Revisiting tropical pasture intake: what has changed in 50 years?

Pasture quality

Tropical forages are noted for their poor nutritive value relative to temperate forages. In part this is attributed to rapid maturation after summer rains, leading to increases in the fibre fraction, and particularly lignin, and concomitant reductions in N content (Minson 1981a, 1981b; Poppi and McLennan 2010). Course fibrous feed compromises the ability of cellulolytic bacteria to degrade structural carbohydrates, leading to reduced digestibility (Minson 1981b) and low rate of passage from the rumen (Forbes 2003). The rate of passage is directly related to intake and this is controlled primarily by the neutral detergent fibre (NDF) content of the diet (Mertens 1987). The close association between the NDF, apparent digestibility and DM intake (DMI) is well understood for mixed diets (NASEM 2016). For pasture-based diets, the plant species (C₃ versus C₄, grass versus legume) can have modifying effects on rate of passage and the relationship between NDF, digestibility and intake (Krizsan et al. 2010).

The importance of digestibility in the control of tropical pasture intake has been established for decades. For example, Topps (1969) presented data showing the characteristic increase in organic-matter (OM) intake as OM digestibility increased in steers of varying bodyweight (BW) grazing veld grasses. However, even in that early study, the author noted that ‘It would seem that the relationship between intake and digestibility for subtropical grasses is not so close as that proposed by Blaxter et al. (1961) for temperate roughages’ (Topps 1969, p. 253). Nevertheless, numerous papers since then have demonstrated that digestibility of tropical forages is central to regulating intake. Within experiments, numerous equations have described this relationship (Hodgson et al. 1977; Coleman et al. 2014). Hodgson et al. (1977) summarised several early papers that suggested that the intake response to increasing digestibility was less above an OM digestibility of 60% (predominantly temperate forages), than it was below 60% OM digestibility (predominantly tropical forages). The authors suggested separate relationships of tropical and temperate forages, as follows:

where OMI is OM intake in g/kg BW and OMD is OM digestibility as a percentage. This suggestion that the relationship between digestibility of tropical or temperate forages and intake differs has been explored further (Freer 1981; Minson 1981a). Minson (1981a) concluded that at any given digestibility, the intake of tropical forages may be as much as 20% higher than that of temperate forages. However, the source of the data used was not cited and was represented by a graph with just seven points (reproduced here as Fig. 1a). It could equally be argued that a single line representing both tropical and temperate forages was just as applicable. Indeed, when data also presented in Minson (1981a) and representing different tropical and temperate grass species are depicted graphically (Fig. 1b) a single relationship is apparent. We therefore question the notion that tropical forages have a higher intake than do temperate forages at the same digestibility. This is important because the higher intake for tropical grasses has been incorporated into the PISC (2007) equation for the determination of intake of tropical forages (PISC 2007). This model has been widely used as the basis for estimating intake of tropical pastures by researchers and extension specialists (McLennan 2014). Minson (1981a) postulated sound morphological reasons why higher intake of a tropical (C₄) grass compared with a temperate (C₃) grass of similar digestibility might be expected. He pointed out that at 60% digestibility, a tropical grass is still vegetative while a temperate grass is reproductive. Differences between C₃- and C₄-dominated pastures in spatial distribution, forage structure, feed-component selectivity, rumen breakdown and rate of passage are all factors that potentially confound relationships between digestibility and intake between tropical and temperate pastures.

Fig. 1.

The relationship from Minson (1981a) supporting the view that (a) intake of tropical forages is 20% higher than that of temperate forages of similar digestibility and (b) the relationship derived from table 8.1 in the same paper. Red squares denote tropical forages, blue circles temperate forages.

At equal NDF, intake of temperate legumes is approximately 20% higher than that of grasses (Freer and Jones 1984). Thornton and Minson (1973) compared eight grasses and six legumes and confirmed that intake of legumes was 28% higher than that of grasses in sheep, due to shorter retention time in the rumen. There are limited data on the relative intake of tropical grasses and legumes. Typically, when legume content of tropical grass-based diets is increased, there is both a legume effect and a diet-quality effect, which cannot be separated (Stifkens et al. 2022). However, Archimède et al. (2011) conducted a meta-analysis of data, comparing 28 tropical C₄ grasses with 12 tropical legumes. Across all diets, legumes were lower in NDF, higher in crude protein (CP), but not higher in intake (Table 1). However, OMD was much lower for legumes (48%) than grasses (62%), which biased the data. Consequently, intake relative to OMD was 35% higher for tropical legumes than for C₄ grasses. One explanation for this is that lignification reduced fibre digestibility of legumes but did not have as marked an effect on intake. Indeed, the disparity between tropical legumes and C₄ grasses in OMI relative to NDF was much less (20%) than in OMI relative to OMD (Archimède et al. 2011).

Research by Poppi et al. (1981) demonstrated that intake of leaf was higher than that of stem in C₄ grasses, even though both were of similar NDF content and digestibility. A similar observation was made for the tropical legume Lablab purpureus by Hendricksen et al. (1981). Both authors attributed this to the faster comminution of leaf than stem particles during rumination, allowing for earlier escape from the rumen via the reticulo-omasal orifice.

Low nutritive value of tropical pastures, especially in the dry season, can be offset to some degree by selective grazing for higher-quality components of the pasture (Topps 1968; Thomas et al. 2010). Fig. 2 shows results over a 12-month grazing trial at Lansdown Research Station in northern Queensland (19°39′S, 146°50′E; E Charmley, P McSweeney, P Greenwood, S Denman, G Bishop-Hurley, unpubl. data). Samples of the standing herbage, harvested at 50 mm above ground level were analysed for CP by using near-infrared (NIR) spectroscopy (Coates and Dixon 2011) and compared with CP measured by NIR in faecal samples. Dietary CP content was consistently higher than pasture CP (Fig. 2a), especially when pasture CP content was low (Fig. 2b), indicating that selectivity is influenced by change in diet quality.

Fig. 2.

Effect of season (a) on pasture (blue circles) and diet (red squares) crude protein concentration measured using NIR and (b) on diet crude protein compared with pasture crude protein (blue circles), measured at varying times over 12 months in 2021. Data from Charmley et al. (unpubl. data).

Nitrogen is essential to provide the building blocks for microbial growth. According to PISC (2007), optimum microbial protein synthesis of 130 g/kg digestible OM (DOM) can be achieved when N supply is above 30 g/kg DOM. Below this concentration, N availability in the diet limits the growth of rumen microbes, slows rate of DM degradability, rate of passage from the rumen, and hence intake (Lazzarini et al. 2009). Supplementation of with N up to approximately 1.3% DM (Batista et al. 2017) increases microbial protein synthesis, which can be accompanied by increased rate of passage and intake (Hunter and Siebert 1985; McLennan et al. 2017). However, Panjaitan et al. (2010) demonstrated that the relationship between forage CP content and efficiency of microbial protein synthesis is not simple, with different responses being observed between tropical grasses of similar NDF. This was possibly related to the interactions among varying values of N, lignin and digestibility among species. In low-N diets, if N content per se elicits an additional influence on intake over and above digestibility, then a curvilinear or broken-stick response on intake to N would be expected. In some datasets, this is apparent (e.g. Hunter and Siebert 1987) but in others it is not. Fig. 3 summarises data from six trials with tropical grasses. It shows the change in OM intake of tropical grasses in relation to increased N content. Over a three-fold range in N content, which covers the expected point at which N limitations on microbial protein synthesis would cease, there was no evidence of curvilinearity or a break point in slopes. For practical purposes, it appears that the digestibility effect on intake is so large that it effectively masks the nuances around supply of N for microbial protein synthesis to promote intake.

Fig. 3.

The relationship between N content of the diet and OM intake (OMI) for a range of tropical forages. The arrow indicates approximate N content range at which the N effect on intake would be nullified (Siebert and Hunter 1977; Poppi and McLennan 1995; Panjaitan et al. 2010; Kennedy and Charmley 2012; Batista et al. 2017; McLennan et al. 2017).

Pasture availability (relative intake)

The adequate supply of pasture and the conditions under which that supply is presented have major implications for voluntary intake under extensive grazing systems of northern Australia. Despite a large amount of research into the ecological and grazing aspects of foraging theory, there is little information on how these influence pasture intake. This section will consider the factors that influence intake, and how they may affect the ability to measure or estimate intake under such conditions.

The relationship between the grazing animal and the sward is complex and defined across spatial and temporal scales (Varva and Ganskopp 1998; Roguet et al. 1998). Nutritional optimisation is but one of several factors that influence intake. Other factors include environmental (paddock size, distance to water, shelter, predators; Hunt et al. 2007), animal (breed, genetics, social hierarchy within the group, cognitive ability; Roguet et al. 1998) and sward (Benvenutti et al. 2009) factors. Roguet et al. (1998) identified four spatiotemporal scales of grazing behaviour, namely, bite, feeding site, feeding station and habitat. Pasture intake is influenced by variables at each scale.

Cumulative intake over a period such as a day is the product of the number and mass of individual bites. Extensive research has demonstrated that bite mass is strongly influenced by diet quality as defined by digestibility, fibre content or leaf:stem ratio. Early research with the tropical grass Setaria sphacelata by Chacon and Stobbs (1976) elegantly demonstrated that cattle graze in horizons, with the uppermost horizon grazed first, comprising mostly leaf. As grazing continues and the upper horizon becomes depleted, cattle graze proportionally more of lower horizons that have a higher proportion of stem and less leaf. Bite mass declines as biomass is progressively removed from the sward, forcing cattle to graze in lower horizons with more stem and less leaf. Cattle initially compensate by increasing bite rate and grazing time. However, ultimately bite rate and grazing time decrease and pasture intake declines. It is not a reduction in nutritive value per se that reduces intake but the avoidance of stem in the bite and the effect of this on bite mass.

At this micro-scale, the potential effect of these morphological differences in grasses on voluntary intake is difficult to assess. Da Silva et al. (2013) used individual grazing paddocks to overcome this issue (Table 2). Grazing pressure in swards of Bothriochloa brizantha was manipulated to maintain the grass at four sward heights (100, 200, 300, and 400 mm) over almost a year. As noted nearly 40 years previously (Chacon and Stobbs 1976), bite mass was highest and bite rate lowest at the highest sward height. However, this study also demonstrated that sward height also influenced pasture intake, LW gain and diet quality (Table 2). An inevitable consequence of the design was the large differences in stocking rates. Thus, there was clear relationship among stocking rate, grazing behaviour at the bite scale, intake and LW gain. In agreement with Chacon and Stobbs (1976), diet quality as determined from CP and OMD was less important than was sward height.

A similar, but short-term experiment was conducted by Benvenutti et al. (2016) with sward heights of 330, 440 and 610 mm. For the three grazing-height treatments, there was a clear break point when residual sward height was approximately 100 mm, at which digestibility, intake and rumination time all declined. Thus, cattle can compensate for increased stemminess and reduced sward height, up to a point, but eventually intake will be affected.

The preceding experiments (Chacon and Stobbs 1976; Da Silva et al. 2013; Benvenutti et al. 2016) all demonstrate a relationship between sward biomass/height and bite mass/rate that appears to be related to the leaf:stem ratio in the grazing horizon. Consequently, pasture intake is reduced through the presence of stemmy material in the grazing horizon slowing the prehension of feed, which may or may not be accompanied by a concomitant reduction in digestibility, potentially further contributing to reduced intake.

A feeding station is defined as the sward surface accessible to the animal without moving its forefeet (Roguet et al. 1998). Depending on the nature of the sward and climatic and edaphic variables, they may be close together or further apart, but groups of stations comprise a feeding site, which in a grazing context is often defined as a patch or uniform area differentiated from surrounding areas due to its nature and appearance (MacArthur and Pianka 1966). From a grazing efficiency perspective, the biomass and quality within a station are important. Abundant leafy grass will result in larger bite mass, than does either lower biomass (Da Silva et al. 2013) or high biomass of stemmy material (Gregorini et al. 2011). Gregorini et al. (2011) examined in detail the relationship between grazing behaviour at the station and site level, by using short-term observations of rumen-evacuated cattle offered swards of low, medium and high sward height/biomass. They noted that as sward height was reduced, intake per feeding station declined, steps per minute and eating distance increased, the area grazed increased, bite rate increased, but bites per feeding station declined. In simple terms, cattle spent more time looking for grass, as the availability per feeding station was reduced. Cattle were optimising their energy budget by visiting more feeding stations but taking fewer and shallower bites at each station. Clearly, these observations are relevant in northern grazing systems. This view is supported by Stobbs (1970), who reported that time spent grazing tropical grasses was 550 versus 360 min per day in more homogenous temperate pastures. Hunt et al. (2007) suggested that animals seek out grazing stations low in biomass but high in quality. The proximity of stations within a patch, and the proximity and size of patches will have important implications on the cumulative DM intake, with many smaller bites reducing intake rate compared with fewer larger bites (Da Silva et al. 2013).

In northern Australian grazing conditions, cattle may have to walk considerable distances between patches and water points and have marked preference for certain patches within a landscape. McGavin et al. (2018) used GPS to relate daily distances walked to paddock size. At a GPS fix rate of 10 s, daily distance travelled increased rapidly as paddock size increased to approximately 50 ha, but thereafter paddock size had little influence on distance travelled, which reached a plateau at approximately 15 km/day. Hunt et al. (2007) demonstrated how cattle utilise patches within large paddocks (9–57 km²) by using GPS tracking. Within the landscape, cattle spent more time close to waterpoints and other attractants in the landscape. Within these large areas, cattle will establish home ranges (Hunt et al. 2007). These show some elasticity with paddock size (Hunt et al. 2007), topography (Bailey et al. 2006), and familiarity (Tomkins and O’Reagain 2007). Hunt et al. (2007) found that home ranges vary, occupying a higher proportion of small paddocks than larger paddocks, consistent with there being a maximum distance travelled limiting the area an animal can access in a day (McGavin et al. 2018). Patches within home ranges become over-grazed due to their proximity to attractants in the landscape. The preference to spend more time close to attractants in the landscape leads to biomass contours, with low biomass close to attractants and increasing biomass further away from attractants. This piosphere effect has been associated with larger paddocks (Manthey and Peper 2010; Frank et al. 2012) and contributes to poor overall pasture utilisation of ~10% standing biomass (Hunt et al. 2007). Coleman (2005), citing Raeburn (1986), demonstrated the concept of relative intake in relation to the maximum intake expected for that class of animal on that quality of pasture. Relative intake was attributed to biomass availability and stocking rate (Fritz and Garine-Wichatitsky 1996). Intake increases rapidly to approximately 90% of maximum when pasture biomass is approximately 2–3 t DM/ha. Under extensive, tropical conditions, it is the patch or home-range biomass that is critical, rather than the whole paddock, and at this scale there is a lack of evidence to support a threshold biomass at which relative intake approaches potential intake.

The highly seasonal nature of diet nutritive value in northern Australia causes fluctuations in the plane of nutrition that results in compensatory gain. Cattle transitioning from a low plane of nutrition to a higher one, such as after the transition from the dry to the wet season (break of season), typically exhibit higher than expected rates of LW gain (e.g. Tomkins et al. 2006; Silva et al. 2022). This compensatory growth has been attributed to increased lean tissue (protein plus associated water) deposition (Berge 1991; Mota et al. 2020), increased proportional digestive tract weight as well as a lower proportional maintenance requirement (Berge 1991). Thus, for a given level of DM intake, cattle will exhibit increased LW gain. In addition, cattle transitioning from a restricted plane of nutrition frequently exhibit increased appetite, with increases in DM intake being considered to average 10–20% (Berge 1991). The changes in feed conversion and intake associated with compensatory growth therefore influence the relationships between intake and both diet nutritive value and LW gain. Estimations of DMI rely on one or both of these variables depending on the estimation method chosen. Thus, in the wet season, DMI would be routinely underestimated to a varying extent and compensatory gain should be accounted for when attempting to predict or estimate wet-season pasture DM intake. However, this is not the case and represents a deficiency in the accurate prediction of DMI.

Lower maintenance requirements for Brahmans have been confirmed with the view that they result from a lower fasting heat production (Vercoe 1970; Chizzotti et al. 2007; Cadenas-Soberanis et al. 2021). Given that there are differences in metabolism between taurine and indicine cattle, presumably as an adaption to their original environments, it follows that there may be differences in voluntary intake. Research suggests that if there are differences, they are small and inconsistent. Hunter and Siebert (1985, 1987) compared voluntary intake of Herefords or Angus steers with Brahman steers. Over 12 treatment comparisons, they observed a 5% higher intake in Brahmans than the British breeds, but this was attributed to one result with young cattle, and when these data were removed, there was no difference. Azevêdo et al. (2010), with feedlot diets, found the opposite response when gain was above 1.4 kg/day, with Bos indicus × Bos taurus cross-cattle having a higher intake than did B. indicus cattle. For slower-growing cattle, given less energy-dense diets, both breed types exhibited the same voluntary intake. It can be concluded that although there are metabolic differences between the two subspecies, voluntary intake of forage-based diets is essentially the same.

Review of empirical equations

The equation for prediction of pasture intake published in PISC (2007) is the most widely used in Australia. Although developed for temperate regions, factors can be included to account for potential differences between C₃ (temperate) and C₄ (tropical) grasses. The DMI predictions from this (PISC 2007) and other equations (Hodgson et al. 1977; Siebert and Hunter 1977; Minson and McDonald 1987; Azevêdo et al. 2010; Coleman et al. 2014; Fernandes et al. 2022; Table 3) were tested against papers with observed measures of intake from tropical forages. Five papers (Kloppenburg et al. 1995; Boval et al. 2007; Krueger et al. 2008; Madruga et al. 2017; Velazco et al. 2017) with observed measures of intake using 14 comparisons that included LW gain data were used to evaluate the equations requiring LW gain. Two additional papers (Ash and McIvor 1998; Andrade et al. 2016) with observed measures of intake contributing 17 additional comparisons were used to evaluate equations that required data only on diet quality.

None of the equations was able to provide precise and accurate predictions of observed intakes (Table 4). The PISC (2007) model showed very poor correlation with observed data, confirming the conclusions made by McLennan (2014) and McLennan et al. (2020). A lack of data on the age and genetics of the animals involved in the published literature may have contributed to poor performance, as these are key inputs required for the model. Further, the PISC model relies on the Allden and McDWhittaker (1970) relationship between herbage availability and rate of intake developed from temperate-pasture grazing studies. This relationship may be quite different for tropical pastures due to high spatial heterogeneity and other sward characteristics mentioned earlier. The Minson and McDonald (1987) equation based on LW and LW gain showed the best agreement, with observed values having the lowest root mean square prediction error (RMSE). The Azevêdo et al. (2010) and Fernandes et al. (2022) equations, also based on LW and LWG and derived from large Brazilian datasets, both performed reasonably well. A major limitation of these methods is the length of time required to record an accurate measure of LW gain, particularly if the diet characteristics are changing, and accurate LW is susceptible to errors arising from compensatory gain. In extensive grazing conditions where diet quality is poor, the energy requirement for activity (walking, grazing, etc.) can be up to 40% of the total energy requirements (McGavin et al. 2018) and affects equations relying on LWG (e.g. Minson and McDonald 1987). Given these methods are easily applied in the field and appear to work quite well, the use of digital technologies to (1) improve LW measurement through development of rolling average LW gain from in-paddock autonomous weighting systems and (2) measure activity from on animal GPS sensors, should further improve their utility for research and commercial application.

Under controlled feeding conditions, response in intake to changes in digestibility or CP has been well characterised and is clearly related to the appropriate balance of nutrients in the rumen, such as, for example, the ME:N ratio (Panjaitan et al. 2010). However, under grazing conditions, availability of pasture supply, the heterogeneity in space (patch grazing), quality (stem, N content, legume versus grasses) and time (change in biomass and quality) effectively reduce the actual intake relative to potential intake (Allden and McDWhittaker 1970; Coleman 2005). Thus, it is not surprising that the relationships derived between intake and N content (Siebert and Hunter 1977) or digestibility (Hodgson et al. 1977) failed to predict intake under grazing conditions. It was expected that equations, including characteristics of both the diet and the animal, would have out-performed other equations. Although the Coleman et al. (2014) equation based on LW and DM digestibility underestimated intake, it responded well to the range in observed intake, having an RMSE of 1.20 kg.

The poor performance of the PISC (2007) model for predicting intake of tropical forages by tropically adapted cattle was addressed by McLennan (2014) and McLennan et al. (2020). They observed that the model overpredicted DMI, especially at higher intakes, i.e. the slope of the relationship was significantly different from 1. These authors revised the equations defining the utilisation of energy by cattle for growth and released a spreadsheet intake calculator (QuikIntake), which utilised the PISC (2007) methodology to predict DMI. The relationships between predicted intakes using QuikIntake and those observed in a number of pen-feeding studies (Mclennan et al. 2020) are shown in Fig. 4. Relative to the PISC (2007) model, changes were made to the maintenance requirements, the efficiency of use of energy for maintenance and gain, the energy value of gain, and the energy density of the diet. With these changes, the QuikIntake model better predicted intake of pen-fed cattle and was defined by the following equation:

where DMI is DM intake in kg/day, r² is the correlation coefficient and SSE is the sum of squared errors. Dove et al. (2010) reviewed the difficulties of accurately estimating intake of low-quality tropical pastures, highlighting the importance of herbage supply and legume content in the prediction of actual intake. They concluded that prediction of intake from pasture and animal characteristics was so fraught with difficulties that a better option was to predict intake from known growth rate relying on accurate prediction of diet ME (from faecal NIR) and efficiencies of energy utilisation, noting that the latter have been called into question by McLennan et al. (2020).

Fig. 4.

Relationships between dry-matter intake using QuikIntake and those observed for cattle fed forage diets (after McLennan et al. 2020). Eqn 1 PISC equation; Eqn 2 is Eqn 1 adjusted for changes in the maintenance requirements, energetic efficiency for maintenance, and the energy value of gain; Eqn 3 is Eqn 2 adjusted for changes in the energy value of the diet; Eqn 4 is Eqn 3 adjusted for changes in the energetic efficiency for gain; and Eqn 5 is Y = X.

We adopted a different approach in attempting to improve the predictive ability of PISC (2007) for DMI, by looking at the relationships between diet characteristics and intake. It was noted that incorporating the ‘C₄ factor’ of 0.16 used in PISC (2007) markedly increased the DMI. The use of this factor in PISC (2007) was attributed to data published by Minson (1981a). In this review, we concluded that the Minson (1981a) data relating intake to digestibility could be represented by a single line as opposed to two separate lines for temperate and tropical diets. Thus, this factor was removed from a revised PISC equation (PISCr) used in our analysis and this resulted in improvements in accuracy but not precision (Table 4).