The study was conducted on smallholder dairy cattle farms in two ecological regions of Benin (arid Sudanian and semi-arid Sudano-Guinean regions). The arid Sudanian region (7° 30-9° 30 N) is characterized by annual rainfall of 800 to 1,100 mm and a vegetation growing season of 145 days, while the semi-arid Sudano-Guinean region (9° 30-12° N) receives annual rainfall of 1,100–1,300 mm and a vegetation growing season of 200 days. The villages of Founougo, Goumori, Bagou and Sori were selected in the arid Sudanian region and the villages of Ouénou, Sirarou, Kika and Béterou in semi-arid Sudano-Guinean region for twelve-month farm monitoring (Figure 1). Dairy farmers of these regions are known for their practice of adaptation to CC based on the integration of animals with trees or shrubs. The herds of farmers who participated in this study were those who had at least two cows at the early lactation and who had participated in the previous study initiated by Assani et al. (2023). The previous study identified four types of smallholder dairy cattle farms namely traditional silvopasture (TSS); Improved Silvopasture (ISS); Small-scale Agrosilvopasture (SIAS) and Large Integrated Agrosilvopasture (LIAS). The characteristics of the four identified type of farmers in dryland areas of Benin are presented in Supplementary Table S1 (Assani et al., 2023). In TSS, farmers had an average of 26 Tropical Livestock Unit (TLU). They did not own land and they used trees and shrubs from the rangelands. They used native trees (Khaya senegalensis, Afzelia Africana and Pterocarpus erinaceus) as feed supplements for dairy cattle in natural rangelands during dry season (Figure 2). In ISS, farmers had a mean of 11 TLU per herd. They owned land (4.2 ha) and associated livestock, forage plants (Panicum maximun C1, Pennisetum purperium and Brachiaria ruziziensis) and trees/shrubs plantation (Khaya senegalensis, Afzelia Africana, Pterocarpus erinaceus, Cajanus cajan, Gliricidia sepium, Acacia auriculiformis, Leucaena leucocephala, Stylosanthes guianensis Mucuna pruriens) in pasturelands (Figure 3). Fodder trees are utilized throughout the year. The SIAS farmers adopted the integration of agriculture, ruminant livestock and trees. They owned a small area of land (3.0 ha) and low size of herd (6 TLU). SIAS farmers used native trees/shrubs fodder (Cajanus cajan, Acacia auriculiformis, Leucaena leucocephala, Stylosanthes guianensis, Mucuna pruriens) and crop residues (Maize stover, rice straw, sorghum straw and Cowpea haulms) as feed supplements for dairy cattle during dry season (Figure 4). In LIAS, farmers tilled large portions of land (9.3 ha) and had a mean of 18 TLU. Leguminous fodder trees (Khaya senegalensis, Afzelia Africana, Pterocarpus erinaceus, Cajanus cajan, Acacia auriculiformis, Leucaena leucocephala, Stylosanthes guianensis, Mucuna pruriens) and crop residues (Maize stover, rice straw, sorghum straw and Cowpea haulms) are utilized to feed ruminants during the dry season (Figure 5).

Monitoring of ten (10) farms per type of farming system identified in dryland areas of Benin i.e. 40 farms was carried out from March 2021 to January 2022. Daily milk yield data were collected from 113, 103, 102 and 129 cows respectively on the TSS, ISS, SIAS and LIAS farms, giving a total of 447 dairy cows. On each farm, dairy cows were selected according to breed (Borgou, Crossbreed, Gudali and White Fulani) and lactation rank (lactations 1, 2, 3, 4, 5 and more). Table 1 gives more details on the number of cows monitored during this study in each type of farming system according to breed and lactation number. The chosen cows were given an identification number or name to facilitate monitoring. All animals lost (mortality, sale, etc.) during monitoring were not counted in this number. Milk was collected once a week during 305 days of lactation. Hand milking is carried out twice a day twice a day, in the morning at 7 a.m. before leaving for pasture, and in the evening at 6 p.m. when return from pasture. Once the milk had been collected, the quantity was measured on a weighing balance and recorded on the data collection sheet. Daily milk yield is the sum of the morning and evening milk collections.

The demographic parameters in TSS, ISS, SIAS and LIAS farming systems are studied on the basis of information collected on the movement of cattle on the farm during the year (initial cattle numbers, birth, purchase, mortality, sale, donation, exchange, slaughter and final cattle numbers), farm cattle composition (female calves, heifers, lactating cows, dry cows, male calves, subadult bulls and reproductive bulls), reproduction parameters (advanced gestation, abortion and calving) and viability-mortality parameters (stillbirths, age-specific mortality and live offspring at weaning). The 40 farms monitored for milk production were also used to collect one-year demographic data (from March 2021 to February 2022). Farm monitoring for demographic parameters data collection was done once a week. During each visit, all events concerning demographics parameters (number of cows present, number of cows in lactation, number of births, number of abortions, number of dead animals, number of animal entries and exits) were recorded using monitoring sheets. The animals were identified beforehand taking into account the name of the animal or identification number, the breed and the date of birth. Demographic parameters were calculated using the formula proposed by Lhoste (2001):

Reproductive rates

Rate of abortions = Number of abortions * 100/Number of cows

Calving rate = Number of calving * 100/Number of cows

Fertility rate = Number of alive born calves * 100/Number of cows

Mortality rates

Stillbirth rate = Number of stillborn calves * 100/Number of calves born

Pre-weaning mortality rate = Number of pre-weaning dead calves * 100/number of calves born alive

Global mortality rate = Number of dead animals * 100/Average herd size

Numerical weaning productivity = number of alive weaned calves * 100/Number of cows

Management rates

Offtake rate (OR) = Number of exploited animals * 100/Average size of the cattle herd

Intake rate (IR) = Number of imported animals * 100/Average size of the cattle herd

Net offtake rate = OR – IR.

To know more about the composition of the most fodder trees/shrubs, forage plants and crop residues used as feed supplements for dairy cattle in each farming system during dry season, the feed samples were collected to determine the chemical composition. Leaves samples were collected from several branches in the canopy. The samples (500 g) for each forage and crop residues were oven-dried at a temperature of 60°C for 72 h for dry matter, then ground and sieved. Analysis was carried out using the AOAC method (AOAC, 1990). The samples were analyzed in triplicate in accordance with approved methods (AOAC, 1990) to determine Dry matter (DM), ash, crude protein (CP) and ether extract (EE) (AOAC procedure 2001.12, 930.05, 978.04 and 920.39 for DM, ash, CP and EE, respectively). The fiber contents (NDF and ADF) were analyzed using the Velp Scientifica™ FIWE 3 Fiber Analyzer according to Van Soest et al. (1991) methods. In vitro evaluation of DM digestibility was based on the two-step technique of Tilley and Terry (1963).

To assess the effects of different small-scale agroforestry systems (TSS, ISS, SIAS and LIAS) on daily milk yield, we used the General Linear Model (GLM) in R.4.2.1 software (R Core Team Development, 2022). The model used was:

Where, Yijk = Response variable (Milk yield);

μ = overall mean;

Si = fixed effect of the silvopastoral system (4 classes; TS = 1, 2, 3, 4);

Pj = fixed effect of season (P = dry season, wet season; 2 classes) or fixed effect of lactation number of cows (P = L1, L2, L3-4, L ≥ 5; 4 classes);

Rk = fixed effect of cow breed (R = Borgou, Crossbreed, Gudali, White Fulani; 4 classes);

Si*Pj = interaction of the silvopastoral system and the season;

Si*Rk = interaction of the silvopastoral system and the cow breed;

Eijk = error.

Average milk yields for each type of farming system were compared using the least significant difference (LSD) test with agricolae package (de Mendiburu, 2021) to see if there was a significant difference at the 5% level.

To understand how different small-scale agroforestry systems impact the demographic parameters, we compared the following variables between farming system (TSS, ISS, SIAS and LIAS): reproductive rates (rate of abortions, calving rate and fertility rate), mortality rates (stillbirth rate, pre-weaning mortality rate, global mortality rate and numerical weaning productivity) and management rates (offtake rate, intake rate and net offtake rate). As the data for each of these variables were not normally distributed, the Kruskal-Wallis test was used for these analyses.

Table 2 shows the chemical composition and nutritional value of tree/shrub leaves, forage plants and crop residues used as feed supplements for dairy cattle in dryland areas of Benin. The crude protein values of tree/shrub leaves, forage plants and crop residues range from 2.80% (Sorghum straw) to 23.80% (C. cajan). Crude protein values for tree/shrub leaves, forage legumes (S. guianensis and M. pruriens) and Cowpea haulms are above 10%. The lowest and highest NDF values were obtained in G. sepium (39.46%) and Sorghum straw (74.60%), respectively. Dry matter digestibility values ranged from 42.32 to 63.20%, respectively.

The daily milk yield was higher in cows from farms adopting ISS compared to those practicing LIAS, SIAS and TSS. In fact, the animals from ISS farms benefit from a supplementation of woody fodder at any time of the year, which allows them to produce more milk. The effects of fodder trees or shrubs in improving milk production have been demonstrated in several studies (Mangesho et al., 2017; Salifou et al., 2017; Agani et al., 2022). Certain trees and shrubs are lactogenic in local dairy cows. In Benin, traditional cattle farmers use roots, leaves, bark, fruits and seeds to stimulate milk production in cows (Salifou et al., 2017). In northern Benin, certain woody fodder such as Afzelia Africana, Khaya senegalensis and Pterocarpus erinaceus are prized but Afzelia africana is the most sought after because it would increase milk production (Houehanou et al., 2008). This result corroborates those of Ibrahim et al. (2005) who reported that dairy cows fed on shrub forage produced an impressive daily milk yield without the use of supplemental concentrates. Furthermore, a study of Cohen-Zinder et al. (2016) on lactating cows revealed that milk production was improved by 2% if cows were fed on Moringa oleifera silage instead of wheat silage. Zhang et al. (2018) showed that the milk production of cows increased by 5% supplementing the diet with 6% Moringa oleifera in lactating multiparous Holstein dairy cows. Leucaena leucocephala supplementation also improved animal performance by increasing their proportion in cattle diet (Stifkens et al., 2022). Fodder trees are fodder resources for animals in dry periods. Their protein and vitamin content makes them the main fodder resource for livestock in the dry season to meet the maintenance and production needs of animals (Njidda and Ikhimioya, 2010; Mebirouk-Boudechiche et al., 2014; Sidi Imorou et al., 2016). In the Agrosilvopastoral territory of Nétéboulou in Senegal, the crude protein content of the leaves of Pterocarpus erinaceus can reach 15 to 20% of the dry matter (Mbaye et al., 2003). Trees and shrubs could then be used as alternative sources of protein for ruminants, which can lead to higher milk yield in native cows (Alam et al., 2009).

The highest daily milk yield obtained in the ISS, LIAS, SIAS farms compared to TSS cattle farm could also be due to the importance of woody fodder in the regulation of the body temperature of dairy cows. In fact, the presence of trees in the pastures can reduce heat stress, with positive effects on food consumption and milk production (de Abreu, 2002).

The amount of milk obtained from each local cattle bred based on tree integration practices in this study is greater than that obtained by Worogo et al. (2021) in traditional dairy cattle in northern Benin. This could be explained by the farming method practiced. In fact, the cattle farmers included in their study did not feed their cows with leaves from trees on their farm. Moreover, the quantity of milk produced by local cows in the TSS systems are similar to those reported by Kassa et al. (2016). This could be explained by the fact that the cows monitored by these authors practice extensive farming with seasonal exploitation of trees and shrubs on natural rangelands. Milk production increases gradually from the 1st to the 4th lactation where it reaches the peak and drops gradually from the 5th lactation. The development of mammary tissues during the reproductive life of the cow could explain this increase in milk according to the rank of birth (Rivière, 1991). From the 5th lactation, the milk production of dairy cows decreases, this could be due to the aging of the mammary tissues (Kassa et al., 2016). The parity number is therefore a physiological factor in the variation of milk production (Rivière, 1991). Several authors also reported the effect of parity number on cow milk production in Benin (Alkoiret et al., 2010; Assani et al., 2015; Worogo et al., 2021).

Monitoring the structure of a dairy herd provides important insights into herd profitability and farm dynamics (Muller, 2018). This study showed that herd structure varied with integrated production systems. The high proportion of lactating cows and dry cows in ISS herds is related to the main production objective of this farm (Assani et al., 2023). In fact, ISS farmers specialize in dairy production and direct their production toward the market or semi-dairies. This could be explained by the large number of dairy cows present on this type of farm. This herd structure, where the proportion of lactating cows is around 50%, confirms the specialization of herds in dairy production. Milk, in fact, represents an essential constituent of the food ration of human populations in Benin and provides a regular income to breeders of this type (Ogodja et al., 1991). Similar results were observed by Akpa et al. (2012) who showed that it is for milk production purpose that farmers raise a greater proportion of cows.

The largest proportions of adult males were found in the LIAS and SIAS types compared to the other types. This could be explained by the fact that the animals of these types are kept by agro-pastoralists and have a high number of draft oxen. Worogo et al. (2021) came up with similar observation where Borgou cattle were widely used for animal-drawn cultivation in northern Benin by agro-pastoralists. The composition of TSS cattle herds is on average 29.2% males and 70.8% females. This result is similar to those obtained in traditional cattle farms in Benin (Dehoux and Hounsou-Ve, 1993; Alkoiret et al., 2010; Assani et al., 2015; Worogo et al., 2021). According to these authors, the herds consist of one male for three females in traditional farms.

The superiority of the reproduction parameters of the ISS herds compared to the other types could be explained by the improvement of practices for the integration of livestock with the installation of fodder plots and the planting of fodder trees and shrubs adopted by farmers of this type. In fact, the presence of plantations of fodder trees and shrubs in livestock farms allows farmers to cope with feed shortages during the dry season, which improves cow productivity (abortion rate, parturition rate and fertility rate). Feeding influences all stages and components of reproduction in females (puberty, cyclicity and heat, mating, gestation, drying off, postpartum, and lactation) and males (puberty, libido, sperm and spermogram) (Meyer, 2009; Martins et al., 2021). In the Sahel, Diawara et al. (2017) observed a deterioration in reproductive performance in less mobile animals that do not receive enough feed supplements. ISS farms were more characterized by high fertility (92.3%), parturition (94%) and lowest abortion rate (1.8%). The results obtained in this type of dairy farming are similar to those obtained by Worogo et al. (2021) at the Okpara breeding farm. On the other hand, Dehoux and Hounsou-Ve (1993) obtained similar values of the fertility rate (65.4%) and the abortion rate (4%) in northern Benin within traditional beef herds corresponding to the performance of TSS.

The mortality parameters were also low in the ISS and SIAS type herds, this could be linked to the improvement of farming practices in this type. The study conducted at the Okpara breeding farm by Youssao et al. (2000) showed the reduction of mortality of the young animals to 2.5% by a program of regrouping of the births, a good feed and a good weaning. On the other hand, another study carried out in South Africa showed that most deaths are often caused by diseases (50%) and drought (34%) (Motiang and Webb, 2016). This is also the observation made in this study where the highest mortalities were obtained in the TSS and LIAS farms. This could be explained by the fact that there is a weak integration of the tree in this farming system, limited only to the exploitation of the trees natural range already very degraded in drylands of Benin which could cause a feeding imbalance and heat stress. In fact, heat stress increases respiration and mortality, reduces fertility, alters animal behavior and suppresses the immune and endocrine systems, thereby increasing the susceptibility of animals to certain diseases (Thornton et al., 2022). Furthermore, the mortality rate results obtained in this study are better than those found by others (Alkoiret et al., 2010; Assani et al., 2015; Worogo et al., 2021) in traditional cattle farming where the mortality rate youth ranged from 14 to 33%. These authors also confirmed that mortality rates are higher in traditional farming due to undernourishment and the absence or insufficiency of sanitary and medical prophylaxis. The main causes of mortality are almost the same for all farming systems in the study area and are of viral, bacterial, parasitic, feed or traumatic origin. Management rates vary depending on the type of farming systems. The superiority of the net offtake rate of the ISS herds would be linked to the low mortality rates recorded and to their objective of producing milk so that the several males are sold. Mortality being higher in the TSS and LIAS type, these herds have fewer animals to sell. The SIAS type consists of small herds, selling very few animals to achieve the goal of increasing numbers. They are also agro-pastoralists, they buy more young male cattle, which could explain the very high intake rate in this type. The numerical exploitation of cattle herds in drylands of Benin is consistent with that of traditional herds (Assani et al., 2015). According to these authors, the numerical exploitation rate of sedentary herds varied between 2 and 9%. Numerical exploitation also varies according to livestock categories: males are often sold very young in ISS farms, while females are kept for a long time for breeding in TSS and LIAS farms (Alkoiret et al., 2010). The low growth rates of herds in TSS and LIAS could be explained by the poor reproduction and mortality parameters associated with high numerical exploitation recorded in these herds. ISS-type herds that had the highest reproductive and exploitation parameters and the lowest mortality parameters had the highest numerical yield. In South Africa, some authors (Scholtz and Bester, 2010; Meissner et al., 2014) also reported that the high herd mortality of small-scale livestock keepers is the main cause of low productivity and low animal off-take rates.

Improved silvopastoral is adaptive to drought because foliage production from trees and shrubs is less affected by varying precipitation, temperature and other climatic variables thus enabling farmers to sustain livestock production even during extreme weather conditions (Papanastasis et al., 2008). The results show that dairy farming can be practiced in the drylands of Benin even during the dry season. The Benin government being aware of the great threat that climate change poses to the country's sustainable development, has drawn up National Action Programs for Adaptation to Climate Change (NAPA). Agrosilvopastoral and silvopastoral practices is one of the priority actions in this context, equally contributing to the adaptation and mitigation of climate change, as well as to food security. It is then necessary to: