Opportunistic Mediterranean agriculture – Using ephemeral pasture legumes to utilize summer rainfall

Agricultural Systems 120 (2013) 76–84

Contents lists available at SciVerse ScienceDirect

Agricultural Systems

journal homepage: www.elsevier .com/locate /agsy

Opportunistic Mediterranean agriculture – Using ephemeral pasturelegumes to utilize summer rainfall

0308-521X/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.agsy.2013.06.001

⇑ Corresponding author. Tel.: +61 8 6488 1936; fax: +61 8 6488 1108.E-mail address: [email protected] (D.L. Nicol).

Dion L. Nicol a,b,⇑, John Finlayson b,c,d, Timothy D. Colmer a, Megan H. Ryan a,b

a School of Plant Biology and Institute of Agriculture (M084), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australiab Future Farm Industries Cooperative Research Centre (M081), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australiac School of Agricultural and Resource Economics (M089), The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australiad EH Graham Centre, Charles Sturt University and Department of Primary Industries, Wagga Wagga Agricultural Institute, Pine Gully Road, Wagga Wagga, NSW 2650, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 March 2012Received in revised form 19 February 2013Accepted 1 June 2013Available online 2 July 2013

Keywords:Low-rainfall environmentsClimatic variabilityCullen speciesNovel plantsCrop and pasture residuesBio-economic modeling

The wet winters and summer droughts of dry Mediterranean-type climates create a highly seasonal sup-ply of feed for livestock. Much of the forage value of winter-active annual pastures and crop residues isrealized as dry feed during summer–autumn. Sporadic summer–autumn rainfall rapidly degrades thequality of dry plant residues. In low rainfall areas of the southern Australian wheatbelt, there are nowell-adapted crops or pastures to convert summer rainfall into high-quality green feed and supplemen-tary feeding is required to maintain livestock condition. We therefore investigated two undomesticatedephemeral legumes (Cullen cinereum and Cullen graveolens). In a field experiment, the ephemerals weredormant in winter–spring and responded strongly to summer rainfall, with 0.45–0.82 t ha–1 of shootdry weight produced over summer. Extrapolation of regional historic rainfall records showed similaror greater summer–autumn rainfall in 40% of years and also suggested that conditions will probablybe too dry for perennial pastures such as Medicago sativa (lucerne) to persist in up to 60% of years. Ananalysis using MIDAS, a bio-economic model, suggested that ephemerals could increase total farm profitand stocking rates (10.3% and 7.7%, respectively), and decrease supplementary feeding of grain by >50%by providing high quality feed in years that summer–autumn rainfall occurs. We suggest there is consid-erable potential for ephemeral legumes to contribute to the sustainability of mixed agriculture in dryMediterranean-type climates by utilizing sporadic summer rainfall whilst complementing existingannual pasture and cropping systems.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction thunderstorms during summer can reduce the digestibility of

Strongly seasonal environments present constraints to livestocksystems due to peaks and troughs in feed quantity and quality. Thepresent study focused on low-rainfall zones in dry Mediterranean-type climates which are characterized by hot, dry summers andmild, wet winters. Although characterized by a long, dry summerperiod, thunderstorms and degraded tropical cyclones can bringepisodic summer–autumn rainfall to these regions, at least in Aus-tralia. Agricultural areas that receive infrequent summer–autumnrainfall events generally do not convert rainfall into additional for-age due to a lack of profitable, commercial summer-forage optionsand poor persistence of summer-active perennial pastures.

Livestock in Mediterranean farming systems that are based onwinter-active annual plants rely on dry plant residues during sum-mer–autumn (Puckridge and French, 1983). When dry plant resi-dues get wet, they rapidly degrade due to microbial activity andleaching of soluble nutrients. Rossiter et al. (1994) found single

high-quality dry pasture residues below those required for main-taining livestock condition. Brown (1977) reported different degra-dation rates among plant types, with high-feed-value legumeresidues degrading the most rapidly once wet. Thus, for livestocksystems, supplementary feeding or alternate high-quality feedsources are required if summer–autumn rainfall occurs.

Although summer–autumn rainfall in dry Mediterranean-typeenvironments such as south-western Australia is considered toprovide little opportunity for plant growth due to high evaporationand temperatures, summer rainfall contributes to weed growththat can often be deleterious to stock (e.g. Tribulus terrestris, Aslaniet al., 2003) or self-sown crop plants that increase the risk of dis-ease carry-over to following crops (Roget et al., 1987). Summer-ac-tive weeds complete their entire life-cycles in dry Mediterranean-type environments in response to episodic rain events during sum-mer–autumn. This demonstrates that at least some forage produc-tion is possible even with very infrequent rainfall, providingsuitably–adapted plants possess adequate forage quality. Plantswith such short growth seasons, and which may be absent mostyears, are often referred to as ephemerals.

D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84 77

We present the concept of using ephemeral summer-activeplants as a means to offset the losses in feed quantity and qualitythat occur due to summer–autumn rainfall events. Our interest isfocussed on environments where perennial pasture options arelimited and/or winter-annual crops are dominant. We use theundomesticated native Australian legumes Cullen cinereum (Lindl.)J. W. Grimes and Cullen graveolens (Domin) J. W. Grimes (previ-ously Psoralea cinerea and P. graveolens) as model ephemeral plantsand the low rainfall, dry Mediterranean-type climate in the easternwheatbelt of Western Australia as a case study. We present datafrom a field experiment and farming system modeling using thebio-economic model MIDAS (Model of an Integrated Dryland Agri-cultural System) (Kingwell and Pannell, 1987). Key opportunitiesfor their potential role are then discussed.

2. Materials and methods

2.1. Field experiment

The field experiment was located approximately 25 km north-east of Mukinbudin (�30.78�, 118.31�) in the low rainfall zone ofthe wheatbelt of Western Australia. Mukinbudin has a dry Medi-terranean-type climate and a mean annual rainfall (MAR) of286 mm (http://www.bom.gov.au/climate/data/). The field site soilwas a duplex, red-brown colored, sandy clay loam over calcareousclay that had been cropped with wheat for the previous 3 years.The pH (CaCl2) ranged from 6.5 near the surface (0–0.1 m) to 8.5in the highly sodic and alkaline subsoil (>0.4 m). The soil contained12 mg kg–1 mineral nitrogen and 12 mg kg–1 bicarbonate extract-able (Colwell) phosphorus (0–0.2 m). The organic carbon contentwas 0.35% w/w (0–0.2 m). No fertilizer was applied. The experi-ment was a randomized block design, with 24 plots consisting of4 replicates of 6 legume species and accessions/cultivars (Table 1).Seeds were scarified where required and inoculated with appropri-ate rhizobia in peat inoculum. The experiment was sown on 5 May2008 by hand (5–15 mm depth), with 16 pre-germinated seeds m–

2 in each 4 � 4 m plot. Only the middle 3 � 3 m of each plot wassampled.

The plots were maintained as monocultures by hand weeding.Little rainfall was received after sowing, so 7.5 mm of irrigationwas applied on 1 and 4 June 2008. On the 15th of each month, pho-tographs were taken from a tripod on a permanent marker in eachplot. These photographs were used to visually compare growth andestimate biomass from dry weights harvested on 15 October 2008(M. truncatula (medic) only), 15 April 2009 (both Cullen spp. and M.sativa (lucerne)) and 15 October 2009 (lucerne only). Shoot dryweights (DW) were obtained by cutting all plants 30–50 mm abovethe base of the stems, followed by drying at 60 �C for 72 h.

2.2. Regional historic rainfall assessments

Historic monthly rainfall data were downloaded (http://bom.-gov.au/climate/data) from 142 weather stations that are dispersed

Table 1The 6 legume species, accessions/cultivars, sources of seed and rhizobia used in the field

Species Accession/cultivar Seed

Cullen cinereum Bill022CC The UC. cinereum Fortescuea MulgC. graveolens AusTRCF320184 (QLD origin) AustrC. graveolens Fortescuea MulgMedicago sativa cv. Sardi10 ComM. truncatula cv. Caliph Com

a Seed and rhizobia of Fortescue accessions were collected from �22.85�, 120.21� by Mfrom soil collected under host plants and sent to Rutherglen Department of Primary Ind

throughout the wheatbelt of Western Australia. These were thenallocated to the agro-climatic zones of the wheatbelt of WesternAustralia described by the Department of Agriculture and Food,Western Australia, for summarizing crop variety suitability(Brown, 1994). These zones are defined in terms of MAR isohyetsand wheatbelt boundaries to give L-low (<325 mm), M-medium(325–450 mm) and H-high (450–750 mm) rainfall zones. Superim-posed on these are north–south divisions reflecting changes in so-lar radiation and temperature gradients giving zones 1–5. The M5and H5 (southern-most high and medium rainfall) zones are di-vided further into west, central and eastern zones due to their widedistribution across the south coast.

The 142 sites were selected on the basis of how complete theywere and to ensure an even spread of data across the zones. Smalldata sets (<30 complete years) were avoided wherever possibleand as many sites as possible were used for each zone. The growingseason for ‘winter’ annual crops and pastures is May–October andsummer–autumn or ‘out-of-season rainfall’ was consideredNovember–April. Years with >100 mm summer–autumn rainfall(similar to the rainfall in the first summer–autumn of the fieldexperiment) were counted at each site. Seasons with any missingdata were omitted from the analysis. The frequency of years with>100 mm summer–autumn rainfall was averaged for each agro-cli-matic zone and presented visually to give an overview of the distri-bution of summer–autumn rainfall across the wheatbelt.

2.3. Farming system modeling

The MIDAS Eastern Wheatbelt Model (EWM) is a whole-farmbio-economic model, which considers the biological, physical,technical and financial aspects of farming systems representativeof farms in the low rainfall (�300 mm MAR) eastern wheatbelt re-gions of Western Australia. The objective function of the model isprofit maximization with farm enterprises being selected on thebasis of assumed yields, system interactions, commodity pricesand overheads (Kingwell and Pannell, 1987). When used in its nor-mal configuration, MIDAS reflects an average farm in an averageseason. Assumptions relating to resource requirements and perfor-mance have been suggested by Department of Agriculture andFood, Western Australia, researchers and advisors based at theMerredin Dryland Research Institute. The major crops included inthe model are bread wheat (Triticum aestivum), barley (Hordeumvulgare), oats (Avena sativa), triticale (� Triticosecale), canola (Bras-sica napus), lupin (Lupinus angustifolius), field pea (Pisum sativum),chickpea (Cicer arietinum) and faba bean (Vicia faba). Annual pas-tures are based on traditional Trifolium subterraneum and annualMedicago spp. ley pastures and livestock systems involve self-replacing flocks of merino sheep farmed for wool and meat.

2.4. Modifications of MIDAS to consider weather scenarios

Due to the high variability in summer conditions in the studyenvironment, weather effects on crop and pasture production,changes in plant residue quality and quantity, and requirements

experiment.

source Rhizobia source

niversity of Western Australia collection Fortescuea

a Research Centre Fortescuea

alian Tropical Crops and Forages GRC Fortescuea

a Research Centre Fortescuea

mercially available AL type commercialmercially available AM type commercial

ulga Research Centre, Curtin University of Technology, WA. Rhizobia were trappedustries, Victoria, Australia for development of peat inoculum.

Table 2Probabilities of alternate weather scenarios, calculated from long-term monthly rainfall data of the L3 agro-climatic zone of the wheatbelt of Western Australia.

Summer scenarios Total

1.<100 mm 2. 100–150 mm 3.>150 mm

Winter scenarios 1. Dry, dry 0.217 0.041 0.043 0.3012. Wet, dry 0.130 0.061 0.070 0.2623. Dry, wet 0.148 0.042 0.041 0.2304. Wet, wet 0.098 0.048 0.061 0.207

Total 0.593 0.192 0.215 1.000

78 D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84

for herbicides were modeled in addition to interactions betweenCullen spp. and other crops/pastures in the same rotation. Themodel was modified to account for 3 summer and 4 winter rainfallscenarios (Table 2). As the field experiment received �100 mmrainfall in the first summer but only 53 mm in the following sum-mer with contrasting results on summer–autumn plant growth,the dry, moderate and wet summers were defined as <100, 100–150 and >150 mm, respectively (Table 2, row 2). We anticipatedthat >150 mm would increase summer–autumn production overthe 100 mm scenario and provide more stored soil moisture for fol-lowing winter crops. To represent the relative variability in cropand annual pasture production, winter scenarios were derivedfrom opening rainfall (April–May) and finishing rainfall (August–September), with a total of 60 mm of rainfall determining if theopening was wet or dry (Table 2, column 2). The probabilities ofeach scenario (winter * summer scenario, n = 12) were calculatedfrom long-term rainfall data for the L3 agro-climatic zone (Sec-tion 2.2) where the field experimental site was located and alsofor the Merredin district where the model specifically applies(Table 2).

The standard form of MIDAS EWM considers the quantity andquality of livestock feed sources for an average summer. To inves-tigate the effect of summer rainfall on dry feed sources, degrada-tion rates were estimated based on results presented by Rossiteret al. (1994) and Brown (1977) (Table 3). The weather scenario ef-fects on arable crop yields were multiplied over winter and sum-mer scenarios (Table 4). Growth patterns of annual pastureswere also modified to reflect different seasonal conditions andmoisture, particularly in consideration of the reduced annual pas-ture production that results from late opening rains. For this,growth in winter scenarios 1 and 3 (<60 mm opening rainfall) were15% in May and 75% in June of standard pasture yields, conserva-tive estimates, and growth in winter scenarios 2 and 4 (>60 mmopening rainfall) were 150% of May and June standard pastureyields.

Table 4Effect of summer and winter scenarios (listed in Table 2) on yields of all crops (% ofaverage yield).

Scenario 1 2 3 4

Summer 90 113 150 –Winter 50 100 100 150

Table 3Effect of ‘summer’ rainfall on the rate of decline in dry feed residue quality (% d�1).

Residue type Summer scenario

1.<100 mm 2. 100–150 mm 3.>150 mm

Cereals 0.10 0.20 0.60Annual pasture 0.40 1.60 6.40Lupins 0.28 0.56 1.12Other arable 0.14 0.28 0.56

2.5. Assumptions about Cullen spp. rotations

Model inputs for Cullen spp. yield were 0, 700 and 1000 kg ha–1

total shoot dry weight (DW) production for summer scenarios 1, 2,and 3, respectively. These were divided across December to Maywith no production assumed from June to November. BetweenDecember and May, losses of Cullen due to senescence and decaywere assumed to be 0.35% per day. Cullen spp. were consideredunavailable in the remainder of the year due to the paddock beingoccupied by winter crops or pastures. The feed values of Cullen spp.were based on reported values from rangeland-grown plants of75% dry matter digestibility (DMD) and 10.75 MJ metabolizable en-ergy (ME) kg–1 (DW) (Jolly, 2009).

In summer scenario 1 (<100 mm), the assumptions were thatCullen spp. do not grow and therefore do not affect crop or pastureproduction. Reductions in crop yields from Cullen spp. in summerscenarios 2 and 3 were estimated to be 20% and 30%, respectively,due to the anticipated removal of stored soil moisture from sum-mer rainfall. The yield of Cullen spp. was assumed to be maximizedin years when it followed annual pasture, with a 10% reduction fol-lowing cereals and a 50% reduction following lucerne, due to ex-pected differences in deep soil moisture. Nitrogen fixation byCullen spp. was not modeled. We assumed Cullen spp. can be grownover the summer, between winter annual pastures or winter cerealcrops, but not in the summer between other winter arable cropssuch as pulses due to potential ‘green bridge’ effects of pest anddisease carryover.

The MIDAS EWM model considers a range of soil types repre-senting an average of 50 farms in the eastern wheatbelt region(Table 5). The soil type in the field experiment is similar to S6(Sawkins, 2009), which in the model represents 475 ha or 12.5%of the 3800 ha model farm. Another 475 ha of a similar soil typeis considered in the model but with applied gypsum amendments(S7). The Cullen spp. are expected to perform best on heavy soilswhich total 25% of the farm.

To solve the model, a genetic algorithm was used in conjunc-tion with 12 linear programming models with each of the mod-els being an instance of the MIDAS EWM specified for aparticular combination of winter and summer weather scenarios.The objective function was calculated as a function of the prob-

Table 5Soil types of the MIDAS model farm in the eastern wheatbelt regions of WesternAustralia with area, percentage of farm area and effect on yield of Cullen spp.

Soil Description Area (ha) % Farm Multiplier of Cullen

S1 Acid sands 760 20 0.2S2 Sandplain 760 20 0.3S3 Gravelly sands 380 10 0.5S4 Duplex 380 10 0.6S5 Medium heavy 570 15 1.0S6 Heavy 475 12.5 1.0S7 Heavy + gypsum 475 12.5 1.0Total 3800

D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84 79

ability of each weather scenario and the profit associated withthe different instances:

Objective function ¼ maxXn

scenario¼1

ðprofit � probabilityÞ

The genetic algorithm maximized the objective function, sub-ject to the constraints that the areas of Cullen spp., pasture, cerealsand sheep numbers were the same for all weather scenarios. Theseconstraints were imposed as optimizing allocations of these sys-tems would require prior knowledge of climatic scenarios andvarying flock size is difficult to alter in the short term. The solutionobtained from the genetic algorithm was subsequently refinedusing a Nelder and Mead simplex algorithm.

3. Results

3.1. Field experiment

Three distinct growth patterns were evident for the legumes inthe field experiment: winter-active annual (Medicago truncatula),temperate perennial (M. sativa) and summer-active ephemeral(Cullen spp.) (Fig. 1). Annual medic (M. truncatula) had the earliestgrowth which peaked at 1.75 t ha–1 shoot DW, with �1.05 t ha–1

accumulating in September alone, prior to senescing in October.Lucerne (M. sativa) was less vigorous than annual medic but con-tinued to grow through spring to early summer and then persistedwell over the first summer, with 0.92 t ha–1 shoot DW harvested inApril 2009. In the second winter growing season, lucerne followeda near-identical production curve to annual medic, producing2 t ha–1 shoot DW (April–October 2009). The second summer(2009–2010) was dry resulting in no lucerne production from

Fig. 1. Rainfall, air temperature and pasture shoot dry weight for the field experiment.bars); summer–autumn rainfall (black bars); June 2008 irrigation (hatched bar); meanPasture production as mean (n = 4, S.E.) standing shoot dry weight (t ha�1) of Cullen cinereQld: AusTRCF 320184), Medicago sativa (lucerne, cv. Sardi10) and M. truncatula (annual mwere cut at 30–50 mm height or from base of stem indicated with an H. NA - April 201

October 2009 onwards with plant deaths observed from Decemberonwards. Cullen spp. did not accumulate DW in the first wintereven though they had germinated, but instead started to grow inNovember–December with rapid growth in February–March with0.45–0.82 t ha–1 shoot DW harvested in April 2009. Survival of Cul-len spp. following the April 2009 harvest was very low in the plots.

In the 2008 winter growing season (May–October), the experi-ment received 153 mm of rainfall. The first summer (November–April) received 114 mm rainfall, of which 20 mm was receivedafter harvest of shoot DW in April 2009. Short dry periods duringthe first summer caused water deficit symptoms in most plantsprior to the late January rainfall (50.8 mm) and February rainfall(26.2 mm). Rapid growth of the Cullen spp. and lucerne occurredthrough to March, but growth then plateaued or declined due toleaf shedding prior to the April harvest in response to the dry con-ditions during March. The 2009–2010 summer–autumn was muchdrier, with negligible growth of Cullen spp. or summer weeds.Abundant recruited seedlings were, nevertheless, observed follow-ing summer rainfall in early 2011 (112 mm; November 2010–April2011). A dense sward of recruited seedlings was again observed inDecember 2012 following �90 mm of rainfall.

3.2. Observations of Cullen spp. in situ

Cullen spp. seedling survival was poor from May until Octoberwhen Cullen spp. were dormant. This resulted in densities low en-ough to negatively impact production. Aphid damage in Septem-ber–October was high in Cullen spp. due to proximity tosenescing crops. The M. sativa and M. truncatula cultivars were bredfor aphid tolerance, unlike the undomesticated Cullen spp., result-ing in substantially lower predation levels. Numerous seedlings of

Total monthly rainfall (mm) for the ‘winter’ growing season (May–October) (graymonthly maximum (solid line) and minimum (dotted line) air temperatures (�C).um (Bill: Bill02CC and Fort: Fortescue accessions), C. graveolens (Fort: Fortescue andedic, cv. Caliph). All species/accessions were sown May 2008. Harvests where plants0 was not harvested due to limited living biomass.

80 D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84

Cullen spp. were infected by alfalfa mosaic virus (AMV) during thistime and their growth was noticeably reduced compared to unin-fected plants. Nonetheless, the vigor of healthy seedlings in sum-mer was still adequate to produce mean shoot DW similar tothat produced by M. sativa. Very few pests were evident in the heatof summer. Many of the recruits of Cullen spp. that first appeared insummer (i.e. December) were indistinguishable in April from unin-fected plants that germinated the previous May, except that theyhad not been affected by insects and viruses during spring. Theserecruits were able to flower within 30 days and set seed within60 days with indeterminate flowering and growth continuing torespond to rainfall. Post-harvest (i.e. cutting at 30–50 mm) survivalof Cullen spp. was poor compared to lucerne, probably due toplants being cut too close to the ground. Unharvested recruits out-side of experimental areas maintained growth through to wintersuggesting suitable conditions and management would allow sig-nificant growth in late autumn to early winter before slowing laterin winter. M. sativa appeared less affected by insect pests, diseaseand cutting than Cullen spp. However, M. sativa appeared morestressed by summer climatic conditions, displaying signs of waterdeficit stress during the heat of summer, even following rainfall(unlike Cullen spp.). M. sativa recovered strongly from cutting inthe second winter.

3.3. Rainfall assessments

Long-term rainfall records from 142 sites across the wheatbeltagro-climatic zones indicate a clear pattern of fewer moderate/wet summers (>100 mm November–April) in the north and moremoderate/wet summers in the south, particularly along the southcoastal regions (Fig. 2). Frequency of moderate/wet summers ap-pears unrelated to MAR (rainfall isohyets), as it has a north–southrelationship not an east–west relationship. The regions H1 and M1had mean frequencies of moderate/wet summers of �20% to 30%.L1, M2 and H3 had 30–40%, with the bulk of the central wheatbelt

Fig. 2. Mean frequencies of moderate and wet ‘summers’ (>100 mm November–April) foare devised by rainfall isohyets (<325, 325–450, 450–750 mm long term mean annual razones, with solar radiation and mean temperature gradients then delineating zones 1–5. Tand eastern due to their wide distribution across the south coast. Degree of shading inmoderate/wet, Table 2).

regions (H2, L2, M3, L3, M4 and L4) having moderate/wet summersin �40% to 50% of years. Other than M5 W (50–60%), all southernM5 and H5 zones had moderate/wet summer frequencies of 60–90%.

3.4. MIDAS assessments of Cullen spp.

Simulations in the MIDAS EWM showed growing Cullen spp.during the summer–autumn (November–April) improved meanfarm profitability by 10.3% (Table 6). Minor changes occurred inthe area of cereal (-1.1%), pasture (+2.1%) and other arable crops(-6.8%), but stocking rate increased by 7.7% and supplementaryfeeding was reduced from 14 to 6 kg (-57%) per dry sheep equiva-lent (DSE) by growing Cullen spp. on 10.1% of the farm area.

Comparing simulated feed sources for different summer scenar-ios revealed that annual pasture residues are the main summerfeed source in the standard farming model output (Fig. 3a and b).However, once this feed source is degraded, stubbles (crop resi-dues) increase in importance and supplementary feeding is re-quired (Fig. 3b and c). A key finding was that Cullen spp. canreduce the requirement for purchased feed and dependence on de-graded stubbles in moderate and wet summers.

4. Discussion

The results from this study support the concept that the sum-mer-active ephemerals C. cinereum and C. graveolens can offsetthe negative effects of sporadic summer rainfall on dry feed in aMediterranean-type climate. The concepts and mechanisms out-lined in this study offer strategies to improve the use of out-of-sea-son rainfall in other strongly seasonal environments, whethersummer or winter dry, to offset dry feed losses that are due torainfall.

r the agro-climatic zones of the Western Australian wheatbelt. Agro-climatic zonesinfall) and wheatbelt boundaries to give low (L), medium (M) and high (H) rainfallhe medium and high rainfall zones M5 and H5 are divided further into west, central

creases with frequency of ‘summer’ (November–April) rainfall being > 100 mm (i.e.

Table 6Effect of including Cullen spp. (i.e. an ephemeral summer–active pasture system) on the proportion of the farm allocated to cropping and pasture systems, stocking rate,supplementary feeding and farm profit ($A – Australian dollars). Mean values calculated from simulations over various climate scenarios weighted by probabilities.

Treatment Cereal(% of farm)

Pasture(% of farm)

Other arable(% of farm)

Cullen spp.(% of farm)

Sheep(DSEa ha�1 of farm)

Supplementary feeding(kg DSE�1)

Mean profit($A total farm)

Without Cullen 51.7 36.6 11.7 0 2.6 14 128,440With Cullenb 51.1 37.4 10.9 10.1 2.8 6 141,740

a DSE – Dry sheep equivalent, livestock feed budgeting figure of a 45 kg wether.b % of farm figures are >100% with Cullen because the same paddock may have Cullen in summer and a winter crop or annual pasture.

Animal intake (TJ ME month-1)

With

Cul

len

W

ithou

t Cul

len

Cullen

Pasture

Purchased

Stubble

Total

Cullen

Pasture

Purchased

Stubble

Total

0.0 1.0 2.0 0.0 1.0 2.0 0.0 1.0 2.0 0.0 1.0 2.0 0.0 1.0 2.0

Cullen

Pasture

Purchased

Stubble

Total

Pasture

Cullen

Purchased

Stubble

Total

W

ith C

ulle

n

With

out C

ulle

n

Cullen

With

Cul

len

With

out C

ulle

n

December January February March April

Purchased

Pasture

Stubbles

Total

Cullen

Pasture

Stubbles

Total

Purchased

(A)

(B)

(C)

Fig. 3. Summer–autumn (December–April) feed sources from simulated animal intake (TJ (Terra-Joules) ME (metabolizable energy) month�1) for three ‘summer’ scenarios:A. Dry (<100 mm); B. Moderate (100–150 mm) and C. Wet (>150 mm November–April rainfall). Data are presented as box plots which display the median (bold line), the 1stand 3rd quartiles (edge of box) and whiskers (min/max) from the four winter scenarios, equivalent to 4 replicates of each summer scenario.

D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84 81

82 D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84

4.1. Current pasture systems are constrained in stronglyMediterranean climates

Annual pasture systems in strongly Mediterranean-type cli-mates have a strong spring growth peak prior to senescence. Forinstance, in this study, September (early spring) contributed 60%of total annual pasture legume growth (Fig. 1). This peak growthmust typically carry livestock over the summer–autumn as dry res-idues (Willoughby, 1959). While these hot, dry summers enablemaintenance of dry feed quality from annual pastures and cropresidues, summer rainfall causes degradation of this dry feed(Rossiter et al., 1994). For instance, in the south coastal regionsof Western Australia where summer rainfall is frequent, Doyleet al. (1996) found 70–90% of pasture residues lost over summerwere not consumed by livestock. Hence summer rainfall increasesthe requirements for supplementary feeding and/or fodder conser-vation. A large number of new annual pasture legume cultivars,and indeed new species, have recently been released in WesternAustralia (Nichols et al., 2007). Whilst these have increased peakproduction and/or provided other farming systems benefits, theyare prone to the same risks from summer rainfall. Thus, the currentannual legume-based pasture systems in Western Australia, andacross the southern Australian cropping zone, require dry summersfor optimum annual pasture utilization.

Temperate perennial pasture species, in contrast, require wettersummers to maximize persistence/survival and growth. M. sativa iscurrently the most successful commercial perennial legume in thelower rainfall areas of the southern Australian cropping zone (Dearet al., 2008; Li et al., 2008). M. sativa has a much longer growingseason than annual pasture legumes, but its persistence/survivalis hampered by frequent summer droughts such as those which oc-curred in the second summer of this field study (Fig. 1). The perfor-mance of M. sativa in dry summers is limited by available storedsoil moisture and thus M. sativa is better suited to high rainfallzones where recharge of soil water deep in the profile occurs morefrequently in winter growing seasons. Adoption of M. sativa in thedrier environments of the Western Australian wheatbelt remainspoor due to the risk of failure (i.e. poor persistence), the high costof M. sativa systems (Bee and Laslett, 2002) and the specialist man-agement required (i.e. rotational grazing) (Lodge, 1991).

Summer cropping (grain or forage) has been attempted in theWestern Australian wheatbelt, but with little success. Summer for-age crops potentially offer an alternative to perennials, but typi-cally require higher summer rainfall than the low rainfall areasexperience, even in ‘wet’ summers. In the south coastal areas ofthe Western Australian wheatbelt with higher summer rainfall(Fig. 2, H5C and E), summer forage crops rarely succeed and areconsidered excessively risky (Robertson et al., 2005). Some of these

(B) Change in

Persistence of

Rel

ativ

e qu

antit

ies

of fe

ed

(A) Hypothetical feed offset

Summer rainfall

Rel

ativ

e qu

antit

ies

of fe

ed

Fig. 4. Hypothetical frameworks for dry feed sources and summer-active pasture growthrainfall increases. (B) The effect on available feed of increasing persistence of summer mpattern for summer-active pasture. (C) The effect of feed source quality on diet composummer-active pasture, . . .. . . total feed).

forage crops are also inflexible as they pose risks such as prussicacid poisoning (e.g. Sorghum bicolor) when sacrificial grazing ismost needed, have higher input requirements than other pasturesystems and impede winter cropping systems. These traits furtherlimit the role summer-active forage crops may play in the region.

4.2. Ephemeral pasture systems

Our results suggest that ephemerals which germinate on sum-mer rains in December/January and set seed by April prior to win-ter crops being sown, may complement annual pastures, and offeran alternative to perennial pastures and summer cropping in envi-ronments with low and/or variable summer rainfall. For instance,in the L3 region long-term rainfall data (Fig. 2) show summers(November–April) receive >100 mm in �40% of years, which islikely to degrade dry feed sources. However, persistence of peren-nial pastures such as M. sativa, and the viability of summer foragecrops, would be limited by the likelihood of failing in the �60% ofsummers with <100 mm of rainfall. The potential strength of anephemeral system in such a region arises from the fact that plants,once established, would persist in the seed bank and at little or nocost provide opportunity from sporadic summer rainfall. Whilstthe ephemeral system could be absent, due to insufficient rainfall,in many or even most seasons, it could respond rapidly to summerrainfall in terms of forage production and restoring the seed bank.Since an ephemeral system requires low inputs (i.e. no additionalfertilizer or seed after the establishment year) risk would also below. This system would complement current annual pasture andcrop-based farming systems and allow livestock systems basedon annual pasture legumes to increase their security of feed insummer-autumn, thereby strengthening the resilience of themixed farming systems to climatic variability.

4.3. Summer dry feed is inversely related to potential growth ofephemerals

Pasture residues, if kept dry, have little if any decline in quality(Brown, 1977; Rossiter et al., 1994). However, pasture residues willrapidly decay once wet (Rossiter et al., 1994). This study illustratedthe importance of the inverse relationship between summer dryfeed decay and opportunity for ephemerals to enhance total farmprofitability and resilience to climatic variability, albeit using lim-ited literature and data. It was shown that stability can come fromclimatic variability, as the loss of dry feed in a wet summer can bereplaced by growth of an ephemeral.

To allow further exploration of this key point, hypothetical rela-tionships between rainfall and relative availability of dry and greenfeed in summer are presented in Fig. 4. In Fig. 4A it is assumed that

available feed

summer moisture

(C) Diet composition

Summer rainfall

Rel

ativ

e qu

antit

ies

of fe

ed

offset. (A) Overview of dry feed replacement by summer-active pastures as summeroisture assuming a linear decay rate for dry feed quantity and a sigmoidal growth

sition of grazing animals of increasing summer rainfall. (——— dry feed sources, – – –

D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84 83

increasing summer rainfall will increase both plant growth and dryfeed decay in a de Wit style replacement relationship. In reality,whilst high evaporation rates and harsh dry conditions in summercan result in rapid loss of soil moisture, summer rain is oftenaccompanied by cooler conditions. Thus, the extent of rainfalland cloudy weather and rate of rainfall will all influence theamount of infiltration and evaporation from the soil surface. As aresult, the persistence of soil moisture in summer will be highlyvariable. However, if moisture does rapidly disappear, degradationof dry feed may not be as severe, whereas if soil moisture persistsfor longer, ephemeral plants are more likely to be productive andoffset the dry feed losses.

In Fig. 4B, the linear decay of pasture residue DW as describedby Brown (1977) is offset by plant growth which is assumed tobe sigmoidal and thus there is an expected lag in the offsettingof dry feed losses after rain and, if rapid soil moisture decline oc-curs after a large rainfall event, there is potential for lower totalfeed. However, the quality of young green feed should be muchhigher than that of the remaining dry feed and could be expectedto be selectively grazed. Thus, as shown in Fig. 4C, the importanceof green summer feed to livestock diet, relative to dry feed, is likelyto be greater than its relative biomass. This presents the opportu-nity to utilize higher quality feed, but requires management to al-low the most desirable species to increase growth for seed-set andforage production. Further research to test these hypothetical rela-tionships is needed.

4.4. Potential of Cullen spp

The system proposed here is based on the field performances ofC. cinereum and C. graveolens. Both of these species occur acrosssome of the most arid and variable rainfall environments in theAustralian continent. Vital to the success of the ephemeral systemis the ability to develop and sustain a seed bank through unsuitableseasons, i.e. dry summers. Potentially high resilience to frequentdry summers is indicated by the fact that C. cinereum naturally oc-curs in the northern wheatbelt of Western Australia (agro-climaticzones H1–L1, see Fig. 2) where summer rainfall frequency is verylow (Bennett et al., 2011). High resilience is also indicated by ourobservations of large numbers of Cullen spp. recruits in Nov2010–Apr 2011, (the third summer after sowing) and November2012–January 2013 (the fifth summer after sowing).

The opportunities with Cullen spp. suggested by this study arelikely to be conservative. The potential production of these specieswould be higher than that captured in this dataset as plant densityin the field experiment was low and occurrence of disease high dueto the winter sowing. However, these undomesticated, unselected,wild plants were shown, through modeling, to increase profitabil-ity of farming systems due to their timely ability to produce highquality shoot biomass under summer conditions which consist ofhigh temperatures, high solar radiation and intermittent soil wateravailability. Other favorable traits include indeterminate flowering,which allows rapid seed set before low management input grazingor removal before cropping, and high winter dormancy which en-ables winter crops and pastures to be sown in the same field andprevents recruits under crops from competing with crops duringtheir winter/spring growing season. Recent screening of C. cinere-um and C. graveolens assessed their potential for development asperennial pasture species, but their strong winter dormancy andlow winter survival, particularly on acidic sandy soils meant theywere overlooked for further development, although C. australasi-cum was found to perform strongly as a potential perennial pasturelegume (Dear et al., 2007; Bennett et al., 2012). The niche of theephemeral Cullen spp. is not currently occupied by a commercialpasture cultivar.

4.5. Future priorities

4.5.1. Development of ephemeral pasturesEphemeral undomesticated plants offer opportunities to

quickly and cheaply develop a novel component that wouldstrengthen current mixed farming systems in Mediterranean-typeenvironments without the need for significant system change. En-hanced resilience to climatic variability would be a key outcome,along with environmental benefits from increased soil cover (Find-later et al., 1990) and increased water use over summer resultingin a reduction in groundwater recharge (Dolling et al., 2006). Anephemeral pasture plant option would also allow greater flexibilityin breeding strategies for perennial pastures for these environ-ments by removing the need for high summer production. Rapiddevelopment of cultivars is possible for C. cinereum as wider collec-tions are available in Australian Genetic Resource Centres (Bennettet al., 2011). Adoption could occur across the southern croppingzone of Australia wherever low summer rainfall and high evapo-transpiration precludes other summer pasture options. More de-tailed investigations are needed to gain greater insight intosystem profitability and the most appropriate breeding/selectionstrategies.

4.5.2. Understanding summer dry feed declineMore rigorous assessments of dry feed decline and summer pas-

ture growth are needed. The findings of this study contrast withprevious research into the value of summer-active pastures inMediterranean-type environments (Moore et al., 2009), as thisstudy found that highly variable production year to year, resultingfrom summer rainfall, adds stability to livestock feed sources andrelated economics. However, assumptions had to be made aboutrates of dry feed decline being offset by Cullen spp. and thus furtherinvestigation into relationships between dry feed decline andgrowth of summer-active pastures would be valuable.

Previous modeling and assessments of summer-active pasturesin Western Australia’s Mediterranean-type climate (e.g. Mooreet al., 2009; Lawes and Robertson, 2008) have failed to considerthe widely understood, but poorly documented, losses of dry feedvalue that occur from out-of-season rainfall. Steady state models,such as MIDAS, are limited in their application to this problem asthey work on assumed average seasonal conditions. Without mod-ification, such models may be unsuitable for assessing summer-ac-tive pastures, as the variability that occurs in summer conditionscannot be addressed with sensitivity analyses. Furthermore, previ-ous bio-economic assessments of summer-active perennials mayneed revising as the economic framework may change substan-tially with consideration of the dry feed offset.

The impact of climatic conditions on various feed sources re-quires calibration for modeling. Although dry feed degradationfrom rainfall/moisture is widely understood among growers, littledata exists for inclusion in models and this component has largelybeen ignored by researchers. To better predict the relationship be-tween dry feed losses and growth of summer pastures, comprehen-sive real world data with models such as APSIM (see Keating et al.,2003) may be suitable for predicting dry feed losses as well asplant growth. Better definition of climate scenarios may then allowgreater interpretation of more complex opportunities for newtechnologies, such as ephemerals, in farming systems.

5. Conclusions

This study demonstrates that the highly opportunistic biologyof summer-active ephemeral legumes (Cullen spp.) can increasethe adaptation of mixed farming systems in strongly Mediterra-nean-type environments through production of high value forage

84 D.L. Nicol et al. / Agricultural Systems 120 (2013) 76–84

in summers when dry feed sources are degraded by summer rain-fall. More thorough exploration of feed source relationships ishampered by a lack of basic information and data on rainfall im-pacts on dry feeds. Further research is warranted on dry feeddynamics through the summer period in Mediterranean-type envi-ronments. Cullen spp. warrant further development as summer-ac-tive, ephemeral pasture legumes for low rainfall zones. Otherundomesticated ephemeral forages may have the potential to en-able quick and cheap development of opportunistic feed sourcesfor livestock systems in other strongly seasonal regions of theworld.

Acknowledgements

Support for this research was provided by the School of PlantBiology, The University of Western Australia, the Future FarmIndustries Cooperative Research Centre and the AW HowardMemorial Trust Inc. We thank Craig & Julie Nicol and Allan Nicolfor providing the land for, and assistance with, the fieldexperiment.

References

Aslani, M.R., Movassaghi, A.R., Mohri, M., Pedram, M., Abavisani, A., 2003.Experimental Tribulus terrestris poisoning in sheep: clinical, laboratory andpathological findings. Vet. Res. Commun. 27, 53–62.

Bee, G.A., Laslett, G., 2002. Development of a rainfed lucerne-based farming systemin the Mediterranean climatic region of southwestern Australia. Agr. WaterManage. 53, 111–116.

Bennett, R., Ryan, M.H., Colmer, T.D., Real, D., 2011. Prioritisation of novel pasturespecies for use in water-limited agriculture: a case study of Cullen in theWestern Australian wheatbelt. Genet. Resour. Crop. Ev. 58, 83–100.

Bennett, R., Colmer, T.D., Real, D., Renton, M., Ryan, M.H., 2012. Phenotypic variationfor productivity and drought tolerance is widespread in germplasm collectionsof Australian Cullen species. Crop. Pasture Sci. 63, 656–671.

Brown, T., 1977. Rate of loss of dry matter and change in chemical composition ofnine pasture species over summer. Aust. J. Exp. Agr. 17, 75–79.

Brown, G., 1994. The crop variety sowing guide for Western Australia. WesternAustralian Department of Agriculture, Bulletin No. 4273.

Dear, B.S., Li, G.D., Hayes, R.C., Hughes, S.J., Charman, N., Ballard, R.A., 2007. Cullenaustralasicum (syn. Psoralea australasica): a review and some preliminarystudies related to its potential as a low rainfall perennial pasture legume.Rangeland J. 29, 121–132.

Dear, B.S., Reed, K.M., Craig, A.D., 2008. Outcomes of the search for new perennialand salt tolerant pasture plants for southern Australia. Aust. J. Exp. Agr. 48, 578–588.

Dolling, P.J., Fillery, I.R.P., Ward, P.R., Asseng, S., Robertson, M.J., 2006. Consequencesof rainfall during summer–autumn fallow on available soil water and

subsequent drainage in annual-based cropping systems. Aust. J. Agr. Res. 57,281–296.

Doyle, P., Carter, D., Speijers, E., Plaisted, T., Hetherington, R., Love, R., 1996. Changesin the amount and nutritive characteristics of annual pastures from late springto autumn on the south coast of Western Australia. Aust. J. Exp. Agr. 36, 791–801.

Findlater, P., Carter, D., Scott, W., 1990. A model to predict the effects of prostrateground cover on wind erosion. Soil Res. 28, 609–622.

Jolly, S., 2009. Best practice nutritional management of grazed pasture plants.<www.productivenutrition.com.au> (retrieved 25.05.11).

Keating, B.A., Carberry, P.S., Hammer, G.L., Probert, M.E., Robertson, M.J., Holzworth,D., Huth, N.I., Hargreaves, J.N.G., Meinke, H., Hochman, Z., McLean, G., Verburg,K., Snow, V., Dimes, J.P., Silburn, M., Wang, E., Brown, S., Bristow, K.L., Asseng, S.,Chapman, S., McCown, R.L., Freebairn, D.M., Smith, C.J., 2003. An overview ofAPSIM, a model designed for farming systems simulation. Eur. J. Agron. 18, 267–288.

Kingwell, R.S., Pannell, D.J. (Eds.), 1987. MIDAS, A Bioeconomic Model of a DrylandFarm System. PUDOC, Wageningen, 207 pp.

Lawes, R.A., Robertson, M.J., 2008. Seasonal variation of Rhodes grass production inthe northern West Australia wheatbelt. In: Proceedings of the 14th AustralianAgronomy Conference, September, 2008, Adelaide, South Australia.

Li, G.D., Lodge, G.M., Moore, G.A., Craig, A.D., Dear, B.S., Boschma, S.P., Albertsen,T.O., Miller, S.M., Harden, S., Hayes, R.C., Hughes, S.J., Snowball, R., Smith, A.B.,Cullis, B.C., 2008. Evaluation of perennial pasture legumes and herbs to identifyspecies with high herbage production and persistence in mixed farming zonesin southern Australia. Aust. J. Exp. Agr. 48, 449–466.

Lodge, G., 1991. Management practices and other factors contributing to the declinein persistence of grazed lucerne in temperate Australia: a review. Aust. J. Exp.Agr. 31, 713–724.

Moore, A.D., Bell, L.W., Revell, D.K., 2009. Feed gaps in mixed-farmingsystems: insights from the Grain and Graze program. Anim. Prod. Sci. 49,736–748.

Nichols, P.G.H., Loi, A., Nutt, B.J., Evans, P.M., Craig, A.D., Pengelly, B.C., Dear,B.S., Lloyd, D.L., Revell, C.K., Nair, R.M., Ewing, M.A., Howieson, J.G., Auricht,G.A., Howie, J.H., Sandral, G.A., Carr, S.J., de Koning, C.T., Hackney, B.F.,Crocker, G.J., Snowball, R., Hughes, S.J., Hall, E.J., Foster, K.J., Skinner, P.W.,Barbetti, M.J., You, M.P., 2007. New annual and short-lived perennial pasturelegumes for Australian agriculture—15 years of revolution. Field Crops. Res.104, 10–23.

Puckridge, D.W., French, R.J., 1983. The annual legume pasture in cereal–ley farmingsystems of southern Australia: a review. Agr. Ecosyst. Environ. 9, 229–267.

Robertson, M.J., Gaydon, D., Hall, D.J.M., Hills, A., Penny, S., 2005. Production risksand water use benefits of summer crop production on the south coast ofWestern Australia. Aust. J. Agr. Res. 56, 597–612.

Roget, D., Venn, N., Rovira, A., 1987. Reduction of Rhizoctonia root rot ofdirect-drilled wheat by short-term chemical fallow. Aust. J. Exp. Agr. 27,425–430.

Rossiter, R., Taylor, G., Klein, L., 1994. Environmental effects, in particular of rainfall,on the digestibility of dry mature subterranean clover. Aust. J. Exp. Agr. 34, 25–32.

Sawkins, D., 2009. Landscapes and soils of the Merredin District. Department ofAgriculture and Food, Western Australia. Bulletin 4788 ISSN 1833–7236.

Willoughby, W., 1959. Limitations to animal production imposed by seasonalfluctuations in pasture and by management procedures. Aust. J. Agr. Res. 10,248–268.