You are on page 1of 14

Agricultural Systems 155 (2017) 1932

Contents lists available at ScienceDirect

Agricultural Systems
journal homepage:

Management options for dairy farms under climate change: Eects of MARK
intensication, adaptation and simplication on pastures, milk production
and protability
Matthew T. Harrisona,, Brendan R. Cullenb, Dan Armstrongc
Tasmanian Institute of Agriculture, University of Tasmania, TAS 7320, Australia
Faculty of Veterinary and Agricultural Sciences, University of Melbourne, VIC 3010, Australia
D-ARM Consulting, 260 Old Telegraph Rd, VIC 3818, Australia


Keywords: There are few holistic analyses of agricultural systems that inclusively consider how the combination of gradual
Drought climate change and increased frequencies of extreme climatic events inuence biophysical variables as well as
Extreme climatic events economic returns. Here we examine how climate change to 2040 inuences pasture growth rates, grazed pasture
Grazing harvested (PH) and protability of three case study (baseline) farms in southern Australia. We applied
Heat wave
development options (or adaptations) to baseline farms in each region that either intensied, simplied or
modied the seasonal distribution of feed supply (Intensify, Simplify or Adapt, respectively) by manipulating
Representative concentration pathway
Simulation several components of the farm system simultaneously, including herd size, liveweight and farm assets.
In general, climate change reduced annual pasture produced. On dryland farms, hotter, drier conditions
reduced growing durations through later autumn breaks and earlier nishes to spring growth, although winter
growth rates were enhanced. On irrigated farms, the magnitude and inter-annual variability of PH was less
inuenced by climate change. Overall, climate change reduced milk production and income, and increased costs
due to additional fodder conservation and more purchased feed.
Current climate variability caused far greater inter-annual variation in PH and prot compared with the long-
term impacts of climate change. This suggests that farm outcomes may be improved by tactically managing for
short-term climatic variability, rather than by making long-term strategic changes in preparation for climate
Future work on adapting dairy businesses to climate change should examine development options that help
maintain or extend growing season length and/or harness the additional winter growth. Our study indicates that
climate change impacts on dairy systems will be regionally-specic; no individual development option was
universally eective in reducing pasture losses to climate change across regions and development options, and
no option consistently increased or decreased PH across sites. Future adaptation strategies should thus take into
account not only local climate variability as well as climate change, but also the existing farming systems already
operating at each site.

1. Introduction employing an estimated 39,000 people (DA, 2015). An important sector

of the dairy industry is based in south-eastern Australia, where the
Increasing global demand for animal products will require strategies majority of production occurs. Similar to their New Zealand counter-
to increase future livestock production. By 2050, consumption of meat parts, the Australian industry remains predominately pasture-based
and dairy is expected to have respectively risen by 76% and 65% with an estimated 7075% of feed requirements met from grazed
against a 200507 baseline, compared with 40% for cereals (FAO, pasture under normal seasonal conditions (DA, 2015), largely because
2012). Currently, the major dairy consumers are China, India, the EU grazed pasture is generally the cheapest source of feed (Chapman et al.,
and the United States; India and China together consume almost 2014a). However, dairy production systems in this region are diverse,
100 million tonnes of milk products per year (Bailey et al., 2014). with some systems implementing grain or alternative concentrate
Australian dairy is a $13.5 billion farm, processing and export industry feeding during milking, and others operating similar systems to those

Corresponding author at: P.O. Box 3523, Burnie, TAS 7320, Australia.
E-mail address: (M.T. Harrison).
Received 24 October 2016; Received in revised form 30 March 2017; Accepted 6 April 2017
0308-521X/ 2017 Elsevier Ltd. All rights reserved.
M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

in Europe and North America, where cows are housed and fed total costs. The aims of the current paper were
mixed rations for most or all of the year (Ho et al., 2013). While pasture
growth is a key driver of prot in pasture-based dairy farms, in practice 1. To examine the eect of climate change on (a) seasonal pasture
farms are complex ecosystems (Bryant and Snow, 2008) with biophy- growth rates, (b) pasture intake per animal and harvested and (c)
sical, management and nancial components (Kalaugher et al., 2017). prot of three case study dairy farms, and
Pasture production is highly dependent on seasonal climatic condi- 2. To examine the inuence of climate change on PH and prot if
tions, with the amount and timing of rainfall being a key factor in the either of three development options that Intensied, Simplied or
seasonal and annual variability of pasture growth (Chapman et al., Adapted the baseline farm were applied in each of the three regions.
2009). Drought conditions have a substantial impact on farm nancial
performance because low pasture growth rates reduce pasture intake,
leading to greater reliance on purchased feeds (grain, concentrates, hay 2. Methods
and silage), which also tend to be more expensive during such times
(Armstrong et al., 2005; Chapman et al., 2014b). Projections for climate 2.1. Overview
change in south-eastern Australia indicate that temperatures will
increase and rainfall is likely to decrease, although large uncertainty An earlier publication detailed the inuence of climate change to
exists in rainfall projections (CSIRO, 2015). Previous studies have 2080 and extreme climatic events on dairy farm production and
shown that warmer and drier climates will lead to increases in pasture protability (Harrison et al., 2016), as well as the new approach used
growth rates in winter and early spring, but with contraction of the to generate the climate data. Here, we rst quantify the inuence of
spring growing season (Cullen et al., 2009; Moore and Ghahramani, climate change between the present and 2040 on pasture production,
2013). However, the future climate scenarios used in those studies did purchased fodder requirement, milk production and protability, then
not capture either (1) the projected eects of rainfall patterns, with elucidate the ability of dierent farm system adaptations in mitigating
more rain forecast to fall in fewer, larger events and longer dry spells or reversing adverse eects of climate change. We simulated these
between events (CSIRO, 2015), or (2) the increase in temperature systems using 38 years of either historical or future climate data and
extremes, including heat waves of greater frequency and magnitude assessed Low, Medium and High climate change trajectories. To enable
(White et al., 2010). Recent work accounting for extreme climatic a more complete insight into the inuence of climate change on
events have revealed an additional detrimental impact to pasture production and prot, we held other variables constant where possible,
production and farm nancial performance compared with analyses such as milk prices and feed costs. In a subsequent report, we will
using only gradual changes in climate (Harrison et al., 2016), with further examine the impacts of implementing farm system adaptations
impacts likely to vary regionally and according to the type of farm at the beginning of wet or dry decades for the historical and future
system implemented. climates, and also incorporate variability in milk and supplementary
Existing climate change research related to dairy production has feed prices. Full details of the methods used to create the future climate
mainly focused on climate impacts rather than adaptation (e.g. data are provided in Harrison et al. (2016).
Klinedinst et al., 1993). For instance, climate projections for South
Africa suggest that dairy cattle over most of the country are likely to be
severely heat stressed at peak annual temperature-humidity index from 2.2. Historical climate data, regional working groups and site characteristics
the year 2046 onwards (Nesamvuni et al., 2012). Although there has
been some research of dairy farm adaptation to climate change in other Australian dairy farming occurs across diverse climatic zones, but
countries (e.g. New Zealand (Kalaughter et al., 2017) and Ireland the bulk of milk production occurs in south-eastern temperate regions.
(Fitzgerald et al., 2009)), there has been little in Australia. What has We undertook case studies of three farms located in regions represent-
been done focuses on adaptation of specic attributes of the farm ing the breadth of dairy farming in this region. These were situated in
systems, such as changes to single factors including pasture character- the Fleurieu Peninsula (Parawa, South Australia), Gippsland (Moe,
istics (Cullen et al., 2014), stocking rates and/or calving dates (Harrison Victoria) and North West (NW) Tasmania (Wynyard). An expert work-
et al., 2014a; Phelan et al., 2015). Such options do not address broader ing group was established in each of the regions, which comprised
questions about future options for the dairy industry by considering, for farmers and local service providers. These groups dened the criteria
example, how whole of farm system changes inuence production and for selecting the case study farm, the farm development options to be
prot resilience to climate change. For example, adaptations that aim to analysed, and provided input into the assumptions used when analysing
intensify production may consider not only increasing stocking rates farm system changes. Regional working group members also contrib-
and concentrate feeding per animal, but also farm infrastructure uted to the interpretation of the results to ensure that the analyses were
investment and total debt. subjected to informed feedback and that a broad range of perspectives
In this study, the impact of future climate scenarios was examined in were considered. Historical daily weather data for each site were
three dairy regions of south-eastern Australia. Case study farms were sourced from meteorological archives (http://www.longpaddock.qld.
modelled as representative baselines in each region, along with three, with baseline data measured from 1 January 1975 to 31
contrasting development options. These represented regionally relevant December 2013. Annual average rainfall of the historical periods at the
systems that (1) Intensify (increased stocking rate with greater reliance three sites ranged from 937 to 995 mm, with a winter-dominant pattern
on o-farm resources and farm infrastructure), (2) Simplify (reduced (Fig. 1). In winter, historical minimum and maximum daily tempera-
stocking rate, fewer inputs such as fertiliser and less reliance on feed tures vary from 4 C to 15 C, respectively, and from 10 C to 26 C,
resources produced external to the farm) or (3) Adapt (re-organise respectively, in summer (Fig. 1). All simulations using historical data
resources around expected changes in pasture growth patterns) of each assumed a baseline atmospheric CO2 concentration of 380 ppm. Case
case study farm. A participatory action research (PAR) approach was study farms used as baselines were located in cool temperate regions
used with regional working groups integral to describing, modifying 12361 m ASL with Grey Dermosol or Red Ferrosol soils (Isbell, 2002)
and validating the modelled results. This paper addresses the impacts of and plant available water capacities of 3280 mm (Table 1). Pasture
climate change on the production and protability of the case study species mainly consisted of perennial ryegrass (Lolium perenne), and
farms, while in a future study we address the impact of climate change were simulated as such in this study. Further site characteristics are
in relation to inherent climate variability in wet and dry periods, the provided in Table 1 in Harrison et al. (2016), and historical climate
transition from the base farm to each development option, and other details are shown in Fig. 1.
sources of variability in the farm business, notably milk prices and feed

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Fig. 1. Monthly average rainfall (bars) and temperature (lines) for the historical ( and ) and 2040 climate scenarios with either Low ( and ), Medium (
and ) or High ( and ) climate change trajectories at (a) Fleurieu Peninsula (SA), (b) Gippsland (Vic) and (c) NW Tas.

2.3. Climate change data the Variable approach in (Harrison et al., 2016)). Our rationale in
using these data was to study the impacts of climate change data
Harrison et al. (2016) describe the methods used to produce the containing more extreme climatic events than historically observed,
future climate data in the present study. A key component of this since such events are predicted to occur more frequently in future
approach was the use of historical climate from each site to create the (IPCC, 2012). The approach documented in Harrison et al. (2016)
future climate data. Importantly, this allowed modication and repla- generated future climate data from global circulation models (GCMs)
cement of climate indices from within the historical sequence, whilst containing both gradual climate change as well as more extreme
concurrently retaining the original time series order (see description of climatic events as predicted by regional climate models (White et al.,

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Table 1 2.4. Simulation model and baseline case study farms

Location and management details, soil characteristics and prices used to model the
baseline dairy farms under all climate scenarios.
A whole farm model (DairyMod; Johnson et al., 2008) was used to
Fleurieu Gippsland NW Tasmania simulate the farm system for each site and climate scenario. Using daily
Peninsula radiation, rainfall, maximum and minimum temperature as inputs,
DairyMod simulates photosynthesis, sward growth and composition,
Latitude/Longitude (S, E) 35.55, 38.20, 146.26 41.00, 145.73
soil biophysics including water balance, organic carbon and nitrogen
Altitude (m above sea level) 361 150 12 pools, grazing animal pasture intake as well as intake from supplemen-
Soil typea Dermosol Dermosol Ferrosol tary feed sources, milk production, mechanical grass cutting, and
Plant available water from the 80 80 32 feedbacks between plant growth and pasture status. Pasture quality is
surface to 400 mm deep an emergent property of the model, varying in response to climate,
plant growth, organic matter inputs to the soil, soil water content and
Saturated hydraulic 150 150 150
conductivity (mm/d)b the fractions of dead and live plant material (pasture quality parameters
Milking area (ha) 208 110 150 in this work were not modied from defaults specied in Johnson
Non-milking area (ha) 44 83 0 (2016)). The ability of DairyMod to reliably reproduce the eects of
Area irrigated (ha) 0 0 100
climate variability and management on pasture growth rates (Chapman
Max. allowable fodder intake 13 10 5
(kg DM/cowday)c et al., 2009; Cullen et al., 2008) and milk production (Johnson et al.,
Min. pasture grazing threshold 1.2 1.3 1.5 2016) in each study region has been well established.
(t DM/ha) For each site, we modelled pasture-based dairy systems using
Max. pasture threshold required 4.0 3.8 4.5 representative case study farms as baselines (Table 1). By interviewing
for cutting (t DM/ha)
farm managers, we collected several years of detailed farm data
Milk price ($/kg fat and 5.40 5.25 5.40
protein) (fertiliser use, milk production, stock numbers, supplementary feeding
Grain cost ($/t DM) 290 335 335 quanta, etc.) and calibrated simulations against these data. Farms at the
Purchased fodder cost ($/t DM) 210 265 265 Fleurieu Peninsula and Gippsland were rainfed, whereas two-thirds of
Irrigation costs ($/ML) NA NA 20
the farm in NW Tasmania was irrigated. Both rainfed farms incorpo-
a rated areas of land that were not grazed by milking cows (Table 1), but
After Isbell (2002).
Value in most limiting layer. were grazed by young stock and/or dry cows, and used for hay/silage
Includes only silage conserved on-farm or purchased. production. Nitrogen fertiliser and irrigation applications were made
using the decision rules dened on each case study farm and were
2010). calibrated in line with historical rates. Within each simulation, xed
Temperature, rainfall and CO2 projections were obtained from the stocking rates were used. Livestock feeding was modelled according to
top quartile of an ensemble of GCMs according to their skill score in animal daily energy demand, rst with grain, second with pasture and
southern Australia (Watterson et al., 2013), and were based on the third with conserved or purchased fodder when animal energy demands
Representative Concentration Pathway (RCP) 8.5 at 2040. Low, Med- were not met by the grain and/or pasture following manager decisions
ium and High climate impact scenarios were developed by taking the on case study farms. We simulated rotational grazing of paddocks until
10th, 50th and 90th percentiles for rainfall and temperature change, pasture dry matter reached a minimum biomass threshold, in line with
respectively, from the ensemble of GCMs; hereafter these data are management conducted on each case property. Pasture biomass was cut
referred to as future climate data or 2040 data. Low impact scenarios to a residual of 1.0 t DM/ha and conserved as silage if biomass reached
increased rainfall in most months (Fig. 1). Averaged over 38 years, a maximum threshold (Table 1). Conserved silage was fed as fodder in
mean monthly rainfall increased by 4%, 6% and 7% at the Fleurieu accord with maximum allowable intakes and wastage rates at each site,
Peninsula, Gippsland and NW Tasmania, respectively. Medium impact and silage/hay was purchased for fodder in years when there was
scenarios had little inuence on average annual rainfall (14% reduc- insucient pasture or on-farm fodder reserves (Table 1). We used the
tion across sites), whereas the High impact scenarios consistently terms pasture harvested (PH) to represent grazed plus cut pasture over
reduced rainfall compared with historical averages (613% reduction the year, and intake for the grazed component only, following Allen
across sites; Fig. 1). All future climate scenarios increased daily average et al. (2011).
temperatures, such that the Low, Medium and High change scenarios
respectively increased temperatures by 4%, 5% and 67% across sites, 2.5. Farm development options
averaged over the 38 years. Increased maximum daily temperatures
mostly contributed to this change, although minimum daily tempera- For each region, we modelled three development options (large-
tures also increased to a lesser extent (Fig. 1). Atmospheric CO2 scale adaptations) that were based on representative farming systems
concentrations for all future climate simulations were set to 489 ppm. dened by the regional working groups. Development options either
Compared with historical data, future climate data had 89% intensied, de-intensied (simplied) or adapted to each baseline
greater continuous dry-day durations (analogous to drought), 2770% farming system to climate change (hereafter we refer to these systems
greater average maximum rainfall events (intense precipitation days), as Intensify, Simplify and Adapt). The changes to baseline systems were
59% higher 90th percentiles of maximum daily temperatures and whole of farm system adjustments, as opposed to changes to individual
131% greater heat waves (measured as the number of consecutive components of the system.
days above the historical 90th percentile); variation within these ranges The Intensify options increased milk production by increasing herd
depended on climate scenario and site location. The overall variability size/stocking rate and mature cow liveweight, as well as the amount of
of future climate data was similarly greater than historical data, with purchased grain or hay/silage fed to animals. These options were
the standard deviation of daily rainfall and temperature being 2678% associated with greater investment in infrastructure and machinery
and 312% greater than respective standard deviations of baseline data. (Table 2). At all sites, the Intensify system reduced the reliance on
Further statistics regarding climate data used in this work are provided pasture as a feed source relative to the case study farm. For the Fleurieu
in Supporting information 1. Peninsula this development option was simulated as a total mixed
ration (TMR) system, with no grazing and 44 ha sold for $550,000 to
help fund the development of infrastructure. In contrast, the Intensify
system in NW Tasmania increased pasture production by increasing

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Table 2
Summary of modelling inputs and assumptions for the base farm and each development option.

Baseline Development options

Fleurieu Peninsula (South Australia) Intensify Adapt Simplify

Milking cows 350 400 400 290
Stocking rate (cows/ha of milking area) 1.7 n/a 1.9 1.4
Mature cow liveweight (kg) 550 600 550 520
Replacements reared per year 97 110 110 85
Calving time MayJuly Year round Split (May-Jun, Aug) May-Jun
Grain (t DM/cowyear) 1.6 3.0 2.2 1.0
Wastage of hay/silage during feeding (%) 10 5 9 10
Irrigation (mm/yearha) 0 0 0 0
Total assets managed ($,000,000) 5.35 5.58 5.59 5.26

Gippsland (Victoria)
Milking cows 352 500 250 200
Stocking rate (cows/ha of milking area) 3.2 4.5 2.3 1.9
Mature cow liveweight (kg) 475 550 550 475
Replacements reared per year 100 125 65 50
Calving time Aug-Sep Split (Apr, Aug) Apr Apr
Grain (t DM/cowyear) 1.1 2.0 1.8 0.5
Wastage of hay/silage during feeding (%) 15 7.5 15 15
Irrigation applied (mm/yearha) 0 0 0 0
Total assets managed ($,000,000) 4.47 5.12 4.32 4.25

North-western Tasmania
Milking cows 450 600 500 350
Stocking rate (cows/ha of milking area) 3.0 4.0 3.3 2.3
Mature cow liveweight (kg) 500 500 500 500
Replacements reared per year 115 155 130 88
Calving time Aug Aug Aug Aug
Grain (t DM/cowyear) 1.1 2.0 1.0 0.5
Wastage of hay/silage during feeding (%) 20 5 20 20
Irrigation applied (mm/yearha)a 205 311 325 205
Total assets managed ($,000,000) 5.92 6.49 6.09 5.77

Values represent averages across the farm area of 150 ha. The entire farm area was irrigated for the Intensify and Adapt options, whereas only 100 ha of the baseline and Simplify
options was irrigated.

irrigated farm area, but also reduced reliance on pasture feed by via interviews. Annual whole farm budgets were developed according
increased feeding of grain and hay. Calving times varied across to Malcolm et al. (2005), and were evaluated using annual prot
development options. For the Fleurieu Peninsula calving time was budgets for each of the 38 years analysed. Total annual costs shown in
year-round to achieve a relatively at supply of milk throughout the Table 4 account for expenses related to herds, sheds, feed requirements,
year, for Gippsland calving time was split equally such that half the cash overheads, depreciation and imputed labour. Operating prot was
herd gave birth in autumn and winter/spring, respectively, and for the calculated as gross income minus variable and overhead costs, before
Intensify option in NW Tasmania calving time was not changed from interest. Average return on total assets managed (ROTA) was used to
the baseline system of winter/spring calving. compare protability across systems, since this metric allowed compar-
The Simplify development options reduced stocking rate and the isons between businesses (development options) having dierent
number of replacements required, and also maintained/reduced mature amounts of assets invested. Capital gains were not included in ROTA
cow liveweight. Simplify systems increased the proportion of feed computations. In this paper we assumed that development options had
derived from both grazed and conserved pasture, and reduced the already been implemented such that systems were in steady-state, with
amount of supplementary feed purchased. By reducing herd size and no expenses incurred by transitioning from baseline systems to each
pasture inputs (such as nitrogen fertiliser), this development option was development option. The same prices per unit input/output were
associated with lower feed and farm machinery costs, lower labour assumed in all years and development options so that trends in ROTA
costs and required no investment in additional infrastructure or reected variability in climates (caused by changes in physical inputs),
machinery. rather than variability caused by temporal price uctuations from year
In general, the Adapt development options were designed to re- to year. The inuence of variation in milk price and feed costs over time
organise resources around expected changes in pasture growth patterns will be also considered in our future work.
without substantial capital investment. For the farms at the Fleurieu
Peninsula and NW Tasmania, the size of the milking herd in this option
was increased, whereas for the farm in Gippsland the number of milking 3. Results
cows was reduced. In Gippsland the calving pattern was changed from
late winter/spring to autumn. In NW Tasmania, pasture production was 3.1. Inuence of climate change on seasonal pasture growth rates
increased by increasing the farm area under irrigation. At the Fleurieu
Peninsula, a partial-mixed ration (PMR) was adopted with investment Regardless of whether sites were rainfed or irrigated, climate
in a basic feedpad. change reduced pasture growth rates from late spring throughout
summer and into the following autumn (OctoberMay), and elevated
growth rates in winter and early spring (JuneSeptember; Fig. 2). These
2.6. Economic analyses trends were amplied as the severity of climate change and extreme
climatic events such as drought and heat waves increased from the Low-
Detailed information on recent income from milk and livestock sales to High-change scenarios, with increasing water decit and higher
and costs of the case study farms was collected from the farm managers temperatures depressing summer and autumn growth, and higher

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

90 160 (b)
80 140

Coefficeint of variation (%)

Pasture growth rates (kg
70 120
DM/ha.d) 100
20 40

10 20
0 0




90 300
(c) (d)
Pasture growth rates (kg


Coefficient of variation (%)

Historical Low 250
70 Medium High
60 200

30 100
0 0









90 40
(e) (f)
Coefficient of variation (%)

80 35
Pasture growth rates (kg

70 30
60 25

10 5
0 0






Fig. 2. Monthly average pasture growth rates and coecient of variation at (a, b) the Fleurieu Peninsula (SA), (c, d) Gippsland (Victoria) and (e, f) North-western Tasmania for the
baseline farm systems under the historical and 2040 climate scenarios.

temperatures and atmospheric concentrations of CO2 causing greater 3.2. Eects of climate change on pasture harvested, milk production and
growth rates in winter (Fig. 2a, c, d; also see Supporting information 1). return on total assets of the baseline farm systems
The reduction in pasture growth rates at the dryland sites (Fleurieu
Peninsula and Gippsland) was largest in spring and early summer, For each of the baseline (current) farming systems, pasture intake
whereas the magnitude of growth depression at NW Tasmania was and milking area PH were mainly reduced with climate change. As for
smaller (Fig. 2c, Table 1). Monthly variation in pasture growth rates pasture growth rates, the reduction became greater from the Low to
was greater for High and Medium climate change scenarios in late High change scenarios and was more severe for the dryland sites (Fig. 3,
spring and summer, indicating the detrimental eects of high tempera- Table 3). There was relatively little eect of climate change on the
tures combined with low rainfall in these months (JanuaryMarch and distribution or variability of pasture intake or harvested at the irrigated
NovemberDecember; Figs. 1, 2). Despite increased variation at the site. However, the variation in pasture intake increased at the dryland
ends of the year, long-term growth rate variation in winter was lower sites in 2040, particularly at the Fleurieu Peninsula, and the proportion
(Fig. 2). of PH by cutting increased. Reduced pasture production in future
climate scenarios increased the need to purchase feed, such that the
requirement for fodder increased as the severity of climate change

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Fig. 3. Milking area PH (t DM/ha) for each development option at (a) the Fleurieu Peninsula (SA), (b) Gippsland (Victoria) and (c) North-western Tasmania under historical ( ), and
2040 Low- ( ), Medium- ( ) and High- ( ) change scenarios. Boxplots show the median, 25th and 75th percentiles in the box, with the 10th and 90th percentiles in the
whiskers and dots showing outliers.

became greater (relative to the historical climate for the baseline, the For all baseline farms, the climate change scenarios required more
High change scenario increased purchased feed requirements by 10%, purchased feed and increased the need to conserve fodder, reduced
22% and 12% at the Fleurieu Peninsula, Gippsland and NW Tasmania, gross income and increased total costs (Tables 3 and 4). The reduction
respectively). Since nitrogen was applied in response to pasture demand in gross income with climate change from the historical system to the
in the biophysical simulations, lower growth with climate change High change scenario was low (0.71.2%) compared with the increase
reduced the requirement to fertilise with N in pasture-based options in costs (0.44.6%, Table 4). Together these eects reduced return on
(Table 3). total assets (ROTA) in absolute terms by 0.5%, 0.8% and 0.4% at the
It should be noted that the biophysical model used to conduct these Fleurieu Peninsula, Gippsland and at NW Tasmania, respectively
simulations (DairyMod/SGS, see Johnson et al., 2008) adjusts animal (Table 4). At the Fleurieu Peninsula and Gippsland, the magnitude of
feed requirement in order to meet metabolic demand, and by doing so ROTA depression increased as climate change became more severe,
largely maintains milk production under dierent climatic inputs. Since whereas there was little change in ROTA between alternative climate
milk production per cow and per farm under dierent climate scenarios scenarios at NW Tasmania since two-thirds of the farm area was
diers little (maximum change of 3% across all sites and options), irrigated (Fig. 3c). Climate change increased ROTA variability at the
changes in gross income due to climate scenario reect this and hence, Fleurieu Peninsula and Gippsland, but had little eect at NW Tasmania
were also small. (Table 4, Fig. 4).

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Table 3
Long-term average annual results for each case study site (percentage coecient of variation in parentheses). Results are shown for the historical and 2040 climate scenarios for each
baseline farm and development option.

Site and development Climate Milk yield per Total milk Pasture Hay/silage Fodder Cut yield Milking area pasture Nitrogen fertiliser
option animal prod. intake intake purchased harvested (= PH) applied
(kg MS/cow) (t MS/ (kg DM/ (kg DM/cow) (t DM/farm) (t DM/ha) (t DM/ha) (kg N/ha)
farm) cow)

Fleurieu Peninsula
Baseline Historical 526 (3) 184 (3) 3326 (13) 1989 (18) 536 (30) 2.0 (7) 7.6 (9) 110 (0)
Low 519 (4) 182 (4) 3260 (14) 2021 (17) 509 (35) 2.0 (12) 7.7 (11) 110 (0)
Medium 520 (3) 182 (3) 3180 (15) 2071 (18) 536 (35) 2.0 (12) 7.5 (12) 110 (0)
High 520 (4) 182 (4) 3025 (18) 2177 (18) 589 (35) 2.2 (12) 7.3 (14) 110 (0)
Intensify Historical 700 (0) 280 (0) 0 4600 (0) 614 (28) 8.1 (11) 8.1 (11) 184 (29)
Low 700 (0) 280 (0) 0 4600 (0) 643 (30) 8.0 (13) 8.0 (13) 170 (34)
Medium 700 (0) 280 (0) 0 4600 (0) 673 (31) 7.8 (14) 7.8 (14) 165 (34)
High 700 (0) 280 (0) 0 4600 (0) 759 (30) 7.3 (16) 7.3 (16) 154 (31)
Adapt Historical 561 (2) 224 (2) 3221 (14) 1730 (22) 559 (38) 1.9 (18) 8.1 (11) 184 (29)
Low 557 (2) 223 (2) 3085 (15) 1832 (21) 569 (40) 2.0 (19) 8.0 (13) 170 (34)
Medium 555 (2) 222 (2) 3001 (15) 1891 (20) 604 (42) 2.0 (23) 7.8 (14) 165 (34)
High 553 (2) 221 (2) 2789 (17) 2051 (19) 695 (38) 2.0 (25) 7.3 (16) 154 (31)
Simplify Historical 473 (4) 137 (4) 3705 (10) 1467 (23) 78 (120) 2.3 (9) 7.4 (8) 130 (0)
Low 468 (4) 136 (4) 3668 (12) 1486 (24) 77 (163) 2.4 (14) 7.5 (10) 130 (0)
Medium 468 (4) 136 (4) 3551 (13) 1558 (24) 52 (197) 2.4 (14) 7.4 (11) 130 (0)
High 471 (4) 137 (4) 3359 (16) 1697 (25) 139 (119) 2.4 (16) 7.1 (14) 130 (0)

Baseline Historical 395 (3) 139 (3) 2815 (15) 1163 (29) 244 (61) 0 (0) 9.0 (7) 189 (21)
Low 397 (4) 140 (4) 2554 (14) 1395 (19) 208 (52) 0.9 (26) 9.1 (12) 199 (22)
Medium 395 (4) 139 (4) 2471 (15) 1447 (17) 237 (48) 0.9 (28) 8.8 (13) 190 (23)
High 392 (4) 138 (4) 2256 (16) 1581 (14) 298 (36) 0.9 (22) 8.1 (14) 172 (28)
Intensify Historical 531 (3) 266 (3) 2141 (20) 2053 (14) 990 (19) 0 (0) 9.7 (20) 224 (15)
Low 525 (3) 263 (3) 2117 (18) 2048 (13) 956 (17) 0 (0) 9.6 (18) 216 (15)
Medium 524 (3) 262 (3) 2066 (19) 2079 (13) 975 (17) 0 (0) 9.4 (19) 215 (15)
High 515 (3) 258 (3) 1883 (18) 2179 (10) 1041 (13) 0 (0) 8.6 (18) 203 (18)
Adapt Historical 512 (4) 128 (4) 3350 (9) 1115 (26) 17 (288) 1.1 (52) 8.7 (10) 206 (32)
Low 516 (3) 129 (3) 3313 (9) 1166 (24) 5 (409) 1.5 (27) 9.0 (10) 218 (26)
Medium 518 (3) 130 (3) 3235 (10) 1233 (24) 5 (434) 1.6 (28) 8.9 (10) 229 (25)
High 518 (2) 130 (2) 3023 (13) 1400 (24) 20 (311) 1.6 (29) 8.5 (13) 239 (21)
Simplify Historical 449 (4) 90 (4) 3699 (8) 1271 (20) 29 (202) 1.1 (48) 7.8 (10) 120 (0)
Low 451 (4) 90 (4) 3737 (9) 1250 (19) 2 (557) 1.5 (32) 8.3 (10) 150 (0)
Medium 450 (4) 90 (4) 3646 (9) 1317 (19) 3 (612) 1.6 (32) 8.2 (10) 150 (0)
High 448 (4) 90 (4) 3401 (12) 1483 (18) 16 (301) 1.6 (30) 7.8 (12) 150 (0)

NW Tasmania
Baseline Historical 496 (3) 223 (3) 3572 (2) 790 (5) 302 (18) 1.7 (17) 12.4 (3) 251 (12)
Low 489 (3) 220 (3) 3548 (2) 800 (6) 311 (18) 1.6 (16) 12.3 (3) 245 (12)
Medium 488 (3) 220 (3) 3540 (2) 806 (6) 319 (17) 1.6 (17) 12.2 (3) 241 (13)
High 489 (3) 220 (3) 3511 (2) 823 (7) 337 (18) 1.6 (17) 12.1 (3) 236 (14)
Intensify Historical 540 (2) 324 (2) 3269 (2) 1006 (3) 853 (03) 0.5 (12) 13.6 (2) 267 (6)
Low 531 (2) 319 (2) 3236 (2) 1015 (5) 856 (04) 0.5 (12) 13.5 (2) 265 (7)
Medium 532 (2) 319 (2) 3227 (2) 1026 (5) 864 (04) 0.5 (14) 13.4 (2) 262 (8)
High 531 (2) 319 (2) 3187 (3) 1047 (6) 877 (05) 0.5 (12) 13.6 (2) 257 (9)
Adapt Historical 513 (2) 257 (2) 3908 (3) 658 (13) 293 (19) 0.9 (11) 13.9 (3) 267 (6)
Low 504 (3) 252 (3) 3883 (3) 662 (12) 314 (20) 0.9 (12) 13.8 (3) 266 (7)
Medium 500 (3) 250 (3) 3864 (3) 666 (13) 295 (20) 0.9 (13) 13.8 (3) 263 (8)
High 501 (3) 251 (3) 3837 (3) 691 (14) 314 (20) 0.9 (16) 13.6 (3) 259 (8)
Simplify Historical 442 (4) 155 (4) 3579 (5) 1074 (6) 221 (40) 2.8 (17) 11.2 (6) 150 (7)
Low 436 (3) 153 (3) 3544 (4) 1091 (6) 232 (46) 2.8 (21) 11.1 (7) 146 (7)
Medium 436 (3) 153 (3) 3516 (4) 1111 (7) 243 (44) 2.8 (21) 11.0 (7) 144 (7)
High 435 (3) 152 (3) 3512 (4) 1116 (6) 253 (39) 2.7 (20) 10.9 (7) 143 (7)

3.3. Inuence of climate change on PH, intake per animal, purchased feeds historical climate increased signicantly for all development options
and ROTA of each development option at the Fleurieu Peninsula (Fig. 3a). All climate scenarios had relatively
little inuence on PH at NW Tasmania due to the use of irrigation at this
Relative to the historical climate, the Low change scenario had little site. Variability of milking area PH of the Simplied system at NW
eect on milking area PH of the Intensify and Adapt options, Tasmania was higher than that of all other systems (Table 1, Fig. 3c;
particularly in NW Tasmania. In some years, there were moderate also see discussion).
increases in milking area PH in the Simplify option at the dryland sites
for the Low change scenario, suggesting a benecial eect of low-level 3.3.1. Fleurieu Peninsula
climate change for these case study farms under this development At the Fleurieu Peninsula, average milking area PH of the Intensify
option (Fig. 3a, b). The High change scenario caused the greatest and Adapt options under the High change scenario decreased by 10%
depression in milking area PH regardless of site or development option, relative to the historical climates, compared with only a 4% reduction
with the exception of the Simplied system at Gippsland, which showed for the Simplify system. The reduction in the High-change grazed
a moderate increase in milking area PH (Fig. 3b). The variability in pasture intake of all development options (except the Intensify option)
milking area PH under the High change scenario relative to the relative to historical climates was similar, at 913%. This resulted in the

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Table 4 the Fleurieu Peninsula caused by the High change scenario relative to
Long-term average annual economic results for each case study site and percentage historical costs was small, because the cost associated with higher
coecient of variation in parentheses. Gross income, total costs, operating prot and
purchased feeds was largely balanced by lower N fertiliser costs
return on total assets managed are shown for the historical and 2040 climate scenarios for
each baseline and development option. (average 16% reduction in N applied under climate change of the
Intensify and Adapt options; Tables 3 and 4).
Site and Climate Gross Total Annual Return on For all farm systems at the Fleurieu Peninsula, climate change
development income variable operating total assets reduced and increased the mean and variability of operating prot,
option ($000) and prot manageda
overhead ($000) (%)
respectively. This indicates that for all options we modelled, climate
costs change will tend to make dairy production more dicult, less protable
($000) and more risky at this site. However, some systems were impacted less
than others. The relative changes in operating prot and the absolute
Fleurieu Peninsula (SA)
change in ROTA of the Simplied system were 23% and 0.4%
Baseline Historical 1049 (3) 929 (4) 121 (34) 2.3
Low 1036 (3) 928 (4) 108 (45) 2.0 respectively, and for the Adapt option these changes were even larger
Medium 1037 (3) 933 (4) 104 (49) 1.9 ( 31% and 0.7% in operating prot and ROTA under the High
High 1037 (3) 945 (4) 92 (61) 1.7 change scenario relative to the historical climate). Operating prot and
Intensify Historical 1575 (0) 1393 (2) 182 (13) 3.3
ROTA of the Intensify system changed the least under the 2040 High
Low 1575 (0) 1391 (2) 184 (14) 3.3
Medium 1575 (0) 1392 (2) 182 (15) 3.3 change scenario, declining by only 3% and 0.1% respectively.
High 1575 (0) 1397 (2) 177 (15) 3.2
Adapt Historical 1275 (2) 1144 (4) 130 (32) 2.3 3.3.2. Gippsland
Low 1265 (2) 1147 (4) 118 (40) 2.1 For all farming systems at Gippsland, there was little eect of the
Medium 1262 (2) 1152 (4) 110 (43) 2.0
High change scenario on gross income relative to the historical climate,
High 1258 (2) 1167 (4) 91 (59) 1.6
Simplify Historical 783 (4) 681 (3) 101 (30) 1.9 although gross income and PH of the Intensied system were reduced
Low 775(4) 687 (4) 88 (39) 1.7 by 3% and 11%, respectively (Table 4). Gross income varied little under
Medium 776 (5) 681 (3) 95 (35) 1.8 climate change due to minimal change in milk production, which was
High 779 (5) 700 (4) 78 (52) 1.5
maintained due to higher use of supplementary feed. In contrast, PH of
Gippsland (Vic) the Adapt system at Gippsland was little inuenced by climate change,
Baseline Historical 760 (3) 632 (7) 128 (37) 2.9 and the Simplify system aorded a moderate gain in this variable,
Low 763 (4) 642 (4) 121 (36) 2.7
Medium 760 (4) 647 (4) 113 (39) 2.5
particularly under Low impact climate scenarios (Fig. 3, Table 3).
High 755 (4) 661 (4) 94 (45) 2.1 Feeding of hay/silage increased signicantly for all development
Intensify Historical 1449 (2) 1316 (4) 132 (62) 2.6 options under the High change scenario (6%, 17% and 26% increase for
Low 1433 (3) 1311 (3) 121 (59) 2.4 the Intensify, Simplify and Adapt options, respectively, relative to
Medium 1429 (3) 1317 (3) 112 (68) 2.2
historical climates; Table 3), as did purchased fodder in the Adapt
High 1407 (3) 1332 (3) 75 (95) 1.5
Adapt Historical 709 (5) 580 (4) 128 (26) 3.0 system. The purchased feed requirement for the Simplied option was
Low 715 (5) 586 (3) 129 (25) 3.0 reduced due to higher pasture utilisation. The amount of silage cut for
Medium 716 (5) 589 (3) 127 (26) 2.9 the Adapt and Simplify systems under the High change scenario
High 709 (5) 596 (3) 112 (33) 2.6 increased in line with greater feeding of hay/silage (45% increase for
Simplify Historical 490 (5) 383 (3) 108 (24) 2.5
Low 502 (6) 388 (2) 114 (23) 2.7
both systems, as opposed to the Intensied system at this site, which
Medium 500 (6) 389 (2) 110 (25) 2.6 had no cutting). Purchasing more fodder increased total costs by 1% for
High 492 (6) 394 (3) 98 (32) 2.3 the Intensify system and 3% for the Adapt system, though higher costs
North-western Tasmania of feeding associated with climate change in these systems were
Baseline Historical 1259 (2) 991 (1) 269 (10) 4.6 partially mitigated by lower nitrogen fertiliser costs. On the other
Low 1243 (2) 992 (1) 251 (11) 4.2 hand, the costs associated with purchased feed and N fertiliser of the
Medium 1242 (3) 992 (1) 249 (12) 4.2 Simplied system were reduced and increased, respectively, due to
High 1244 (3) 995 (1) 249 (11) 4.2
Intensify Historical 1821 (2) 1527 (0) 294 (9) 4.5
higher dependence of this system on home-grown feed.
Low 1794 (1) 1528 (0) 266 (9) 4.1 Relative to the historical climate, lower gross income and higher
Medium 1795 (2) 1530 (0) 266 (12) 4.1 total costs under the High change scenario resulted in large reductions
High 1792 (2) 1532 (0) 260 (11) 4.0 in operating prot and ROTA for all farm systems at Gippsland. Average
Adapt Historical 1447 (2) 1067 (1) 379 (8) 6.2
operating prots (absolute ROTA) were reduced by 9% (0.2%), 13%
Low 1413 (3) 1072 (1) 342 (10) 5.6
Medium 1411 (2) 1068 (1) 343 (8) 5.6 (0.4%) and 44% (1.1%) for the Simplify, Adapt and Intensify systems,
High 1413 (3) 1072 (1) 342 (10) 5.6 respectively. These results indicate that (if ROTA were used to contrast
Simplify Historical 878 (3) 713 (2) 165 (19) 2.9 the inuence of climate change), Intensied systems at Gippsland may
Low 866 (3) 715 (2) 151 (20) 2.6 fare worse than farming systems that have less investment in infra-
Medium 866 (3) 717 (2) 149 (17) 2.6
High 864 (3) 718 (2) 146 (21) 2.5
structure and lower dependence on purchased feeds.

Coecient of variation values for ROTA equal those of operating prot. 3.3.3. NW Tasmania
Under the High change scenario, the change in gross income of each
respective need to increase feeding of hay/silage and purchased feed by development option relative to the historical climate of the Intensify
19% and 24% for the Adapt system, and by 16% and 77% for the and Simplify options was similar, with reductions of 1.6% and
Simplied system. The Intensify option at the Fleurieu Peninsula was 1.5%, respectively, whereas the drop in gross income of the Adapt
based on a TMR system and hence did not include directly grazed system was somewhat larger ( 2.3%).
pasture, however pasture was harvested indirectly through feeding of Although grazed pasture per animal was reduced by ~3% for all
hay/silage. The purchased fodder requirement for this option increased systems under the High change scenario (Table 3), the distribution of
by 24% due to lower pasture growth under climate change (cutting milking area PH of all options was only slightly reduced, because
yields decreased by 10% under the High change scenario). The change irrigation of pastures largely buered the inuence of low production in
in costs associated with climate change of all development options at drought years (Fig. 3c). The increase in hay/silage feeding of all
systems under the High change scenario was also small (45% relative

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

Fig. 4. Return on total assets (ROTA) for each development option at (a) the Fleurieu Peninsula (SA), (b) Gippsland (Vic) and (c) North-western Tasmania under historical ( ), and
2040 Low ( ), Medium ( ) and High ( ) change scenarios. Boxplots show the median, 25th and 75th percentiles in the box, with the 10th and 90th percentiles in the
whiskers and dots showing outliers.

to the historical climate; Table 3). However, the purchased feed study farms. The rst aim was to examine the eect of climate change
requirement in the baseline and Simplied systems under the High on seasonal pasture growth rates, PH and prot of baseline farms in
change scenario increased somewhat more than other options (12% and three regions. The second aim was to examine the inuence of climate
14% respectively, relative to the historical climate), since one-third of change on milking area PH and prot if development options that either
the farm area in these systems was rain-fed, and since these systems had intensied, simplied or adapted were applied to the baseline farms.
greater reliance on home-grown feed compared with the other systems.
The corollary of higher purchased feed requirements in the Simplied 4.1. Inuence of climate change on seasonal growth rates, pasture grazed
system was higher costs and a more signicant eect of climate change and prot of the baseline farms
on protability (Table 4). Relative to historical climates, costs of the
Simplied option increased by 0.7%. Across options, climate change Consistent with work by Cullen et al. (2009) for mainland Australia,
reduced absolute ROTA relative to the historical climates by 0.40.6%. our study predicted increased winter and early spring pasture growth
rates, but this was counteracted by both later autumn breaks and
4. Discussion shorter spring growing seasons, regardless of climate scenario (Fig. 3).
However, Cullen et al. (2008) also predicted that annual pasture
In this study we addressed two aims through analyses of three case production in Victoria (Ellinbank) and Tasmania (Elliot) in 2030 would

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

be 410% greater than that produced under their 19712000 baseline, as the emissions scenario moved from Low to High, indicating that
in contrast to the lower total annual pasture production simulated here climate change will not only make overall protability of dairy systems
(relative to our historical baseline of 19752013, annual yields in 2040 lower, but also more variable from year to year (Table 4). These results
under the High change scenario at Gippsland and NW Tasmania were represent applied examples supporting the observation by Nardone
decreased by 10% and 3%, respectively). Phelan et al. (2015) also et al. (2010) that increased future climatic variability will exert a strong
forecast a benecial inuence of climate change on pasture production inuence on livestock systems, even though they have developed the
in Tasmania, intimating that mean annual pasture growth would capability to cope and adapt to existing climate uncertainty. However,
increase by 16% in 2050. for conditions that deviate many degrees from a coping range,
The dierence between results of the current study and those above livestock systems will become vulnerable if there is no adaptive
may be due to several reasons. The most likely explanation is the capacity. Increased climate variability that increases the inter-annual
dierence in climate data. A key novelty of the present study was that and seasonal variation of forage availability will contribute to reduce
we used climate projections that incorporated the higher likelihood of the overall sustainability of each farm system, both from socialeco-
more extreme climatic events (ECEs) expected to occur with climate nomic and ecological perspectives (Nardone et al., 2010), particularly
change (Seneviratne et al., 2012). In contrast, past studies such as those for farms that are totally reliant on rainfall for pasture production (e.g.
above and others (Harrison et al., 2014b; Jones and Thornton, 2003; Fig. 2a, b).
Parry et al., 2005) have typically modelled crop and livestock impacts
of gradual or average climate change, as derived from one or more GCM 4.2. How did climate change impact milking area PH and prot for farm
or emissions scenario (Soussana et al., 2010). Indeed the focus of the systems that intensied, simplied or adapted existing management
great majority of previous climate change impact studies has been on practices?
changes in mean climate, since these changes are more robust to predict
than changes in climate variability (see review by Thornton et al., 2014, Our results demonstrate that no development option was univer-
and references therein). Climate data produced in this way are sally better across regions at abating the inuence of climate change
generally xed according to the method chosen and are not subject to on PH or protability. Nor did any development option result in
modication to account for regional dierences in projected ECEs due universally lower long-term variability in PH or prot, although the
to local geography or climatic variability at a given location (Harrison Simplify option had a tendency for lower median impact and variability
et al., 2016). However, by concentrating on changes in climate means, on the dryland farms. For instance, although average PH of the Intensify
the full impacts of climate change on biological and human systems are systems at Gippsland and the Fleurieu Peninsula was reduced by ~10%
probably being seriously underestimated (Thornton et al., 2014). The under the High change scenario relative to the historical climate, the
climate scenarios used in the present study were developed by increas- impact of climate change on PH of the Intensify systems in Tasmania
ing the frequencies of extreme rainfall events, dry-day durations was relatively negligible (Fig. 3). In the same vein, PH in the Simplify
(analogous to drought) and heat waves by altering the regularity of system at Gippsland was positively benetted under all climate
climatic events falling in the tails of the long-term historical distribu- scenarios, whereas PH at the other sites in the Simplify system under
tion observed at each site (we used the Variable method as described the High change scenario dropped by 4% on average, relative to
in Harrison et al., 2016). Accounting for increased frequencies of ECEs historical climates. Further, whilst the High change scenario signi-
was likely a major reason for the result that climate change had a cantly increased the variability of PH of all systems at the Fleurieu
detrimental impact on annual pasture growth, as opposed to the Peninsula, this scenario had relatively little inuence on the variance in
positive impact found in previous studies above. Indeed, a fundamental PH at the other sites. These results are not surprising since pasture
conclusion in the paper by Harrison et al. (2016) was that biophysical management of each adaptation diered across sites and none of the
modelling that explicitly accounted for the increased climate variability adaptations explicitly changed pasture management in isolation.
associated with more frequent ECEs had impacts on livestock systems We found that climate change had a detrimental inuence on
over and above those of gradual climate change. The dierences protability under the High impact climate scenario, regardless of
between the historical and 2040 climates, including information on development option (Fig. 4). This indicates that no farming systems
extremes, are shown in Supporting information 1. adaptation examined in the present study was capable of fully alleviat-
Our results demonstrate that climate change increased year to year ing the negative impact of climate change on prot. However, some
variability in grazed pasture consumption and farm protability, in line systems were less impacted than others, as shown by comparing the
with recent observations that inclusion of more frequent ECEs in climate-induced reduction in ROTA for each development option
climate change analyses increases inter-annual standard deviation of relative to the same reduction under climate change for the baseline
biophysical and economic metrics (Harrison et al., 2016; Nardone et al., farm at each site. For the Fleurieu Peninsula, climate change reduced
2010). Lower annual pasture production of baseline farms with climate ROTA of the baseline farm by 0.53%. At this site the Intensify option
change was reected in lower grazed pasture per animal and per area was the least impacted under the High impact scenario ( 0.08%
(Table 3 and Fig. 3), notably for the baseline farm at Gippsland, change in ROTA), whereas ROTA of the Adapt option decreased by
although the High change scenario caused large increases in the 0.77%, suggesting that the Intensify option at the Fleurieu Peninsula
variability of pasture utilised at the Fleurieu Peninsula (Fig. 3a). The may be less impacted and more resilient to climate change. For
impact of climate change on PH of the base farm in Tasmania was lower Gippsland, the Simplify option was least impacted by climate change
than that of other sites since the Tasmanian site was irrigated. Eects of (0.22% reduction), and the Intensify option the most (1.13% reduc-
climate change on PH of the baseline farms were generally translated tion). In Tasmania, all development options had similar reductions in
into similar trends in ROTA1 (cf. Figs. 3 and 4); for example, the largest ROTA under climate change ( 0.37% to 0.62%), bracketing the
reduction in mean ROTA occurred at Gippsland as climate change climate-induced reduction of the baseline farm ( 0.39%).
severity increased. For all baseline farms, a consistent inuence of Together these conclusions indicate that studies designed to in-
climate change was lower gross income, higher costs and lower vestigate alternative farm systems as adaptations to climate change
operating prot. As well, the inter-annual ROTA variation increased should be regionally adapted, ground-truthed with local experts, and
devised using participatory-action research (PAR) with case study
farmers. Although PAR has been used to study climate change adapta-
Here we used constant milk prices and feed costs for all seasons analysed, similar to
tion in the cropping, livestock and natural resource management
past work designed to isolate the inuence of climate on dairy farming systems and the sectors (Bridle and Price, 2009; Mapfumo et al., 2013), little work
implications of dierent adaptations (e.g. see Kalaugher et al. 2013). using PAR has been conducted to study the inuence of longer term

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

adaptations to climate change for the Australian dairy industry. the gross value of dairy production was observed (Leblanc et al., 2012).
However, research of dairy farm adaptation to climate change has By the close of the century, increased drought frequency, lower rainfall
been conducted in other countries. Kalaugher et al. (2017) analysed six and nearly complete absence of runo is likely to have terminated
pasture-based dairy farms in New Zealand and suggested that without irrigated agriculture in this region (Garnaut, 2008).
adaptation, a negative impact of climate change would likely occur in
all regions. However, the level of impact was largely dependent on local 4.3. Impacts of climate change relative to dierences across development
climate variability as well as on the management practices of each farm. options
These results agree with results found in the present paper, indicating
that widespread studies applying a generic management scenario and/ Although the development options analysed in the present paper
or farm details to several regions should be treated with caution. represented some of the variability possible in dairy farm management
Similar to the work presented by Kalaugher et al. (2017) and Dynes and infrastructure (e.g. we included simultaneous changes to stocking
et al. (2010), we found that some adaptations had the potential to rates, herd sizes, calving times, total assets managed etc.), there are
reduce climate change impact compared with the baseline. In Ireland, several other factors that we did not consider, such as the dynamic
Fitzgerald et al. (2009) found that (1) grass-based dairy production was variation in milk price, feed costs or management over time (e.g.
able to adapt to climate change, (2) that there were dierences between climate resilience; Kalaugher et al., 2013). Thus, we could have
farms on poorly and well-drained soils, and (3) that there would be investigated several other development options, but we considered
more feed in spring and autumn. In East Africa, Claessens et al. (2012) those simulated here as a foremost priority, since they were selected by
showed that mean agricultural income of smaller-holder dairy farmers expert regional working groups. Some of the other development options
could be increased by 331% by planting sweet potato or an improved that may warrant investigation in future work are suggested in the
maize variety, depending on location and species. Future Ready Dairy Systems Climate Toolkit produced by Dairy
Although the negative inuence of climate change on dairy farms Australia (see
was a common result to all systems we examined, causes of this result australia/adapting-western-australian-dairy-farms-to-climate-change/
diered across sites and development options. Climate change reduced [Veried 13 February 2017]). In response to higher winter growth
the duration of the growing season at all sites and required a greater rates, pastures could be sown earlier, annual pasture species with
proportion of fodder to be ensiled as opposed to directly grazed, which greater winter productivity could be introduced into existing perennial
increased costs associated with fodder conservation. For the dryland pastures, and additional fertiliser could be applied in winter/early
sites, costs were often further increased under future scenarios because spring so plants could better utilise the spring ush. In response to
shorter growing seasons signicantly reduced PH, increasing the earlier termination of spring growth with climate change, farmers could
requirement to purchase supplementary feed (Tables 3 and 4). For cut silage earlier, sow (and allow grazing of) summer forage crops, or
the irrigated site, water availability during summer increased total introduce more drought tolerant, deep-rooted species suited to hotter,
pasture production so climate change had less eect on PH compared drier spring and summer conditions (e.g. C4 species such as kikuyu and
with dryland farms (Figs. 2 and 3), but the eect of climate change on maize). As global warming will increase vapour pressure decit and
milk production was still present due to a reduction in pasture intake evapotranspiration rates, farmers with irrigation infrastructure may
and quality. Averaged over each 38-year simulation, pasture intake and have increasing dependence on reliable water supply. Any or all of
metabolisable energy intake (MEI) under the historical climate and improvements in irrigation water-use eciency, on-farm water harvest
2040 High scenario were respectively reduced from 9.77 to 9.61 kg methods and the spreading of dairy-shed euent may help overcome
pasture DM/animalday, and from 166.5 to 165.2 MJ/animalday. lower water allocations.
Together, lower pasture intake and quality reduced daily lactation
from 1.350 to 1.339 kg milk solids/animalday. When cumulated over 4.4. Managing for climate change versus climate variability implications
the simulation, these subtle dierences led to larger dierences in for the dairy industry
average annual farm milk production between the historical and 2040
High scenario (e.g. for the irrigated baseline farm this was 1.4%); This paper has revealed that the impact of climate variability was
although these changes were lower than the average annual variability much greater than that of climate change for all development options,
of 3.0%. For both dryland and irrigated dairy farms, the inuence of climate change scenarios and sites. This result can be seen in Figs. 3 and
either lower PH or lower pasture quality reduced milk production and 4 where the variability of pasture harvested and ROTA within each
thus gross income. Together with higher costs, lower income reduced development option is greater than the reduction within development
operating prot and ROTA; this result was consistent across climate options across climate scenarios (even before accounting for extreme
scenarios, sites and development options. years shown by outliers). These trends are similarly reected in pasture
In implementing adaptations, the inuence of climate change on and hay/silage intake per animal, fodder purchased, cut yields and
irrigation requirements also should be considered, as more water will nitrogen fertiliser, as shown by comparing the CV values within
be required per unit area under drier conditions, and peak irrigation development options and across climate scenarios versus the CV values
demands will rise due to more severe heat waves (Parry, 2000). In the across options at a given site. The variability of any metric simulated
Tasmanian case study, we assumed that water supply was non-limited would depend on the number of years used to conduct the modelling,
(and would be non-limiting in 2040), in line with the large catchment and since variability generally increases as more years are included in
area and storage volume available for irrigation on this farm. Compared the analysis, this indicates that including more years (rather than less)
with the historical climates, the high change scenario increased is a necessary protocol to properly account for climate variability at a
irrigation requirements by 56% per year. While there is potential for given site (Harle et al., 2007).
the farm in NW Tasmania to expand irrigated area as an adaptation to Altogether the complexity of livestock farming systems dictates that
climate change, this option does not exist for the majority of the the implications of climate change on individual dairy farms may be
Australian dairy industry. For example, major irrigated dairy regions in dicult to predict, given there will be several (possibly interacting)
the southern Murray-Darling Basin are very likely to have reduced impacts. Higher temperatures increase heat stress on cows, which is
volumes of water for irrigation due to projections for higher tempera- likely to reduce dry matter intake, gross milk yield and eciency of
tures, lower rainfall and less run-o in future, and greater allocations of milk yield (West, 2003). Higher temperatures may encourage growth of
water to achieve environmental objectives (Leblanc et al., 2012). summer grasses that occur in tropical northern Australia (e.g. kikuyu;
Indeed, during the Millennium drought in the Murray-Darling Basin Bell et al., 2013), which would result in lower pasture digestibility and
region from the mid-1990s to 2009 a reduction in dairy water use and metabolisable energy. Increased atmospheric CO2 concentration can

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

increase plant water-use eciency and contribute to additional growth options, nor did any option consistently increase or decrease pasture
(Harrison et al., 2014b; Harrison et al., 2010), but if more nitrogen harvested across sites. However, some options appeared more prefer-
fertiliser is not applied, leaf nitrogen content is likely to be diluted by able than others; for example, at dryland sites, median pasture
additional non-structural carbohydrate (Harrison et al., 2009), leading harvested of simplied farming systems was less impacted by climate
to lower forage quality (Bell et al., 2013; Howden et al., 2008). Lower change than that of other options. Similarly on farms with capability for
rainfall may lead to lower irrigation allocations, since the majority of irrigation, increasing irrigated area and minimal additional investment
irrigation water in Australian dairy is derived from self-extracted in other infrastructure consistently resulted in high pasture harvested
sources and irrigation service providers (DA, 2016b). Indeed, the and prot over the long-term, as well as low variability. Future work on
combination of post-deregulation restructuring and the impacts of a adapting dairy businesses to climate change should aim to optimise
1-in-100 year drought has already had signicant impacts on both production and prot by examining development options that include
milk production and the condence of farmers in past events (Khan raising water-use eciency, helping maintain growing season length
et al., 2010). Since the combined inuence of the above factors on and/or take advantage of additional growth expected in winter.
individual farms will depend on current farming systems as well as
environmental factors, the Australian dairy industry advises that farm- Acknowledgements
ers should manage for climate variability (e.g. week-to-week and year-
to-year) rather than for climate change (e.g. long-term change expected This project was supported by The University of Melbourne and The
in pasture growth over the next 40 years) (DA, 2016a). Our results Tasmanian Institute of Agriculture through funding from the Australian
unequivocally reinforce this advice, showing much greater inter-annual Government Department of Agriculture and Dairy Australia. We thank
variability compared with median change expected over the next the Dairy Businesses for Future Climates (R0021803) team members
40 years. Nevertheless, our results also demonstrate that some con- (Gillian Hayman, Warren Mason, Margaret Ayre, Nicole Reichelt, Cathy
sideration should be given to climate change when planning longer- Phelps, Monique White, Ruth Nettle, Richard Rawnsley, Rachel Brown
term strategic investment. and Ruth Beilin) for their work in the project. We acknowledge Natalie
Further research should consider the regional-level cost and envir- Doran-Browne for support in conducting DairyMod simulations. We
onmental inuences if one adaptation becomes the preferred option of thank the case study farmers and the regional working group members
several farms within a given location. A cluster of farms following an for their contribution to the project.
intensication trend (e.g. as for dairy farming in New Zealand; Foote
and Joy, 2014) would cause greater demands for purchased feeds, Appendix A. Supplementary data
livestock water (and if available) irrigation water, which at the regional
level would be expected to elevate the costs of feed and/or water. At the Supplementary data to this article can be found online at http://dx.
same time, an oversupply of milk could deate milk price. Regional-
level intensication could also result in environmental impacts such as
erosion and salinization (Kirchmann and Thorvaldsson, 2000), as well References
as nitrogen pollution via groundwater leaching and nitrous oxide
emissions resulting from more nitrogenous fertilisers and livestock Allen, V.G., Batello, C., Berretta, E.J., Hodgson, J., Kothmann, M., Li, X., McIvor, J.,
excrement. Such cost and environmental inuences of agricultural Milne, J., Morris, C., Peeters, A., Sanderson, M., The, F., Grazing Terminology, C.,
2011. An international terminology for grazing lands and grazing animals. Grass
adaption at the regional level should be carefully considered when Forage Sci. 66, 228.
forming future political and legislative guidelines. Armstrong, D., Ho, C., Doyle, P., Malcolm, B., Gibb, I., Brown, S., 2005. Analysing
drought impacts and recovery options by adapting a dairy farming systems modelling
approach. Aust. Farm Bus. J. 2, 1116.
5. Conclusions Bailey, R., Froggatt, A., Wellesley, L., 2014. Livestock Climate change's Forgotten Sector
Global Public Opinion on Meat and Dairy Consumption. Chatham House, The Royal
We have shown that climate change impacts on pasture growth Institute of International Aairs Available:.
rates and dairy farm nancial performance may be highly regionally 20141203LivestockClimateChangeBaileyFroggattWellesley.pdf (Accessed 7
specic. An increase in extreme weather events coupled with gradual September 2016).
climate change will increase water shortages and regional irrigation Bell, M.J., Eckard, R.J., Harrison, M.T., Neal, J.S., Cullen, B.R., 2013. Eect of warming
on the productivity of perennial ryegrass and kikuyu pastures in south-eastern
demand with high regional variability (Dll and Siebert, 2002). Unlike
Australia. Crop Pasture Sci. 64, 6170.
past work, the present paper made explicit treatment of increased Bridle, K.L., Price, R.J., 2009. Undertaking participatory research at a national scale: the
frequencies of extreme climatic events in studying climate change biodiversity in grain graze approach. Anim. Prod. Sci. 49, 916927.
eects on agricultural production systems (e.g. see Harrison et al., Bryant, J.R., Snow, V., 2008. Modelling pastoral farm agro-ecosystems: a review. N. Z. J.
Agric. Res. 51, 349363.
2016), from climate to milk production to protability. Including Chapman, D., Beca, D., Hill, J., Tharmaraj, J., Jacobs, J., Cullen, B., 2014a. Increasing
extreme events may be one reason why our study found a reduction home-grown forage consumption and prot in non-irrigated dairy systems. 4.
in dairy productivity and protability under high-level climate change, Economic performance. Anim. Prod. Sci. 54, 256262.
Chapman, D.F., Cullen, B.R., Johnson, I.R., Beca, D., 2009. Interannual variation in
regardless of location. pasture growth rate in Australian and New Zealand dairy regions and its
We found that climate change to 2040 will reduce pasture harvest consequences for system management. Anim. Prod. Sci. 49, 10711079.
yields and cause greater yield variability in late spring and summer. For Chapman, D.F., Hill, J., Tharmaraj, J., Beca, D., Kenny, S.N., Jacobs, J.L., 2014b.
Increasing home-grown forage consumption and prot in non-irrigated dairy systems.
the case study farms analysed, the High climate change scenario will 1. Rationale, systems design and management. Anim. Prod. Sci. 54, 221233.
reduce pasture growing season durations. While winter pasture growth Claessens, L., Antle, J.M., Stoorvogel, J.J., Valdivia, R.O., Thornton, P.K., Herrero, M.,
rates increased relative to the historical climate, total annual pasture 2012. A method for evaluating climate change adaptation strategies for small-scale
farmers using survey, experimental and modeled data. Agric. Syst. 111, 8595.
production and consumption were reduced and year-to-year variability CSIRO, 2015. Climate change in Australia. In: Information for Australia's Natural
increased. Further, greater expenditure on supplementary feed and Resource Management Regions: Technical Report. CSIRO and Bureau of Meteorology
fodder conservation was required to maintain milk production, which Australia.
Cullen, B., Rawnsley, R., Eckard, R., Christie, K., Bell, M., 2014. Use of modelling to
in turn led to reduced protability and increased risk for dairy farms.
identify perennial ryegrass plant traits for future warmer and drier climates. Crop
Climate-induced reductions in pasture harvested or prot may Pasture Sci. 65, 758766.
potentially be reversed by implementing certain farm adaptations, Cullen, B.R., Eckard, R.J., Callow, M.N., Johnson, I.R., Chapman, D.F., Rawnsley, R.P.,
assuming such adaptations are practicable. None of the development Garcia, S.C., White, T., Snow, V.O., 2008. Simulating pasture growth rates in
Australian and New Zealand grazing systems. Aust. J. Agric. Res. 59, 761768.
options analysed here was universally more eective in reducing Cullen, B.R., Johnson, I.R., Eckard, R.J., Lodge, G.M., Walker, R.G., Rawnsley, R.P.,
pasture losses to climate change across regions and development

M.T. Harrison et al. Agricultural Systems 155 (2017) 1932

McCaskill, M.R., 2009. Climate change eects on pasture systems in south-eastern Johnson, I.R., France, J., Cullen, B.R., 2016. A model of milk production in lactating dairy
Australia. Crop Pasture Sci. 60, 933942. cows in relation to energy and nitrogen dynamics. J. Dairy Sci. 99 (2), 16051618
DA, 2015. Australian Dairy Industry in Focus 2015. Dairy Australia, Melbourne Available February.
at:. Jones, P.G., Thornton, P.K., 2003. The potential impacts of climate change on maize
%20markets/Australian%20Dairy%20Industry%20In%20Focus/Australian%20Dairy production in Africa and Latin America in 2055. Glob. Environ. Chang. 13, 5159.
%20Industry%20In%20Focus%202015.pdf (Accessed 25 February 2016). Kalaugher, E., Beukes, P., Bornman, J.F., Clark, A., Campbell, D.I., 2017. Modelling farm-
DA, 2016a. Climate Variability. Dairy Australia Limited Available:. http://www. level adaptation of temperate, pasture-based dairy farms to climate change. Agric. Syst. 153, 5368.
Climate-impacts-and-responses/Adapting-to-climate-change/Climate-variability.aspx Kalaugher, E., Bornman, J.F., Clark, A., Beukes, P., 2013. An integrated biophysical and
(Accessed 17 April 2016). socio-economic framework for analysis of climate change adaptation strategies: the
DA, 2016b. Dairy Australia (2016). The Australian Dairy Industry Available:. http:// case of a New Zealand dairy farming system. Environ. Model. Softw. 39, 176187. Khan, S., Abbas, A., Rana, T., Carroll, J., 2010. Dairy Water use in Australian Dairy Farms:
(Accessed 21 May 2016). Past Trends and Future Prospects. CSIRO: Water for a Healthy Country National
Dll, P., Siebert, S., 2002. Global modeling of irrigation water requirements. Water Research Flagship.
Resour. Res. 38 (8-1-8-10). Kirchmann, H., Thorvaldsson, G., 2000. Challenging targets for future agriculture. Eur. J.
Dynes, R., Payne, T., Brown, H., Bryant, J., Newton, P., Snow, V., Lieering, M., Wilson, Agron. 12, 145161.
D., Beets, P., 2010. New Zealand's land-based primary industries & climate change: Klinedinst, P., Wilhite, D., Hahn, G.L., Hubbard, K., 1993. The potential eects of climate
assessing adaptation through scenario-based modelling. In: Nottage, R.A.C. (Ed.), change on summer season dairy cattle milk production and reproduction. Clim.
Climate Change Adaptation in New Zealand: Future Scenarios and Some Sectoral Chang. 23, 2136.
Perspectives. NZ Climate Change Centre, Wellington, NZ, pp. 4455. Leblanc, M., Tweed, S., Van Dijk, A., Timbal, B., 2012. A review of historic and future
FAO, 2012. World Agriculture Towards 2030/2050: The 2012 Revision, ESA Working hydrological changes in the Murray-Darling Basin. Glob. Planet. Chang. 8081,
Paper No. 12-03. FAO, Rome. 226246.
Fitzgerald, J.B., Brereton, A.J., Holden, N.M., 2009. Assessment of the adaptation Malcolm, B., Makeham, J., Wright, V., 2005. The Farming Game Agricultural
potential of grass-based dairy systems to climate change in Ireland-the maximised Management and Marketing, second ed. Cambridge University Press, Melbourne.
production scenario. Agric. For. Meteorol. 149, 244255. Mapfumo, P., Adjei-Nsiah, S., Mtambanengwe, F., Chikowo, R., Giller, K.E., 2013.
Foote, K., Joy, M., 2014. The true cost of milk: environmental deterioration vs. prot in Participatory action research (PAR) as an entry point for supporting climate change
the New Zealand dairy industry. In: Paper Presented at the 2014 NZARES Conference, adaptation by smallholder farmers in Africa. Environ. Dev. 5, 622.
Nelson, New Zealnd, August 2829, Available:. Moore, A.D., Ghahramani, A., 2013. Climate change and broadacre livestock production
bitstream/187496/2/FooteJoy_2014.pdf (Accessed 15 February 2017). across southern Australia. 1. Impacts of climate change on pasture and livestock
Garnaut, R., 2008. Chapter 6 Climate change impacts on Australia. In: The Garnaut productivity, and on sustainable levels of protability. Glob. Chang. Biol. 19,
Climate Change Review. Cambridge University Press (ISBN 978-0-521-74444-7. 14401455.
Retrieved 13-02-2017. Available). Nardone, A., Ronchi, B., Lacetera, N., Ranieri, M.S., Bernabucci, U., 2010. Eects of
Harle, K.J., Howden, S.M., Hunt, L.P., Dunlop, M., 2007. The potential impact of climate climate changes on animal production and sustainability of livestock systems. Livest.
change on the Australian wool industry by 2030. Agric. Syst. 93, 6189. Sci. 130, 5769.
Harrison, M.T., Christie, K.M., Rawnsley, R.P., Eckard, R.J., 2014a. Pasture management Nesamvuni, E., Lekalakala, R., Norris, D., Ngambi, J.W., 2012. Eects of climate change
and livestock genotype interventions to improve whole farm productivity and reduce on dairy cattle, South Africa. Afr. J. Agric. Res. 7, 38673872.
greenhouse gas emissions intensities. Anim. Prod. Sci. 54, 20182088. Parry, M., Rosenzweig, C., Livermore, M., 2005. Climate change, global food supply and
Harrison, M.T., Cullen, B.R., Rawnsley, R.P., 2016. Modelling the sensitivity of risk of hunger. Philos. Trans. R. Soc. Lond. B Biol. Sci. 360, 21252138.
agricultural systems to climate change and extreme climatic events. Agric. Syst. 148, Parry, M.L., 2000. Assessment of potential eects and adaptations for climate change in
135148. Europe. In: The Europe ACACIA Project. Jackson Environment Institute, University of
Harrison, M.T., Edwards, E.J., Farquhar, G.D., Nicotra, A.B., Evans, J.R., 2009. Nitrogen East Anglia, Norwich, UK.
in cell walls of sclerophyllous leaves accounts for little of the variation in Phelan, D.C., Harrison, M.T., Kemmerer, E.P., Parsons, D., 2015. Management
photosynthetic nitrogen-use eciency. Plant Cell Environ. 32, 259270. opportunities for boosting productivity of cool-temperate dairy farms under climate
Harrison, M.T., Kelman, W.M., Moore AD Evans, J.R., 2010. Grazing winter wheat change. Agric. Syst. 138, 4654.
relieves plant water stress and transiently enhances photosynthesis. Funct. Plant Biol. Seneviratne, S.I., Nicholls, N., Easterling, D., Goodess, C.M., Kanae, S., Kossin, J., Luo, Y.,
37, 726736. Marengo, J., McInnes, K., Rahimi, M., Reichstein, M., Sorteberg, A., Vera, C., Zhang,
Harrison, M.T., Tardieu, F., Dong, Z., Messina, C.D., Hammer, G.L., 2014b. Characterizing X., 2012. Changes in climate extremes and their impacts on the natural physical
drought stress and trait inuence on maize yield under current and future conditions. environment. In: Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J., Ebi, K.L.,
Glob. Chang. Biol. 20, 867878. Mastrandrea, M.D., Mach, K.J., Plattner, G.K., Allen, S.K., Tignor, M., Midgley, P.M.
Ho, C., Newman, M., Dalley, D., Little, S., Wales, W., 2013. Performance, return and risk (Eds.), Managing the Risks of Extreme Events and Disasters to Advance Climate
of dierent dairy systems in Australian and New Zealand. Anim. Prod. Sci. 53, Change Adaptation: A Special Report of Working Groups I and II of the
894906. Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press,
Howden, S.M., Crimp, S.J., Stokes, C.J., 2008. Climate change and Australian livestock Cambridge, UK, and New York, NY, USA, pp. 109230. Available. http://ipcc-wg2.
systems: impacts, research and policy issues. Aust. J. Exp. Agric. 48, 780788. gov/SREX/images/uploads/SREX-Chap3_FINAL.pdf.
IPCC, 2012. Managing the risks of extreme events and disasters to advance climate Soussana, J.-F., Graux, A.-I., Tubiello, F.N., 2010. Improving the use of modelling for
change adaptation. In: Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J., Ebi, projections of climate change impacts on crops and pastures. J. Exp. Bot. 61,
K.L., Mastrandrea, M.D., Mach, K.J., Plattner, G.-K., Allen, S.K., Tignor, M., Midgley, 22172228.
P.M. (Eds.), A Special Report of Working Groups I and II of the Intergovernmental Thornton, P.K., Ericksen, P.J., Herrero, M., Challinor, A.J., 2014. Climate variability and
Panel on Climate Change. Cambridge University Press, Cambridge, UK, and New vulnerability to climate change: a review. Glob. Chang. Biol. 20, 33133328.
York, NY, USA, pp. 582. Watterson, I., Hirst, A., Rotstayn, L., 2013. A skill score based evaluation of simulated
Isbell, R.F., 2002. The Australian Soil Classication, Revised, rst ed. CSIRO Publishing, Australian climate. Aust. Meteorol. Oceanogr. J. 63, 181190.
Melbourne. West, J.W., 2003. Eects of heat-stress on production in dairy cattle. J. Dairy Sci. 86,
Johnson, I.R., 2016. DairyMod and the SGS Pasture Model: A Mathematical Description of 21312144.
the Biophysical Model Structure. IMJ Consultants, Dorrigo, NSW, Australia. White, C.J., Grose, M.R., Corney, S.P., Bennett, J.C., Holz, G.K., Sanabria, L.A., McInnes,
Johnson, I.R., Chapman, D.F., Snow, V.O., Eckard, R.J., Parsons, A.J., Lambert, M.G., K.L., Cechet, R.P., Gaynor, S.M., Bindo, N.L., 2010. Climate Futures for Tasmania:
Cullen, B.R., 2008. DairyMod and EcoMod: biophysical pasture-simulation models for Extreme Events Technical Report. Antarctic Climate and Ecosystems Cooperative
Australia and new Zealand. Aust. J. Exp. Agric. 48, 621631. Research Centre, Hobart, Tasmania.