Factors influencing the variability in performance of cattle grazing tropical pasture

Site

A 276-ha area, comprising three adjoining and interconnected paddocks, was selected for the trial at the Lansdown Research Station (19°39′S, 146°50′E). The site comprised areas of cleared and open woodland with riparian areas. The naturalised pasture was dominated by the grasses Indian couch (Bothriochloa pertusa), black speargrass (Heteropogon contortus), and Queensland bluegrass (Dicanthium sericium), and by the legumes seca stylo (Stylosanthes scabra) and verano stylo (S. hamata). Soils were predominantly loam and sandy loam and the annual rainfall for Lansdown is 902 mm.

The trial site was set stocked for 23 months, encompassing two dry seasons and two wet seasons. Rainfall patterns corresponding to dry and wet seasons are given in Fig. 1. Three water points were provided

Fig. 1.

(a) Monthly rainfall totals and (b) definition of dry and wet seasons throughout the trial.

Animals and treatments

Ninety Droughtmaster steers (Bos indicus × B. taurus) with an average age of ~6 months were purchased for the trial locally, and quarantined for 10 days before allocating to the trial on 28 August 2015 with a LW (mean ± s.d.) of 188 ± 30 kg. Cattle were set-stocked on the area for 698 days, at which time they were sent for slaughter on 26 July 2017 at an abattoir 50 km from the trial site and carcase data from the abattoir were obtained for all animals. Cattle were mustered at pre-defined periods (Table 1) for measurement of LW and sampling of rumen fluid, blood and faeces by using industry-standard cattle yards and crush. One animal died during the trial of unknown causes. The LW (mean ± s.d.) at the end of the trial was 554 ± 47 kg. The trial was approved for animal ethics under the Brisbane Animal Ethics Committee Application A 12 2015.

Animals were assigned to the following two groups post hoc according to their overall LW gain: low performance (LP, 20 animals with the lowest LW gain), and high performance (HP, 20 animals with the highest LW gain). The remaining 48 animals were excluded from the statistical analysis of animal variables. Animal and pasture variables were grouped according to season, resulting in a statistical comparison including high versus low performance and wet versus dry season and their interaction. The years of study were defined by rainfall patterns, with Year 1 being from August 2015 to June 2016 and Year 2 from June 2016 to July 2017. Data from each year were treated independently and analysed separately.

Sampling

Cattle were weighed at the beginning, end, and every 2–3 months during the trial at a cattle-handling facility ~1 km from the grazing area. All weights were non-curfewed and recorded within 4 h of mustering. Concurrently with the weight sessions, all animals were sampled for blood via jugular venipuncture, rumen fluid by using an oral stomach tube and faeces by rectal grab sampling. The animals were restrained in a commercial cattle crush with the animal’s head secured in a head bail. For rumen sampling a reinforced plastic suction tube was inserted into the mouth through a protective black polypropylene pipe that was passed over the epiglottis and into the oesophagus before entering the rumen. Approximately a volume of 100–200 mL of rumen fluid was removed via the tube by using negative pressure from a hand-operated pump. The sample was immediately subdivided for pH determination (at the time of sampling by using a portable pH meter), and preservation and storage at −20°C in 20% metaphosphoric acid for volatile fatty acid (VFA) and ammonia-N analysis. A faecal sample for faecal near-infrared (NIR) analysis (100–200 g) was collected by rectal palpation and stored at −20° C within 2 h. A 10 mL blood sample was obtained from the jugular vein and held at room temperature for 30 min before storing on ice for 1.5 h. Samples were then centrifuged at 5°C for 20 min at 4000g. Plasma was analysed for ammonia-N.

Animal behaviour

Up to 12 animals were randomly selected to be fitted with Commonwealth Scientific and Industrial Research Organisation (CSIRO) designed collars (Arablouei et al. 2023) for determination of distance travelled and spatial distribution throughout the grazing area. These collars capture and process sensor data. The spatial distribution of cattle was collected using global positioning system (GPS) technology at 1 s intervals. GPS units were successfully fitted to these animals on three occasions, namely 3/6/2016 to 5/7/2016 (early-dry), 11/8/2016 to 8/9/2016 (mid-dry) and 8/12/2016 to 23/12/2016 (late-dry). The collar included an artificial intelligence of things (AIoT) device (CSIRO, Pullenvale, Australia), battery pack and six photovoltaic modules for harvesting solar energy. The AIoT device featured a Texas Instruments CC2650F128 system-on-chip with an Arm Cortex-M3 CPU at 48 MHz stored at random access memory (RAM) and read-only memory (ROM). It also included a MPU9250 9-axis micro-electro-mechanical (MEMS) inertial measurement unit containing a tri-axial accelerometer sensor to measure acceleration in three axes. The power was supplied by lithium-ion battery pack charged by solar panels. The collars were fitted around the animal’s neck with a strap. The GPS unit was held on top of the neck by means of a counter-balance underneath the neck. As a result of collar failure, data were successfully retrieved from six, seven and five animals in the early-, mid-, and late-dry periods respectively. Collar failure was attributed to moisture entering the electronics housing, collars rotating through 180°, leading to signal failure and loss of collars. Data for analysis were used only on days when all selected collars were generating continuous 1 s fix data.

After each study period, data were downloaded from micro-SD cards on collars and entered and stored in CSV files. GPS data were sorted on Rstudio (Version 2023.06.2). Data were filtered to eliminate points geolocated with less than three satellites and where distance travelled >3.3 m/s. Daily mean distances were included in the analysis when >99% of potential GPS fixes (86,400) were successfully achieved. Daily distances travelled were calculated from fixes every 10 and 30 s to reduce overestimation owing to the frequency of measurement error (McGavin et al. 2018). After filtering, six, seven and five animals were selected from early-dry, mid-dry, late-dry periods respectively. GPS points were imported to QGIS (3.34.3) for visualising preferential use of the grazing area. Preference ratios were calculated as the proportion of GPS points in an area divided by proportion of that area in the total area (Hobbs and Bowden 1982; Tomkins and O’Reagain 2007). In this case, areas were within 200 m of a waterpoint or one of three interconnected paddocks.

Pasture evaluation

Pasture biomass, legume proportion and bare ground (Year 2 only) were assessed using the Botanal technique (Tothill et al. 1992). Three trained individuals each collected data from ~100 0.25-m² quadrats taken along three parallel trajectories in each of the three interconnected paddocks representing the grazing area. For statistical analysis, 10–15 individual observations were averaged to produce 10 mean values per person within each paddock. Sampling sites were not geo-referenced. Within 1 week of Botanal assessment, 10 random sites were selected within each of the three paddocks and the pasture was sampled by harvesting a 0.25-m² quadrat to within 50 mm of the ground for NIR analysis of nitrogen, DM digestibility, neutral detergent fibre and acid detergent fibre.

Analyses

The nutritive value of pasture and the diet were scanned by NIR by using a monochromator (Model 6500 NIRSystems, Silver Spring, MD, USA) at the CSIRO Floreat laboratory, according to the method of Coates and Dixon (2011). Pasture samples were dried at 65°C to a constant weight in a forced-air oven and ground though a cyclonic mill (Retsch Twister, Retsch BMGH, Dusseldorf, Germany) with a 1-mm sieve and analysed for nitrogen (N), neutral detergent fibre (NDF), acid detergent fibre (ADF) and DM digestibility (DMD). CP was estimated as N × 6.25 and hemicellulose as the difference between NDF and ADF. Diet nutritive value was determined by NIR spectroscopy of faeces by using the full reference dataset of over 1000 samples from 264 diet–faecal pairs (Coates and Dixon 2011). Faecal samples prepared as described for pasture were analysed for dietary N, faecal N, NDF, ADF, DM digestibility, and organic-matter (OM) digestibility. Random samples of pasture and faeces were analysed for N by using a Leco CN628 N analyser (Leco, St Joseph, MI USA), with the following relationships:

The method of Coates and Dixon (2011) was used to estimate the percentage of non-grass (equivalent to C₃ plants) in the diet from faecal NIR predictions. VFA analysis of rumen fluid was performed using the method of Gagen et al. (2014) and ammonia N by using the colorimetric method of Chaney and Marbach (1962).

Statistical analyses

Statistical analyses were performed using SAS general linear models (SAS Institute, Cary, NC, USA). All data for the two successive years (August 2015 to June 2016 and June 2016 to July 2017) were analysed separately. Significance was declared at the 5% level, with trends being declared at the 10% level. The experimental design for pasture characteristics used a general linear model with the fixed effect of season (dry versus wet), paddock and, where appropriate, operator. Operator and paddock effects were not significant and statistical comparisons were made at the season × paddock interaction. For all animal data, the experimental unit was the animal (n = 20 for the LP and HP groups). The fixed effects were season (dry versus wet) and performance group, and the means were compared at the interaction level of season × performance group (low versus high).

Pasture characteristics

Biomass varied between 1.4 and 4.4 t DM/ha (Table 2). In Year 1, biomass was higher in the wet than the dry season as expected (P < 0.001); however, in Year 2, the opposite was observed, with biomass being lower in the wet season (P < 0.001). The legume content averaged 34%, and even though there was no difference between seasons in Year 2 (P > 0.05), the legume content was higher in the dry than the wet season in Year 1 (P = 0.018). The percentage of bare ground was measured only in Year 2 and averaged 34% across both dry and wet seasons (P > 0.05). The N content of pasture was higher in the wet than the dry season in both years (Year 1, P = 0.01; Year 2k P = 0.002). In terms of CP, dry-season values averaged just under 7% DM, whereas wet-season CP was over 9% DM. DMD was very low in both dry seasons, but significantly (P < 0.001) higher in the wet seasons of both years. NDF and ADF followed predictable patterns in Year 1, being lower in the wet than the dry season (NDF, P = 0.07; ADF, P < 0.001). In Year 2, NDF was not significantly (P > 0.05) different between seasons, but ADF content was lower in the wet season (P = 0.05). Because of the seasonal differences in NDF and ADF, hemicellulose was consistently higher in the wet seasons (Year 1, P = 0.01; Year 2, P < 0.001).

Diet characteristics

Diet characteristics were estimated from faecal NIR samples (Table 3). The responses to season, performance group and their interactions were inconsistent between the 2 years.

Seasonal effects in Year 1 were highly significant (P < 0.001) for dietary N (and CP), faecal N, DMD, and OM digestibility, with values being greater in the wet than the dry season. In Year 2, dietary N (and CP) were again higher in the wet than the dry season (P < 0.001) and tended to be higher for faecal N (P = 0.085). However, DM and OM digestibility were lower in the wet season than the dry season (P < 0.001). The proportion of C₃ species in the diet, mainly represented by legumes, was not affected by season in Year 1 (P > 0.05), but was lower in the wet season in Year 2 (P = 0.043).

The nutritive value of diets selected by LP and HP cattle were not significantly (P > 0.05) different. However, for several diet-quality characteristics, there were significant interactions with season. In Year 1, significant (P < 0.05) or trend (P < 0.10) interactions were seen for dietary N (and CP; P = 0.063), faecal N (P = 0.009) and OM digestibility (P = 0.012). These interactions implied that the HP group consumed a higher-quality diet in the wet season, than in the dry season, but this response was less pronounced for the LP group. By contrast, in Year 2, significant interactions were attributed to the LP group selecting a higher-protein diet than did the HP group, as evidenced by interactions for N (and CP) and faecal N (P = 0.01). However, there were no interactions with performance group for DM and OM digestibility (P > 0.05).

Grazing behaviour

Pasture samples collected close to the periods (Table 1) when collars were on cattle showed a significant time-of-year effect for biomass (P < 0.01) and, in West paddock, legume % (P = 0.037; Table 4). Daily distance travelled by cattle was significantly (P < 0.001) affected by time-of-year, with the distance travelled in the late-dry season (December) being shorter than that in the earlier months of June and August. Travel activity was greatest between 06:00 hours and 10:00 hours and again between 12:00 hours and 18:00 hours in June and August (Fig. 2). In December, activity was much reduced and peak activity was earlier in the morning (04:00 hours to 07:00 hours) and later in the evening (17:00 hours and 20:00 hours) than in earlier periods. Fig. 3 shows the frequency with which cattle spent time in different areas of the three interconnected paddocks (West, Middle, East). In June, cattle were found predominantly within the Middle paddock, whereas in August they were equally dispersed between the West and Middle paddocks. In December, cattle were predominantly found closer to water points. Preference ratios supported the visual evidence (Table 4). In the early-dry season, cattle showed a marked preference (PR = 1.69) for Middle paddock, but 2 months later, preference had shifted to West paddock (PR = 1.64). Throughout the three measurement periods, cattle showed a marked avoidance of the East paddock. Preference ratios showed that cattle spent a disproportionate amount of time (PR on average > 2.0) within 200 m of water points, but the preference for an individual waterpoint changed over time.

Fig. 2.

Diurnal variation (mean ± s.e.) in hourly distance travelled (10 s fix interval) for seven cattle in June and August and five cattle in December.

Fig. 3.

Heat maps (using 20-m radius) representing cattle location in June (red), August (blue) and December (green) 2016 in three connected paddocks totalling 276 ha.

Animal performance

The average initial LW of the HP group was 22 kg heavier (P = 0.028) than that of the LP group, but final LW was 99 kg heavier for the HP group (P < 0.001), resulting in higher LWG of these cattle than of the LP group (P < 0.001; Table 5). These differences in performance resulted in increased carcase weight and value at slaughter (P < 0001). There was no effect of performance group on backfat, muscle score or killing out percentage (P > 0.05). Cattle gained more LW in the wet than dry seasons (P < 0.001; Table 6) and in Year 1, the HP group outgained the LP group (P = 0.041). In Year 2, there was an interaction trend (P = 0.059) between season and performance group, with the HP group having double the increase in LWG (+0.44 kg/day) in the wet compared with the dry season, whereas this response was only +0.22 kg/day in the LP group.

Rumen fermentation and blood urea-N

Rumen fermentation was highly influenced by season, but performance group effects were small or not apparent (Table 7).

Pasture and diet characteristics

Pasture biomass in northern Australia is highly responsive to rainfall and stocking rates. Typically, biomass accumulates rapidly in the wet season, to reach a peak of between 4 and 8 t/ha depending on the location, rainfall, and previous and current stocking rates (Bortolussi et al. 2005). This pattern was observed in the first year of the study, but not in the second year, where biomass was lower in the wet season that the dry season. The study was conducted during a period of lower-than-average rainfall. The long-term mean for Lansdown is 901 mm/year, but over the 3 years in which the study occurred, rainfall was 460, 624 and 842 mm, with just 1344 mm falling during the 23-month study period. The pasture area in this almost 2-year study was stocked with 89 growing cattle, with average initial and final LW of 188 and 554 kg respectively. Thus, stocking rate increased over the duration of the study (as animals increased in size) from ~5 ha per adult equivalent (AE) to 3 ha/AE in the following year. We postulate that this led to overutilisation of the pastures in the latter months of the trial, during this lower-rainfall period. Despite this, LWG remained higher in the wet than the dry season in Year 2, although the response to season was smaller than in Year 1.

Throughout the study, legume content in the pasture remained above 25%, being as high as 41% in the second dry season when total pasture biomass was low. Undoubtedly, this contributed to the acceptable rates of gain (overall 0.52 kg/day) observed because tropical legumes are generally of higher nutritive value than are C₄ grasses (Archimède et al. 2011) and increase voluntary intake (Kennedy and Charmley 2012). Faecal NIR showed that intake of C₃ non-grass (corresponding to legume in the diet) was higher than the proportion of legume measured in pasture by using Botanal techniques. This suggested that cattle were preferentially selecting legume over grass in the diet, as noted by Dixon and Coates (2008). The preference for legume was reflected in the CP content of the diet that was two percentage units higher than in the pasture. It was also apparent that the legume and CP contents of pasture and diets were greater in Year 2 than Year 1, reflective of the lower total biomass in Year 2 having a concentrating effect on legume proportion.

Pasture DMD in the dry season of Year 1 was extremely low (21%). However, only a single sampling was taken for this measurement at the very end of the dry season (December), so the data may not accurately reflect the digestibility of the pasture over the preceding grazing period (August to December). In the following wet season, the expected increase in digestibility was apparent as new vegetative growth replaced the mature and dead material of the dry season (Tothill and Gillies 1992). The seasonal response in pasture digestibility was broadly similar from Year 2 to Year 1.

Diet digestibility from faecal NIR did not follow the same pattern as that observed for digestibility in pasture samples. In Year 1, diet digestibility from NIR was predictably higher than that of the pasture and was higher in the wet than the dry season. These data reflected the selection by cattle for legumes and less stemmy, higher-digestibility grasses (Benvenutti et al. 2015). In contrast, DM digestibility in Year 2 was lower in the wet than dry seasons. It is hypothesised that the low biomass availability restricted the animal’s ability to select the more digestible leaf and green stems of grasses and digestibility was compromised (Chacon and Stobbs 1976).

Grazing behaviour

The relationship between the animal and pasture is critical to both the sustainability of the pasture and the performance of the animals grazing that pasture. We deployed GPS collars periodically throughout the dry season in 2016 to follow the patterns of grazing behaviour as pasture biomass and quality declined. It was not possible to successfully deploy collars in the wet season because many collars failed owing to the wet conditions. Given the small proportion of collard cattle (6–8%), the sampled cattle may not have represented the behaviour of the herd. However, research with virtual fencing technology has shown that the herding instinct in cattle is strong and successful animal control can be achieved with a small proportion of collared cattle (Campbell et al. 2021). Other research has suggested that even a single animal may be sufficient (Moritz et al. 2012). Cattle walked about 14 km/day (at 10 s fix rate) when biomass was above 2 t/ha, but activity dropped off markedly in December when biomass dropped below this value and temperatures increased. Foraging theory would predict that as the feed resource declines, then so too does activity (walking and grazing), to conserve energy for maintenance and production (Owen-Smith et al. 2010). Studies of grazing behaviour (Chacon and Stobbs 1976; da Silva et al. 2013) suggest that initially as biomass declines, cattle will graze longer to maintain intake but ultimately grazing time (and associated activity) will decrease in response to biomass availability. Although it was not possible to differentiate between grazing and walking, it is likely that much of the observed activity was to maximise feed intake. It was speculated that activity (as measured by distance travelled) may be related to growth performance. However, the limited activity data gathered did not allow this theory to be tested.

The determination of accurate travel distance by cattle is difficult. Very high fix rates magnify the inherent GPS error and include the micro-movements of the animal. Very low fix rates increase interpolation error (McGavin et al. 2018); that is the path tortuosity is not captured to the same degree as the length of time between points increases. McGavin et al. (2018) compared distance travelled with fix rates varying from 1 to 600 s and concluded that for cattle in northern Australia values of 5 to 10 s were most appropriate to use.

Cattle do not graze uniformly and in large paddocks will select certain areas over others. Reasons for how animals choose to utilise an area are difficult to ascertain, but factors such as water points and supplement locations, shade, pasture biomass, species and quality, soil type and steepness of terrain are all important factors (Tomkins and O’Reagain 2007; Charmley et al. 2023). In the current study, cattle avoided the East paddock for inexplicable reasons, as biomass and legume content appeared adequate. The highest legume content was seen in Middle paddock (47%) and this paddock showed the highest preference ratio in two of the three measurement periods. In the other period, cattle dispersed into West paddock, again when legume content was high (52%). Although it is ill-advised to make too much of the link between legume content and preference, the data do support the known preference for legumes over grasses in mixed swards (Rutter 2006).

As the season advanced and pasture biomass and quality declined, cattle spent an increasing proportion of their time near waterpoints. When the data from both waterpoints are combined the preference ratio was 1.0, 1.5 and 3.4 for the early-, mid- and late-season measurement periods. The likelihood of overgrazing near these waterpoints is high (Hunt et al. 2007), and overgrazing may explain why cattle switched preference between waterpoints; as biomass became depleted at one point, cattle switched to a point with higher biomass. Cattle are known to have two major grazing periods throughout the day, one from dawn to mid-morning and a second in the afternoon, typically after 16:00 hours (Tomkins and O’Reagain 2007; Tomkins et al. 2009). The current data partially support this observation, although in the early- and mid-dry seasons, peaks were broader than those observed by Tomkins and O’Reagain (2007) and Tomkins et al. (2009). In the late-dry season, peak activity was more clearly defined and linked to dawn and dusk, suggesting that as biomass became depleted, cattle spent less time grazing and more time resting, a supposition reinforced by the higher preference to be near waterpoints in December.

Animal performance

Cattle were sourced from a local herd without data on birth date, breeding values or sires. Thus, it was not possible to take these into account when selecting the LP and HP groups at the completion of the grazing trial. The 22-kg higher initial LW of the HP group could have affected subsequent performance. However, it is unlikely that any pre-trial management, such as age at weaning or compensatory gain, would have persisted throughout the 23 months of study. Such differences usually disappear over time (Berge 1991). Genetic variation in production phenotypes such as feed efficiency or feed intake could have influenced performance both before and during the trial. In a review by Cantalapiedra-Hijar et al. (2018) it was suggested that differences in the rumen microbiome could influence rumen volume and absorptive capacity and, in so doing, affect feed intake. Differences in lean to fat ratio in the carcase and protein turnover rates have also been implicated in affecting feed efficiency (Kenny et al. 2018).

We hypothesised that higher performance by cattle could be attributed to differences in either diet selection, digestion, or animal grazing behaviour, acknowledging that influence on one factor could elicit a response in the others. Consequently, statistical comparisons were drawn between the best- and worst-performing animals as measured by LW gain. Over the 23 months of the trial, annual LW gain was 190 kg/ha, somewhat above the norm for this area of 111–190 kg/ha (Bortolussi et al. 2005). This can be attributed to the high proportion of legume in the diet (Charmley et al. 2023). LW gains of the LP and HP groups were significantly different, as was expected, allowing for a meaningful comparison between the two groups. As a consequence of higher carcase weight, the value of beef sold from the highest-performing cattle was 20% greater than the value of the carcases from the lowest-performing cattle, indicating the importance of selecting for better-performing animals and understanding and managing the reasons for the difference (Silva et al. 2022).

When the diet nutritive value was compared between LP and HP cattle, overall there was no effect. However, when the interaction with season is considered, it is apparent that in the Year 1 wet season, there was some evidence that HP cattle selected diets with higher nutritive value (higher OM digestibility and a trend for higher N content). This was not apparent when cattle were older in Year 2, where the reverse was observed. The VFA concentration of HP cattle was lower than that of LP cattle in Year 1 when forage availability was non-limiting. Although there were no significant effects on acetate and propionate proportions, pH was consistently higher in HP cattle.

The wet-season response in rumen fermentation in Year 1 was contrary to expectations. The improved nutritive value was expected to increase rumen fermentation, resulting in higher total VFA and a trend away from acetate towards propionate fermentation as the fibre content of the diet was lower in the wet season. However, in the first year of our study, total VFA concentration was reduced and the acetate:propionate ratio increased in the wet season. Bowen et al. (2017) also observed a reduction in VFA concentration in the wet season compared with the dry season in steers grazing black speargrass (Heteropogon contortus)-dominated pasture. Similarly, Martinez-Fernandez et al. (2020) observed lower VFA concentration in pregnant heifers grazing Mitchell grass (Bothriochloa erwartiana) in the wet season than in the dry season. These data (Bowen et al. 2017; Martinez-Fernandez et al. 2020), together with data presented in this paper, suggest that lower VFA concentration in the wet season may be due to higher rumen outflow rates (Bowen et al. 2017) which could also imply higher intake (Hendrickson et al. 1981). In support of this theory, Perry et al. (2017) compared tropical grasses of similar nutritive value (~9% CP and ~62% DM digestibility) either as conserved baled hay or cut and carry pasture. They observed lower (nearly halved) VFA concentration and significantly higher rumen outflow rates in the fresh cut and carry forage than in hay. Increased rumen turnover would reduce the instantaneous balance between VFA production and absorption. Other factors such as grazing events and relative sampling times (Playne and Kennedy 1976), and water intake (Rodríguez-Prado et al. 2012) from high-moisture pasture plants may also influence rumen VFA concentration. If we contend that low VFA concentration is linked to higher rumen turnover and intake, then it is possible that HP cattle were achieving higher intake through increased rate of passage from the rumen (Forbes 2003). HP cattle exhibited a marginally greater response to diet nutritive value in the wet season than did LP cattle, especially in Year 1. This suggests that increased LW gain of HP cattle may have been influenced by selection for more nutrient-rich pasture components contributing to higher intake of energy and protein. Additionally, the markedly lower LW gain of the high-intake group in Year 2 (0.68 kg/day) than in Year 1 (0.98 kg/day) demonstrates the added constraint of pasture availability on performance.

In low-protein diets, Silva et al. (2022) contended that variation in N-use efficiency is sufficiently large to affect dry-season performance. The dietary CP contents in the current trial were relatively high for tropical pastures (Hunter and Siebert 1985; Panjaitan et al. 2010), presumably owing to the high proportion of legume in the diet. These concentrations were not likely to limit microbial protein synthesis (Bowen et al. 2017) and rumen ammonia concentrations were above 5 mg/dL, considered to be the minimum to maintain optimum microbial protein synthesis in the rumen (Satter and Slyter 1974). The proportion of branch-chain fatty acids were higher in the wet seasons, in agreement with trials reported by Playne and Kennedy (1976) and Martinez-Fernandez et al. (2020). As branch-chain fatty acids are derived from the deamination of dietary amino acids, higher concentrations in the wet season demonstrate improved N availability in the rumen, even though concentrations were considered adequate in the dry season.

The marked seasonal pattern in animal growth rate was typical of northern Queensland, with low rates of gain in the dry season contrasted with high gains in the wet season. Although this was particularly apparent in Year 1, with gains increasing from ~0.1 in the dry to ~0.9 kg/day in the wet season, it was less so in the second year. Year 2 was characterised by a poor response in pasture biomass production to wet-season rainfall, potentially owing to overstocking in Year 1 during a period of lower-than-average rainfall. Although it was not possible to measure intake in this study, it can be predicted from growth rate, animal weight and activity (McLennan et al. 2020). This approach sums the metabolisable-energy (ME) requirements for maintenance, growth and activity and relates this to the ME content of the diet (from faecal NIR). Dry-season DM intakes for HP and LP cattle were estimated to be 23 g/kg LW in both years. However, wet-season intake was estimated at 34 g/kg LW in Year 1, but only 23 g/kg LW in Year 2. These values suggest that intake was compromised in the second wet season because of low digestibility (~40%) and low biomass availability (1.6 t/ha). Coleman (2005) noted that voluntary intake was constrained when biomass was less than ~1.7 t/ha. Also, Forbes and Coleman (1993) concluded that intake per bite was restricted when green leaf biomass was less than 1.1 t/ha. It was concluded that availability of green biomass was restricting voluntary intake and, hence, LW gain in the second wet season. These observations are supported by da Silva et al. (2013) who manipulated stocking rate to influence biomass, digestibility and intake and showed that biomass had a greater influence on animal performance than did digestibility.