Research, development and adoption for the north Australian beef cattle breeding industry: an analysis of needs and gaps

Keywords: adoption, calf mortality, environmental sustainability, grazing management, heat stress, nutrient supplementation, performance metrics, reproductive efficiency.

The north Australian beef industry spans Queensland, Northern Territory, and the Pilbara and Kimberley regions of the north of Western Australia. It comprises production enterprises that account for ∼64% of Australia’s national beef herd of some 23 million cattle (ABARES 2019), together with transport, processing and other supply chain infrastructure, including facilities for live cattle export. The total value of the industry is estimated to be ∼A$5 billion per annum, mostly generated by export of processed product and live cattle, making it the most valuable agricultural sector in the northern half of the country. Importantly, much of this income is generated on land that cannot be used for other agricultural purposes, and the pastoralists involved have responsibility for stewardship of natural resources across ∼60% of the land base of northern Australia, in addition to management of their beef enterprises.

As described by Greenwood et al. (2018) and Chilcott et al. (2020), the northern pastoral zone and its beef enterprises are diverse in terms of climate, soils, native and introduced pasture species, cattle genotypes, scale of enterprise, management systems and business ownership. Nevertheless, some general features and challenges clearly distinguish the northern beef industry from its southern counterpart. These include often extreme variation in climate within and between wet and dry seasons, with consequences for the feedbase, which features lower quality C₄ grasses and few legumes, and widespread phosphorus (P) deficiency. This limits cattle growth rates and carrying capacity, necessitating large-scale, extensive operations with a lower degree of stock control, exacerbated by the cost and limited availability of labour due to geographic distance. Heat stress and other factors limit stock-work during the summer wet season, with particular challenges for management of breeding herds. Other challenges include access to markets, constrained by distance and transport infrastructure, and the vulnerability of supply chains to external forces, most notably affecting the live export market on which much of the industry in Far North Queensland, the Northern Territory and northern Western Australia presently depends. Also, the widespread use of Brahman and composite breeds because of their disease resistance and ability to perform in harsh environments has brought trade-offs in terms of generally lower fertility and, to some extent, meat quality. The combination of climate, remoteness, cattle genotypes and market access, together with the need to protect against cattle ticks and numerous arboviruses, has resulted in internationally unique production systems that require specialised attention from the Australian scientific community.

Assessment of enterprise and industry performance has been approached in two ways. The first approach, and that traditionally used in earlier studies, was to collect biological and production data from herds in a region or several regions and to describe numerical trends. Since about 2014, large datasets from many individual properties have been collected. This includes studies in the Beef Cooperative Research Centre (CRC) (see Bunter et al. 2014) and in the CashCow project (see McGowan et al. 2014). In addition to biological and production data, business has been the focus of other studies (see Holmes et al. 2017), in which the authors used powerful statistical methods on their large datasets to describe parameters including production metrics and economic outcomes. The results are typically expressed as region-specific median values and the spread as quartile boundaries (e.g. top 25%). Impacts of different factors are expressed as percentage point impact or odds ratio. These studies benefit from the fact that they are built on systematically collected data and also from the power of the statistical analyses and the inferences so generated. The studies of Holmes et al. (2017) are limited to family-owned enterprises. Importantly, they report EBIT (Equity Before Interest and Tax) as a measure of herd productivity. This ideally serves the purpose of the present review because EBITs remove the complexity associated with land and animal asset value as well as market fluctuations, which are outside the control of farm management. A limitation of such an approach is that it requires data on a large number of animals (typically thousands) from industry-representative enterprises across multiple years. Also, these observational studies are inevitably retrospective in nature.

The second approach is to model a typical farm enterprise. The enterprise data used are a composite of data from observational studies such as those described above. These inputs are used in mathematical models that include well-developed simulations of pasture growth and animal performance to predict outcomes. Examples relevant to northern breeding herds include papers by Ash et al. (2015) and Bowen and Chudleigh (2021). An advantage of this approach is that the simulations can be run for decades so that the data generated are prospective. Therefore, researchers can ask ‘what if’ questions to investigate the predicted impact of a management change. However, a disadvantage is that the input values need to be accurate. Further, the stochastic nature of animal production means that models may not yet be sufficiently nuanced to enable reliable predictions. Outputs from models are usually validated in the field and the model is iteratively modified to try to reflect reality.

The challenges described above, together with factors beyond the influence of producers, processors and other industry participants, such as increasing government regulation and volatile international terms of trade over recent decades, have led to the assessment that well over half of northern production enterprises are economically unviable in the long term (Holmes et al. 2017; McLean et al. 2018). On a more optimistic note, those researchers have also reported a 10-fold difference in long-term profitability between the average and top 25% of beef production enterprises ($6 vs $62 per adult equivalent (AE)) (McLean and Holmes 2015). These analyses led to the identification of key factors that distinguished the top performers, as summarised by Fitzpatrick (2020):

• higher income per AE through greater productivity (kg beef/AE) as determined by

  1. – higher reproduction rates

  2. – lower mortality rates

  3. – higher sale weights

• lower enterprise expenses per AE, indicating more targeted herd expenditures

• lower overhead expenses per AE, due mostly to better labour efficiency

• lower asset values per AE.

The profit drivers listed above were ranked in the order reproduction, mortality, weight gain and cost of production (McLean and Holmes 2015), with reproduction found to be twice as important as mortality and turn-off weight, as long as cow mortality is not substantially higher than 2–3% (Holmes et al. 2017). Improvement of these production indexes has for decades been the goal of many well-funded research projects (Holroyd and O’Rourke 1989; Hasker 2000). This again raises the question of barriers to adoption of research and development (R&D) by the northern beef industry despite numerous successful research outcomes and their incorporation into feasible extension programs. Accordingly, the recently published ‘Northern Australia beef situation analysis’ has concluded that failure to adopt best management practices is the industry’s major impediment to lifting performance (Chilcott et al. 2020). However, despite its comprehensive coverage of R&D challenges and opportunities, this large report did not do much to substantiate its conclusion about adoption with specific examples or analyses, nor did it seek to offer remedies to barriers to adoption.

With regard to breeding herd performance, the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) has collected data on branding rates; those from Queensland and Northern Territory over 40 years are presented in Fig. 1 (ABARES 2019). A small upward trend in branding percentages occurred in the late 1980s and early 1990s, but percentages later flattened to lie between 60% and 75%. Year-to-year variation is especially notable.

Fig. 1.  Branding rates in Queensland (grey dots) and Northern Territory (black dots) between 1977 and 2019 (adapted from ABARES 2019).

This review evaluates the present state of knowledge in three areas deemed by industry leaders and supporting scientists to be important drivers of productivity and profitability that are amenable to translation and adoption of existing and future research findings. These are: (1) breeding herd management, (2) feedbase management, and (3) environmental sustainability. Much of the content is drawn from a report commissioned and recently published by Meat & Livestock Australia (MLA) to inform the MLA-sponsored Northern Breeding Business (NB2) strategic partnership (Bell and Sangster 2022). The background, vision, objectives and work plan for this initiative, involving industry participants and multiple providers of scientific support for the northern beef sector, are detailed in another MLA report (Fitzpatrick 2020).

Beginning in the 1950s, the rapid and widespread dissemination of Bos indicus (mostly Brahman or Brahman cross) cattle across northern Australia was driven by growing appreciation of their ability to survive and produce in challenging tropical environments. However, this initiative became tempered by concerns about their reproductive performance, initially based on anecdotal reports from producers and later supported by empirical research evidence (e.g. Seebeck 1973; Holroyd et al. 1979).

This section is prefaced by a summary of subsequent industry surveys of breeding herd performance, culminating in the relatively recent, comprehensive CashCow investigation (McGowan et al. 2014; McCosker et al. 2020a). This is followed by a review of current and past R&D on reproductive physiology, performance and management, including consideration of opportunities and challenges for genetic improvement of reproductive performance in northern herds. Management practices to improve reproductive performance and reduce mortality in the breeding herd are then discussed, with a particular focus on R&D needed to understand and reduce causes of calf mortality.

Since the late 1980s, four major surveys of breeding herd performance in northern Australia have been undertaken. Of these, three were funded by MLA or its predecessors (Holroyd and O’Rourke 1989; O’Rourke et al. 1992; McGowan et al. 2014) and the other by CSIRO (Bortolussi et al. 2005a). For methodological details and results of these surveys, readers are referred to the original reports, and for a more comprehensive review, to our recent MLA Final Report (Bell and Sangster 2022).

The data of Holroyd and O’Rourke (1989) are of historical interest but are of limited use as a baseline for assessing changes in industry performance because of the widely varying time frame of their collection during a period of major changes in breed structures, operational changes and market opportunities in the northern industry.

The next survey by O’Rourke et al. (1992) was based on responses of producers to a detailed questionnaire sent in December 1990 to all beef enterprises in northern Australia that normally carried at least 300 cattle. This produced the best snapshot to that date of breeding management and other aspects of the north Australian beef industry and a useful baseline for judging future changes. However, the picture painted may have been somewhat rosier than that for the industry as a whole because the sample of respondents, representing about 33% of the total number of producers contacted, was self-selected and possibly biased towards the more progressive end of the industry.

The third survey of northern breeding herd performance and management was conducted in 1996 and 1997 by Bortolussi et al. (2005b), and reflected growth in live exports since the previous report. The authors’ claim that their survey was representative of the whole northern industry is hard to verify because of lack of information on the process for selection of participating properties; certainly, the process was not random because of criteria described by Bortolussi et al. (2005a). The conclusion of this report noted the considerable variation in reproductive performance within and between regions but was generally optimistic about previous improvements and future opportunities to increase performance of breeding herds across northern Australia.

The most recent and detailed survey of reproductive performance in north Australian beef herds was the CashCow study (McGowan et al. 2014; McCosker et al. 2020a, 2020b), which reported data gathered on 72 commercial properties across four types of country: Southern Forest, Central Forest, Northern Downs and Northern Forest. Findings are further discussed in following subsections.

The above surveys have also been reviewed by Chilcott et al. (2020), who concluded that early progress in improvement of reproductive performance of northern breeding herds appears to have stalled. We cannot confidently endorse this claim because of important variations in sample selection, regions represented, and data analysis among the surveys of O’Rourke et al. (1992), Bortolussi et al. (2005a), Bunter et al. (2014) and McGowan et al. (2014). This point was reinforced in a recent meta-analysis of reproductive wastage (Chang et al. 2020a), which considered 43 articles on wastage published between 1936 and 2014. As indicated above, the metadata contained a wide range of data types, collection times and sampling periods, which limited the statistical power of the conclusions. However, if the commercially achievable level of performance can be represented by performance of the 75th percentile of mobs or cows, as proposed by McGowan et al. (2014), then clearly, regardless of historical trends, there is ample room for improvement in most indexes of breeding herd performance across northern Australia. In broad terms across the industry, a shift from the median to the 75th percentile could be expected to improve weaning rate from 67% to 80% (McGowan et al. 2014).

Median values for performance variables, summarised for all eligible females across the four land types, are presented in Table 1. This summary demonstrates the variation among regions and the extreme of the Northern Forest, which is detailed in the following subsection. The other three regions are quite similar to each other for rates of pregnancy within 4 months of calving (P4M), annual pregnancy and calf loss. The weaning rate is 72–77% (branding rates per se were not reported in the CashCow survey but can be estimated by subtracting fetal/calf loss percentage from annual pregnancy rate percentage). As expected, these estimated values are similar to, or slightly lower than, reported values for weaner contribution and, considering differences among surveys in definition of regions or country type, not markedly different from the values for branding percentage reported by Bortolussi et al. (2005b). Liveweight production per cow is cited here as an enterprise measure of breeding herd productivity.

Table 1.  Median values for reproductive performance of all cows by country type.

The Northern Forest includes parts of Western Australia, Northern Territory and Queensland, approximately north of a line between Proserpine and Broome, and includes country where eucalypt forest predominates across a range of landforms. The Northern Forest represents the environmental extreme typified by a native grass resource vulnerable to degradation, low rainfall over a range from arid regions to wet tropics, and high temperatures. Most cattle are exposed to tick and buffalo fly infestation. The Northern Forest is typified by low stocking rates, large properties (top 25% of sampled properties carry 11 000–22 000 AE, compared with 1100–6800 AE in other regions), challenges of remote management, and long distances. Year-round mating is common but maiden heifers are commonly run as a separate group, so tighter calving periods occur for this cohort.

In terms of median values for the four regions (see Table 1), the Northern Forest has the lowest values for P4M (15%), pregnancy rate (66%), and cows producing a weaner (53%). At the same time, the fetal/calf mortality is the highest (12.9%), as is cow mortality (10.6%). Liveweight production is lowest in the Northern Forest, at 89 kg/cow retained. Although the top 25% of Northern Forest properties sampled are superior to the region’s median in every metric, each is worse than the median in all other regions. The Northern Forest appears to represent a quite different production system to the other regions and is an outlier in the CashCow analysis. For example, across the whole study, variation in P4M accounts for 60% of variation in breeding rates, but this does not hold for the Northern Forest. The environment affects the approach to management in fundamental ways.

Lower feed availability and N deficiency occur for more than half the year during the dry season, and P availability is marginal to severely deficient in most soils. The environment best suits cattle with predominantly Brahman content; other breeds do not thrive. Comparisons of the performance of Brahmans versus Tropical Composite (TCOMP) cattle have not been conducted in the Northern Forest owing to the low number of TCOMP cattle, but in the other three regions, the mean P4M was significantly greater in cows with <50% than those with >75% B. indicus content (68.3% vs 50.7%). Beef CRC research showed that Brahmans take longer to develop the first corpus luteum (mean 751 days) than TCOMPs (mean 651 days). In the second breeding season, Brahman herds had lower pregnancy rates (59% vs 76%), longer lactational anoestrus interval (134 vs 84 days) and greater number of days to calving (363 vs 344 days) than TCOMP herds (Johnston et al. 2009). In the Northern Forest, breed factors predicated by the environment lead to fewer herd pregnancies.

It is assumed that the higher median rate of calf loss (12.9%, Table 1) in this region is also associated with the environment and the constraints to management placed on properties under these conditions. In a further illustration, in Northern Forest herds, unlike herds from other regions, fetal/calf loss is unaffected by temperature–humidity index (THI; see McGowan et al. 2014). Calf loss is not significantly different among regions if THI is >79 for at least 2 weeks during the expected month of calving, whereas under lower THI conditions (≤79 for at least 2 weeks in this period), calf losses are reduced in all other regions. These data suggest that comparatively high fetal/calf loss in Northern Forest herds arises from the suite of factors prevalent in Northern Forest properties including heat and humidity. The meta-analysis of Chang et al. (2020a) calculated a calf mortality rate of 21.6% (compared with 12.9% in CashCow). Possible reasons for this discrepancy are (1) the CashCow data are a more contemporary assessment since the widespread introduction of Brahman cattle, and (2) the selection of properties and (necessary) human interventions during conduct of the CashCow trials may have led to improved herd performance.

The Northern Forest environment clearly constrains production. However, despite the challenges of cattle production in this harsh environment, a business comparison across the northern beef industry concluded that the financial performance of properties in this area follows the broader industry pattern (Holmes et al. 2017). For example, mean EBITs of the top 25% properties in the Northern Forest ($4.0–8.6/AE) were in the same range as those in the other three northern regions considered in CashCow ($3.6–9.7/AE). This somewhat unexpected finding may be partly due to the greater scale of operation on most Northern Forest properties.

The variation in data and analytical approaches cited above highlights that data on productivity should be collected and recorded in a practical and systematic way. Such data can be used to: set benchmarks; identify risks and patterns; understand opportunities for improvement; and measure progress or change over time. For the north Australian breeding industry, the major focus is on the breeding females in the herd and their calves. Examples of information collected include herd data (numbers, weights, body condition score (BCS) of each sex and age class) at particular times (weaning, branding, purchase/sale). Financial data such as business income, labour costs, depreciation and variable costs are also useful in understanding whole-of-enterprise positions. A challenge in northern Australia is that the environment makes the consistency of data and data collection difficult. A call for standardisation of data protocols (Chang et al. 2020a) was largely addressed in the CashCow project (McGowan et al. 2014; McCosker et al. 2020a). The ABARES data collection framework is also standardised.

Table 2 provides examples of the most commonly used physical measurements to derive the metrics that usefully describe breeding herd productivity. This table also defines several metrics used in research, in enterprise benchmarking, and for genetic selection, which is referred to later in this review. Note that weaning (or branding) percentage is the most useful single variable for assessing rates of pregnancy and calf loss. Branding rates are derived from whole-of-industry producer survey data collected annually by ABARES. It is important to adopt consistent use of metrics such as weaning rates, not only on individual properties but also across the industry. Standardisation and demonstrated use of appropriate and practical metrics should be a fundamental goal of programs such as NB2 that are designed to promote adoption of R&D outcomes. These parameters are targeted for use at enterprise level, so their value is in making year-to-year or cohort-to-cohort comparisons.

Table 2.  Measures that are important in recording and decision-making.

Data capture should be practical to the situation, collected in a timely fashion, recorded, and acted upon as appropriate. Data should be identified to individual animals by using National Livestock Identification System (NLIS) tags and linked to Data Capture Systems located crush-side and coupled to weighing machines. Results such as pregnancy diagnosis and BCS should also be entered into the data system. Subsequent analysis is achieved by downloading records and processing through management software.

In addition to the measures listed in Table 2, other data are relevant to productivity, such as:

• bull semen and soundness evaluations to ensure successful mating by bulls

• plasma inorganic P (PiP) to assess the need to supplement animals with P in the wet season

• data on disease investigations and vaccination history at a herd level

• pasture feed availability and budgeting ‘on the ground’ or via remote systems

• breeding records such as rates of females per bull, and purchase of bulls with known estimated breeding values (EBVs) chosen to meet productivity goals.

Economic modelling of northern beef productivity in different regions predicted the potential benefit from improved weaning rate on farm returns. Ash et al. (2015) predicted that an increase in weaning rate of 5% achieved through genetic gain would increase profits by 13% in South East Queensland, by 20% in northern Queensland, and by 27% in the Victoria River District of Northern Territory. By contrast, in the modelling study of Bowen and Chudleigh (2021), a 6% increase in weaning rate was predicted to provide a small economic gain in the Northern Gulf region of Queensland and no gain in other regions. The modest results in the latter study were attributed to the model’s inclusion of the cost of transition to superior genetics, such as the purchase of elite bulls, which was not accounted for in the model of Ash et al. (2015). Nevertheless, longer term productivity benefits are expected to accrue from genetic improvement in reproductive traits.

Genetic improvement of reproduction in the tropically adapted beef breeds used in northern Australia has been limited by difficulties in recording appropriate performance metrics and slow selection responses (Johnston 2013), as well as the long-term inclination of seedstock and commercial breeders to select for type rather than performance traits. In addition, heritability of the key maternal production trait, lifetime weaning rate, is low (Meyer et al. 1990). An important, overarching limitation continues to be nutritional and other environmental constraints on genetic expression of selected traits. Nevertheless, selection-line experiments have demonstrated that significant improvement in reproduction rates is possible in Droughtmaster (Davis et al. 1993) and Brahman (Schatz et al. 2010) breeds, as has the practical experience of astute and well-informed producers such as Alf Collins (Anon 2019). Recently, the genetic bases for these outcomes have become much better understood through research conducted by the CRC for Beef Genetic Technologies, particularly on defining the degree of genetic control for component physiological traits that underpin overall reproductive performance.

The Beef CRC Northern Breeding Project was a large, multi-site investigation of the genetics of growth and reproductive performance of Brahman and TCOMP cattle that are broadly representative of genotypes of beef cattle used in northern Australia. For full details of the objectives, scope and design of the project, see Barwick et al. (2009a, 2009b) and Johnston et al. (2009). Genetic analyses of performance of females confirmed previous reports of the low heritability of lifetime reproduction traits, with estimates of 0.11 and 0.07 for lifetime annual weaning rate in Brahmans and TCOMPs, respectively (Table 3; Johnston et al. 2014a). However, component traits of early reproductive performance each had moderate to high heritability, especially in Brahmans. These included age at puberty as determined by age at first observation of a corpus luteum (Table 3; Johnston et al. 2009) and, most notably, length of the post-partum anoestrous interval in 3-year-old cows (Table 3; Johnston et al. 2014a). Importantly, genetic correlations between early-in-life measures and lifetime traits were moderate to high, particularly the correlation between post-partum anoestrous interval and lifetime annual weaning rate (Table 3; Johnston et al. 2014a). The authors concluded that these results highlight an important opportunity for genetic improvement of weaning rates in tropically adapted beef cows by focusing recording and selection on early-in-life reproduction traits, particularly in Brahmans for traits associated with post-partum anoestrus.

Table 3.  Heritabilities (estimated value ± approximate standard error) of lifetime annual weaning rate, age at puberty and post-partum anoestrous interval, and genetic correlations between early-in-life traits and lifetime weaning rate, in Brahman and Tropical Composite (TCOMP) cows.

The Beef CRC project also included evaluation of reproduction traits in young bulls up to 24 months of age (Corbet et al. 2013). Scrotal circumference was among the most highly heritable traits in both Brahmans and TCOMPs, but genetic correlation of this trait with semen quality traits, including percent normal sperm, varied with breed and age. Thus, a single, reliable indicator of bull fertility was not identified. However, the lack of antagonism among bull traits means that selection for improved semen quality should not adversely affect other production traits. Genetic associations between reproductive traits of young bulls and female traits also were investigated (Johnston et al. 2014b). Semen quality traits were genetically correlated with short duration of post-partum anoestrus in first-lactation cows and lifetime cow reproduction traits in both genotypes, but magnitudes of relationships varied with bull age. Thus, inclusion of some bull measures in selection indexes may indirectly help to improve female reproduction in tropical breeds.

Cessation of the Beef CRC in 2012 was followed by the northern beef Repronomics™ project, a large, MLA-funded breeding and genotyping study that combined intensive recording of early-in-life female reproduction phenotypes with dense genotyping of all project animals (Johnston et al. 2017). The project used Brahman, Droughtmaster and Santa Gertrudis breeds at three different research stations, located in central and northern Queensland, and Northern Territory, respectively. The overarching goal of the project was to drive the development of new, genomics-enhanced BREEDPLAN evaluations specific to the most numerous tropically adapted beef breeds across typical northern environments.

Preliminary reports published have focused on development of conventional genetic and genomic approaches to enhance rates of genetic improvement for female reproductive traits. These include elaboration of the potential use in BREEDPLAN of more heritable early-in-life reproductive measures in bulls and cows, discussed above, as correlated traits to predict the easily recorded trait of days to calving (Johnston and Moore 2019), and the use of intensively recorded phenotypic reference data and genotypes to increase the accuracy of genomic selection of young bulls (Moore et al. 2019). An important, recently completed milestone was the whole-genome sequencing of 55 sires from the three focal breeds that, cumulatively, have >3300 progeny born and recorded in the project. Another example of the substitution of easily measured traits to overcome the challenges of acquiring sufficient, accurate phenotypic data for primary traits in B. indicus and B. indicus-infused cattle is the use of reproductive maturity score as a proxy for age at puberty, the primary trait for which is age at appearance of the first corpus luteum (Engle et al. 2019).

Successful completion, translation and commercial demonstration of the applicability of the above research on development of genomic EBVs will be especially important to restore industry confidence in genetic technologies that has been lost due to inaccuracy and variability of quantitative trait EBVs in BREEDPLAN.

Wastage can occur at many points in the breeding cycle. Aspects such as attainment of puberty and ovulation are processes driven by a range of factors including genetics and nutrition. From that point on, there are multiple sources of reproductive inefficiency that can affect a cow’s ability to wean a calf. These inefficiencies can be broken down into fertilisation failure, embryo/fetal mortality and perinatal/postnatal calf loss as summarised by Burns et al. (2010), and viewed in the context of overall breeding performance (branding or weaning percentage), they have not changed in northern Australia during the past 30 years (Fig. 1; ABARES 2019).

Fertilisation failure can be assessed only through analysis of oestrus cycle components, but it is a significant component of loss. Historical estimates of losses in Queensland range from 12% to 19%. Subsequently, lactational anoestrus significantly reduces pregnancy rates, especially in first-lactation heifers and older cows that have lost condition during pregnancy and lactation.

Variation in bull fertility related to semen quality in terms of percentage of normal spermatozoa (Fitzpatrick et al. 2002), libido intensity (Bertram et al. 2002) or structural defects (McGowan et al. 2002) also contributes to variation in cow fertilisation rates in B. indicus herds in northern Australia.

Embryo mortality covers the period between fertilisation and Day 24 (‘early’), and from Day 25 to Day 45 (‘late’). For this period, reported losses are highly variable in the northern environment and means range from 17% to 75%. Early loss appears to contribute in the range 25–30% and late loss 10–15%. In some cases, especially with early losses, cows may be able to return to service and successfully become pregnant. Fetal losses (after Day 42) account for 2–8%.

In commercial herds, fertility and embryo/fetal losses are difficult to measure, but 100 minus the pregnancy rate (%) in joined females taken at pregnancy diagnosis provides an estimate of wastage to that point (McGowan et al. 2014). As shown in Table 1, median pregnancy rates (excluding Northern Forest) across years in all age classes of females are in the range 80–85%, suggesting that the practical degree of loss on an annual basis is 15–20%. The top 25th percentiles in these categories are 90–92%, which suggests that 10% loss is the minimum effect, a level which is common in the beef industry internationally (Burns et al. 2010).

Opportunities for improvement include puberty management, nutrition management (rising plane to initiate ovulation), genetic selection and disease management. In many cases these factors are fixed or slow to change; therefore, the more immediate gains on offer are through reducing calf loss. The goal of reducing calf loss can be justified on several fronts including the potential financial benefits, that the risk factors have been quantified, that some aspects are under management control, and the opportunity to improve animal welfare. An additional source of wasted productivity is cow mortality, which, across the whole northern industry, is believed to average 9% per annum.

As an example of the financial benefits (in kg/AE), McLean and Holmes (2015) used farm economic data to assess the relative impact costs of three proposed improvements. For the stated improvements, the equivalent benefit is estimated as follows:

• A 1% increase in reproduction rate leads to a 1.5 kg/AE response.

• A 1% reduction in mortality leads to a 2.28 kg/AE response.

• A 1 kg increase in turn-off leads to a 0.18 kg/AE response.

How these benefits accrue is worth exploring under different scenarios. A cow that produces a live calf to weaning is the ideal and adds value to all three elements. A non-pregnant cow can gain weight and, without a calf to raise, can often reach a BCS sufficient to breed in the next year, providing a delayed but positive benefit. Cows that lose a calf may gain additional weight and condition and have a quicker return to oestrus because they are not producing milk. Dead cows and calves not only cost the enterprise inputs but also provide no return. As in the biological analysis, the financial analysis indicates that the major benefits will be achieved through reductions in both cow and calf mortality. Cow mortality and calf mortality are discussed in turn below.

Estimating cow mortality is difficult. In CashCow, the ‘number of cows missing at muster’ was used as a practical approach to estimating mortality (Table 1). Percentage mortality rates vary among regions, seasons and properties, but accounting for other factors, a mean of ∼9% is a conservative estimate across the four regions surveyed in CashCow (Fordyce et al. 2022). Currently, the number of cattle in northern Australia is estimated to be ∼14 million, and assuming that half of those are cows, then >630 000 die each year. Of course, every cow will die in time, but the aim is to cull cows before they become unproductive. Furthermore, breeding old cows reduces the potential for genetic gain in a herd.

Mortality rates from multi-season studies reveal two trends: herds in harsher environments have higher mean mortality (e.g. Brigalow Station 2.3%, Kidman Springs 11.3%; Mayer et al. 2012); and rates differ between age classes of cows (2 years of age 27.5%, 3–7 years 2.8–8.9%, 8 years 11.9%, 9 years 14.2%; O’Rourke et al. 1995). The CashCow data (Table 1) reported a median cow mortality of 8.4%, although the multivariate analysis in that study (McGowan et al. 2014) predicted means of 8.9–18.1%. The major determinants of loss appear to be location and nutrition:

• Harsh vs milder conditions (e.g. Northern Forest vs Northern Downs) can account for 9.2% points.

• Mortality is highest in first-calf heifers and cows >7 years of age.

• The difference between BCS 1 and 5 accounts for 7.8% points.

• Low dry-season biomass of <2000 vs >2000 kg/ha accounts for 5.4% points.

• Time to follow-up rain >30 days at the start of the wet accounts for 4% points.

The gaps between the median and 25th percentile (see data and discussion below in Potential gains in addressing calf loss: benchmarking the top 25%) show that there is opportunity for gains across the industry in the order of 5 percentage points. In order to address heifer mortality, separate management of this cohort to improve nutrition and BCS is essential, as is providing the best possible conditions for calving. Opportunities to reduce mortality of aged cows include providing nutrition to increase BCS, and culling for age (>7 years), bottle teats and failure to raise a calf. An alternative is to remove the pressure of pregnancy on older cows by sterilising them, and then fattening them for sale.

Data from successive surveys suggest that rates of calf loss in northern beef herds have been unchanged for several decades. Although industry-wide fetal/calf loss is recorded as 9%, it is known to be as high as 20% in some areas in certain years, especially in heifers. Burns et al. (2010) summarised data on losses from studies undertaken in regions of northern Australia between 1983 and 2009. The extremes for fetal/calf loss were from Queensland’s Brigalow Belt (7–18%) and regions of Northern Territory (several regions >20%, with others 3.4–14%). Perinatal loss (within 48 h of birth) accounts for 2–12 percentage points and postnatal mortality (between 48 h and weaning) contributes 0.3–15 percentage points. Holroyd (1987) considered that 12% would be an acceptable level of loss, comprising 5% prenatal, 4% perinatal and 3% postnatal losses. Losses averaged 9.5% across the five Queensland research stations sampled in the Beef CRC trials (Bunter et al. 2014), which may underestimate losses on commercial properties.

Two independent studies published in 2014 drew conclusions from analyses of large datasets (Table 4). Although these studies provide excellent industry-wide views, it should be noted that individual properties may have a subset of the risks, or the impact of each risk may differ from the regional trend.

Table 4.  Odds ratio (OR) of a factor affecting calf mortality, or estimated impact on mortality.

Bunter et al. (2014) analysed factors contributing to calf mortality over 9 years in Brahman and TCOMP breeds on research stations across different regions in Queensland. The data comprised a range of animal metrics and breeding timings, and udder and teat scores were included. Using multivariate analysis, an odds ratio (OR) was generated for each factor. In this context, an OR of 2, for example, indicates that a factor increases the probability of calf mortality two-fold. The lack of association of calf loss with maternal BCS was attributed to better management on research farms than on commercial properties, where BCS was likely to be more variable (Bunter et al. 2014) and therefore limiting. Factors such as location, adverse environment and year, which were significant sources of variability, are not listed in Table 4 because these factors are not under management control. Several husbandry activities were not considered, probably owing to them being regarded as standard procedures (e.g. castration) or where data were fragmentary (e.g. cow spaying).

The CashCow study ran over 3–4 years consecutively and collected animal and property data on mainly commercial properties across four regions (McGowan et al. 2014). This allowed a large number of parameters to be calculated and many factors to be estimated through statistical analysis and models using multivariate analysis. The factors are ascribed percentage points of effect on calf loss. From these data, a ranking of effects, clustered with likely origin across the industry, was:

1. deficient nutrition, which also has direct impact on BCS

2. heifer management issues such as mismothering, abandonment of calves, low-birthweight calves

3. risks such as bottle teats that could be resolved by culling cows

4. temperature/humidity impacts

5. aspects amenable to genetic control.

Causal webs of calf loss have been generated by McGowan et al. (2014, 2017). Although they are helpful in general understanding of possible impacts, those acting at property level may vary in importance. Despite these and earlier investigations, the causes of calf mortality remain poorly understood, partly because mothers cannot be observed around the time of calving. Fetuses and dead or weakened calves are rarely found, let alone made available for investigation of the cause of death. Even when observations are possible, it has been estimated that about one-third of the deaths have an unknown cause (O’Rourke et al. 1995). Similarly, Bunter et al. (2014) stated that ‘despite regular observation the reason for mortality … was still essentially unknown’. As a result, most presumed causes are not known or estimates that exist are based on a handful of studies or are assumed to arise from a particular risk factor.

Causes of prenatal fetal loss are especially difficult to identify. Some infectious agents that cause infertility or early abortions include Campylobacter fetus, Neospora caninum, Tritrichomonas foetus, Akabane virus, bovine herpes virus, bovine pestivirus and bovine ephemeral fever virus (see summary in Burns et al. 2010). Disease agents that tend to affect older fetuses, including causing abortions in late pregnancy, are Leptospira spp., Neospora caninum and Akabane virus.

The majority of deaths occur during or within 48 h of birth. An estimated 67% of deaths observed during the Beef CRC longitudinal study occurred within a day of calving (Bunter et al. 2014). Based on limited observations, some causes of death during this period included dystocia, congenital defects, cow mortality, sick and weakened calves that failed to suckle (some due to bottle teats), heat stress and predation, with 43% unknown (Holroyd 1987). Calves with low vigour are less likely to suckle in the perinatal period, resulting in dehydration, starvation, and low immunity due to lack of colostrum.

In the later postnatal period, between 48 h and weaning, death may occur through predation, and wound infection following castration, branding and dehorning. However, although anecdotal reports exist of significant losses to dogs on individual properties, survey data suggest that predation is not a major industry-wide issue (Allen et al. 2020).

To address the major gap in direct evidence of causation, research is needed that can monitor herds during the breeding season, mother up cows and calves, find dead calves and fetuses, and diagnose their conditions in order to categorise causes of death, over several seasons and, ideally, across multiple regions. A further challenge is that research methods should minimise disruption of cow–calf bonds while collecting data. The aim of such work is to link causation to risk factors and use that information to develop and evaluate interventions for improving calf survival. In place of direct physical observation that has been used to date, modern telemetry and positioning systems offer the opportunity to undertake remote surveillance to determine causation.

Remote surveillance of cattle is a developing technology. This discussion covers several techniques categorised into fixed and mobile data collection systems. The main focus is the use of technology in a research environment to study causation of mortality. However, in time, such technologies may be deployed as useful herd-management tools.

Fixed systems include static data-collection devices coupled with sophisticated data analysis. The best example is walk-over weighing (WOW), which enables cows to be weighed as they enter or leave watering points and the data to be used to estimate weights, growth rates, mothering up and potential calving events (Menzies et al. 2018a, 2018b). These systems generally require little maintenance, operate for long periods and are autonomous. They are ideal for determining herd trends, but are only as good as their frequency of use by cows and the sophistication of the algorithms. They do not allow attribution of causes of calf loss but have been used to estimate herd BCS and other relevant aspects of the herd.

Mobile systems include ear tags or collars that allow monitoring of an individual cow’s physical position and behaviour over time by sending information back to researchers. GPS-enabled ear tags are available that signal position at various intervals (e.g. 15 min) and communicate via the internet or satellite. Ear tags incorporating accelerometers can inform on additional activities such as calving behaviours (Chang et al. 2020b). An intravaginal device, Calf Alert, that is ejected at birth, can signal time of birth (Stephen and Norman 2021) and location can be estimated from triangulation of signals, or the GPS signal from the cow’s ear tag, if fitted. Using methods such as these, it is possible (but still challenging) to record the cow’s identity, find the birth location and time, then tag and clinically evaluate the calf. If the calf is dead, the cause of death can be investigated by laboratory post-mortem. Stationary GPS signals can also indicate cow death (or lost tags).

Even with well-resourced research programs, the measuring of causes of calf loss in a typical production system remains a challenge. Some challenges are physical (e.g. internet coverage, frequency of reporting, loss of tags), others are human (e.g. coverage at night and in rugged terrain), or due to the animal (e.g. an aggressive cow). Herd size is another challenge. At least 200 cows need to be tagged and tracked over several months. This number is required because there is some loss of data and, even in the best circumstances, only causes with an incidence >5% can be measured with sufficient accuracy. Seasonal variation and sporadic events such as disease outbreaks means that several seasons of data are required.

These methods could be supplemented with multifunction ear tags (location, behaviour) and technologies such as drone-based predator spotting. At present, these are cutting-edge research tools, but as future costs fall, some of the components may be deployed in commercial herds to assist management. Research work has commenced in this area with the Calf Watch project (FutureBeef 2020), which is building on the earlier trialling of WOW, accelerometers and intravaginal devices to identify causation, especially in the perinatal period.

As observed earlier, property data, aggregated regional data and longitudinal data collected from the cattle industry are highly variable. Because the calf-loss datasets discussed above are large and have been collected in a consistent fashion, the reported trends have a sound statistical basis, and it has been common to present those data and their analyses as 25th and 75th percentiles about a median. The 75th percentile represents the best performing properties in terms of productivity (or 25th percentile for mortality data) and is considered an achievable benchmark for properties with physical similarities in the same region. A thorough industry analysis (Holmes et al. 2017) examined financial performance of properties against productivity in northern beef enterprises. Three production items recurred as typical characteristics of northern beef properties in the top 25%. These were: (1) higher reproductive rate, (2) lower mortality rate, and (3) higher sale weight. These insights broadly align with the physical herd data from the CashCow project (McGowan et al. 2014) where farm productivity drivers were: (1) cows pregnant (%); (2) fetal/calf loss (%) (or weight of weaned calves); (3) liveweight change in cows/heifers; and (4) herd mortality (%). Some examples of the 25th percentile target for mortality are shown in Table 5.

Table 5.  Target 25th percentile fetal/calf loss values for different ages of females and cow mortality cited as the aspirational management goal, and the deviation from median (in parentheses) as a measure of the required potential gain (from McGowan et al. 2014).

Knowledge of risks and causes and how to mitigate them should focus on opportunities to reduce calf loss. For the industry, risks that contribute 2 percentage points of loss, which are associated with an OR of >1.5, should be a high priority for mitigation. On the other hand, the percentage point losses and OR of the risks in Table 4 are industry trends, which may not be translated to a particular property where some risks may be more important or more readily mitigated than others. Nevertheless, the set of aspirational benefits are increases in productivity of:

• 6% in heifers across all regions (reducing the ‘heifer effect’)

• 3.2% for second-lactation cows in Southern, Central and Northern Forests (with opportunities to improve return to in-calf the priority in all regions)

• 2.4–4.1% for older cows in all regions.

Holmes et al. (2017) claim that half of northern beef production comes from properties already at or above the 75th percentile level, and that even these would benefit from gains. Achieving such gains across half of the industry, including those properties placed between the 25th and 75th percentile, could lead to a 1–2 percentage point reduction in calf loss.

The influence of maternal genetics on rates of calf loss was studied in the Beef CRC (Bunter and Johnston 2014). Heritability of calf death before weaning was low in both Brahman (0.09) and TCOMP (0.02) cattle. However, much higher values for heritability of maternal traits contributing to calf mortality were obtained, including birthweight (0.48), udder score (0.49) and teat score (0.38). Therefore, the authors recommended selection for the maternal contribution to birthweight, while avoiding very high birthweights that may predispose to dystocia, with the more accessible, genetically correlated measurement of weaning weight considered an acceptable proxy when birthweight is not known. They further recommended that selection for birth and weaning weights should be accompanied by recording of teat and udder characteristics to assist in preventing undesired correlated effects on teat or udder size, which can also have detrimental outcomes for calf survival. Genetic links to maternal behaviour, including flight time and mothering score, as well as calf vigour traits, were not deemed to be useful traits on which to base selection to improve calf survival (Johnston et al. 2019).

The beef industry in northern Australia depends heavily on a feedbase consisting of mostly native perennial grasses growing in a range of semi-arid to tropical savanna rangeland environments as categorised by Tothill and Gillies (1992) and Bortolussi et al. (2005c). These environments feature generally poor soils and wide variation in rainfall within and between seasons, resulting in highly variable quantity and nutritional quality of pastures. Length of the non-growing season and quality of pasture are especially important influences. As discussed in the previous section, these environmental factors and the availability of adequate nutrition for cattle have significant impacts on the reproductive performance of breeding females, as well as on the growth patterns and market readiness of animals destined for slaughter.

This section briefly reviews past and present R&D on the production and utilisation of native and naturalised forages, and considers options for improving quality and management of the northern feedbase. In particular, the primary importance of appropriate grazing management practices to maintain and, if necessary, rehabilitate natural grasslands is emphasised. Additional options to increase pasture production and breeding herd performance in more favoured regions are also considered, including establishment and management of perennial legumes; potential for irrigation to increase forage production at the enterprise level; development of feasible, cost-effective supplementation strategies; and potential for greater use of conserved forages.

A simple but important concept in grazing management is that of ‘pasture utilisation’, defined as the percentage of pasture growth per unit time (usually season or year) that is consumed by cattle (Chilcott et al. 2020). Understanding and application of this concept are central to best practice in all types of grazing systems. However, its importance is critical to the sustainable management and long-term productivity of perennial native grasses in northern Australia, which are especially sensitive to overgrazing (Hunt 2008).

Almost all studies of pasture utilisation in northern Australia have used cattle growth rate as the animal production performance metric, including those discussed below. However, a large modelling study is under way to determine levels of pasture utilisation required for optimal breeding-herd performance in northern Australia as judged by pregnancy rate, P4M, calf mortality and weaning percentage (Cowley et al. 2019). This project aims to relate cow performance datasets from 28 commercial properties across Northern Territory and northern Queensland to rates of pasture utilisation predicted by the GRASP pasture growth model and to use the CSIRO’s Crop Livestock Enterprise Model (CLEM) (Meier et al. 2019) to predict bioeconomic outcomes.

In intensively managed grazing systems, pasture growth, the denominator in the calculation of pasture utilisation, has been estimated by applying technologies that use sward height as a proxy for pasture biomass, such as rising plate meters, pasture sleds and, more recently, multispectral sensors mounted on unmanned aerial vehicles (Alvarez-Hess et al. 2021). However, these technologies are inapplicable on extensively managed northern beef properties because of the heterogenous distribution of ground cover, pasture species diversity and vast paddock sizes. A promising alternative approach to remote sensing of pasture biomass and ground cover may be high-resolution satellite imagery, initially developed by the CSIRO and Department of Agriculture and Food Western Australia (Donald 2021), and since refined and commercialised by Cibo Labs (https://www.cibolabs.com.au/). Using 2000 field calibration sites, Cibo Labs provides an Australia-wide service for estimating feed on offer at 10-m resolution every 5 days with a median prediction error in pasture biomass of 295 ± 8 kg/ha derived from 100 training episodes involving ∼16 000 estimates (Donald 2021).

A recent study conducted under extensive commercial conditions in the Victoria River Downs region of Northern Territory found significant positive relations between various indexes of pasture availability assessed by satellite imagery and liveweight change of breeding cows assessed remotely by WOW over a 2-year period (Pearson et al. 2021). Further, machine-learning predictive modelling was used to show that liveweight change could be predicted with reasonable confidence by a combination of information on pasture availability, calendar date and rainfall.

Numerous modelling studies and long-term grazing trials have established the principles of sustainable grazing management, as reviewed by Hunt et al. (2014) (Table 6). Among these studies, the ‘Wambiana’ grazing trial stands out because its 20-year time span has allowed collection of comprehensive biological and economic data over the gamut of climatic events and market variations likely to be encountered by the northern beef sector (O’Reagain et al. 2018). The key findings of this study, conducted on a commercial property near Charters Towers, were that a fixed, moderate stocking rate at long-term carrying capacity for growing steers allowed pasture and land condition to be maintained and that it maximised individual animal production. Over the long term, it also was more profitable than fixed, heavy stocking. In general, these findings confirmed and reinforced those of earlier grazing trials conducted in the Victoria River District of Northern Territory (Dyer et al. 2003) and in Central Queensland (Burrows et al. 2010). It also is encouraging to note that outcomes of the Wambiana trial were predicted with reasonable accuracy by an earlier modelling study that simulated a hypothetical property in the Charters Towers district (MacLeod et al. 2004). However, unlike the Wambiana trial, economic assessment in that study was based on breeding herd performance as well as steer growth rates.

Table 6.  Principles and guidelines for grazing management in northern Australia (adapted from Hunt et al. 2014).

A trial into long-term grazing strategies has been ongoing since 2010 at Old Man Plains Research Station, south-west of Alice Springs (FutureBeef 2021a). Final results have yet to be published but the researchers are encouraged that long-term carrying capacity determined by using Grazing Land Management methodology appears to have been central to the maintenance of good land condition and consistent production over a range of seasonal extremes in this arid/semi-arid environment.

Technical guides to best management practices for grazing management to optimise land condition, animal production and profitability in the Barkly Tablelands (Walsh and Cowley 2014a), Victoria River Downs (Walsh and Cowley 2014b) and Alice Springs (Walsh et al. 2014) regions of Northern Territory are generally aligned with the recommendations of Hunt et al. (2014). These publications additionally identify specific knowledge gaps related to stocking rates, pasture spelling, landscape restoration, prescribed burning and infrastructure development that are directly relevant to issues discussed in the rest of this section. The steps required to determine long-term carrying capacity, including assessment of land condition and estimation of safe rates of pasture utilisation, have recently been summarised in practical terms (Walsh and Paton 2020).

Despite the clarity and consistency of the advice cited above, its widespread communication to northern beef producers, and well-documented examples of the successful adoption of advice to commercial enterprises (e.g. Walsh and Cowley 2016), overgrazing continues to be a major concern, with ongoing rangeland degradation and declining profitability of beef enterprises as demonstrable negative consequences. Factors contributing to the mismatching of cattle stocking rate and the native forage resource were analysed by Stafford Smith et al. (2007). These included the unpredictability of short- and long-term variations in both climatic and market conditions, and lack of knowledge of technical and other (e.g. risk-management) options to aid decision-making about grazing pressure. Ironically, although understanding of the impacts of grazing pressure has increased, innovations such as the introduction of B. indicus cattle and feed supplementation have enabled greater rates of pasture utilisation and perceived carrying capacity, with long-term detriment to land condition and enterprise profitability (Stockwell et al. 1991; Ash et al. 2011). Macro-industry factors that place a high value on the herd, such as property valuation and bank lending practices, also contribute to the mismatching of stocking rates and carrying capacity (Bowen and Chudleigh 2021). The focus of many producers (and some advisors) on production per hectare rather than production per animal exacerbates this problem (Holmes 2022).

Another key practical issue is the tendency of many graziers to retain stock for too long during a drought event before taking action to sell or agist cattle (Landsberg et al. 1998). The lead author of that paper and other progressive graziers have successfully managed this risk by setting hard turn-off dates if it has not rained by a certain predetermined date. Other options for managing grazing pressure are discussed below.

The concept of wet-season spelling of native pastures is based on observations of the particular sensitivity of tall-grass communities in northern Australia to grazing selectivity of cattle and defoliation during early forage growth (Ash and McIvor 1998). Effects of wet-season resting on vegetation dynamics and land condition were examined at three sites in north-eastern Queensland with differing levels of soil fertility and two contrasting classes of land condition (Ash et al. 2011). This comprehensive study clearly showed that either conservative stocking (25% pasture utilisation) year-round or moderate stocking (50% pasture utilisation) with some wet-season resting maintained land in a desirable state or helped transition from a less desirable to a more desirable state for sustainable production and rangeland condition. Results of other studies into the effects of resting have been more equivocal; however, in some cases, they were confounded by effects of various other factors, or lacked controls. After reviewing all of the available literature, Hunt et al. (2014) concluded that, in most circumstances, resting pastures during the early growing season will have positive effects on subsequent growth and botanical composition of native pastures. They recommended that the rest period should commence immediately after rainfall sufficient to initiate forage growth (i.e. 38–50 mm) at the beginning of the growing season or, if paddock access is difficult after rain, before the wet season starts (Table 6).

Longer term spelling of rangeland pastures over one or more growing seasons also has been advocated as a means of sustaining their productivity and ecological stability or remediating degraded land (Hunt et al. 2014), based on research findings (Orr and Paton 1997; Post et al. 2006) and the experiences of commercial practitioners (e.g. Landsberg et al. 1998). The recommended duration of resting varies with initial land condition and seasonal growing conditions.

On extensively managed northern properties, distribution of grazing pressure within very large paddocks with few water points also may be an issue for optimising pasture use and cattle production, and minimising land degradation (Hunt et al. 2007). Proximity of water points may be especially important because, although cattle can range large distances from water, activity declines markedly beyond 3–4 km (Fisher 2001; Hunt et al. 2013; Cowley et al. 2020). This can lead to overgrazing near water points and underutilisation of more remote pasture. Therefore, Hunt et al. (2014) concluded that a grazing radius of 2.5–3 km (i.e. ∼5–6 km between water points) should ensure acceptable levels of forage utilisation across the landscape and reduce the overgrazing of pastures near water points, as long as the number of cattle per water point is <300 (Table 6).

Hunt et al. (2014) also found that reduction of paddock size can improve grazing distribution; however, the cost of fencing increases markedly for paddock sizes less than ∼30 km². Accordingly, those authors recommended that, on more extensive northern properties, producers should aim for paddocks of 30–40 km² in area with two water points, and a maximum distance to water of 3–4 km. For more intensively managed properties in north-eastern Australia, the recommendation was paddocks of 20 km² with two water points to optimise grazing distribution (Table 6).

The principles established by Hunt et al. (2014) are being applied in a current MLA-funded project aiming to assess the influence of paddock area and distance to water on reproductive performance and calf wastage in beef heifers on two commercial properties in the Barkly Tableland and north-western Queensland (Walsh and McCosker 2019). The ultimate goal of the project is to refine and test a user-friendly spreadsheet tool to enable producers to compare the benefits and costs of different infrastructure options on their own properties.

Rotational or cell grazing continues to be promoted as a means of more efficiently using rangeland pastures in northern Australia despite an abundance of research findings to the contrary. For example, 30 years ago, O’Reagain and Turner (1992) concluded that there was little difference between continuous and rotational grazing systems in terms of effects on rangeland condition or animal production in South Africa. More recent Australian studies have supported this conclusion (Hunt et al. 2013; Hall et al. 2014, 2016; Schatz 2019), as have comprehensive reviews of the international literature (Briske et al. 2008; Hawkins 2016). Therefore, it is not surprising that cell grazing was found to be much less profitable than set stocking because of its additional capital and operational costs, including the opportunity cost of labour (Hunt et al. 2013). Nevertheless, rotational/cell grazing continues to be practised and advocated by experienced commercial producers in northern Australia and elsewhere. Possible reasons for the dichotomy between experimental evidence and producer experience have been reviewed by Teague et al. (2013), who offered three hypothetical explanations: (1) failure of experimental treatments to take account of ‘accepted’ principles of plant health and animal intake; (2) lack of commercial scale of controlled grazing trials; and (3) failure of researchers to optimise the integration of ecological, economic and social contexts and conditions. This suggests that quality of management is a central issue that should not be discounted when assessing the relative merits of continuous versus rotational grazing systems.

Self-herding has been proposed as a less expensive potential alternative to paddock-based rotational grazing. This approach, which uses feed rewards linked to visual, auditory and olfactory cues to modify cattle grazing behaviour and distribution, was evaluated recently in a project conducted at Kidman Springs, Northern Territory (Revell 2019). Positive outcomes included the attraction of cattle into previously undergrazed areas, more even distribution of grazing pressure in large paddocks, improvement of feed quality by removing dry, rank grasses, and associated reduction of fire risk. This trial used breeding cows of mixed age without deliberate inclusion of bulls. However, stray bulls that were intermittently present used the self-herding attractant stations together with the breeders. Further research is required to confirm and extend these findings under a range of conditions before self-herding telemetry technology can be offered as a reliable alternative to conventional rotational grazing systems.

Most native grasses in northern Australia are tropically adapted C₄ species that feature more efficient photosynthetic processes than temperate C₃ grasses and relatively high biomass production during the growing season. However, the feeding value of these grasses is poorer owing to their lower concentration of soluble nutrients, especially non-structural carbohydrates, and lower digestibility associated with their more fibrous leaf structures (Wilson and Hacker 1987; Van Soest 1994). With these species, seasonal variation in nutritional value can be as important as pasture abundance, particularly the decline in nitrogen (N) content and dry matter (DM) digestibility during the dry season in dry tropical regions. The utility of faecal near-infrared reflectance spectroscopy as a tool to assess pasture quality and improve the nutritional management of northern breeding herds has been reviewed by Dixon and Coates (2007).

To varying extent, the native feedbase has been augmented by deliberate or inadvertent introduction of exotic grass and legume species, beginning as early as the late 19th Century, as reviewed by Clements and Henzell (2010). Notable early examples include buffel grass (Cenchrus ciliaris), Townsville stylo (Stylosanthes humilis) and leucaena (Leucaena leucocephala). These and related species can, in more favourable regions of northern Australia and with appropriate management, offer substantial opportunities for increased productivity of beef herds. However, their introduction should be considered only where conditions allow and after the basic principles of grazing management of native pastures have been implemented, as discussed above. The relative productivity of a range of introduced pasture species in Queensland is summarised in Table 7.

Table 7.  Average commercial steer performance on a range of pasture systems in Queensland (adapted from Noble et al. 2000).

Early production responses to the sowing of introduced grasses, particularly buffel grass on fertile soils cleared of brigalow and gidgee scrub, were substantial and have continued to boost the productivity and profitability of beef enterprises in regions such as Central Queensland (Table 7; Peck et al. 2011). However, since the late 1980s, it has been apparent that the productivity of sown grass pastures declines over time, mostly due to a progressive decrease in available soil N (Myers and Robbins 1991; Tothill and Gillies 1992). For example, cattle liveweight gains for 6 months from June 1989 at Brian Pastures Research Station in South East Queensland were 78 kg/head on 2-year-old pasture, 37 kg/head on 5-year-old pasture, and 20 kg/head on 8-year-old pasture (Myers and Robbins 1991).

A more recent review of this problem estimated that pasture decline reduces production by ∼50%, with a projected industry cost exceeding $17 billion over 30 years, and concluded that the best long-term solution is to establish a range of adapted legumes in the existing grass-dominant pastures (Peck et al. 2011). A main concern was restoration of the productivity of buffel grass pastures, which were estimated to be ‘dominant’ on 5.8 Mha and ‘common’ on a further 25.9 Mha in Queensland, mostly in central and southern regions.

Pasture dieback affecting introduced and some native grasses has been identified as an emerging problem across wide areas of northern and eastern Queensland. Early trials have demonstrated that perennial legumes are unaffected and should be part of the solution to restoring pasture productivity in affected areas (FutureBeef 2021c).

Although most of the recent R&D on new pasture species and cultivars for the tropics and subtropics has centred on adapted perennial legumes, ongoing work in North Queensland is evaluating a range of promising, recently developed Panicum grasses, with an initial focus on optimising seed production (Cox et al. 2019).

The most widespread tropical legumes in northern Australia are the Stylosanthes species, most notably S. scabra cv. Seca (shrubby stylo) and tetraploid S. hamata cv. Verano, which by the end of the 20th Century were being regularly oversown into about 1 Mha of native pastures (Noble et al. 2000). Both of these cultivars are adapted to seasonally dry environments and most soil types of northern Australia other than heavy clays. In addition, both are relatively tolerant of Australian strains of the fungus Colletotrichum gloeosporioides, the cause of anthracnose disease, which devastated previously well-established and widespread communities of Townsville stylo (S. humilis) and common stylo (S. guianensis) in the 1970s (Edye 1997).

Grazing studies during the early 1990s demonstrated that oversowing of either or both stylo cultivars (cvv. Seca and Verano) across a range of northern environments consistently supported higher cattle growth rates than native pastures alone and extended cattle growth for several months beyond that achieved by many native pastures (Coates et al. 1997). For example, at Lansdown, south of Townsville, steers grazing native pastures oversown with Verano gained more than steers grazing native grasses alone, by 50 kg/head.year, during the period 6–12 years after pasture establishment (McCaskill and McIvor 1993). Similar responses to inclusion of Seca in native pastures were observed at two sites in Central Queensland studied over periods of 4 or 5 years (Middleton et al. 1993). These results support the conclusion of Coates et al. (1997) that stylo-based pastures have the potential to allow feeder or grass-finished cattle to meet market specifications and markedly improve breeder performance. However, positive responses to stylo inclusion may be reduced by prolonged drought or pre-existing adequate levels of soil N (e.g. Jones et al. 2000).

The destruction of S. humilis and S. guianensis cultivars by anthracnose in the 1970s led to an integrated research program over two decades to improve anthracnose resistance in stylos. Much of this work focused on development of durable resistance in S. scabra, using cross-breeding of lines carrying different resistance genes (Cameron et al. 1996; Chakraborty 2004). Australian research in this area then lapsed. However, genetic lines of S. seabrana and S. scabra have been recently reselected from old evaluation sites in the humid tropics of North Queensland and are being screened for anthracnose resistance (Gorman et al. 2019). Preliminary results suggest that at least two of the 19 new lines tested appear considerably more resistant than the other new lines and commercially available cultivars.

In contrast to the relatively widespread use of stylo legumes by north Australian beef producers, adoption of leucaena, a highly nutritious and productive tree legume, has been slow despite the sustained research efforts of scientific proponents (e.g. Shelton 2019) and the positive experiences of leading producers (e.g. Heatley 2019). Thus, despite the availability of grazing cultivars since the 1960s and repeated research demonstrations of the production, profitability and environmental benefits of the legume (e.g. Bowen et al. 2018), the total area sown to leucaena across northern Australia is estimated to be no more than ∼130 000 ha, mostly in Central and southern Queensland (Buck et al. 2019a). This is miniscule considering a conservative estimate that >8 Mha of land in Queensland alone is potentially suitable for growing leucaena (Peck et al. 2011).

The slow rate of adoption has been attributed to multiple factors including: lack of awareness of or confidence in the plant’s productive potential; concerns about the negative effects of mimosine toxicity on cattle health and performance; high rates of crop failure related to inadequate knowledge of the environmental and agronomic requirements for the successful establishment and management of leucaena; and high upfront cost of establishment and ongoing cost of management (Buck et al. 2019a, 2019b). Additional factors include susceptibility of the most used leucaena cultivars to the psyllid insect Heteropsylla cubana, especially in more humid growing regions (Lemin et al. 2019), and environmental concerns about the potential of leucaena to establish as a weed in native ecosystems (Campbell et al. 2019; Revell et al. 2019).

The problem of mimosine toxicity in non-adapted Australian cattle was believed to have been solved by the discovery of a rumen bacterium, Synergistes jonesii, that effectively degraded mimosine and its toxic ruminal metabolites, 3,4-dihydroxypyridine (3,4-DHP) and 2,3-DHP (Jones and Megarrity 1986). Cultured strains of S. jonesii were introduced into Australia in 1982, leading to the development and commercial release in 1995 of a mixed-culture bacterial inoculum that could be administered as an oral drench (Klieve et al. 2002). Doubts have been raised about the efficacy of the inoculum (Halliday et al. 2019) and the necessity of its use in non-adapted cattle (Shelton et al. 2019). However, most cattle fed leucaena in Queensland continue to perform well and do not display clinical symptoms of mimosine/DHP toxicity. In the absence of conclusive evidence for widespread reduction in efficacy of the bacterial inoculum, it would seem premature to discontinue the practice of inoculation, particularly for cattle not previously exposed to leucaena.

Past and present research on leucaena breeding in Australia, and possible future opportunities, have been recently reviewed by Dalzell (2019). The review highlighted the use of interspecific hybridisation among the 24 known species of the genus Leucaena to improve psyllid resistance and cold and frost tolerance of cultivars used for grazing, without sacrificing forage yield and cattle growth performance. Ongoing research to breed sterile cultivars of leucaena also was briefly reviewed.

Redlands, a psyllid-resistant hybrid leucaena cultivar bred by University of Queensland scientists, was commercially released in 2017 after 15 years of research. The cultivar was developed by progressive backcrossing of psyllid-resistant lines of L. pallida to the commonly used commercial cultivar L. leucocephala ssp. glabrata cv. Wondergraze. Resulting breeding lines were assessed for psyllid resistance and in vitro forage quality. Performance of cv. Redlands in terms of psyllid resistance and cattle growth is being assessed against Wondergaze in a large grazing trial at ‘Pinnarendi’ station in the Atherton Tablelands (Lemin et al. 2019). Results so far have confirmed the psyllid resistance of Redlands and demonstrated that its ability to support cattle growth matches that of Wondergraze. This trial also has identified challenges to the establishment of leucaena in less-than-ideal environments.

In some regions of Australia, including pastoral leasehold land in Western Australia and Northern Territory, establishment of commercial plantations of leucaena is forbidden, owing to concerns about environmental weed risk. In other regions, such as parts of Northern Territory and northern New South Wales, it has been discouraged for the same reason. Development of sterile varieties of leucaena would obviate this risk and enable expanded opportunities for graziers to take advantage of the benefits of leucaena–grass pasture systems. Current efforts to breed sterile cultivars involve development of male or female sterility via mutagenesis (McMillan et al. 2019) or gene editing to prevent flowering (Real et al. 2019). Interspecific hybridisation to produce sterile triploids also is being evaluated (Real et al. 2019). As well as reducing or eliminating the weed potential of leucaena cultivars, sterility may enhance forage yield because plant resources will not be diverted from vegetative growth to seed production (Dalzell 2019).

Stylos are the preferred legume option for broadacre pasture improvement on lighter northern soil types, and leucaena is best suited to the more fertile soils of Central and southern Queensland. However, until recently, the vast areas of semi-arid clay soil rangelands of northern Australia, such as the Mitchell Grass Downs Bioregion, have had no commercially available or adapted sown pasture legume (Gardiner 2016). Among numerous legumes tested, Desmanthus species were found to be among the best long-term survivors on cracking clay soils in this region (Hall and Walker 2005). Earlier grazing trials had demonstrated high DM production and grazing tolerance of several accessions of D. virgatus grown on heavy soils over 7 years in subtropical subcoastal Queensland (Jones and Brandon 1998). More recently, increased liveweight gains have been observed in steers grazing mixed buffel grass–desmanthus pastures compared with those grazing buffel grass only during the dry season in Central Queensland (Gardiner and Parker 2012; Collins et al. 2016). However, these promising preliminary studies need to be extended and replicated under the harsher northern conditions in which it is hoped desmanthus will be of greatest value. As with leucaena and other tropical legumes, establishment of desmanthus can be challenging and producers are advised to adhere strictly to seed manufacturers’ guidelines, preferably with professional agronomic advice.

The practice of re-visiting abandoned former pasture evaluation sites to assess long-term survival and persistence of tropical legumes has led to the discovery, selection, further evaluation and commercial release of several varieties of desmanthus that are able to persist under heavy grazing on northern clay soils (Gardiner 2016). The most notable example is Progardes™, a composite of five cultivars (JCU 1–5) derived from three species of Desmanthus (D. virgata, D. bicornutus, D. leptophyllus) that was commercially released in 2012 by Agrimix. By 2019, some 35 000 ha had been sown to Progardes™, mostly across Queensland but also in northern New South Wales and Northern Territory, with the targeted soils being Vertosols and related neutral to alkaline clay soils in semi-arid environments (Gardiner et al. 2019); this short update also noted progress with development of four new desmanthus cultivars (JCU 6–9) and the discovery and evaluation of further well-adapted accessions across inland northern inland Australia. Other research has included intra- and interspecific crossing of Desmanthus species to yield novel plants with softer, more erect growth, later maturity and greater cold tolerance (Stuart and Kempe 2017).

During the latter half of the 20th Century, there was a sustained effort by Australian scientists to assemble, characterise and evaluate a tropical forages resource collection, which, by 1996, contained ∼17 000 legume accessions and almost 5000 grass accessions (Hacker 1997). However, after the mid-1990s, this work essentially ceased due to perceived diminishing returns on R&D investment in exotic tropical forages, especially legumes, and by the early 2000s both the germplasm repository in the Australian Tropical Forages Collection (ATFC) and the scientific knowledge base were in danger of being lost. To mitigate the latter risk, an online interactive tool, Tropical Forages (see www.tropicalforages.info) was developed in 2005 that allowed access to information on 180 tropical and subtropical forage species, their adaptation and potential. Content has been revised and updated, with modernised IT access (Cook et al. 2020). This resource was complemented by a comprehensive stocktake and analysis of legume evaluations for tropical pastures in Australia, which was collated into a database with >180 000 individual records collected from 567 sites across northern Australia (Bell et al. 2016). In 2000, custody of the ATFC was transferred from the CSIRO to the Queensland Department of Primary Industries, and then, in 2014, to the newly created Australian Pastures Genebank, in Adelaide (Hughes et al. 2017). This resource, together with plants extant in earlier trial plots, can provide plant stocks that may be useful in particular environments and potentially suitable for changing and variable climates.

The project undertaken by Bell et al. (2016) was prompted by the renewed interest of the northern beef industry in expanding the range of adapted tropical legumes available for use in different northern soils and climatic environments, particularly those currently devoid of sown-pasture options. The approach involved both re-evaluation of previous work, including re-visiting abandoned pasture evaluation trial sites, and identification of opportunities to develop new, elite cultivars. The authors concluded that the highest priorities for further legume development are: legumes that persist in competitive grass pastures in the subtropical semi-arid inland; legumes for clay soils in the northern tropical regions; legumes for lighter sandy and duplex soils in the inland subtropics; and more robust ley legume options for use in mixed farming systems. Several species and accessions were identified that had previously shown promising advantages over existing commercial varieties but have yet to be commercialised. These included cultivars of Desmanthus, Stylosanthes, Macroptilium and Aeschynomene.

Re-examination of old sites of plant evaluation also has led to commercial release of cultivars of several other legume species for pastures on clay soils, including Clitoria ternatea, Macroptilium bracteatum and Stylosanthes seabrana (Cox 2016). Each of these cultivars is intended to occupy different production niches according to climate, soil type and grazing strategy. However, adoption of these cultivars has been slowed by lack of promotion, mismatch of seed supply and demand, and difficulty of establishing legumes in pastures dominated by some key grass species.

The Australian Beef Sustainability Framework defines sustainability as the production of beef in a manner that is socially, environmentally, and economically responsible (Red Meat Advisory Council 2021). Social aspects include public attitudes to animal welfare with implications for the social licence to produce beef. Examples of opportunities to improve current practices in the northern breeding herd are discussed earlier, in the section Breeding herd management, including alternatives to aversive practices such as surgical dehorning, and reducing the incidence of calf and cow mortality. Opportunities to reduce the negative effects of poor nutrition are discussed in the section above (Feedbase management), and heat stress is considered in the present section. The profitability of northern beef enterprises is clearly an important aspect of sustainability, and recently has been assessed by experts in farm business management and northern beef production (McLean et al. 2020). Therefore, this section focuses on aspects of environmental sustainability.

The north Australian beef industry is challenged by a natural environment that is characterised by a highly variable and unpredictable climate, nutrient-deficient soils, animal pests, weeds and toxic plants. The predicted influence of climate change on average temperatures, frequency and severity of droughts, and incidence of extreme weather events is likely to exacerbate these natural environmental challenges. The built or managed environment includes the further challenges of long distances to markets and to necessary services, and lack of supply-chain infrastructure. This section addresses management strategies to promote resilience and offset the impacts of these factors on the productivity and health of breeding herds in the north, as well as to ensure the sustainability, long-term productivity and biodiversity of the rangeland systems in which they graze. These strategies are discussed in the context of the environmental priorities of Red Meat 2030, the current strategic plan of the Australian red meat industries (Red Meat Advisory Council 2019).

Predicted changes in the climate of northern Australia will accentuate the already pressing need to increase the genetic resilience of both pasture plants and the cattle that graze them. Features of climate change projected for northern Australia include increased average temperatures for all seasons with greater incidence of extremely hot spells in the summer wet season. Rainfall patterns also are predicted to change with increased incidence of extreme rainfall events, interspersed with prolonged and severe droughts.

Heat tolerance and drought resistance have long been priorities for the genetic selection or accession of tropically adapted grasses and legumes, and this will need to be further emphasised to combat a hotter and intermittently drier climate. These adaptive characteristics, together with tolerance of low N, are more evident in the C₄ grasses that predominate in northern pastoral systems. The likely key driver of climate change, elevated levels of atmospheric CO₂, also may influence the future balance of C₄ versus C₃ pasture species, particularly in intermediate and subtropical pastures. Early studies indicated that the ability of C₄ plants to increase rates of photosynthesis in response to elevated CO₂ is much less than that of C₃ plants (Ehleringer et al. 1997). However, those experiments were short-term, and more recent results of studies conducted over 20 years have shown that this pattern of response was reversed after 12 years (Reich et al. 2018). Part of this later increase in CO₂-induced growth of C₄ grasses was attributed to increased rates of mineralisation of soil N by C₄ but not C₃ plants. Thus, there is evidence that long-term effects of climate change on northern pasture communities may change the present balance between C₄ and C₃ plants to favour C₄ grasses further. This has implications for biomass production (likely positive), forage quality (likely negative), and, perhaps, the successful establishment and maintenance of C₃ legumes (possibly negative).

The most significant non-nutritional environmental stressor affecting the productivity of breeding herds in northern Australia is heat stress. Exposure of cattle to high ambient temperatures around mating and during early pregnancy can have direct negative effects on development and quality of male and female gametes and on early embryo development and survival (Hansen 2009, 2013; Abdelatty et al. 2018), although such effects appear to be less pronounced in B. indicus than B. taurus breeds (Rocha et al. 1998). Heat exposure during the first half of pregnancy also can impair placental growth and functional development, with consequences for later fetal growth, birthweight and calf survival (Ouellet et al. 2021). Emerging evidence additionally suggests that heat stress of cows during late pregnancy may negatively affect mammogenesis, lactogenesis and later milk yield, at least in dairy breeds, and also may have a negative carryover effect on milk yield of heifer progeny during their first lactation (Ouellet et al. 2021). Finally, whereas most of the above effects are considered to be mediated by direct effects of heat stress on cellular and molecular functions in the testis, ovary, placenta and mammary gland, the secondary effects of heat-induced inappetence and reduced energy balance must be considered, as well as largely independent effects on post-absorptive metabolism (Baumgard and Rhoads 2013).

There is considerable genetic variance for heat tolerance between and within cattle breeds. The substantially greater tolerance of B. indicus than B. taurus breeds is well known (e.g. Beatty et al. 2006; Gaughan et al. 2010a) and, together with tick resistance, was a major reason for the introduction of Brahmans to northern Australia in the mid-20th Century. However, taurine breeds such as the Shorthorn and Hereford that were selected for their ability to produce in the tropics became substantially more heat-tolerant than their unselected counterparts, albeit less so than indicine genotypes (Frisch 1981). This heat tolerance and genetic improvement in other aspects of tropical adaptability, including tick resistance, led to the development of a synthetic taurine breed, the Adaptaur, at ‘Belmont’, the CSIRO’s Cattle Research Station near Rockhampton, during the 1980s and 1990s (O’Neill et al. 1998). However, despite impressive reproductive and growth performance (O’Neill and Frisch 1998), industry uptake of the Adaptaur was poor and the breed is now almost defunct (CJ O’Neill, pers. comm.).

Other examples of tropically adapted taurine breeds available to northern beef producers include the Belmont Red and Senepol, which were developed by infusion of African B. taurus genetics into British breeds (Belmont Red: Africander × Hereford × Shorthorn, O’Neill and Frisch 1998; Senepol: N’Dama × Red Poll, O’Neill et al. 2010), and the Tuli, a pure African Sanga breed. In cross-breeding studies comparing Tuli × Hereford with Brahman × Hereford and Boran × Hereford steers, the heat tolerance of the Tuli crossbred steers was shown to be similar to that of the two indicine crossbreeds (Gaughan et al. 1999).

The above breeds are widely used by the major pastoral companies to create their own proprietary tropical composite breeds based roughly on the combination of one-third Brahman (and other B. indicus breeds), one-third tropically adapted taurine, and one-third British and European breeds (mostly Shorthorn and Charolais but with increasing interest in Angus and Wagyu) (Porto-Neto et al. 2014). However, it is recognised that higher proportions of B. indicus genetics are likely to be optimal for animals in harsher environments (Burrow 2012). In a modelling study, the economic values of using such tropically adapted composite genotypes or a terminal crossbreeding system based on Brahman cows were compared to that of a straightbred Brahman herd as used by much of the northern industry (Burrow et al. 2003). The simulated composite herd was predicted to be considerably more profitable than the crossbred enterprise, with both outperforming the straightbred Brahman herd. Much of the superior performance of the composites was attributed to their higher weaning and turnoff rates, the latter due to greatly increased growth rates of progeny.

Despite experimental evidence for the value of using TCOMP genotypes, and their successful utilisation by the pastoral companies, adoption of crossbred cattle by the broader northern industry has been limited. Reasons for this are not clear but possibly include lack of access to the professional genetic advice necessary for successful introduction and maintenance of complex cross-breeding programs.

An important consideration when selecting for adaptability traits such as heat tolerance is the possibility of antagonistic relationships between adaptive and productive traits (Burrow 2012). However, genetic correlations between resistance to heat stress and reproduction traits are generally positive (Turner 1982; Burrow 2001), indicating that selection for heat tolerance is likely to be associated with improved reproductive performance. Also, a large study of genetic associations between adaptive and productive traits, including growth rate and age at puberty, in Brahman and TCOMP heifers led to the conclusion that selecting for reduced age at puberty is unlikely to have negative effects heat tolerance or other tropically adaptive traits (Prayaga et al. 2009).

Non-genetic management options for mitigating heat stress in cattle include provision of shade, ensuring an adequate water supply, and nutritional manipulation (Henry et al. 2012). Research on physiological, behavioural and production responses to application of these options has focused mainly on intensively managed dairy and feedlot beef systems (Blackshaw and Blackshaw 1994; Gaughan et al. 2010b). For extensively managed northern cattle, provision of adequately spaced watering points with good supply has already been discussed in relation to grazing management. Provision of natural or artificial shade at or near these points and ensuring that paddocks have sufficient tree cover would appear to be achievable management interventions. However, this assumes that the firmly held industry paradigm that ‘trees and cattle don’t mix’ (Arbuckle 2009) can be overcome by existing research evidence that open woodland with a grass–legume feedbase provides both effective shade (González et al. 2013) and superior animal performance (O’Neill et al. 2013). Other management options proposed to reduce environmental or metabolic heat load in dairy and beef feedlot enterprises, such as artificial cooling or dietary manipulation (Gaughan et al. 2008a; Gonzalez-Rivas et al. 2018), are infeasible for extensively managed northern beef operations.

A major challenge for northern beef producers is managing the risk posed by year-to-year variation in the onset, intensity and duration of the summer wet season; this situation is likely to be exacerbated by emerging climate change (Cobon et al. 2020a). A key issue affecting cattle productivity is the number of days in the year when pasture is green, which, according to a recent analysis, is affected more by the number of days of rainfall than by total precipitation because the latter is dominated by extreme events when most rainfall is lost as runoff (Brown et al. 2019). That study concluded that an ideal forecasting system would predict the number of rain days when the soil is dry; incorporate soil moisture content at the beginning of the wet season; determine the probability of an early break to the wet season; and establish the interaction of El Niño Southern Oscillation (ENSO) with other climate modes. The Northern Rainfall Onset (NRO) model has been used by the Australian Bureau of Meteorology since 2015 to predict the date when an accumulation of 50 mm of rainfall is reached after the beginning of September. More recently, the Bureau has developed a multi-week to seasonal model, ACCESS-S1, which became operational in 2018. A recent evaluation of this model found a significant improvement in its ability to forecast interannual variation in the NRO, with further improvements expected in the forecasting of high-frequency rainfall events as the wet season progresses (Cowan et al. 2020). The economic value of using more skilful and accurate forecasting models to set cattle stocking rates before onset of the wet season was assessed to be most important when pasture availability was low at the end of the dry season (Cobon et al. 2020b).

The largest gap in delivering productivity gains in northern Australia is in translation of research through adoption of beneficial practices. In the context of breeding herds, an important step is to identify practices for which research has shown potential for a significant quantum of improvement, that are practical in their application, and that offer an attractive return on investment. The process of adoption should be underpinned by knowledge exchange, skills development, demonstration in the region, and the application of metrics and tools to demonstrate benefits, for example, on calf survival rates.

Successful adoption of novel technologies and practices has occurred at different rates in the industry, and many initiatives have taken a long time to achieve peak adoption. Some were top-down, such as the botulism vaccine and solar pumps with telemetric tank level monitoring, which were driven by commercial interests. Others were more organic and followed the efforts of pioneering producers. Examples are helicopter mustering, yard feeding of weaners and growing steers for live export.

Over time, MLA has invested in adoption and, in the process, has a developed a framework comprising five elements. Four of these are designed to demonstrate sequential improvements in producer engagement and sophistication; the fifth is an enabling initiative. They provide a range of service-delivery options that can be tailored to personal preference and needs, including:

• resource materials such as printed guides, decision tools, websites (e.g. Future Beef https://futurebeef.com.au/)

• awareness activities such as BeefUp Forums, which are also used to direct producers towards adoption opportunities

• short-term skills training courses such as the EDGE suite of programs

• participatory projects and mentoring activities, such as Producer Demonstration Sites and Profitable Grazing Systems courses offering benefit though groups of producers working together.

The enabling activity is intended to build livestock advisory capacity through advisor updates, training of new consultants, future consultant sponsorship and mentoring by experienced operators.

Nevertheless, rates of adoption have been slow and there is a need to identify and develop means to overcome barriers to adoption. Specific barriers are diverse and vary among individual enterprises and over time. They can be loosely classified as social, technical and financial. Whereas technical and financial aspects may be more easily defined, social aspects can be broad and complex, including default bias, education/understanding and time pressures. Such issues can only be addressed by application of social science skills and insights, which are outside the scope of this review. However, in the MLA report by Bell and Sangster (2022) from which much of the content of this paper is drawn, we offered examples of adoption of new practices and technologies by the northern beef industry, including perceived barriers to adoption. Several examples follow.

• Leucaena has proven benefits for cattle performance when grown in suitable geographic regions. Trial plantings have had variable success, affecting return on investment and the time delay for that return, so that the barriers to adoption are financial. Thus, a major objective of The Leucaena Network initiative has been to lower the financial barrier to adoption by sharing knowledge and collaborative producer support.

• Phosphorus supplementation in the wet season has clear financial benefits in areas of P deficiency. MLA has recently funded a Producer Demonstration Site to demonstrate benefits and ‘ease’ of feeding methods. This exposed that a critical barrier to adoption is the practicability of PiP testing to diagnose the P status of cattle and thereby aid decision-making. A resulting goal is to develop a simple, robust crush-side testing system. Modelling of the financial benefit of herd-level interventions for the Katherine region in Northern Territory suggest that benefits from P supplementation are likely to be greatest in the Northern Forest region, where there is widespread incidence of moderate to severe deficiency of P in its soils (Bowen et al. 2020).

• The potential benefits of remote feedbase monitoring on rangeland management are large. Past barriers to adoption were technical, arising from limitations to the system’s predictive power. There is now an opportunity to sell the benefits and user-friendliness of more advanced systems by educating producers and advisors in the use of tools and addressing inadequate internet connections. This technology has potential benefits in all north Australian regions.

• Measurement and mitigation of environmental impacts on fertility, and demonstration of the benefits of intervention, are difficult and complex issues. Despite ongoing demonstration trials and model development, barriers such as implementation of changes in management, cost of building infrastructure such as fences and watering points, and the property-specific nature of possible changes will be difficult to overcome for many producers.

Research projects that have produced important data and insights into northern breeding herd management include:

• The Wambiana Grazing Trial, which has been running for >20 years and has provided valuable data on stocking rates and land condition across seasons including three droughts, as well as economic analysis of the enterprise. Major insights have been that excessive stocking rates lead to lower productivity and profitability over time and land condition also suffers.

• CashCow and the Beef CRC studies, which benchmarked breeder productivity metrics and identified risks associated with cow and calf mortality. Some interventions to improve survival rates and enhance breeder productivity have been prompted by these data including P-supplementation, feedbase management and heifer management.

The rest of this section considers future opportunities to improve adoption of R&D on breeding herd management, feedbase management and management of environmental sustainability.

Many of the potential gains in metrics such as pregnancy rates and mortality could be influenced by improvement in feedbase management as previously discussed. Research has shown that improving pregnancy rates can be tackled through breed selection and genetic improvement, especially bull selection. Genetic gain in the north has yet to take advantage of the new technologies that have been developed, including robust EBVs suitable for genomic testing in Santa Gertrudis and Brahman breeds, and those under way for Droughtmaster. Rapid and effective deployment of superior genetics through bull selection is an urgent requirement, and a plan needs to be promulgated. In addition, there is a direct adverse genetic effect of low birthweight, large teats and small udders on calf growth and survival (Bunter et al. 2014). Progress towards improved birth/weaning weight and teat and udder size as selection goals in cows is expected to be rapid because of the high heritability values for birth (0.48) and weaning (0.39) weights, and udder (0.49) and teat (0.38) scores. Genetic links to behaviour, including flight time and mothering score, as well as calf vigour traits were not found to be useful traits on which to base selection (Johnston et al. 2019). An important caveat for investment in genetic improvement of the northern breeding herd is the overriding imperative to ensure that feedbase management and thus, cow nutrition, is adequate to allow genetic expression of desirable reproductive and other traits.

Mitigation of calf loss can leverage the knowledge that is now available on risks. Studies on causation are under way as described above, and it is hoped that they will close the knowledge gap. Together, they will point to areas where improvement through reducing risks and removing causes can occur. The major research, development and adoption (RD&A) tasks ahead for the industry are in mitigation of calf loss through practical and cost-effective interventions. Interventions should be designed by considering several questions such as: are they researchable, is a large quantum of effect expected, are they practical to apply, is there a clear dollar benefit and return on investment, and do they also improve animal welfare? In terms of quantum, we have already noted (Table 4) that removing some individual risks may yield benefits of 3 or 4 percentage points across a region, and possibly more on a single property. A gap in research lies in demonstrating the return on investment that can be achieved so that the barrier of high infrastructure costs of changes in herd structure can be justified.

Examples of interventions that could have impact are:

• improving shade to reduce effects of THI and improve maternal effects

• smaller paddock size or shorter distance to water to reduce heat impacts

• herd management with a special focus on heifer management, but also managing older cow cohorts

• reducing the mating period to optimise feed utilisation and cluster calving times.

Practices to increase fertility and enhance pregnancy rates also provide opportunities to increase weaning rates. These parameters are linked to improved BCS and quicker return to oestrus, which themselves hinge on feedbase management, especially of the native pasture resources. Quantity, quality and timing are important aspects and, where financially viable, feeding supplements. A rising plane of nutrition, referred to as flushing, can also be used to induce ovulation. For some properties, the management of breeding focuses around the ‘green date’ when significant rain leads to a build-up of feed after onset of the wet season.

As discussed earlier, the issue of decision-making strategies for supplementation of breeding cows is complex and certainly not amenable to a ‘one-size-fits-all’ approach. Nevertheless, strategies to increase weight at mating and reduce time to return to oestrus in first-calf heifers should provide benefits of increased rates and better timing of conception in most cases. In particular, the benefits for delivering calves at the right time of year and not missing a season can have large impacts on weaning rates. We note that a recent modelling analysis predicted that supplementary feeding of heifers will provide little economic benefit to breeding enterprises across northern Australia (Bowen and Chudleigh 2021). This conclusion is inconsistent with conclusions of most empirical studies and industry observations, highlighting the need for further studies to endorse or refute the modelling predictions.

It has been difficult to find a quantitative assessment of the adoption of advice to use moderate stocking rates based on long-term carrying capacity, despite the abundance of persuasive empirical evidence and the considerable efforts by researchers and extension specialists to demonstrate and disseminate the economic and environmental benefits of such practices. However, scientific leaders involved in these efforts concur that across the northern beef industry, the overall level of adoption has been disappointing (e.g. P. O’Reagain and D. Smith, pers. comm.). We do not purport to have the answers to this problem, which surely must be influenced by the complex array of factors that determine decision-making of primary producers, including logic, intuition and emotion (Nicholson et al. 2015). However, we believe that finding solutions to this challenge should be especially amenable to the facilitated small-group learning approach, particularly if groups include respected producers who have had long-term success with adoption of more conservative stocking rates (e.g. Landsberg et al. 1998). For example, for the NB2 initiative, a feedbase learning program has been prepared that includes a list of achievements expected of participating producers, such as ability to assess ABCD land condition, estimate groundcover percentage, and identify key pasture species, as well as other key skills related to feedbase and grazing management (D. Walsh, pers. comm.).

It is generally agreed that the realisation of opportunities for great improvement in the performance of breeding and growing cattle in northern Australia by including tropically adapted perennial legumes in native or sown grass pastures has not fulfilled the promise offered by many research trials and extension demonstrations. This applies particularly to leucaena, which has a passionate following among its protagonists in the scientific and production communities but has yet to be widely adopted by the industry at large. The multiple reasons for this disappointing level of adoption have been discussed in the earlier Feedbase management section (Buck et al. 2019a, 2019b), the principal of which are inadequate knowledge of the environmental and agronomic requirements for, and cost of, successful establishment and long-term maintenance of leucaena plantations. Clearly, exposure of producer groups to peers who have overcome these obstacles should help to overcome negative attitudes such as ‘I’m a grazier not a farmer’. Relevant case studies include the positive experiences of North Queensland producers such as Don Heatley (Heatley 2019) and Brett and Theresa Blennerhasset (Meat & Livestock Australia 2020). However, exposure to these positive examples should be accompanied by financial literacy training and technical support to ensure that producers have (1) a clear view of financial risks and benefits, and (2) access to appropriate agronomic expertise. Some of the required tools and sources of advice are already publicly available; however, future provision of such assistance should be a significant opportunity for private consultancies.

Despite the considerable evidence that P deficiency is perhaps the most significant limitation to productivity of breeding and growing cattle in many parts of northern Australia, especially the Northern Forest region, the level of adoption of wet-season supplementation remains unsatisfactory (Bowen et al. 2020). Part of the problem appears to be lack of awareness of the substantial economic benefits of supplementing deficient animals despite evidence from case studies of commercial properties (e.g. Jackson et al. 2012) and regionally relevant modelling studies of long-term business productivity and profitability (Bowen et al. 2020). Identification of the needs of individual enterprises should lend itself to assessment of the benefits of P supplementation at the property level. This should begin with assessment of soil P levels, assisted by improved analytical and mapping techniques (Zund et al. 2022) and animal P status, preferably by blood collection and laboratory measurement of PiP levels (Dixon et al. 2020), followed by expert advice on supplementation strategy and assessment of responses. Development of a crush-side test for PiP would be an important innovation where access to preparation, refrigerated transport and remote laboratory analysis of blood is a serious limitation. The diagnostic data generated should then be used in benefit–cost ratio analyses for the enterprise in question. A specific practical issue that may require further technical development is the devising of effective strategies for feeding out P supplements during the wet season when P is likely to be the first limiting nutrient in cattle grazing abundant forages that provide relatively high levels of N and energy but are deficient in P. This might include design of home-made lick sheds, provision of online tools for calculating target P intakes and managing the cost of supplementation, and advice on training animals to eat supplements (FutureBeef 2021b). Recommendations should focus on feeding P for as much of the wet season as feasible, and offering some P during the rest of the year to replenish animal bone reserves.

In this review, it has been convenient to discuss separately the R&D needs and gaps for management of the breeding herd, feedbase and environmental sustainability. However, the multifactorial influences on breeding herd performance and sustainability that cross over in all three areas are most apparent when considering environmental issues and management. For example, the need to match stocking rates to appropriately estimated long-term carrying capacity of native and naturalised pastures, discussed in the Feedbase management section, is of central importance to both animal nutrition and the environmental sustainability of rangelands. Likewise, goals for reduction in GHG emissions intensity can be addressed largely by changes in management to improve reproductive efficiency and turnoff rates. Also, aspirations for the genetic improvement of reproductive performance of the northern breeding herd need to be integrated with those for genetic improvement of heat tolerance and other aspects of environmental adaptability. Finally, management options to mitigate heat stress should be considered together with broader aspects of grazing management, such as spacing of watering points, paddock size, and provision of natural and artificial shade.

More specific opportunities to reduce impacts of environment on breeding herd performance include (1) technologies for remote assessment of effects of environmental stressors, especially heat, on animal physiology and production under extensive pastoral conditions; and (2) meteorological tools to predict climatic variation, including the occurrence of extreme weather events and the timing of wet-season onset. The technologies for remote assessment are still under development but producers should be made aware of the benefits of being able to gauge the status of their cattle remotely during extreme heat and other climatic events. Regarding meteorological tools, there has been recent, significant improvement in meteorological models to predict climatic variation, especially timing of the onset of the wet season (Cowan et al. 2020). Raising awareness of these tools and providing training in their operation could be considered as an example of ‘low-hanging fruit’ to promote their adoption.

The following points should be borne in mind when determining priorities for future funding of R&D and strategies for increasing adoption of new technologies and practices by the northern breeding industry:

• Increasing adoption of existing research outputs should be the highest priority and has the potential to deliver the quickest gains.

• Opportunities for participatory research involving producers should be central to all phases of the RD&A process, starting from identification of the problem through to delivery of the solution. This bottom-up approach will need to be complemented by guidance and scientific input from trusted researchers and extension professionals with intimate knowledge of the northern breeding industry.

• Partnership with private companies should be especially important for the ongoing development and adoption of technologies such as satellite monitoring of pasture availability and land condition, and remote sensing of cattle behaviour and performance. Future uptake of these promising technologies will require significant improvements in their ease of use, integration and demonstrated economic value, which should not be a high priority for public funding.

• Marketing-based approaches to address the differing needs and appetite for change of individual enterprises could include profiling producers, matching profiles to preferred communication methods, understanding work routines, and using appropriate mentoring and coaching techniques, including peer-to-peer influencing, to promote change. Meeting the needs of tech-savvy younger producers will present particular challenges and opportunities related to data analysis and management, and development of user-friendly decision-making tools.

In our recently published MLA report (Bell and Sangster 2022), we detailed selected examples of adoption activities in relation to research background, aims and potential, adoption plan and execution, measurement of success (or otherwise), and barriers to adoption. These examples were intended to illustrate possible approaches to extension and adoption of research findings rather than prescribe specific priority topics for adoption. They included:

• phosphorus supplementation in the wet season

• remote sensing technologies for feedbase budgeting

• planting of regionally suitable legumes to enhance nutrition of breeding cows

• genetic selection of polled cattle

• mitigation of environmental impacts on fertility and mortality.

Data for these examples were obtained from MLA project reports, including that of The Centre for International Economics, AgStrat Associates Pty Ltd and ISJ Investments Pty Ltd (2016), and publicly available websites such as http://mla.com.au, http://futurebeef.com.au and http://genetics.mla.com.au. Where available, assessment of ex-ante and ex-post benefit–cost ratios were included.

The following six recommendations are intended to address priority needs, some of which could have short-term impact, whereas others, even if immediately adopted and implemented, will take longer to generate impact but do offer tangible benefits. They are not necessarily listed in order of importance because needs and applicability will vary among regions and individual businesses.

1. The goal of identifying and implementing appropriate and uniform metrics for assessing breeding herd performance is strongly advocated. This will be essential to the establishment of baseline data from which responses to changes in management can be measured and benchmarked against peer enterprises. Accurate measurement of herd performance is also essential to sound economic management and will be central to producer involvement in research activities.

2. Studies on the causes of calf and cow mortality should be continued to improve understanding of the linkages between risk, causation and death, and thus enable the design of interventions to break the linkages. This work should include field-testing of practical methods for electronic tracking of cow movement and behaviour, initially for research usage but with a goal of eventual commercial application for remote management of breeding herds. Interventions to mitigate calf loss should be designed with the following questions in mind:

   1. Are they researchable?

   2. What quantum of effect is expected?

   3. Are they practical to apply?

   4. Is there a clear economic benefit and return on investment?

   5. Will there be publicly demonstrable improvements in animal welfare?

3. Strategies to promote the demonstrated economic and environmental benefits of sustainable grazing management practices should be the primary focus of work on feedbase RD&A. These should include the upskilling of producers to assess ABCD land condition, and to estimate pasture biomass and other key variables. Specific R&D priorities should include: (i) further development and deployment of tools for remote sensing of land condition, feed on offer and pasture growth rate, including iterative involvement of end users to ensure that the tools are user-friendly and clearly beneficial; and (ii) integration of the principles of sustainable grazing into best management practices for the northern breeding herd (e.g. MLA ‘Sweet Spot’ project).

4. Promotion of P supplementation in P-deficient regions/locations should be considered as a high priority example of ‘low hanging fruit’ because of the clear evidence of major, measurable effects and early impacts on breeding herd performance. Key elements of approaches to increasing adoption should include: (i) increasing awareness of, and access to, diagnostic tools for identifying soil/plant and animal deficiencies (an R&D objective could be development of a crush-side test for PiP); (ii) development and demonstration of practical and effective wet-season feed-out practices (e.g. MLA ‘Easy P’ strategy); and (iii) use of appropriate production metrics and economic analysis to demonstrate impacts on herd performance and profitability.

5. There is opportunity to broaden levels of adoption of research-proven management practices that have already been successfully implemented by segments of the breeding industry. These include:

   1. controlled mating to enable seasonal breeding

   2. evidence-based culling of older cows to reduce cow and calf mortality

   3. early weaning to enhance cow fertility by reducing the duration of post-partum anoestrous

   4. further development of supplementary feeding practices such as spike feeding to achieve target mating weights and reduce post-partum anoestrus.

   5. Advertising of case studies such as those posted on the MLA website and use of commercial demonstration sites should be part of the process. For some practices, these should be accompanied by access to professional expertise to evaluate benefit–cost ratios and whole-farm impacts of changing specific management practices on individual properties. For example, a decision to introduce early weaning would need to be informed by understanding of the needs and capacity to feed adequately and otherwise manage younger calves, such as access to good-quality hay. On the other hand, development of general and specific guidelines for culling cows should be a high priority because the gains will be immediate and readily apparent.

6. Research, development and adoption of genomic selection to improve genetic traits such as reproductive efficiency, heat tolerance, tick resistance and polledness should continue to be a priority for the northern breeding industry. While there has been some impressive research success, the path to market and adoption of the genomic tools developed so far, and those yet to come, needs to be more clearly defined and implemented. For example, realistic modelling of anticipated rates of progress and economic benefits, especially for northern seedstock enterprises, should be a key element. Once again, this highlights the imperative to collect and compile accurate data on animal performance.

The recommendations listed below include additional, potentially positive opportunities for RD&A as well as advice on aspects that are not considered to warrant further industry investment at this stage.

1. Testing of approaches to herd management and risk assessment in the face of an increasingly variable climate should include development and dissemination of clear guidelines to aid decision-making for drought management. These should be accompanied by promotion of awareness and utilisation of increasingly robust tools for predicting important meteorological and climatic events such as timing of the onset of the wet season.

2. R&D on mitigation of heat stress in breeding herds should include development of tools for remote assessment of the physiological status of sentinel animals and benefit–cost analysis of various infrastructure options to provide shade and access to water.

3. Among the existing options to use tropically adapted perennial legumes to enhance the feedbase and improve nutrition of the breeding herd, the stylos, desmanthus and leucaena stand out, with each suited to different regions across northern Australia. Efforts to fulfil the potential of these legumes should include increasing producer awareness of both their productive benefits and potential challenges to their establishment and maintenance. Access to professional agronomic advice and demonstrations of successful adoption should be important components of this work.

4. There are few compelling data that suggest further investment in animal disease research or control of predators will significantly benefit the northern breeding industry. Therefore, such work should be a lower priority.

5. Development of practical methods to reduce ruminal methane production with efficacious dietary additives is warranted. Also, awareness of the win–win benefits of improved reproductive efficiency and increased turn-off rates on emissions intensity at the herd or enterprise level should be emphasised and promoted.

Data sharing is not applicable because no new data were generated or analysed during this study.

Both authors were involved in writing a related industry report that was commissioned by MLA but have no other no conflicts of interest.

This research was funded by MLA Project B.GBP.0055. MLA is funded by Australian red meat producers through a production levy and by the Australian Government’s matching contribution for eligible R&D activities.