on-farm experiment, following the participatory livestock technology development approach, was performed in Yuriria, state of Guanajuato 20° 12' 51" N and 100° 08' 19" W in central Mexico. The region is a plateau with a mean altitude of 2000 masl, a semi-warm sub-humid climate with rains in summer and a dry season in winter, and a mean rainfall of 600-700 mm/year.

Sunflower-chickpea hay

The sunflower intercropped with chickpea was sown after a corn crop on 1.0 ha on September 2^nd 2018, towards the end of the rainy season and harvested on December 20^th at 109 days post sowing. The sowing rate was 10 kg seed/ha (estimated density: 80,000 plants/ha) of sunflower (cv. Tiacaque) and 75 kg/ha (90,000 plants/ha) of a local landrace chickpea seed. The distance between rows was 80 cm, and distance between plants was 15 cm. At harvest, the sunflower was at the R8 phenological stage (^{Schneiter and Miller, 1981}) and chickpea at the R7 stage (Herrera-Flores et al., 201), and both were left to dry in the field (for 96 h), after which all material was ground through a 2.5 cm sieve with a tractor-driven hammer mill. The final proportion of each forage was 60 sunflower and 40% chickpea. Crop yield was approximately 7.5 tonnes DM of field-cured dried sunflower-chickpea hay (SFCP).

Animals

The experiment was conducted with 28 multiparous Saanen dairy goats, randomly divided into two groups of 14. Goats were in late lactation (195±11 days in milk).

Initial conditions were as follows: 58±3.2 kg live weight, 1.2±0.2 kg daily milk yield, and 6 body condition score (BCS) on a scale from 1 (very thin) to 9 (very fat) (^{Aumont et al. 1994}). Analysis of variance was performed on goat live weight before starting the experiment to determine if there were significant differences between both groups; no differences were observed; therefore, a completely randomized design was deemed adequate.

The animals were housed in open pens (one pen per treatment group) with dirt floor and roofed in one third of the area, covering above the feeding troughs, which ran the whole width of the pen. The feeding area had a 1.0 m concrete floor along the feeding troughs. Each pen was 7.5 x 8.0 m, providing ample space for each goat group (4.3 m²/goat).

The forage treatments and lucerne hay were fed to each treatment group collectively, and the concentrate supplement fed individually to each goat at milking. Goats were milked by hand once a day at 7:00 h.

Two weeks prior to the commencement of the experiment, all goats in the flock received an adaptation diet that included the concentrate supplement to be used during the experiment at 400 g/goat/day, and 200 g/goat/day of lucerne hay, and 600 g/goat/day of a 50:50 (fresh basis) mixture of corn straw and sunflower-chickpea hay, plus minerals and water ad libitum.

The experimental treatments MZST or SFCPT were implemented in each of the groups at the beginning of the trial. Milk yield for each goat was recorded daily, and a 20-ml milk sample from each goat was daily taken and refrigerated (3 °C), with a weekly pooled sample analyzed for milk composition. Mean values were used for analyses.

Live weight (LW in kg) was recorded on the first and last day of the experiment with a hook digital scale, and body condition score (BCS) was assessed simultaneously by the same observer during both weight recordings.

The experiment took place from June 10^th to July 9^th, 2019 (30 days). The length of the experiment was decided in consultation with the participating farmer taking into consideration his flock management needs as well as sufficient time to evaluate the effect of the experimental feeding treatments. The response to a diet by dairy ruminants can be observed after a few days of the diet introduction, and short experimental periods in feeding experiments for dairy goats are well validated and accepted in the scientific literature, as demonstrated by Charpentier et al. (2019). Thus, a 30-day experiment was appropriate to detect potential effects of the treatments.

Diets

Treatment MZST, the conventional diet (control) consisted of concentrate, lucerne hay, and corn straw; whilst the experimental treatment (SFCPT) was concentrate, lucerne hay, and sunflower-chickpea hay replacing the corn straw. The forages (lucerne hay, corn straw and sunflower-chickpea hay) were ground with a hammer mill through a 2.54 cm screen.

Diets provided 67% forage (lucerne hay plus corn straw or sunflower-chickpea hay) and 33% concentrate, with water and a commercial mineral plus vitamin mix for goats provided ad libitum.

The concentrate supplement was a homemade mix of commercial compound concentrate for dairy goats with 22% CP (65% fresh weight), ground white corn grain (20% fresh weight), and ground sorghum grain (15% fresh weight), with 18% final CP content for the mixture.

The concentrate was provided individually to each goat at milking, and the ground lucerne hay and experimental forages (corn straw or sunflower-chickpea hay) were thoroughly mixed before allocation to the goats to minimize selection and offered collectively to each experimental group of goats (Table 1), offering half of the forage ration in the morning and the other half in the afternoon. Every morning, refusals were collected before a new allocation of forages, then weighed, and a sample taken for determination of dry matter content.

Samples of the experimental diets (forage plus concentrate) were taken daily before allocated, and samples of the individual ingredients taken weekly for chemical analyses.

Both treatments met the requirements for energy and protein of milking goats (NRC, 2007). However, the diets were not isoproteic nor isoenergetic since the objective of the study was not to compare the two forage sources (MZST or SFCP) on a nutritional basis. The objective was in line with what ^{Rao et al. (2015}) described as the need to improve feeding strategies for small-scale livestock systems based on quality forages to increase productivity to better contribute to sustainable livelihoods and to meet world challenges for livestock production.

Chemical composition of feeds

Samples were dried at 55 °C for 72 hours in a draught oven and then milled through a 1 mm sieve. Samples of all feeds were analyzed for ash by incineration at 550 °C (AOAC, 1990) to determine organic matter (OM), crude protein (CP) by the Kjeldahl method (N x 6.25), and ether extract (EE), following AOAC (1990) procedures. Neutral detergent fiber (NDF), and acid detergent fiber (ADF) following ^{Van Soest et al. (1991}). In vitro dry matter digestibility (IVDMD) followed the method of ^{Tilley and Terry (1963}), with rumen fluid obtained through a stomach tube from five donor female goats.

Milk sampling and analyses

Milk yield for each goat was recorded daily. A 20-ml sample of milk from each goat was taken daily, pooled by treatment, and refrigerated (3 °C) before being analyzed for milk composition, determined with an automatic ultra-sound milk analyzer (Lactoscan MCC).

Samples were pooled by treatment due to restraints in the on-farm adaptive experiment with the collaborating farmer as the milk analyzer was facilitated by the farmers’ association only once per week. This illustrates one of the limitations of the adaptive research approach followed in this experiment, with the goal of adapting a given technology to local conditions through in-farm experiments (^{Flor et al. 2017}).

Adaptive research not only applies existing knowledge; research with farmers investigates how results adapt to the farmer objectives and productive conditions to facilitate dissemination and adoption of results (^{Stroup et al. 1993}).

Compared to experimental centers, on-farm research with small-scale farmers faces limitations like small land holdings, small herds or flocks, and management constraints to carry out the experiments; but these are offset by the benefits of participatory research, as stated by ^{Stroup et al. (1993}).

Statistical analyses

Variables for goat performance and milk composition were analyzed with ANOVA (Minitab 14 statistical software) following a completely randomized design with the following model:

Y_ij= µ_i + t_j e_ij (1)

Where: µ= general mean, t= effect of treatments (i = 1, 2) and e= residual variation. Analyses for variables were on the mean values. Significant differences were declared at p≤0.05.

Chemical composition of feeds

Sunflower-chickpea hay combines an oil plant and a legume, resulting in a forage high in crude protein content and high ether extract (lipid) (Table 2). The total diet (SFCPT) was thus 33% higher in CP, 75% higher in EE, and 15% higher in IVDMD than the control diet (MZST).

Animal variables

Milk yield in SFCPT with sunflower-chickpea hay was significantly higher (P<0.001) at 10% more than the control diet (MZST), and there was a significant increase (P<0.05) in protein, total solids, and non-fat solids content over the conventional ration (MZST) (Table 3).

There were no significant (P>0.05) changes in body condition score, but the SFCPT treatment showed a significant (P<0.009) 1.02 kg higher live weight gain during the experiment than that for goats on the MZST treatment.

Total dry matter intake was 11.2% higher (P<0.001) in the conventional MZST ration than in the SFCPT treatment, and refusals were also significantly (P<0.001) higher at 0.24 kg DM/goat/day in the SFCPT treatment compared to 0.12 kg DM/goat/day in MZST.

Chemical composition of diets

Corn straw has been described as a roughage high in fiber, with low digestibility and poor protein content (^{Fuentes et al. 2001}), as observed in this experiment.

On the other hand, sunflower-chickpea hay had better nutritive quality, with less NDF and ADF compared to corn straw, and a higher content of CP and EE resulting in 28% higher IVDMD (Table 2).

Ether extract and CP contents of sunflower-chickpea hay were lower compared to other reports evaluating sunflower as a forage source (^{Rodrigues-Gandra et al. 2017}; ^{Sainz-Ramírez et al. 2020}), in spite of being harvested at similar stage of maturity; but fiber contents were lower to values reported for sunflower silage by ^{Guney et al. (2012}) and ^{Aragadvay-Yungán et al. (2015}).

Sunflower forage is high in fiber and has low digestibility (^{Demirel et al. 2009}) compared to other forages. Forages with high fiber content reduce ration digestibility, synthesis of rumen microbial protein and energy supply (^{Gottardo et al. 2017}). The association of sunflower with chickpea, however, resulted in lower fiber content than sunflower in monoculture (Table 2) given that chickpea forage is low in fiber (^{Herrera-Flores et al. 2019}). Nonetheless, in spite of its lower fiber content, the experimental SFCPT ration showed lower dry matter intake, perhaps due to a higher concentration of lipids in the diet (28.4 g/kg DM in MZST vs. 113.7 g/kg DM in SFCPT).

Milk yield and composition

A report by ^{Sanz-Sampelayo et al. (2007}) suggested that increased dietary lipids do not affect net energy intake or milk yield in goats, but positively affect milk fat content, which was not observed in the current experiment. Dietary inclusion of energy foods increases microbial protein synthesis in the rumen as well as propionic acid concentration, thus increasing milk production (^{Hills et al. 2015}; ^{Vicente et al. 2017}). In spite of lower DM intakes, this effect was observed in the SFCPT ration where sunflower seeds in hay with high lipid content (^{Rodrigues-Gandra et al, 2017}) may have favored the increase of milk yield in the SFCPT diet.

Contrary to reports by ^{Chilliard et al. (2003}), who did not observe differences in milk yield when supplementing the diet of milking goats with 3% fish oil in mid and late lactation, milk yield in the present study was higher in the SFCPT treatment.

Diet composition does influence milk composition in dairy goats due to factors such as energy intake (^{Sanz-Sampelayo et al. 2007}; ^{Sanz-Ceballos et al. 2009}). High-energy feeds increase fat and protein content in milk (^{Kalač and Samkora, 2010}), although only the increase in protein was significant in the present study, with no significant effect on milk fat.

Milk fat is the component most susceptible to changes in the diet and of great importance for assessing milk quality. ^{Kalač and Samkora (2010}) and ^{Gottardo et al. (2017}) found a positive correlation between dietary protein and fat content in milk. Nevertheless, no significant differences between treatments for milk fat content were observed in the present study even though SFCPT had over 47% higher CP content than MZST.

The effect of supplementary fat in the diet on fat secretion in milk has a minor effect in mid and late lactation compared to early lactation (^{Chilliard et al. 2003}) due to anabolic enzymes effects on the adipose tissue involved in de novo milk-fat synthesis, as well as to lipase lipoproteins activity after the peak of lactation (^{Soryal et al. 2004}; ^{Deshwala et al. 2020}).

Results from the present experiment were similar in terms of milk fat and protein to a report by ^{Zucali et al. (2007}) supplementing sunflower seeds to dairy goats at the peak of lactation; whilst ^{Arco-Pérez et al. (2017}) reported higher milk fat and protein contents in goats after parturition when including sunflower oil in the diet.

It was also notable that goats on SFCPT showed a liveweight gain 38.6% higher than on MZST, although no change in body condition was observed.

In conclusion, our results show that feeding dairy goats with a SFCPT diet based on sunflower-chickpea hay increases milk yield and live weight gain.