Introduction
Detailed representation of digesta outflow from the rumen is critical for improving the modelling of rumen function and herbage intake of grazing ruminants. A body of empirical data and mathematical interpretation of processes determining the outflow of rumen digesta has been accumulated and summarized by Poppi et al. (Reference Poppi, France, McLennan, Theodorou and France2000) and Faichney (Reference Faichney, Forbes and France2005). Attempts to model outflow of rumen digesta have been diverse (Waldo et al., Reference Waldo, Smith and Cox1972; Barboza and Bowyer, Reference Barboza and Bowyer2000; Seo et al., Reference Seo, Lanzas, Tedeschi, Pell and Fox2009). However, models describing this process do not account for all significant sources of variation, nor do they control rumen fill and digesta flow without fractional outflow rates or retention time defined by the user (Kennedy, Reference Kennedy, Dijkman, Forbes and France2005). Moreover, most models of rumen digestion do not include the foraging process and oral processing of the forage, which is the first step of digestion. Thus, the condition in which forage is received by the rumen is neither simulated nor predicted, making these models incomplete (Prinz and Lucas, Reference Prinz and Lucas1997).
MINDY is a mechanistic model, incorporating diurnal patterns of foraging, digestion and metabolism, dietary choice, excretion and production of a grazing ruminant (Gregorini et al., Reference Gregorini, Beukes, Romera, Levy and Hanigan2013; Reference Gregorini, Villalba, Provenza, Beukes and Forbes2015b, Reference Gregorini, Villalba, Chilibroste and Provenza2017). It is a cluster of six models, including the rumen sub-model of Molly (Baldwin, Reference Baldwin1995, modified by Gregorini et al., Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a). Although MINDY includes ingestive actions and rumination time in response to animal and sward state and condition, oral processing and the resultant changes in particle size and distribution of ingesta are not included. Moreover, the current rumen sub-model in MINDY does not consider the dynamics of water ingestion with the forage (i.e. ingesta dry matter [DM]) and the fractional passage of liquid is fixed. This limits MINDY's mechanistic and dynamic capability to exploration of foraging situations where the dynamics of ingestion, modulated either passively (by sward features) or actively (by management), are expected to alter patterns of forage digestion and nutrient supply, and in turn affect animal performance and excretory (e.g. urination; Gregorini et al., Reference Gregorini, Provenza, Villalba, Beukes and Forbes2018) behaviour.
The objective of the current work was to formalize implicit interactions between ingestion and digestion processes in MINDY by including explicit representation of oral processing of ingesta and simulating, rather than setting, the fractional passage of liquid in the rumen in response to foraging context. The advances here integrate functional relationships between forage ingestion, oral physiology and rumen digestion responsible for variations in ingesta characteristics and digesta outflow rates from the rumen. The current paper describes changes in the structure and function of the model and assesses it in terms of responses to foraging conditions.
Model description and rationale
Dynamics of digesta outflow from the fore-stomach of ruminants depends on: (1) sward structure, patterns of forage intake and associated oral processing of ingesta determining particle size distribution of swallowed boli; (2) rumen function and contents and (3) regulation of flow and retention of liquid in the rumen (Pérez-Barbería and Gordon, Reference Pérez-Barbería and Gordon1998; Lechner et al., Reference Lechner, Barboza, Collins, Fritz, Günther, Hattendorf, Hummel, Südekum and Clauss2010; Gregorini et al., Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a). Cattle pass large amounts of fluid through the rumen, which enhances the stratification and thereby selective retention time of digesta in the rumen (Clauss and Lechner-Doll, Reference Clauss and Lechner-Doll2001; Clauss et al., Reference Clauss, Hume and Hummel2010). Particulate retention in the rumen facilitates ruminal digestion, while high liquid passage increases ruminal bacterial yield (Dove and Milne, Reference Dove and Milne1994; Meng et al., Reference Meng, Kerley, Ludden and Belyea1999; Dewhurst et al., Reference Dewhurst, Davies and Merry2000; Clauss et al., Reference Clauss, Hume and Hummel2010) and serves as transport for both particulate (small) and soluble nutrients (Poppi et al., Reference Poppi, Minson and Ternouth1981; Faichney, Reference Faichney, Forbes and France2005). As a result, digesta outflow from the rumen is a function of fluid flow and concentration of DM in the fluid (Ulyatt et al., Reference Ulyatt, Waghorn, John, Reid and Monro1984; Kennedy and Murphy, Reference Kennedy and Murphy1988). Consequently, to better represent patterns of forage intake, digestion and nutrient supply from the rumen of grazing ruminants, the passage of solids must be linked to dynamic fluctuations of particle size distribution of the ingesta and fluid outflow from the rumen.
The current work implicitly represents (1) oral processing of ingesta determining particle size distribution of swallowed boli associated with temporal patterns of feed intake; (2) dynamic regulation of flow and retention of liquid in the rumen; and (3) the association of 1 and 2 with particulate passage throughout the rumen. Modifications to the original particle size pools of the rumen model used in MINDY and related changes in digestive parameters have been presented by Gregorini et al. (Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a).
The relevant factors, pools and functional relationships are illustrated in Figs 1 and 2. The code was developed and simulations were conducted using ACSLXtreme (Aegis Technologies Group, Austin, TX, USA). Numerical integration was conducted using a fourth-order, fixed-step, Runge–Kutta method. The maximum integration interval was set to 0.001 d. Results were collected after 5 d of simulation to ensure the model had reached a stable state. The order in which the structure of model development is presented follows the natural path of forage ingestion, oral processing of ingesta, inflows of ingesta to the rumen and outflow of digesta from the rumen.
Fig. 1. Diagram (Conceptual model) of functional relationships responsible for variations in rumen digesta outflows in grazing ruminants. White boxes with solid lines are compartments; white boxes with dashed lines are pools; black boxes with solid lines are processes. Arrows indicate effects on pools, processes and fluxes are indicated by triangle valves.
Fig. 2. A scheme of the oral processing module introduced into the MINDY cow model. Boxes represent pools or compartments and arrows represent fluxes.
Oral processing and particle size distribution of ingesta
The distribution of ingesta through different particle size pools in the rumen is a function of ingestive actions (severing, handling and salivation) and oral processing (mastication and salivation) of a cluster of bites forming the bolus to be swallowed (Moseley and Jones, Reference Moseley and Jones1984; Spalinger et al., Reference Spalinger, Robbins and Hanley1986; Pond et al., Reference Pond, Ellis, Lascano and Akin1987; Prinz and Lucas, Reference Prinz and Lucas1997). Particle size distribution in the swallowed bolus is related closely to forage species, sward structure, herbage chemical composition and plant phenology (Wilson and Kennedy, Reference Wilson and Kennedy1996; Poppi et al., Reference Poppi, France, McLennan, Theodorou and France2000; Kennedy, Reference Kennedy, Dijkman, Forbes and France2005). The drive to eat (i.e. hunger) also influences particle size distribution in the swallowed bolus (Demment and Greenwood, Reference Demment and Greenwood1988; Greenwood and Demment, Reference Greenwood and Demment1988; Gregorini, Reference Gregorini2012). Hunger reduces oral processing of ingesta through a reduction in mastication as a compensatory mechanism to increase short-term forage DM intake rate (Greenwood and Demment, Reference Greenwood and Demment1988; Demment and Laca, Reference Demment and Laca1994). In addition to the severing jaw movements (bite), mastication initiates the breakdown of forage physical structure, releasing around 0.65 of the water and soluble cell contents and exposing cell walls to microbial enzymatic action (Hogan et al., Reference Hogan, Keeney, Weston, Wheeler, Pearson and Robards1985). Ultimately, and in conjunction with a parallel salivation process, mastication determines the physical characteristics of the swallowed bolus (i.e. particle size distribution), the rate at which boli are swallowed (Bailey and Balch, Reference Bailey and Balch1961; Saunders, Reference Saunders1980; Stuth and Angell, Reference Stuth and Angell1982; Bailey et al., Reference Bailey, Erdman, Smith and Sharma1990; Prinz and Lucas, Reference Prinz and Lucas1997; Lucas et al., Reference Lucas, Prinz, Agrawal and Bruce2002) and thereby the rate of digestion and nutrient availability in the rumen (Poppi et al., Reference Poppi, France, McLennan, Theodorou and France2000; Chilibroste et al., Reference Chilibroste, Soca, Mattiauda, Bentancur and Robinson2007; Reference Chilibroste, Dijkstra, Robinson and Tamminga2008; Gregorini, Reference Gregorini, Cangiano and Brizuela2011). Therefore, models must include oral processing and its impact on the characteristics of the bolus to be swallowed.
Oral processing of forage in the present development of MINDY is based on mastication and swallowing models of mammals (Hutchings and Lillford, Reference Hutchings and Lillford1988; Prinz and Lucas, Reference Prinz and Lucas1997), concepts of perception/anticipation of feeding, food texture, particle agglomeration in the oral cavity and surface tension and viscosity of saliva (Prinz and Lucas, Reference Prinz and Lucas1997), the models of chewing efficacy of Pérez-Barbería and Gordon (Reference Pérez-Barbería and Gordon1998) and Shipley et al. (Reference Shipley, Gross, Spalinger, Hobbs and Wunder1994), and a comprehensive data set of ingestive boluses and salivation reported by Balch (Reference Balch1958), Prinz and Lucas (Reference Prinz and Lucas1997), Gill et al. (Reference Gill, Campling and Westgarth1966), Stuth and Angell (Reference Stuth and Angell1982), Boudon et al. (Reference Boudon, Acosta, Delagarde and Peyraud2006) and Acosta et al. (Reference Acosta, Boudon and Peyraud2007).
Particle size of ingested herbage (and supplemental feed)
First, the model calculates the distribution of particle size of herbage harvested as pasture or consumed as supplementary food. For grazing, MINDY calculates bite depth (cm) based on sward canopy structure (see Gregorini et al., Reference Gregorini, Beukes, Romera, Levy and Hanigan2013 for details). Bite depth depends on sward surface height (cm), and changes with herbage depletion (i.e. reductions in sward surface height). Based on a normal distribution of particle sizes per grazing stratum of the canopy (P Gregorini, unpublished data), it is assumed that the mean particle size of herbage harvested while grazing is half of the bite depth. Then the ingested particles of herbage harvested while grazing are ‘allocated’ to one of a set of 14 ‘bins’ (i.e. pools), according to their size. Mesh aperture size of each bin is doubled over the range 0.0375–153.6 mm (Note: the model allows for a variable number of bins, and the range of mesh aperture size is taken from wet sieving particle size wet analyses methodology). When feeding fresh cut herbage, silages or grains, data on the ratio between small and large particles (<1.2 and >4.8 mm, respectively), known as Psf factor (see Baldwin et al., Reference Baldwin, Thornley and Beever1987 for details) of the particular feed are provided to the model as input. Particle size distribution of ingested particles fed as supplements is then calculated using the equation of Pond et al. (Reference Pond, Tolley, Ellis, Matis and Kennedy1984) adapted for Molly by Gregorini et al. (Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a) and such a distribution of particles is then allocated to the set of the 14 ‘bins’, according to their size.
Bolus size and frequency of swallowing
The size of the bolus to be swallowed is calculated as a cluster of a variable number of bites. During grazing, bite characteristics (mass, volume and mastication jaw movements per bolus) are variable and depend on sward structure and condition (Laca and Demment, Reference Laca, Demment, Palo and Robbins1991; Laca et al., Reference Laca, Ungar, Seligman and Demment1992; Reference Laca, Ungar and Demment1994), as well as the animal's internal state (Gregorini, Reference Gregorini, Cangiano and Brizuela2011). MINDY accounts for those characteristics and dependencies, and in the present calculations of bolus size, mastication jaw movements according to bite mass, fibre content of the bite mass, plant species (i.e. C3, C4, herbs and legumes) and hunger level of the animal (Wilson and Kennedy, Reference Wilson and Kennedy1996; Baumont et al., Reference Baumont, Cohen-Salmon, Prache and Sauvant2004; Kennedy, Reference Kennedy, Dijkman, Forbes and France2005; Gregorini et al., Reference Gregorini, Soder and Kensinger2009b) are also considered. Bolus size and frequency of swallowing are represented by the following calculations.
A maximum bolus size (MaxBolusWeightFresh, grams of fresh matter excluding saliva) was assumed, limited by the volume of the oral cavity [the oral cavity is three-dimensional; hence a linear relationship with cow weight was assumed (e.g. if a cow is X2 wide, X2 long and X2 deep, then both oral cavity and weight will be X8) (Stuth and Angell, Reference Stuth and Angell1982; Prinz and Lucas, Reference Prinz and Lucas1997)]. It was also assumed that grazing ruminants maximize their short-term herbage intake rate (Bergman et al., Reference Bergman, Fryxell, Gates and Fortin2001; Fortin et al., Reference Fortin, Fryxell and Pilote2002) and thereby bolus size and swallowing rate (Stuth and Angell, Reference Stuth and Angell1982). Thus, in the short-term, grazing ruminants (and thereby MINDY) try to reach MaxBolusWeightFresh, unless extra saliva is released in the harvesting and handling actions. Faster salivation rates cause the developing bolus to disintegrate, which promotes early swallowing (Prinz and Lucas, Reference Prinz and Lucas1997; Lucas et al., Reference Lucas, Prinz, Agrawal and Bruce2002). The latter is supported by the results with grazing and non-grazing cattle reported by Boudon et al. (Reference Boudon, Acosta, Delagarde and Peyraud2006) and previous works of Gill et al. (Reference Gill, Campling and Westgarth1966) with cattle fed indoors.
It was assumed that salivation (BaseSalivaPerJawMovement, g saliva) is proportional to the number of jaw movements (severing and mastication), and is modulated by forage species (feed), time into the meal and hunger (Balch, Reference Balch1958; Bailey and Balch, Reference Bailey and Balch1961; Gill et al., Reference Gill, Campling and Westgarth1966). Hence, greater intake rate, where a total number of severing and mastication jaw movements per unit of feed harvested is reduced, will lead to less salivation (see Boudon et al., Reference Boudon, Acosta, Delagarde and Peyraud2006). The first boluses would be smaller than those following due to extra salivation at the start of the meal. The latter increases with increasing levels of hunger (Gill et al., Reference Gill, Campling and Westgarth1966).
Bolus size and frequency of swallowing are then calculated as a function of jaw movements and time per bolus, which are derived from severing (i.e. bites) and mastication jaw movements through a feedback loop (Fig. 2) using the following set of equations:
$$\eqalign{MaxBolusWeightFresh & = StandardCowMaxBolusWeightFresh\; \cr & \quad \times CorrectedBW}$$where StandardCowMaxBolusWeightFresh is 200 g (for a cow of 550 kg live weight). CorrectedBW is the empty body weight of the cow (no pregnancy and gut fill).
$$\eqalign{BaseSalivaPerJawMovement & = BaseSalivationRate \cr & \quad \times JawMovementRate}$$
$$BaseSalivationRate = StandardSalivation \times \; BW^{0{\cdot}75}$$BaseSalivaPerJawMovement is the base amount of saliva secreted per jaw movement, severing and mastication, StandardSalivation is 355 g/min derived from Gill et al. (Reference Gill, Campling and Westgarth1966) and BW0.75 the metabolic weight of the animal.
$$TargetBolusWeightFresh = BolusWeightFresh$$Due to the following circular reference:
$$\left\{ {Saliva =\gt BolusWeightFresh_{ = \gt MasticationJawMovement = \gt Saliva}^{ = \gt SeveringJawMovement = \gt Saliva}} \right\} ,$$an initial and arbitrary fresh weight of the bolus (TargetBolusWeightFresh, g fresh matter) is used to solve for the BolusFreshWeight iteratively in successive approximations (REPEAT, see below).
REPEAT:
$$BitesPerBolus{\rm \;} = \displaystyle{{TargetBolusWeightFresh} \over {BiteMassFresh}}$$
$$ActualBolusWeightFresh = BitesPerBolus \times BiteMassFresh$$
$$\eqalign{& MasticationJawMovementPerBolus \cr & \quad = DryFeedF/TimeIntoMealF \times BolusSizeF^{xBolusMov} \cr & \qquad \times BaseMastJawMovPerBolus \cr & \qquad \times HungerEffectOnMastication \times NDFF \times SpeciesF}$$
$$\eqalign{& TotalJawMovementsPerBolus = BitesperBolus \cr & \quad + MasticationJawMovementsPerBolus}$$
$$\eqalign{TimePerBolus & = BitesPerBolus\; \times TimePerBite \cr & \quad + MasticationJawMovementsPerBolus \cr & \quad \times TimePerMastication}$$
$$BolusSwallowingFrequency = \displaystyle{{60} \over {TimePerBolus}}$$
$$\eqalign{& SalivaPerMasticationJawMovement \cr & \quad = BaseSalivaPerJawMovement \times DroolingSalivaF \cr & \quad \times BolusSizeF^{xBolusSaliva} \times NDFF^{xNDFSaliva} \times DryFeedF}$$
$$\eqalign{& SalivaPerSeveringJawMovement \cr & \quad = BaseSalivaPerJawMovement \times kSalivaSevering \cr & \quad \times TimeIntoMealF \times NDFF^{xNDFSaliva}}$$
$$\eqalign{SalivaPerBolus & = BitesPerBolus\; \cr & \quad \times SalivaPerSeveringJawMovement \cr & \quad + MasticationJawMovementsPerBolus \cr & \quad \times SalivaPerMasticaitonJawMovement}$$
$$\eqalign{BolusWeightFresh & = \; TimeIntoMealF^2 \cr & \quad \times MaxBolusWeightFresh \cr & \quad \times \lpar {1 - LiquidizingSaliva - 0{\cdot}25} \rpar /0{\cdot}25}$$Then,
$$BolusWeightDry = BolusWeightFresh \times FeedDMContent$$
$$WholeBolusWeight = BolusWeightFresh + SalivaPerBolus$$where in Eqn 7, DryFeedF (unitless) increases saliva per mastication and number of mastications for dry feed, and decreases for moist feed (0.2 is the base FeedDMContent, which gives neutral effect in this equation) and is represented as follows:
$$\eqalign{DryFeedF & = 1\; + \; \lpar {FeedDMContent - \; 0{\cdot}2} \rpar \cr & \quad \times xFDMMastication}$$xFDMMastication is the slope of the linear relationship between FeedDMContent and extra salivation, having a value of 0.179.
TimeIntoMealFactor is a unitless factor forcing a smaller bolus at the beginning of a meal for all foods and late in the meal for dry food, as reported by Gill et al. (Reference Gill, Campling and Westgarth1966). For example,
$$\eqalign{& TimeIntoMealFactor_{BeginningOfMeal} \cr & \quad = \; 1\; - \; 300\; \times \; \lpar {0{\cdot}025 - TimeIntoMeal} \rpar ^2}$$
$$\eqalign{TimeIntoMealFactor_{RestOfMeal} & = \; 1 - \; 600 \cr & \quad \times \lpar {FeedDMContent{\rm \;} - \; 0{\cdot}3} \rpar \cr & \quad \times \lpar {TimeIntoMeal - 0{\cdot}025} \rpar ^2}$$BolusSizeF is a factor (BolusWeightDry/20) increasing salivation during mastication as BolusWeigthDry increases, and xBolusMov (0.6, unitless) is the sensitivity of a number of mastications to bolus size. BaseMasJawMovPerBolus is a constant, 27, derived from the literature (Gill et al., Reference Gill, Campling and Westgarth1966; Boudon et al., Reference Boudon, Acosta, Delagarde and Peyraud2006), representing a base number of total jaw movements per bolus. HungerEffectOnMasticaiton (Unitless) modulates mastication in response to hunger, and it is derived as follows:
$$\eqalign{HungerEffectOnMastication & = \; HungerMasticationF\; \cr & \quad + \; \lpar {1\; - \; HungerMasticationF} \rpar \cr & \quad \times e^{e^{\lpar { - kHungerMastication\; \times \; Hunger} \rpar }}}$$HungerMasticationF is a constant equal to 0.63. For details on the set of equations representing motivation to feed (i.e. hunger, unitless), refer to Gregorini et al. (Reference Gregorini, Beukes, Romera, Levy and Hanigan2013, Reference Gregorini, Villalba, Provenza, Beukes and Forbes2015b). kHungerMastication is a decay constant (6.8, unitless) of mastication time as a function of hunger. kHungerMastication can represent differences due to breed, age and dental efficacy as suggested by Pérez-Barbería and Gordon (Reference Pérez-Barbería and Gordon1998). NDFF represents the effect [linear adjustment, according to Baumont et al. (Reference Baumont, Cohen-Salmon, Prache and Sauvant2004)] of fibre on mastication and salivation and is calculated as ActualNDF/BaseNDF. Base NDF is 590 g/kg of forage DM, and actual NDF is the forage NDF content that MINDY is grazing at a particular time and space. Forage chemical composition in MINDY changes during the day and between sward canopy strata (Gregorini, Reference Gregorini2012; Gregorini et al., Reference Gregorini, Beukes, Romera, Levy and Hanigan2013). SpeciesF (1, 0.6, 0.4 and 1.4, unitless, for C3, legumes, herbs and C4 forages, respectively) is the inherent ease of breakdown (i.e. toughness) of the feed, independently of the NDF factor (Wilson and Mertens, Reference Wilson and Mertens1995; Wilson and Kennedy, Reference Wilson and Kennedy1996).
TimePerBite and TimePerMastication, in Eqn 9, are calculated as follows:
$$\eqalign{TimePerBite & = TimePerJawMovementBase{\rm \;} \cr & \quad \times {\rm \;} HungerEffectOnBiteRate}$$
$$\eqalign{TimePerMasticaiton & = TimePerJawMovementBase{\rm \;} \cr & \quad \times {\rm \;} HungerEffectOnMastication}$$TimePerJawMovementBase (s) is assumed to be equivalent for severing and mastication movements [0.35 × CorrectedBW0.125, (Illius and Gordon, Reference Illius and Gordon1987, Reference Illius and Gordon1992)]. HungerEffectOnBiteRate is a factor (unitless) reducing the time per bite (severing jaw movement) as hunger increases and is calculated as the difference between HungerEffectOnMastication and kHungerBiteRate.
In Eqn 11, BaseSalivaPerJawMovement is the salivation rate (g saliva per severing or mastication jaw movement). xBolusSaliva and xNDFSaliva are unitless constants, 0.595 and 0.3 respectively, representing the sensitivity of salivation during mastication to bolus size and NDF content of ingesta. The factor DroolingSalivaF is calculated as TimeIntoMealFxSalivaTimeIntoMeal, where xSalivaTimeIntoMeal is a constant (1.77, unitless) representing the sensitivity to salivation as meal progresses.
In Eqn 12, the kSalivaSevering is a constant, 0.31, representing the ratio of saliva per severing to saliva per mastication movement. In Eqn 14, LiquidizingSaliva is the proportion of non-absorbed saliva in the bolus, and derived as follows:
$$\eqalign{LiquidizingSaliva =\;& \lcub {Min\lsqb {0{\cdot}49,\; Max\lpar {0{\cdot}3,\; SalivaInBolus }}}\cr & \times \lpar {1 - AbsorvedSaliva \rpar \rpar \rsqb } \rcub}$$The SalivaInBolus and Absorbed Saliva are calculated as
$$SalivaInBolus = \displaystyle{{SalivaPerBolus} \over {SalivaPerBolus + \left( {\displaystyle{{BolusWeigtDry} \over {FeedDMContent}}} \right)}}$$
$$\eqalign{AbsorbedSaliva = & MaxSalivaAbsorption \cr & \times FeedDMContent \cr & + MinSalivaAbsorption \cr & \times \lpar {1 - FeedDMContent} \rpar}$$Equations 24 and 25 are biological assumptions and more data are required. However, these assumptions are supported by the fact that their incorporation in the module improved the representation of both dry feed and grazing/moist feed bolus size (Gill et al., Reference Gill, Campling and Westgarth1966; Boudon et al., Reference Boudon, Acosta, Delagarde and Peyraud2006; Acosta et al., Reference Acosta, Boudon and Peyraud2007). Saliva plays a role (reducing target bolus size) only when its proportion in the bolus is above 0.3 and does not seem to go beyond 0.5 of the bolus, presumably because the bolus would become ‘too liquid’ (Prinz and Lucas, Reference Prinz and Lucas1997; Lucas et al., Reference Lucas, Prinz, Agrawal and Bruce2002).
Particle size distribution of ingesta in the swallowed bolus
Particle size distribution is derived from particle sizes of ingesta in the masticated and salivated swallowed bolus, MasticationJawMovementsPerBolus, and kComminute(i). The latter is the proportion of the content of each bin that is halved upon one mastication jaw movement and therefore moves (is transferred) to the next smaller ‘bin’, assuming that the breakdown is proportional to the particle size.
$$\eqalign{kComminute\; \lpar i \rpar = & \; kComminuteMin\; \cr & + \; \lpar {kComminuteMax\; \ndash \; kComminuteMin} \rpar \cr & \times \displaystyle{{BinMinMeshSize\lpar i \rpar } \over {BinMinMeshSize\; \lpar {14} \rpar }}}$$KComminuteMin and KComminuteMax are functions of ease of particle size breakdown, which depends on the fibre content of the feed and plant species. Note that kComminute[1] represents kComminuteMin, but is not in use (nothing leaves the smallest particle bin) and kComminute[14] represents kComminuteMax.
The model then predicts the SwallowedDistribution (i), which is the distribution (proportion) over the 14 bins totalling to 1. It is the final particle size distribution of ingesta after X number of mastications per bolus, i.e. bolus particle size distribution. The SwallowedDistribution is then collapsed into three proportions of particle sizes to feed the three particle size pools in the rumen (Gregorini et al., Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a), LPart, MPart and Spart (large, medium and small particle size, respectively). The critical size for particles to escape from the rumen has been observed to be 1.2 mm (Poppi et al., Reference Poppi, Minson and Ternouth1981), so SPart represents the proportion of particles of 1.2 mm or less, while MPart and SPart represent the proportion of the particles between 1.2 and 4.8 and 4.8 mm or greater, respectively.
The particle size distribution of swallowed ingesta boluses (P SwallowedLPart, P SallowedMPart and P SwallowedSPart) is used with the DM intake rate (FdRat):
$$FdRat = BolusWeightDry\; \times BolusSwallowingFrequency$$and nutrient fractions of the feed to calculate insoluble nutrient flow (f SwallowedLPart, f SallowedMPart and f SwallowedSPart) into each of the three particle size pools flowing into the rumen:
$$\eqalign{f_{SwallowedLPart} =\;& FdRat \times P_{Swallowed\; LPart} \times \lpar {\,f_{StFd} - f_{StSFd} }\cr&+\; f_{HcFd} + f_{CeFd} + f_{PiFd} + f_{LgFd} + f_{AiFd} \rpar} $$
$$\eqalign{f_{SwallowedMPart} =\;& FdRat \times P_{Swallowed\; MPart} \times \lpar {\,f_{StFd} - f_{StSFd} } \cr &+\; f_{HcFd} + f_{CeFd} + f_{PiFd} + f_{LgFd} + f_{AiFd} \rpar} \; $$
$$\eqalign{f_{SwallowedSPart} =\;& FdRat \times P_{Swallowed\; SPart} \times \lpar {\,f_{StFd} - f_{StSFd} }\cr &+\; f_{HcFd} + f_{CeFd} + f_{PiFd} + f_{LgFd} + f_{AiFd} \rpar}$$fStFd, fStSFd, fHcFd, fCeFd, fPiFd, fLgFd and fAiFd represent the fractional proportions of total starch, soluble starch, hemicelluloses, cellulose, insoluble protein, lignin and insoluble ash, respectively. These inputs are as described previously (Hanigan et al., Reference Hanigan, Bateman, Fadel, McNamara, Smith, Kebreab, Dijkstra, Bannink, Gerrits and France2006; Reference Hanigan, Appuhamy and Gregorini2013). The inclusion of starch and insoluble protein in the LPart pool is a deviation from the original description by Baldwin (Reference Baldwin1995), which seems warranted given that the pool represents both the ruminal mat and also, to a certain extent, non-fermenting fractions of feed which may be caused by a delay in nutrient wetting after food enters the rumen. Also, an insoluble protein associated with the cell wall is not subject to fermentation until it is released from the cell wall matrix (Jung and Allen, Reference Jung and Allen1995).
Particle sizes, pools and passage through the rumen
As described before, LPart, MPart and SPart flow into the rumen after ingestion and oral processing; both processes depending on sward state and condition, forage species and the internal state of the animal. This inflow of particles feeds the three-pool scheme in the rumen of MINDY (see Gregorini et al., Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a for detail of this development in Molly). Each particle size pool in the rumen is subjected to variable rates of rumen digesta outflow and differential degradation.
Passage of MPart and SPart are assumed to be a fractional function of the liquid passage rate and the concentration of particulate DM in ruminal liquid. Initial fractional passage rates (kPMpart and kPSpart) were set to 0.1 and 0.75 at prevailing particulate DM concentrations per litre of total liquid passed. Initial assessment of the use of values closer to 1 for kPSpart resulted in non-biological rumen functions. The use of these values (<1) is supported by observations that particulate matter is retained and not freely flowing with the liquid phase even in cattle consuming lush herbage (Clauss et al., Reference Clauss, Hummel and Streich2006; Reference Clauss, Hume and Hummel2010, Reference Clauss, Lechner, Barboza, Collins, Tervoort, Südekum, Codron and Hummel2011; Lechner et al., Reference Lechner, Barboza, Collins, Fritz, Günther, Hattendorf, Hummel, Südekum and Clauss2010). Implicitly, this reflects the sieving action of the mat, the inability of particles to migrate to the region of the omasal orifice as rapidly as the fluid phase, or a mechanical action within the omasum or abomasum that acts to retard particle flow. Passage of particulate matter from the rumen, PPart, is represented as a weighted average passage rate:
$$\eqalign{PPart = &\; \displaystyle{{QMPart} \over {QMPart + QSPart}} \times kPMPart \cr & + \displaystyle{{QSPart} \over {QMPart + QSPart}} \times kPSPart}$$where QMPart and QSPart are the rumen pools (kg) of MPart and SPart, respectively.
Because of the large differences in passage rates for the two pools, shifts in the distribution of bolus particle sizes due to either the sward features or its interaction with the internal state of the animal (e.g. hunger), ingestion pattern and oral processing will have significant effects on ruminal retention times and thus diet digestibility. In the model, a portion of the microbes is associated with particulate matter. Therefore, increased passage rates associated with reduced particle size will also result in greater passage rates of attached microbes.
Liquid passage from the rumen
At first glance, predicting liquid passage through the rumen seems relatively simple, being equal to the sum of water and salivation inflows, fluid outflow and the net balance of fluid across the rumen wall. However, the significant variations of particle size distribution and flow of saliva with individual ingestive boluses in response to feed and oral processing (Gill et al., Reference Gill, Campling and Westgarth1966; Pérez-Barbería and Gordon, Reference Pérez-Barbería and Gordon1998), passage rate of particles and rumen fermentation pattern within and between meals (Gill et al., Reference Gill, Robinson and Kennelly1999; Gregorini, Reference Gregorini, Cangiano and Brizuela2011, Reference Gregorini2012), as well as diurnal fluctuations of rumen pools and fluid outflows (Dove et al., Reference Dove, Milne, Sibbald, Lamb and McCormack1988; Gill et al., Reference Gill, Robinson and Kennelly1999; Taweel et al., Reference Taweel, Tas, Dijkstra and Tamminga2004), indicate that predicting this phenomenon is complicated and needs a more mechanistic and dynamic approach. Based on the model of ruminal water balance (l/day) incorporated into Molly by Argyle and Baldwin (Reference Argyle and Baldwin1988) and modified by Gregorini et al. (Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a), the passage of liquid through the rumen, RumenLiquidOutflow (l/day), was represented more mechanistically and dynamically as follows:
$$\eqalign{RumenLiquidOutflow = & RumenDigesta \times fLiquidOutflow \cr & \times \; \lpar {1 - RumenDM} \rpar}$$and rumen DilutionRate (%/h), known as the fractional passage of liquid, was calculated as:
$$DilutionRate = \displaystyle{{RumenLiquidOutflow} \over {RumenLiquidVolume}}\; /24$$
$$\eqalign{RumenLiquidVolume & = \; RumenLiquidInflow\; \cr & - \; RumenLiquidOutflow \cr & + \; OsmWater}$$The inflow of liquid to the rumen, RumenLiquidInflow (l), is the sum of water ingestion (moisture content of the feed) and water imbibed. The latter is described in detail by Gregorini et al. (Reference Gregorini, Provenza, Villalba, Beukes and Forbes2018), in a parallel development of MINDY to simulate diurnal patterns of urination and drinking. RumenLiquidInflow also includes saliva, as a product of severing bites and oral processing (as described before), plus salivation while ruminating and resting as described in Gregorini et al. (Reference Gregorini, Beukes, Romera, Levy and Hanigan2013) and Gregorini et al. (Reference Gregorini, Provenza, Villalba, Beukes and Forbes2018). OsmolalWater (l, Eqn 35) is the water flowing in and out of the rumen through the rumen wall as a response to the difference between RumenOsmolaity and blood osmolality (López et al., Reference López, Hovell and MacLeod1994). RumenOsmolality represents ruminal milliosmolality and blood milliosmolality is 280, plus the intercept of Eqn 35, 20 l/day. RumenOsmolality is calculated as the molar sum of soluble carbohydrate (Cs), ammonia (Am), VFA (Acetic, Ac; Propionic, Pr; Butiric, Bu), lactate (La), amino acids (Aa), and soluble ash (As) divided by RumenLiquidVolume (in ml). Moles of soluble ash were calculated by dividing the weight of soluble ash by the molecular weight of sodium bicarbonate and multiplying by an osmolality factor of 1.7, which was derived empirically. The remaining metabolites are predicted from the rumen of MINDY. Water moves from the rumen to blood when RumenOsmolality is <225 and the reverse when it is >225.
$$\eqalign{OsmolalWater =\;& 70 \times \lsqb {\lpar {RumenOsmolality - {\rm BloodOsmolality}} \rpar } \cr & \times 1000 \rsqb + 20}$$In Eqn 32, RumenDigesta (kg) is the sum the rumen DM and liquid [(assuming a density of rumen liquid of 1), RumenLiquidVolume, litres)] contents. Ruminal fluid outflow has been reported to be dictated by total rumen content (Okine et al., Reference Okine, Mathison and Hardin1989; Chilibroste, Reference Chilibroste1999; Schettini et al., Reference Schettini, Prigge and Nestor1999). RumenDM is the proportion of DM of the rumen digesta and fLiquidOutflow is a multifactorial function described as follows:
$$\eqalign{& fLiquidOutflow\; = \cr & \; fLiquidOutMin + \cr & \displaystyle{{\,fLiquidOutMax - fLiquidOutMin} \over {\lcub {1 + e^{[\lpar {kLiquidOutCurvature \times \lpar {kRumenInflection\; - \; RumenLiquidVolume/CorrectedBW} \rpar } \rsqb }} \rcub }} \cr & \times RumenDMF \times BehaviorF\; \times PregnancyF \times SerotoninF} $$fLiquidOutMin and fLiquidOutMax in Eqn 38 represent the lower (0.052) and upper (0.4) fractional (proportion/h) of liquid passage rate through the rumen. kLiquidOutCurvature (179, unitless) is a constant controlling the curvature of the sigmoid dependency of fractional liquid passage rate on RumenDigesta. kRumenInflection is also a stabilizing constant (0.13), representing the proportion of the rumen liquid volume flowing out of the rumen around which rate of liquid passage is growing fastest. RumenDMF is a factor representing how much to slow down fLiquidOutflow when the rumen is ‘dry’, giving a value of 1 when the RumenDM equal 0.14 (RumenDMBase), or a value of 0.89 when RumenDM reaches its driest point (~0.18, RumenDMUpper) and drinking is triggered [see Gregorini et al. (Reference Gregorini, Provenza, Villalba, Beukes and Forbes2018) for details on the representation of drinking behaviour in MINDY]. Simply, RumenDMF represents the slowing down of the liquid passage rate as the rumen becomes drier and is calculated as follows:
$$\eqalign{& RumenDMF = RumenDMFMin + \lpar {1 - RumenDMFMin} \rpar \cr & \quad \times Max\lcub {0,\lsqb {Min\lpar {1,\; RumenDMUpper - RumenDM} \rpar } \rsqb } \rcub / \cr & \quad\lpar {RumenDMUpper - RumenDMBase} \rpar}$$As demonstrated by Balch and Campling (Reference Balch and Campling1962) digesta does not flow through the rumen with the rumen in stasis. See also Owens et al. (Reference Owens, Secrist, Hill and Gill1998), which indicates, as suggested by Deswysen et al. (Reference Deswysen, Ellis and Pond1987), Ulyatt (Reference Ulyatt, Wallace and Bell1983) and Wilson and Kennedy (Reference Wilson and Kennedy1996), that ruminal contractions are necessary for digesta to pass through the rumen (Thiago et al., Reference Thiago, Gill and Sissons1992; Gill et al., Reference Gill, Robinson and Kennelly1999). Seo et al. (Reference Seo, Lanzas, Tedeschi and Fox2007) compiled data from 19 experiments, showing that frequency of rumen contractions differs with behavioural activity: eating, ruminating and idling, had 1.56, 1.12 and 1.13 rumen contractions per min., respectively. Thus, BehaviourF is a representation of this behavioural modulation. BehaviourF takes the values of Seo et al. (Reference Seo, Lanzas, Tedeschi and Fox2007) divided by 1.56. This modulation factor helps to represent fluctuation of liquid outflow from the rumen (and particulate matter carried in it) throughout the day, as response to feeding management and or diet that alter behavioural (i.e. eating time, ruminating and idling) time budgets.
The PregnancyF (unitless) and SerotoninF (unitless) in Eqn 36 are factors representing the effect of pregnancy stage on rumen capacity and the effect of serotonin on motivation to eat and peristaltic movements of the rumen. The former accelerates RumenLiquidOutflow during eating as the gravid uterus increases in volume, causing a faster liquid passage rate of digesta. This is supported by the results of Vanzant et al. (Reference Vanzant, Cochran and Johnson1991), who reported a faster passage rate of the rumen liquid for pregnant and lactating than for non-lactating cattle at the same forage digestibility; by the report of Hanks et al. (Reference Hanks, Judkins, McCracken, Holcombe, Krysl and Park1993), who showed that pregnancy leads to a greater particle passage rate, a shorter ruminal and total gastrointestinal retention time in cattle; and by the results of Coffey et al. (Reference Coffey, Paterson, Saul, Coffey, Turner and Bowman1989) and Gunter et al. (Reference Gunter, Judkins, Krysl, Broesder, Barton, Rueda, Hallford and Holcombe1990) with sheep, showing increments of particle passage rates with advancing pregnancy.
$$PregnancyF = \displaystyle{1 \over {{\left( {\displaystyle{1 \over {GravidUiterusWeight}}} \right)}^{kPregnancyRumen}}}$$where kPregnancyRumen (0.61, unitless) is a constant controlling the acceleration of liquid passage rate as the foetus grows (due to increased pressure on the rumen).
SerotoninF is calculated as follows:
$$SerotoninF{\rm \;} = 1 + kLiquidSerotonin \times \lpar {1 + Serotonin} \rpar $$where kLiquidSerotonin is a constant (0.5, proportion), representing the proportional change of liquid passage rate due to serotonin fluctuations. In MINDY, Serotonin (0–1, unitless) is a proxy of the effect of the diurnal fluctuations of serotonin and its effect on feeding motivation and rumen function (Gregorini, Reference Gregorini, Cangiano and Brizuela2011, Reference Gregorini2012). The effect diurnal fluctuation of light intensity on the hypothalamic suprachiasmatic nucleus has been hypothesized to be related to the secretion of melatonin (Gregorini, Reference Gregorini2012). Diurnal fluctuations in melatonin release (greater during the dark and lower during light periods) have been documented in domestic and wild ruminants (Gregorini, Reference Gregorini2012). Melatonin is synthesized from tryptophan derived from serotonin, which explains the diel rhythmic patterns of serotonin depletion during the late afternoon to early evening and replenishment from dawn onwards. Increased levels of serotonin may inhibit the reward functions at the mesolimbic system, diminishing motivation to feed (Pittroff and Soca, Reference Pittroff, Soca and Bels2006). Serotonin has also been related to gastric emptying dynamics in ruminants through the effects on cholecystokinin release (Pittroff and Soca, Reference Pittroff, Soca and Bels2006). Serotonin has been related to reductions in gastric emptying by augmenting the secretion of, and response to, cholecystokinin (Hayes et al., Reference Hayes, Moore, Shah and Covasa2004; Li et al., Reference Li, Wu and Owyang2004). Cholecystokinin reduces the intensity and frequency of the reticulum-rumen contractions (Bruce and Huber, Reference Bruce and Huber1973) and reduces the opening size of the pyloric orifice, increasing the retention time of digesta in the rumen and satiety signals from it, thereby reducing ingestion rates (Pittroff and Soca, Reference Pittroff, Soca and Bels2006). Therefore, rhythmic diel variations in serotonin depletion counteract the effects of cholecystokinin. This phenomenon is supported by the results of Dove et al. (Reference Dove, Milne, Sibbald, Lamb and McCormack1988) with grazing ewes, who reported the fastest digesta flow during and immediately after the dusk grazing.
$$Serotonin = \left[ {\displaystyle{{\lpar {DielRythm + 1} \rpar } \over 2}} \right]^{kSerotonin}$$
$$\eqalign{CircadianRythm =\; & Sine \times Pi \times \lsqb {2 \times Pi \times \lpar {RelativeTime }} \cr & - DielRythmLag \rpar \rsqb}$$where kSerotonin is 1.81 (unitless) modifying the sine wave shape. RelativeTime is the time of day (d) relative to the time of sunrise and sunset (d); and, DielRythmLag is a constant, 0.045 (d) set to 1 h to create a delay due to build-up time of serotonin and melatonin during day/night, respectively.
Illustrations
The present work focuses on formulating and describing the structure and function of the new development in MINDY, with a preliminary conceptual validation (Rykiel, Reference Rykiel1996) conducted for different feeding scenarios. Validation means that the model is acceptable for its intended purpose because it meets specified performance requirements (Rykiel, Reference Rykiel1996). The purpose of the present development was to integrate existing knowledge of interactions between ingestion and digestion processes in MINDY by including explicit oral processing of ingesta and simulating – rather than setting – fractional passage of liquid in the rumen as a response to behaviour/ foraging context. The context for the model is a foraging ruminant. The theoretical model performance was assessed subjectively, as suggested by Rykiel (Reference Rykiel1996), by MINDY's ability to simulate realistic patterns of oral processing of ingesta in response to various grazing scenarios, commonly present in intensive pastoral dairy farms.
For illustrative purposes, MINDY was confronted with: (1) herbage of contrasting forage species, only differing in SpeciesF (a proxy for forage species toughness); (2) a period of restriction in available grazing time, with or without a maize grain meal before allocation of a new pasture strip; and (3) a factorial arrangement of herbage allowances and sward surface heights. In other words, the goal was to illustrate how MINDY's diurnal patterns of ingestion and oral processing of ingesta are sensitive to the toughness of herbage, hunger and sward structure, and how, would that change digesta outflows dynamics from the rumen. In all simulations, the outputs required from MINDY were: intake rates, masticatory behaviour, ingestive bolus weight, particle size distribution in the bolus and rumen dilution rate.
The effect of forage species with contrasting features for oral comminution
The distribution of ingesta through different particle size pools in the rumen digesta is a function of ingestive actions and oral processing forming the swallowed bolus (Moseley and Jones, Reference Moseley and Jones1984; Spalinger et al., Reference Spalinger, Robbins and Hanley1986; Pond et al., Reference Pond, Ellis, Lascano and Akin1987; Prinz and Lucas, Reference Prinz and Lucas1997). Thus, forage species, sward structure, herbage chemical composition and plant phenology are key factors determining particle size distribution of the swallowed bolus (Wilson and Kennedy, Reference Wilson and Kennedy1996; Poppi et al., Reference Poppi, France, McLennan, Theodorou and France2000; Kennedy, Reference Kennedy, Dijkman, Forbes and France2005). Although this phenomenon is well documented in the literature and some models have attempted to include it implicitly (Sauvant et al., Reference Sauvant, Baumont and Faverdin1996; Baumont et al., Reference Baumont, Cohen-Salmon, Prache and Sauvant2004), there is a lack of information on the effect of plant-related comminution properties (e.g. toughness) on patterns of bolus size, particle size distribution, intake rate and rumen dilution rate. The current work simulated a scenario where MINDY [initialized as a pregnant Friesian dairy cow (500 kg liveweight) was in mid-lactation (150 days in milk)] strip-grazing monoculture swards of Lolium perenne, Medicago sativa and Pennisetum clandestinum. All swards had a height of 30 cm and herbage mass of 3000 kg DM per ha. The grazing area allocated to MINDY was 100 m2 and pasture was allocated after the morning milking (08:00 h).
Figure 3 presents MINDY's diurnal patterns of ingestive bolus particle size distribution, bolus weight, mastications per bolus, bolus swallowing frequency, rumen dilution rate and herbage intake rate. These results indicate different ingestive-oral processing dynamics as a function of forage species. This dynamic in turn determines the variations in intake pattern and rumen dilution rate among forage species.
Fig. 3. Predicted effect of herbage toughness, i.e. SpeciesF, on ingestive (swallowed) boluses particle size distribution, boluses weight, mastications per bolus, bolus swallowing frequency, rumen dilution rate and herbage intake rate. In a–c: FLPartSwal, FMPartSwal and FSPartSwal are large, medium and small particles, respectively. In d–h: Solid line is Lolium perenne; dotted line is Pennisetum clandestinum; and dashed line is Medicago sativa.
As constraints in forage comminution increase, i.e. Medicago sativa, Lolium perenne and Pennisetum clandestinum, respectively, the proportion of swallowed large particles (FLPartSwal) increases, with a significant reduction of the proportion of medium and small particles in the swallowed boluses (Fig. 3, a–c). These results are supported by the classical works of Poppi et al. (Reference Poppi, Minson and Ternouth1981), McLeod and Minson (Reference McLeod and Minson1988), Luginbuhl et al. (Reference Luginbuhl, Pond, Burns and Russ1989) and Bailey et al. (Reference Bailey, Erdman, Smith and Sharma1990), who studied the particle size reduction during mastication of different forages (C3, C4 and legumes) and plant parts (leaf and stem) by cattle and sheep. The within, and between, meal variations in bolus particle size distribution reveals the capabilities of MINDY to simulate the effect of these plant characteristic throughout the day, in conjunction with grazing management and the internal state of the animal. Mastication dynamic within a meal is also a function of an animal's motivation to feed (Gregorini, Reference Gregorini, Cangiano and Brizuela2011) and diurnal arrangement of meals, thus it also depends on grazing management (Gregorini, Reference Gregorini2012), as presented in the next section.
MINDY predicts differences in bolus size (DM) according to forage species (Fig. 3e). These model outputs are consistent with the reports of Kennedy and Murphy (Reference Kennedy and Murphy1988), Wilson and Kennedy (Reference Wilson and Kennedy1996) and Kennedy (Reference Kennedy, Dijkman, Forbes and France2005), reporting the effect of forage species on mastication dynamics. Both bolus size and mastication dynamics determine the bolus swallowing frequency (Prinz and Lucas, Reference Prinz and Lucas1997). Swallowing frequency reflects one of the main ingestive constraints to herbage intake, which is posed by comminution characteristics of forages, as demonstrated by MINDY's intake rate dynamic and diurnal pattern of herbage intake (Fig. 3h).
Ruminal contractions increase almost exponentially with eating rate (Freer and Campling, Reference Freer and Campling1965) and accelerate liquid and particulate passage rates through the rumen as demonstrated by Okine et al. (Reference Okine, Mathison and Hardin1989), and reported and modelled by Seo et al. (Reference Seo, Lanzas, Tedeschi and Fox2007). Thus, in conjunction with oral processing, herbage intake pattern has a major influence on digesta outflow from the rumen (Gregorini et al., Reference Gregorini, Gunter and Beck2008; Gregorini, Reference Gregorini2012), as indicated by MINDY's rumen dilution rate outputs (Fig. 3d). Liquid flowing out of the rumen is the medium by which solids flow out of the rumen and is the way particle outflow was modelled in the current development of MINDY and the latest development of Molly (Gregorini et al., Reference Gregorini, Beukes, Waghorn, Pacheco and Hanigan2015a). The present results of these modelling illustrations support the concept that forage species, as characterized by chemical and biomechanical features, influence ingestion and thereby digestion dynamics, which influences herbage intake and its patterns and thereby rumen fermentation patterns. The latter is of particular interest from environmental (e.g. enteric methane emission, N excretion), nutritional (nutrient supply to the host animal) and animal welfare (minimal total discomfort and rumen health) standpoints (Gregorini et al., Reference Gregorini, Villalba, Chilibroste and Provenza2017).
The effect of hunger on oral processing of ingesta and rumen dilution rate
In a review of behavioural adaptations of dairy cows to changes in grazing management, Chilibroste et al. (Reference Chilibroste, Gibb, Soca and Mattiauda2015) concluded that most of the available information focusing on short-term ingestive responses (i.e. herbage intake rate) lacked essential links with the internal state of the animal, i.e. hunger, and post-ingestive behaviour such as rumen function. Hunger level influences feeding motivation (Forbes and Gregorini, Reference Forbes and Gregorini2015). The latter modulates the dynamics of ingestive tactics, including intake rate and mastication, and consequently digestive patterns within and between meals. Hungry animals reduce mastication (i.e. oral processing) to increase herbage intake rate, swallowing boluses with larger particles and thus increasing rumen retention time of digesta (Greenwood and Demment, Reference Greenwood and Demment1988; Chilibroste et al., Reference Chilibroste, Soca, Mattiauda, Bentancur and Robinson2007; Gregorini, Reference Gregorini, Cangiano and Brizuela2011).
To evaluate MINDY's oral processing of ingesta and digesta outflow from the rumen in response to hunger, four scenarios were simulated where MINDY [initialized as a pregnant Friesian dairy cow (500 kg liveweight) was in mid-lactation (150 days in milk)]: (a) Non-fasted, strip-grazing a sward of Lolium perenne with a surface height of 30 cm and an herbage mass of 3000 kg DM per ha, allocated 100 m2 after the morning milking (08:00 h); (b) Fasted, strip-grazing the same sward as in (a) but allocated 100 m2 after the afternoon milking (4 pm) for only 4 h; (c) Non-fasted plus supplement, strip-grazing the same sward and being fed three kg DM of maize grain during the afternoon milking; and (d) Fasted plus supplement, strip-grazing the same sward as in (b) and being fed 3 kg DM of maize grain during the afternoon milking. Figure 4a-g present the effect of hunger (as set by scenarios a–d) on MINDY's diurnal patterns of ingestive bolus particle size distribution, bolus weight, mastications per bolus, rumen dilution rate and herbage intake rate. These results indicate different dynamics in the oral processing in response to hunger, which in turn results in variations in intake pattern and rumen dilution rate.
Fig. 4. Predicted effect of hunger (fasting) as modulated by feeding management (supplementation, cracked maize grain), on diurnal pattern of swallowed boluses particle size distribution, boluses weight, mastications per bolus, intake rate and rumen dilution rate. Left panel (a–d): FLPartSwal, FMPartSwal and FSPartSwal are large, medium and small particles, respectively. Right panel (e–h): Doted line is pasture allocation at 8 am; Dashed line is pasture allocation at 4 pm with stand-off between milkings; Solid line is pasture allocation at 8 am with a meal of 3 kg DM of cracked maize grain during the afternoon milking; and Small dotted line is pasture allocation at 4 pm, with stand-off (fasting time off the pasture) between milkings and a meal of 3 kg DM of cracked maize grain during the afternoon milking.
As hunger levels increase, i.e. from scenario c, to a, to d and b, the proportion of swallowed large particles (FLPartSwal) increases, with a reduction of the proportion of small particles in the swallowed boluses. These results are supported by Greenwood and Demment (Reference Greenwood and Demment1988) and Chilibroste (Reference Chilibroste1999), who report an increase in the particle size of forage flowing into the rumen of beef and dairy cattle, respectively, as a response to increments in fasting periods, i.e. hunger. Figure 4g helps explain these differences, which are especially marked at the beginning of the meal after the afternoon milking. The ‘hungrier’ MINDY ‘felt’ before a meal, the fewer mastications per ingestive bolus were made. This phenomenon is also evident during supplement consumption. The boli were heavier and much less masticated in scenario d than c. Gregorini et al. (Reference Gregorini, Soder and Kensinger2009b) reported a marked reduction in oral processing jaw movements including mastication as hunger level of dairy cows foraging Dactylis glomarata increased. These and similar results reported by Gregorini et al. (Reference Gregorini, Gunter, Masino and Beck2007) for beef cattle grazing Cynodon dactilon swards support these outputs of MINDY.
Oral processing, and thereby particle size of ingestive boluses, and eating activity have a strong influence on rumen digesta outflow. Daily mean digesta passage rates through the rumen in scenarios c, a, d and b were, 0.249, 0.218, 0.255 and 0.207 kg of dry digesta per hour, respectively; while daily mean rumen dilution rate was 0.214, 0.210, 0.163 and 0.160 of the rumen liquid pool per hour, respectively. The diurnal patterns of the different rumen dilution rates are evident in Fig. 4g, where marked differences exist between fasted (slower) and non-fasted scenarios. The particular scenario d illustrates the links between oral processing (boluses particle size, eating activity) rumen peristaltic movements and digesta outflow from the rumen. Although MINDY was fasted for 20 h, by the time it was allocated to the pasture, it had just consumed 3 kg of maize at the greatest rate (it was ‘hungry’) with the lowest masticatory rate and heaviest boluses; collectively, reducing hunger and motivation to graze (especially at the beginning of the meal) as shown in Fig. 4h. Because of this reduced hunger level, each grass bolus was masticated more (Fig. 4f) compared with the other scenarios, and boluses with a lower proportion of large and greater proportions of medium and small particles (Fig. 4a-d) were swallowed. Moreover, almost all of the time at pasture was spent grazing, but slower at the beginning of the meal (as a response to a lower level of hunger). Both scenarios help explain the rumen dilution rate dynamic during the meal, and the faster (compared with the other scenarios) solid digesta outflow rate during eating and daily mean (Fig. 5).
Fig. 5. Predicted effect of hunger (Fasting) as modulated by feeding management (cracked maize grain supplementation) on diurnal pattern of rumen digesta outflow rate. Small dotted line is pasture allocation at 8 am; Solid line is pasture allocation at 4 pm with stand-off between milkings; Small dashed line is pasture allocation at 8 am with a meal of 3 kg DM of cracked maize grain in the afternoon milking; and Dashed line is pasture allocation at 4 pm, with stand-off (fasting time off the pasture) between milkings and a meal of 3 kg DM of maize in the afternoon milking.
Together, these results show that MINDY, as in real ruminants (Laca et al., Reference Laca, Ungar and Demment1994), reduces mastication as a compensatory mechanism to increase intake rate, but swallows longer/larger particles of herbage, as shown with cattle by Chilibroste (Reference Chilibroste1999) and Greenwood and Demment (Reference Greenwood and Demment1988), which, in turn, increases rumen retention time of digesta and slows down the rumen dilution rate as hunger increases. Such slowing down of rumen dilution rate is supported by Gregorini et al. (Reference Gregorini, Gunter and Beck2008) for beef cattle and Gregorini et al. (Reference Gregorini, Villalba, Chilibroste and Provenza2017) for dairy cows.
The effect of herbage allowance and sward surface height
The new developments in MINDY include the addition of oral processing and the resultant changes in bolus particle size distribution, along with mechanistic and dynamic water ingestion and the fractional passage of liquid through the rumen. Collectively, they enhance the model's capability to explore foraging situations where the dynamics of herbage ingestion and oral processing is expected to alter patterns of intake and digestion, as well as nutrient supply to the host animal from the rumen. This development, in turn, improves simulations in which variations of animal performance and nutrient excretion are expected to be modulated, passively, by sward features, or actively by grazing management (Gregorini et al., Reference Gregorini, Provenza, Villalba, Beukes and Forbes2018). These hypotheses were challenged by setting a factorial arrangement of scenarios between four herbage allowances [25, 30, 35 and 40 kg DM (above-ground)/cow/day] and three sward surface height (15, 22 and 30 cm, extended tiller height). In each scenario, MINDY [initialized as a pregnant Friesian dairy cow (500 kg liveweight) in mid-lactation (150 days in milk)] strip-grazed a sward of Lolium perenne with a herbage mass of 3000 kg DM/ha, allocated after the morning milking (08:00 h). Herbage allowances were created by changing the daily area allocated to MINDY: 83,100, 116, and 133 m2.
Figure 6a-h present the effect of herbage allowance and sward surface height on daily herbage intake, rumen function, methane yield and urinary N excretion. These results indicate:
Fig. 6. Predicted effect of herbage allowance (kg DM above ground/cow/day) and sward surface height, on (a) herbage DM intake, (b) herbage intake rate, (c) rumen digestibility, (d) Digesta outflow rate, (e) post-grazing herbage mass, (f) milk yield, (g) rumen. Low (dotted lines), 15 cm; Medium (dashed lines), 22.5 cm and High (solid lines), 30 cm sward surface height (SSH) as extended tiller height. (h) Methane yield (thick lines) and urinary N excretion (thin lines) at Low (dotted lines), Medium (dashed lines) and High (solid lines) SSH.
First, that increasing the herbage allowance leads to greater herbage intake with a variable diminishing response but lower grazing efficiency (Fig. 6a and e, respectively). This response was expected, being well-known (Pérez-Prieto and Delagarde, Reference Pérez-Prieto and Delagarde2012; Reference Pérez-Prieto and Delagarde2013). The variability of such a response depended on the sward surface height, indicating that at the same herbage allowance with the same herbage mass, sward structure (i.e. sward canopy height) will determine herbage intake, its rate and harvesting efficiency. The current results suggest that increments in sward surface height, at the same available herbage mass, can increase herbage accessibility and facilitate the harvesting process, leading to greater intake rates and daily herbage intakes. Such a response relates to changes in sward canopy structure and vertical distribution of herbage mass and its morphological components. These outputs of the model are supported by literature reporting investigations of the functional response of large ruminants under relatively homogeneous swards (Laca et al., Reference Laca, Ungar, Seligman and Demment1992; Reference Laca, Ungar and Demment1994; Demment and Laca, Reference Demment and Laca1994; Gregorini et al., Reference Gregorini, Gunter, Beck, Caldwell, Bowman and Coblentz2009a; Reference Gregorini, Gunter, Bowman, Caldwell, Masino, Coblentz and Beck2011; Carvalho et al., Reference Carvalho, Bremm, Mezzalira, Fonseca, da Trindade, Bonnet, Tischler, Genro, Nabinger and Laca2015).
Secondly, digestive (and excretory) responses are linked strongly to ingestive patterns and grazing management. Green leaf content is the sward component that promotes herbage intake (Burns and Sollenberger, Reference Burns and Sollenberger2002) because the amount of green leaf is better correlated with bite than herbage mass per se (Wade and Carvalho, Reference Wade, Carvalho, Lemaire, Hodgson, Moraes, Nabinger and Carvalho2000). Moreover, accessibility to leaves is closely related to the N content of herbage consumed in each bite (Bailey et al., Reference Bailey, Gross, Laca, Rittenhouse, Coughenour, Swift and Sims1996; Drescher, Reference Drescher2003). According to Waite (Reference Waite, Worden, Sellers and Tribe1963) the upper strata of the sward canopy is removed first; therefore, the quantity and quality (e.g. protein content) of the diet selected and eaten by grazing cattle depends on the level of herbage depletion and sward structure (Chacon and Stobbs, Reference Chacon and Stobbs1976). In MINDY, increments of sward surface height at the same herbage mass lead to a square (rather than triangular with the base at the bottom) distribution of herbage mass in sward canopy, increasing the leaf content of the upper grazing strata. Moreover, in MINDY, quality of herbage and leaf content of the sward canopy strata diminish from top to bottom (for details see equations in Gregorini et al., Reference Gregorini, Beukes, Romera, Levy and Hanigan2013). Thus, and as reported in the literature (Wade and Carvalho, Reference Wade, Carvalho, Lemaire, Hodgson, Moraes, Nabinger and Carvalho2000), in temperate swards, increments in herbage allowance and leaf accessibility (e.g. through an increment in sward height, Gregorini et al., Reference Gregorini, Gunter, Beck, Caldwell, Bowman and Coblentz2009a) would increase leaf proportion in the diet. The latter is supported by the reduction in grazing efficiency (Fig. 6e) and thus increments if herbage DM (Fig. 6a) and thereby N intake rate.
Although an increased herbage allowance increases DM intake and its rate, it reduces rumen digestibility. The latter is explained by the faster digesta outflow from the rumen (Fig. 6d) and rumen dilution rate (Fig. 6g). These relationships are documented in the literature (Poppi et al., Reference Poppi, France, McLennan, Theodorou and France2000) and show the benefits of this new development in MINDY. Reductions in rumen retention time diminish methane yield, which adds to the benefits of greater milk production (Fig. 6h). However, and as a product of greater N intake, urinary N excretion increases (Fig. 6h). The magnitude and pattern of this trade-off varies, and quite significantly, with sward surface height. Thus, the three-dimensional arrangement of herbage mass should not be ignored in managing grazing and or selecting forage cultivars with low N. The latter is important for temperate grazing systems; known by the excess of N supply to inefficient N users such as cattle (cattle use of N rarely exceeds 0.30; Dijkstra et al., Reference Dijkstra, Oenema, Van Groenigen, Spek, Van Vuuren and Bannink2013; Gregorini et al., Reference Gregorini, Beukes, Dalley and Romera2016). In these systems an excess of urinary N load onto pastures increases N leaching, leading to pollution of water resources.
The pollution swapping between methane emission and urinary N excretion is always present (Dijkstra et al., Reference Dijkstra, Oenema and Bannink2011; Gregorini et al., Reference Gregorini, Beukes, Dalley and Romera2016), but increments in the environmental impact of increased methane yield are easily offset by reductions in urinary N excretions (Dijkstra et al., Reference Dijkstra, Oenema, Van Groenigen, Spek, Van Vuuren and Bannink2013; Gregorini et al., Reference Gregorini, Villalba, Chilibroste and Provenza2017). Moreover, increments in herbage allowance and accessibility reduce rumen digestion (Fig. 6c) and thereby fibre digestibility, the cheapest source of nutrient in pastoral systems. Within this context, the present model outputs suggest that, at the same herbage mass and allowance, increments of sward surface height of forage species with similar chemical composition should be re-considered and further evaluated.
Summary and conclusions
The model development presented in the current paper makes explicit the functional relationships among direct and indirect controls of ingestion and rumen digestion. Although additional statistical evaluations are required, and more data may be needed to further define some parameters, MINDY theoretical validation indicates that patterns of herbage intake and oral processing of ingesta are reproduced realistically, achieving the sensible (and realistic) effect on rumen digesta outflow from a grazing dairy cow consistent with basic knowledge reported in the literature. The model's representation of those functional relationships allows simulating grazing management and its effects more comprehensively and realistically. Therefore, the new concepts encoded in MINDY capture many of the underlying biological mechanisms that influence and link the effect of ingesta oral processing and digesta outflow from the rumen. MINDY's new development can then help in advancing the understanding and nutritional ecology of foraging, grazing patterns and their management for environmental protection. Previous modelling efforts on forage ingestion and digesta outflow from the rumen have been either purely empirical or not comprehensive enough to include these more complex concepts. Therefore, this current iteration of MINDY represents a step forward, but the model offers promise as a heuristic tool for feed intake and grazing process research and as an informative tool for grazing and cow management decisions.
Acknowledgements
We gratefully acknowledge Drs. Alvaro Romera and Elena Minee (DairyNZ, New Zealand) and Rachael Bryant (Lincoln University, New Zealand) for reviewing the typescript, providing valuable comments, suggestions and constructive criticism.
Financial support
This model development was funded by New Zealand dairy farmers through DairyNZ Inc.
Conflicts of interest
None
Ethical standards
The authors declare that this work complies with ethical standards required by DairynZ Inc.
