Dynamics of tall fescue (Festuca arundinacea) and Kikuyu grass (Cenchrus clandestinus) pastures associated with white clover (Trifolium repens) in small-scale dairy systems in the highlands of central Mexico

An on-farm participatory study was undertaken to assess changes over seasons on the availability, botanical and morphological composition, and nutritive value of herbage from two pastures in small-scale dairy systems in the highlands of central Mexico. One pasture (TF-33) was originally sown with tall fescue, while the second pasture (KY) was naturally invaded by Kikuyu grass; both associated with white clover and over-sown in winter with annual ryegrass. Sampling was every 28 days for a year. Variables were sward height, net herbage accumulation, soil cover, tiller density, chemical composition, and in vitro digestibility. A split-plot design was used. Sward height was greater for KY. There were no differences (p > 0.05) for herbage accumulation. TF-33 was 53% live tissue, 30% dead tissue, and 17% other plant material, whilst KY was 50% live tissue, 30% dead tissue, and 18% other plant material. TF-33 showed a higher leaf to stem proportion, while stem was higher in KY. Kikuyu grass and annual ryegrass performed better when associated with TF-33 than in KY pasture. Seasonal changes significantly affected sward height, herbage mass and net herbage accumulation, whereby TF-33 performed better than KY in winter. The multispecies association of diverse grasses and legumes may be complementary at different times of the year due to plasticity among species, and is a feasible option for small-scale dairy systems.

In Mexico, over 88% of all farms with cattle are categorised as small-scale dairy systems (SSDS) (INEGI 2018).Small-scale dairy systems typically have a low hectarage and herds between three and 35 cows and are considered 'small' (Fadul-Pacheco et al. 2013).The sale of milk is the main source of income, and these farms base their operation on family labour (Fadul-Pacheco et al. 2013), thereby aiding families in rural areas in overcoming poverty (Espinoza-Ortega et al. 2007).
In Mexico, SSDS contribute more than 30% of the national milk production (Espinoza-Ortega et al. 2007).Small-scale dairy systems represent 80% of all dairy farms around the world (17 million farms), and farm 35% of the arable land in Latin America producing 67% of milk production in the region (FIDA 2019).The economic scale of SSDS is the limiting factor in the sustainability and longevity of these farms, hindered by high feeding costs that reduce their profitability (Fadul-Pacheco et al. 2013;Prospero-Bernal et al. 2017).Prospero-Bernal et al. (2017) demonstrated that by implementing grazing of cultivated pastures in SSDS in the highlands of central Mexico, feeding costs were reduced, profitability increased, and sustainability of the small-scale farm improved.
Grazing is reliant on seasonal changes, particularly in temperate sub-humid climates such as in the highlands of central Mexico, where significant changes in rainfall and temperature occur between wet and dry seasons (Plata-Reyes et al. 2018;2021).Pastures require irrigation in the dry season when temperatures are high, as well as in spring when evapotranspiration is high (Plata-Reyes et al. 2018).Frosts occur in winter when temperatures and rainfall are low (Plata-Reyes et al. 2021).
As perennial ryegrass (Lolium perenne L.) grows slower at temperatures above 25°C and does not tolerate droughts, it is the grass of choice for temperate pastures (Lee et al. 2013;Parsons and Chapman 2000).Tall fescue (Festuca arundinacea [Schreb] Darbysh), in contrast, is more resistant to water deficits and high temperatures (Plata-Reyes et al. 2018).Endophyte-free tall fescue herbage growth and nutritive value are similar to perennial ryegrass in the highlands of Mexico (Plata-Reyes et al. 2018;Marín-Santana et al. 2020) , Carlos Galdino Martínez-García 1 , Omar Hernández-Mendo 2 and Carlos Manuel Arriaga-Jordán 1 * drought conditions with growth peaks in the rainy season, (Plata-Reyes et al. 2018, 2021).White clover (Trifolium repens L.) is a temperate legume that increases dry matter (DM) yields, the nutritive value of herbage and reduces nitrogen fertiliser requirements due to biological fixation of atmospheric N 2 (Lüscher et al. 2014;Guy et al. 2020).This legume is well adapted to irrigated pastures in the highlands of Mexico, which improves pastures of SSDS with low fertiliser use (Pozo-Leyva et al. 2019).
The animal performance when fed on Kikuyu grass pastures have similar results to temperate grass pastures, but with low establishment costs, since Kikuyu is known to invade pastures and agricultural plots through stolons and rhizomes (Marais 2001).Kikuyu grass is therefore a viable option for pasture in SSDS (Marín-Santana et al. 2020).However, low winter temperatures limit its year-round production since Kikuyu grass, as a subtropical species, stops growing at temperatures below 8 °C, and is susceptible to frost when temperatures fall below zero for long periods (Marais 2001).
Temperate grasses such as the hardy tall fescue, in contrast, are a viable option for pastures in winter, if it is capable of overcoming the invasion by Kikuyu grass (Plata-Reyes et al. 2021).A synergistic relationship between the subtropical Kikuyu grass and the temperate tall fescue would result in a multispecies pasture that could a high-performance pasture year-round.Therefore, there is need to study and understand the pasture dynamics of the Kikuyu grass-tall fescue association to enhance management practices that result in optimal pasture performance for SSDS.
The growth of pasture grasses and legumes is determined by management decisions such as fertilisation, stocking rate and grazing system (Bernal and Espinosa 2003).Biophysical factors such as temperature, rainfall, solar radiation, evapotranspiration and soil type may directly influence yearly herbage availability and, owing to their uncontrollable nature, may be difficult to forecast in the field (Swanepoel et al. 2017;Claffey et al. 2020;Piña et al. 2020;Venter et al. 2021).Knowledge of pasture dynamics related to seasonal changes will significantly improve the management of pastures in grazing-based livestock systems, in order to obtain maximum yields and the highest performance, as has been reported from Australia, Ireland, Mexico and South Africa (van Wyngaard et al. 2018;Benvenutti et al. 2020;Guy et al. 2020;Plata-Reyes et al. 2021).It is therefore necessary to know the development of different varieties of grasses and legumes over time under field conditions.
The objective of this study was to assess the effect that seasonal changes have on the botanical and morphological composition, accumulation and nutritive value of herbage of (1) tall fescue (F.arundinacea) invaded by Kikuyu grass (C.clandestinus) and (2) Kikuyu grass pastures associated with white clover (T.repens) over one year.The work followed a participatory livestock technology development approach (Conroy 2005) aimed at identifying and developing strategies that participating farmers can adopt to improve production through on-farm research by two small-scale dairy farmers that manage their land as a single unit.

Study site
The experiment was carried out in two SSDS headed by two collaborating farmers who voluntarily and actively participated in this research.The collaborating farmers had knowledge of the objectives of this work.They provided informed consent, and their decisions were respected at all times.They were permitted to opt out of this study at any time.
Both study farms are within the municipal boundaries of Aculco, State of Mexico, Mexico, located between 20°00′ and 20°17′ N and between 99°40′ and 100°00′ W, at a mean altitude of 2 440 m.The climate is sub-humid temperate with mean temperatures between 10 and 18°C, with an annual rainfall of 700-1 000 mm (Celis-Álvarez et al. 2016).There is a well-defined rainy season from June to October, and a dry season from November to April with high evapotranspiration (Figure 1, Supplementary Figure S1; Plata-Reyes et al. 2021), as well as frosts in winter (from late November to mid-February).Estimated evapotranspiration was calculated as 121 ± 10.32 mm per period (Segura-Castruita and Ortiz-Solorio 2017).The soil in these pastures were acidic with a pH of 5.9 ± 0.34, had a sandy loam texture (46% sand), low organic matter content (3.34 ± 1.30%) and low apparent density.
The herbage in the experimental KY and TF-33 pastures in winter (KY) and spring (KY) are shown in Supplementary Figures 2 to 5.

Study design
Two grass/legume pastures of 1.0 ha each were delineated using a hand-held GPS and delimited using an electric fence.The evaluated pastures were: • KY = A pasture naturally invaded by Kikuyu grass (KY) associated with white clover (WC) and oversown in autumn 2019 with annual ryegrass (ARG).• TF-33 = A pasture originally sown in December 2015 to tall fescue (TF) and WC at a sowing rate of 30 kg tall fescue ha −1 and 3 kg white clover ha −1 , respectively, which had been invaded by KY.This pasture was also oversown in autumn 2019 with ARG.These two pastures were nominally divided into two 0.5-ha sub-plots as replicate sampling units (Carrillo-Hernández at al. 2020), from which herbage sampling was every four weeks with a total of 12 sampling periods over the course of one year (April 2019 to March 2020), according to procedures described by Reeves et al. (1996) and Álvarez García et al. (2020).Sample collection and data recording were on the last day of each sampling period.
Six randomly located grazing exclusion cages (a 0.5 × 0.5 m metal frame) were placed in each pasture treatment, i.e. three cages per 0.5 ha sub-plot, to measure the non-grazed contingent of the herbage mass and nett herbage accumulation.

Treatments
Sixteen dairy cows day-grazed the two evaluated pastures continuously for 8.0-9.0 h day −1 .Ten of the 16 cows were milking cows with a mean live weight of 506 ± 53.73 kg and a mean daily yield of 15.5 ± 0.50 kg milk cow −1 day −1 .Dry cows (n = 6) had a mean live weight of 485 ± 42 kg.
Pastures were initially fertilised at sowing with 60:80:60 kg N:P:K ha −1 and thereafter at the end of June (wet season) and December (winter dry season) using diammonium phosphate, urea, and potassium chloride.Additionally, 23 kg N ha −1 as urea was applied every 28 ± 5 days.In the dry season (November to April) the pastures were flood irrigated about every four weeks with approximately 1 350 m 3 N ha −1 .

Pasture variables
Compressed sward height (cm) was recorded with a rising plate aluminium grass meter at the end of each period in a W-shaped pattern following Hodgson (1990).Fifteen recordings per 0.5-ha sub-plot were taken, therefore 30 recordings were taken for each of the two pastures.
The NHA per unit of time (sampling period or per day) and herbage mass are indicators of available forage and grazing conditions, which vary according to plant species, season, climatic conditions, soil fertility and management factors, such as stocking rate (Hodgson 1990;Mayne et al. 2000).Herbage was cut to ground level on the first day (day 0) and at the end (day 28) of each sampling period using an electric grass shearing machine.Herbage samples were dried at 55 °C in a draught oven till constant weight, and expressed in kg DM ha −1 (Reeves et al. 1996;Hoogendoorn et al. 2016).Herbage mass was recorded on day 0 outside of the exclusion cages (Álvarez García et al. 2020;Carrillo-Hernández et al. 2020).Net herbage accumulation (NHA) was calculated as the difference of herbage (kg DM ha −1 day −1 ) in the exclusion cages at the end of each 28-day sampling period with the herbage cut outside the cage on day 1.
To determine botanical and morphological composition, five 50 g herbage samples were taken randomly from each pasture (25 g per 0.5 ha sub-plot), and manually separated into pasture species (KY, TF-33, ARG, WC and other species) and morphology (leaves and stems).Live and dead material were also recorded (Dennis et al. 2012;Sanderson et al. 2016).The separated samples were dried in a draught oven at 55°C up to constant weight.
Ground cover (%) was determined for each pasture by visually observing the plant species present within five random a 0.5 × 0.5 m sites subdivided into a 5.0 × 5.0 cm grid, according to the procedures described by Fenetahun et al. (2020).The proportion of bare soil, KY, TF-33, ARG, WC and other plant species were recorded.
Live tiller density in grasses and live rooted nodes for WC were counted from five core samples (5 cm diameter × 15 cm depth) taken within five 0.25 m 2 frames (25 cores samples in total) that were demarcated in each pasture at random, following the method by Lush and Franz (1991).The core samples were processed on the sampling date and sorted by species (KY, TF-33, ARG, WC and other species), expressing results in live grass tiller or clover nodes m 2 ha −1 .
The chemical composition of herbage was determined in 200 g samples at random from the sampling plots, which were hand-plucked to simulate grazing (Cook 1964).The dry matter (DM), organic matter (OM) and crude protein (CP) were measured following the standard laboratory methods described in Anaya-Ortega et al. (2009).Neutral detergent fibre (NDF) and acid detergent fibre (ADF) were determined using the alpha-amylase and filter-bag technique for A200 (method 6 method 5, respectively; Ankom Technology 2014a, b).The cow ruminal fluid samples were dehydrated for 48 h in an Ankom Daisy incubator, from which the in vitro organic matter digestibility (IVOMD) was ascertained (Tilley and Terry 1963).The IVOMD was used to calculate the digestible organic matter in the dry matter (DOMD) (AFRC 1993).The metabolisable energy (ME) content was calculated using the DOMD value (AFRC 1993).

Experimental design and statistical analyses
Simple linear regressions (Kaps and Lamberson 2004) were fitted with the data on sward height next to the exclusion cages (day 1), inside exclusion cages (day 28), as well as herbage mass (day 1 and 28) to generate predictive equations in which sward height (cm) was the independent variable (x) and herbage mass (kg DM ha −1 ) the dependent variable (y).
Other pasture variables were analysed with a randomised split-plot design where pasture type (KY or TF-33) was the fixed effect (main plots) and measurement periods were random effects (split plots).This design was recommended by Stroup et al. (1993) for on-farm experiments where replicates are limited.The analysis of variance (ANOVA) was performed with the following model (Kaps and Lamberson 2004): where: μ = general mean; r i = effect due to replicates i = 1 and 2, T j = effect due to pasture treatments (main plots) j = 1 and 2, E k = error term for main plots, p i = effect due to measurement periods (split plots) k = 1…12, Tp jk = effect due to interactions between treatments and measurement periods, e ijk = experimental residual variation.The results were analysed with Minitab statistical software (version 14; Minitab 2010).

Results
The mean maximum (25 °C), overall mean (19 °C) and mean minimum (8 °C) temperatures (Figure 1), as well as the rainfall and evapotranspiration (Figure 2) were recorded for each of the 12 monthly sampling periods from April 2019 to March 2020.The wet season occurred from June (P3) to October (P7).Total rainfall during the one-year period was 715 mm, with most of the rainfall (250 mm) occurring in July (P4).
Simple linear regression analysis, based on predictive equations, between sward height (cm; independent variable) and forage herbage mass (kg DM ha −1 ; dependent variable) are shown in Table 1.The correlation coefficient was low at 33.6% (p < 0.05); the compressed sward height outside of the exclusion cages on day zero accounted for 11.3% of the variation for mean herbage mass on day zero.The  correlation coefficient between compressed sward height and herbage mass inside exclusion cages on day 28 was slightly higher at of 43.0% (p < 0.05), accounting for 18.4% of the variation for mean herbage mass values on day 28.Sward height (cm) of pastures between treatment and sampling period (Table 2) showed statistically significant interaction (p < 0.05), indicating that the pasture variability was influenced by changes in temperature and rainfall with season.There were no significant differences (p > 0.05) for herbage mass for pasture treatments with a mean of 842 ± 358 kg DM ha -1 per period for TF-33 pasture and 626 ± 426 kg DM ha -1 per period for KY (Table 2).
The leaf:stem ratio was higher in the TF-33 pasture with a leaf proportion of 73% for TF-33 (Figure 3) and 57% for KY (Figure 4).The morphological composition of ARG indicated a larger proportion of leaves (Figure 3).In the KY pasture, WC had a larger proportion of leaves, as was also the case for ARG (Figure 4).
The tiller density of ARG decreased in both pastures (KY and TF-33) as the experiment progressed (Table 5).The decrease was more pronounced in TF-33 from.KY and TF-33 tillers were higher in the KY pasture than the TF-33 pasture throughout the experimental period (Table 5).The tiller density of WC and ARG, in contrast, was higher in the TF-33 pasture than the KY pasture.There were also few KY live tillers in the early sampling periods (P1-P6) in the KY pasture (Table 5), but KY recovered in the wet season.In both the TF-33 and KY pastures, TF-33 and WC were present in high numbers throughout the year.However, the statistically significant interactions for tiller and node density in both pasture and sampling period (p < 0.05) for all species indicates a changing dynamic of pasture components over time due to changing seasons and meteorological conditions (Table 5; Figures 1 and 2), e.g. the increased rainfall in the wet season from June to October 2019 increased KY grass in both pastures in P3 to P7.

Chemical composition of herbage
There were significant interactions (p < 0.05) between treatments and sampling periods for the chemical composition of herbage from the treatment pastures for contents of DM, CP, NDF, IVOMD and ME (Table 6).A significantly larger content (p < 0.05) of CP was noticed in P1 and P4, which were the largest CP contents in the TF-33 pasture due to the higher leaf:stem proportion (Figure 3).The NDF content increased in both the KY and TF-33 pastures as the experiment progressed, and the mean NDF content was significantly higher (p < 0.05) for KY than the TF-33 pasture.There were differences between periods for ADF content (p < 0.05), but no statistically significant differences between pastures (p > 0.05) The IVOMD and estimated ME content were highest in both pastures during the beginning of the experiment (P1-P5) in late spring and early summer (wet season), but soon declined thereafter in the latter sampling periods towards the end of autumn (dry season).However, IVOMD in the TF-33 pasture increased again in late winter (P10), whilst IVOMD in the KY pasture remained low.Mean ME content of pastures indicated a good quality herbage, although the TF-33 pasture showed more stable values across the study period than the KY pasture.These variations within and between pastures represented a significant interaction (p < 0.05) such that the nutritive value of the two pasture treatments performed differently following seasonal and meteorological changes.

Discussion
Based on pasture management applied by SSDS farmers, this study sought to ascertain the dynamics of a KY and TF-33 pasture over a year by ascertaining their botanical and morphological composition, herbage mass and accumulation, as well as chemical composition.
Rainfall in the subhumid highlands is higher in the wet season from June to October; but, is offset by high evapotranspiration from November to May.Furthermore, the evapotranspiration is not offset by the limited irrigation The compressed sward height is an indirect indicator of grazing conditions.The low mean values for TF-33 (2.7 cm) and KY (3.0 cm) indicate poor grazing conditions because the forage is close to ground level.The optimal sward height for continuous grazing systems is between 5 and 8 cm (Mayne et al. 2000).The short sward height can be attributed to a decrease in plant growth owing to prolonged dry periods with high temperatures (>25 °C) and high evapotranspiration in the study area (Figures 1 and 2).In such temperate climates, grass species slow their growth by decreasing the leaf:stem ratio (Chapman et al. 2014).Since hydric stress reduces the forage photosynthesis rate thereby prolonging the recovery time between each defoliation, carbohydrate reserves in roots and stems are hence depleted affecting the growth of axillary meristems (Pereira et al. 2015;Mendoza Pedroza et al. 2018).The additional and constant defoliation due to high stocking and grazing rates further prolongs recovery and growth.
Net herbage accumulation is the result of growth and decomposition of the plant material present per land surface unit under grazing.The NHA for TF-33 (30.1 ± 0.4 kg DM ha −1 d −1 ) and KY (22.3 ± 0.5 kg DM ha −1 d −1 ) were similar to González-Alcántara et al. ( 2020) and Marín-Santana et al. (2020) during the dry season and at the end of the wet season, respectively, with a mean NHA of 26.25 ± 5.4 kg DM ha −1 d −1 for the same two grass species in the same study area.Merino et al. (2020) reported that NHA values of 17-25 kg DM ha −1 d −1 under intensive rotational grazing in Chile were suitable for forage and had no negative implications for grass quality.In a three-year study in Ireland, Guy et al. (2020) reported an average herbage yield between 15 and 17 tonnes DM ha −1 year −1 in ARG pastures associated with WC, representing a NHA ~60% higher than in the TF-33 pasture and ~200% higher than in the KY pasture in this study (Table 2), and further demonstrates the notable impact that the dry season has on limiting pasture growth.
The persistence in terms of botanical and morphological composition, soil cover and tiller (or node) density of ARG and KY decreased over the course of the experiment, while TF-33 showed characteristics of resistance and adaptability to the environment conditions by increasing growth over the winter periods (P10, P11 and P12), indicating that KY becomes inactive as temperature drops in autumn-winter (Marais 2001).
The increase in the mean proportion of leaves in the morphological composition is attributed to the response to defoliation and could be related to the speed of plant regrowth that varies according to season and pasture management (Mendoza Pedroza et al. 2018;Schmitt et al. 2019;Benvenutti et al. 2020).Furthermore, under high stocking rates, intervals between defoliations decrease.Lemaire and Belanger (2020) reported that the proportion of leaves declines as regrowth periods are reduced due to a depletion of carbohydrate reserves.The leaf:stem ratio is also an indicator of the possibilities and limitations of grazing as a higher proportion of stems affects the size of bites since stems act as a vertical barrier for grazing causing slow regrowth of forage and weed invasion (Venter et al. 2021).
Similar to the mean tiller density in this study, Pulido and Leaver (2001) reported a mean of 3 602 ± 1 648 tillers m −1 ha −1 for perennial ryegrass (Lolium perenne) pastures continuously grazed in Great Britain, with similar regrowth performance as indicated by Benvenutti et al. (2020).High tiller density is a common feature of continuous grazing, as evidenced in this experiment, and lower sward heights increase tiller density (Baker and Leaver 1986).The higher tiller density of KY, TF and ARG grasses, as well as WC node density, and the higher NHA of KY in the TF-33 pasture, may indicate a possible synergy and complementarity in that multispecies pasture (Muciño-Álvarez et al. 2021).
Despite sward heights below 8 cm in this study, the largest proportion of leaves in both treatments may have influenced the chemical composition of herbage as reported by Piña et al. (2020).Therefore, the larger the proportion of leaves to stems, the better the herbage quality (Barbehenn et al. 2004;Zanini et al. 2012).In this regard, Difante et al. (2009) 6: Chemical composition of herbage by treatment and sampling period regrowth on the vegetative stage.Over the study periods herein, the best nutritional quality of herbage was obtained when the compressed sward height was lower, while the proportion of the more nutritious leaves was higher than stems (Botha et al. 2008;Sanderson et al. 2016).
Regarding the morphological components, leaves use the upper strata, while stems and dead material use the lower strata (Difante et al. 2009;Nuñez et al. 2022).Therefore, young leaves have a larger concentration of carbohydrates and enzymes, and a lower concentration of structural components (Zhao et al. 2008;Chapman et al. 2014).Piña et al. (2020) reported a higher content of CP in shorter grasses, which concurs with the early experimental stages in this study.The higher CP content in the TF-33 herbage was, therefore, likely due to a higher leaf proportion and shorter compressed height.
Organic matter digestibility and ME decreased as the experiment progressed, and was attributed to increased cell wall contents in herbage (Barbehenn et al. 2004;Dodd et al. 2019).By and large, the nutritional quality values presented in this study are similar to those previously reported by García et al. (2014), Plata-Reyes et al. (2018) and Marín-Santana et al. (2020).

Conclusion
Pasture performance under grazing analysed in this study showed that TF-33 developed better in winter, while KY grass was latent in cold weather, and presents a feasible option for small-scale dairy systems.Both KY and ARG performed better when associated with TF-33 in the TF-33 pasture than in the KY pasture due to possible synergies and complementarity among species.The findings in this work can be used to improve implementation strategies designed to compensate herbage deficits with alternate management approaches, such as the seasonal complementarity of the multispecies association between diverse grasses and legumes.
Further studies are required to ascertain the persistence of different pasture species over longer periods of time to determine the optimal combination of these species with respect to greater availability and nutritional value of herbage in SSDS.
. Kikuyu grass (Cenchrus clandestinus [Hochst.ex Chiov] Morrone, previously Pennisetum clandestinum) is a subtropical grass with a strong seasonal growth, naturalised in the temperate and subtropical regions of the highlands and tolerant to Introduction Dynamics of tall fescue (Festuca arundinacea) and Kikuyu grass (Cenchrus clandestinus) pastures associated with white clover (Trifolium repens) in small-scale dairy systems in the highlands of central Mexico Dalia Andrea Plata-Reyes 1

Figure 1 :
Figure 1: Temperature ranges during the 12-month experiment from April 2019 to March 2020
stated that the variations in chemical composition are more related to sward height than season or the age of