FUNGAL MICROBIOLOGY

Seasonal and Diurnal Patterns of Spore Release CanSignificantly Affect the Proportion of Spores Expectedto Undergo Long-Distance Dispersal

David Savage & Martin J. Barbetti &William J. MacLeod & Moin U. Salam & Michael Renton

Received: 15 August 2011 /Accepted: 20 September 2011 /Published online: 4 October 2011# Springer Science+Business Media, LLC 2011

Abstract Many of the fungal pathogens that threatenagricultural and natural systems undergo wind-assisteddispersal. During turbulent wind conditions, long-distancedispersal can occur, and airborne spores are carried overdistances greater than the mean. The occurrence of long-distance dispersal is an important ecological process, as itcan drastically increase the extent to which pathogenepidemics spread across a landscape, result in rapidtransmission of disease to previously uninfected areas, and

influence the spatial structure of pathogen populations infragmented landscapes. Since the timing of spore releasedetermines the wind conditions that prevail over a dispersalevent, this timing is likely to affect the probability of long-distance dispersal occurring. Using a Lagrangian stochasticmodel, we test the effect of seasonal and diurnal variationin the release of spores on wind-assisted dispersal. Sporesreleased during the hottest part of the day are shown to bemore likely to undergo long-distance dispersal than thosereleased at other times. Furthermore, interactions are shownto occur between seasonal and diurnal patterns of release.These results have important consequences for furthermodelling of wind-assisted dispersal and the use of modelsto predict the spread of fungal pathogens and resultingpopulation and epidemic dynamics.

Introduction

Long-distance dispersal is an important ecological processthat influences the long-term survival and genetic structureof populations, particularly within fragmented environ-ments [6–8, 20, 27]. For wind-dispersed fungal pathogens,long-distance dispersal can play a highly significant role inthe transmission of disease to previously uninfected areasby allowing migration across uninhabitable regions [6, 8,35]. Long-distance dispersal also facilitates genetic interac-tion between spatially separated populations, and cantherefore result in the introduction of new virulent allelesinto existing populations [35]. Understanding the drivers oflong-distance dispersal can therefore provide insights intothe management of diseases caused by fungal pathogens,and is therefore of significant importance to agriculture,human and animal health, biosecurity, conservation ofnatural ecosystems and pest management.

Microb Ecol (2012) 63:578–585DOI 10.1007/s00248-011-9949-x

D. Savage :M. J. Barbetti :W. J. MacLeod :M. RentonSchool of Plant Biology and the UWA Institute of Agriculture,The University of Western Australia,35 Stirling Highway,Crawley, WA 6009, Australia

D. Savage :M. J. Barbetti :W. J. MacLeod :M. RentonCooperative Research Centre for National Plant Biosecurity,University of Canberra,Level 2, Building 22, Innovation Centre, University Drive,Bruce ACT 2617, Australia

W. J. MacLeod :M. U. SalamDepartment of Agriculture and Food Western Australia,Locked Bag 4, Bentley Delivery Centre,Perth, WA 6983, Australia

M. RentonCSIRO Ecosystem Sciences,Floreat, WA 6014, Australia

M. RentonCentre of Excellence for ClimateChange and Woodland and Forest Health, Murdoch University,Perth, WA, Australia

D. Savage (*)23 Ingrams Rd, Research,Melbourne, VIC, Australia 3095e-mail: david.savage@mac.com

In general, organisms undergoing wind-assisted dispersalare most likely to experience long-distance dispersal duringperiods of unstable wind conditions [19, 23, 26, 37, 39]. Suchconditions include the presence of turbulent updrafts, and inparticular, the auto-correlation of wind gusts in an upwarddirection, which is crucial to long-distance dispersal [37, 39].Sustained updraft is able to keep even relatively large seedsaloft over long periods of time [37, 39], and periods offrequent updraft have been shown to correlate with increasedmean horizontal wind speeds and, to a lesser extent, warmertemperatures, which lead to an increased sensible heat flux[22, 37].

The timing of propagule release in many fungal pathogensthat undergo wind-assisted dispersal has been shown to behighly non-random [1, 4, 5, 14, 25, 28, 29, 31–33, 40]. Forexample, in some species, passive mechanisms of release areobserved, with spores liberated by wind gusts strong enoughto sweep away the slow-moving boundary layer and entrainthe spore in more turbulent airflow (e.g. [4, 5]). Alternatively,release of spores may be active, and is often triggered by theoccurrence of particular environmental conditions, such asrainfall [25, 31–33] or darkness [28]. As a result, the activerelease of spores may follow particular seasonal and diurnalpatterns, with release more likely to occur at particular timesof the year, or within particular hours of the day (e.g. [14, 25,28, 29, 31–33]). These patterns are common in many fungalspecies of agricultural and economic importance (e.g., [14,25, 28, 29, 31–33]), and understanding the manner in whichthe timing of spore release affects the spread of these speciescan lead to more efficient and effective management practices[25, 32, 33, 36].

Previous research has shown that for the wind-assisteddispersal of tree seeds, release of seeds during warmer periodsof the day can increase the proportion of seeds undergoinglong-distance dispersal [22]. The buoyant movement ofwarm air during these periods results in the generation ofturbulent updrafts, which are capable of carrying tree seedsover relatively large distances [22, 37]. It seems likely thatthis result would also hold for the wind-assisted dispersal ofsmaller propagules such as fungal spores, and those species,or individuals within a species that release their spores intowarmer conditions could be expected to have a higherproportion of their offspring undergo long-distance dispersal.If this is indeed the case, then, a particular seasonal ordiurnal pattern of release would affect the proportion of anindividual’s offspring that undergo long-distance dispersal.This would represent a potential mechanism for selection toact on an individual, selecting for or against mechanisms ofrelease timing corresponding to a greater probability of long-distance dispersal. Therefore, the overall purpose of thispaper is to test the effect of variation in seasonal and diurnalrelease on the long-distance dispersal of fungal spores.Specifically, we test whether the month of the year, and the

hour of the day in which release occurs are significant indetermining the proportion of spores undergoing long-distance dispersal. Of particular interest is the release ofspores during the warmest part of the day, and whether theproportion of spores undergoing long-distance dispersalincreases with temperature, as is the case for the wind-assisted dispersal of tree seeds [22].

In determining the proportion of spores undergoing long-distance dispersal, we consider release occurring in twodifferent months, and also at various times throughout theday. We employ an existing Lagrangian model [37] tosimulate the dispersal of wind-borne fungal spores at eachof these different times. The results of these simulationsshow that for the cases considered, a greater proportion ofspores undergo long-distance dispersal when release occursat particular times of the year, or during a particular hour ofthe day. We suggest that this result can be generalised to alarge number of fungal species that undergo wind-assisteddispersal.

Materials and Methods

A Lagrangian Model of Wind-Assisted Dispersal

We used a slightly modified version of the previouslypublished Atmospheric Stability Correction model [37] tosimulate dispersal. This model is a stochastic Lagrangianmodel that employs a statistical description of turbulent airflow, with the horizontal and vertical wind speed modelledas auto-correlated random variables. The model simulatesthe dispersal of a specified number of individual spores,with each spore treated as an inert, passive particle with agiven terminal velocity. The path of each spore is simulatedby calculating the instantaneous wind velocity experiencedby the spore during each time step, and moving the sporeaccording to this velocity and the effect of gravity. Incalculating the wind velocity, the model takes buoyantlyproduced turbulence into account, which results from theheating of the earth’s surface and the subsequent rising ofwarm air. The model therefore requires as part of its inputmeasurements of surface temperature and sensible heatflux, which describes the transfer of energy from the heatedsurface of the earth into atmospheric airflows. The modelalso requires as input measurements of the mean horizontalwind speed and also the height of vegetation, the verticalprofile for leaf area index within the vegetation and the dragcoefficient for the vegetation over which simulation occurs.This vegetation is assumed to homogeneously cover theentire region of simulation.

We used the model as previously described [37] exceptfor one minor modification. In the original formulation ofthe model, friction velocity, which describes the logarithmic

Timing of Spore Release Affects Long-Distance Dispersal 579

profile of the mean horizontal wind speed close to theearth’s surface, is calculated within the model, and is nottaken as an explicit input. However, since friction velocitywas available along with other meteorological data, wechose to input this variable directly.

Model Inputs

Meteorological inputs were obtained using The Air PollutionModel (TAPM), which includes an atmospheric sub-modelcapable of simulating the mean horizontal wind speed, frictionvelocity, temperature and sensible heat flux over a landscapearea [17]. Based on large-scale synoptic data, which isprovided by CSIRO with the model, TAPM models a set ofmeteorological variables for a given location over a specifiedtime period. Meteorological variables are calculated at 5-minintervals and output as the hourly mean at each point in aregular mesh over the computational domain. The model hasbeen well-validated [15, 16], and represents a suitable sourceof data in the absence of sufficiently detailed observations.

As a case study, we used TAPM to simulate meteorology atNortham (31.40 S, 116.9 E) and Merredin (31.29 S, 118.13 E)in Western Australia, over early winter and early spring 2007.In Western Australia, these periods correspond to June andSeptember respectively, and will be referred to in this waythroughout this paper. Wind conditions over these timeperiods at Northam have been previously shown to be withinthe expected norm and thus represent a suitable test case forthis investigation [34]. TAPM simulates meteorologicalvariables over a regular grid, and we chose the central gridpoint to be our reference. Mean wind speed at this grid pointis output at a reference height of 10 m, and this informationwas also input into the Lagrangian model. The frictionvelocity, temperature and sensible heat flux are generated asatmosphere–land surface variables by TAPM and these werealso used as input into the Lagrangian model.

The Lagrangian model requires four meteorologicalinput values, namely wind speed, sensible heat flux, frictionvelocity and temperature. For each of these four variables,we calculated the mean value for each hour of the dayacross each month. This was done separately for Northamand Merredin, giving four sets of daily profiles. For eachset of profiles, we determined those hours corresponding tosunrise, sunset, maximum temperature and minimumtemperature. Sunrise and minimum daily temperature oftencoincided; however, as this was not always the case, both ofthese times were simulated separately. At both locations,sunrise and sunset were taken to be 7:00 AM and 5:00 PM inJune and 6:00 AM and 6:00 PM in September. For each ofthe selected hours, we took the four meteorologicalvariables from their daily profiles and used these as inputinto the Lagrangian model. For our case study, weparameterised the model to represent the wind-assisted

dispersal of spores released from an infected agricultural crop.This included a terminal velocity of 0.01 ms−1 and a releaseheight of 0.24 m. The terminal velocity chosen is consistentwith published estimates for a number of fungal pathogens[12, 13, 24, 41] while the height of release represents asuitable height for a number of fungal pathogens that infectagricultural crops [3, 18, 30]. We assume that spores arereleased individually, and ignore any clustering of spores,which has been shown to occur in some fungal pathogenssuch as smuts, rusts and powdery mildews [10]. Whileclustered spores have an increased terminal velocity [10], themagnitude of this increase is not likely to significantly affectour results [21]. For the parameterisation of vegetation, weused the values for grassland given in Soons et al. [37],which we assumed to be a suitable representation for anagricultural environment. Each model run was performedwith two million spores. Spores reaching a horizontaldistance of greater than 10 km were assumed to continuetheir journey, but simulation did not occur past this point.

Results

Logarithmic plots of the probability density for distancestravelled by fungal spores released at Northam are shown inFig. 1 and for Merredin in Fig. 2. Since raw values were usedrather than fitted functions, the stochastic nature of thesimulations and the low density of deposited spores at largerdistances results in jagged tails. For each time period tested,at both locations, dispersal stemming from release during thehottest part of the day can be clearly seen to result in a higherprobability of dispersal to greater distances.

Percentages of spores travelling distances greater than 1and 10 km, respectively, are shown in Table 1. Comparisonsbetween different times of the day show that in general,dispersal over 1 and 10 km is more likely during the hottestpart of the day. Release occurring during September at eithersunset or during the hottest part of the day is more likely toresult in dispersal over 1 and 10 km than equivalent releasein June, while the opposite is true for release at sunrise or thecoldest part of the day.

The diurnal pattern of mean horizontal wind speed, frictionvelocity, temperature and sensible heat flux is shown for bothlocations in Figs. 3 and 4, respectively. All four variables canbe seen to peak between 10:00 AM and 5:00 PM during bothJune and September at both locations. At both locations,higher mean wind speeds are observed overnight and in themorning hours in June compared to September. At Merredin,the magnitude of difference in temperature and sensible heatflux between June and September is greater than that forNortham, and Merredin experiences lower maximum tem-peratures over June and a higher maximum sensible heat fluxin September.

580 D. Savage et al.

Discussion

Results presented show that the proportion of fungal sporesundergoing long-distance dispersal changes depending on themonth of the year and the hour of the day at which spores arereleased. This is due to the interaction of seasonal and diurnalpatterns of release with seasonal and diurnal variation inrelevant meteorological variables, in particular, mean hori-zontal wind speed and sensible heat flux.

In the case study presented, the probability of dispersalover large distances is clearly affected by the hour in whichspores are released. Spores released during the hottest part ofthe day are more likely to undergo long-distance dispersalthan those spores released at other times of the day (Figs. 1and 2, Table 1). This is a direct result of peak sensible heatflux and mean horizontal wind speed occurring at, or around,the hottest time of the day (Figs. 3 and 4). Increases in bothof these variables have previously been shown to result in alarger standard deviation in vertical wind speed and adecreased auto-correlation time, increasing the probability

of long-distance dispersal occurring [37]. For tree seedsundergoing wind-assisted dispersal, the proportion of seedsthat experience long-distance dispersal has previously beenshown to be positively correlated with temperature [22], andthis would appear to be the case for fungal spores as well.

For spores released at sunrise or during the coolest partof the day, at both locations used in this study, a greaterproportion of the spores released experienced long-distancedispersal in June than in September (Figs. 1 and 2, Table 1).These greater proportions result from the occurrence ofhigher mean horizontal wind speed during the morninghours in June compared to September (Figs. 3 and 4). Thehigher mean horizontal wind speeds in June overrides theoccurrence of higher sensible heat flux during these sametimes in September, due to the fact that mean horizontalwind speed has a stronger effect on deviations in verticalwind speeds (i.e. turbulence) than sensible heat flux [37].

Spores dispersed at Merredin during the hour ofmaximum temperature can be seen to be more likely toundergo long-distance dispersal than those released at

Figure 1 Logarithmic plot ofthe probability density for thedistance travelled by fungalspores released during the hoursof minimum and maximum tem-perature and sunrise and sunset atNortham, Western Australia, overJune and September 2007

Figure 2 Logarithmic plot of theprobability density for the dis-tance travelled by fungal sporesreleased during the hours of min-imum and maximum temperatureand sunrise and sunset at Merre-din, Western Australia, over Juneand September 2007

Timing of Spore Release Affects Long-Distance Dispersal 581

Northam at the analogous time (Figs. 1 and 2, Table 1).Comparison of the daily meteorological profiles betweenthe two locations (Figs. 3 and 4) shows that maximumtemperature and wind speed over June are far more closelyaligned at Merredin than Northam, resulting in a higherproportion of spores undergoing long-distance dispersal atMerredin than at Northam. Over September, Merredinexperiences slightly higher wind speeds and much highersensible heat flux, again resulting in a higher proportion ofspores undergoing long-distance dispersal. At Merredin, thedifference between the June and September profiles fortemperature and sensible heat flux are much greater than for

Northam (Figs. 3 and 4), and this explains the much greaterincrease in the proportion of spores undergoing long-dispersal between June and September at Merredin(Fig. 2, Table 1).

Results presented in this paper show that seasonal anddiurnal variation in the release of simulated spores canaffect the proportion of the spores released by a particularindividual that would be expected to undergo long-distancedispersal. This is due to the interaction with seasonal anddiurnal variation in relevant meteorological variables. Wespeculate that in reality, since an individual can only releasea limited number of spores, this would represent adifferential capability for long-distance dispersal betweenindividuals who release their spores at different times ofyear and at different times of the day. Consequently, apopulation that benefits from long-distance dispersal wouldbe likely to face selection pressure for release during aspecific month, or specific hours of the day, that maximisesthe proportion, and thus the total number, of sporesundergoing long-distance dispersal. Conversely, a popula-tion that undergoes selection for a particular temporalpattern of release for reasons unrelated to dispersal distance(e.g. differing ability of spores to survive in the light) mayincidentally be shifted towards an increased or decreasedprobability of long-distance dispersal.

If fungal pathogens do indeed undergo adaptations forrelease of spores at particular times, this would be in additionto any other adaptations that influence the likelihood that a

Table 1 Percentage of fungal spores dispersing over a distancegreater than 1 and 10 km, respectively, at Northam and Merredin,Western Australia, in June and September 2007

Northam Merredin

1 km 10 km 1 km 10 km

June Min. temp 3.8 <1 3.2 <1

Max. temp 27.7 2.6 34.6 3.0

Sunrise 3.8 <1 3.2 <1

Sunset 6.8 <1 9.0 1.0

September Min. temp <1 <1 <1 <1

Max. temp 69.2 6.2 99.2 9.0

Sunrise <1 <1 <1 <1

Sunset 7.8 <1 10.2 1.2

Figure 3 Diurnal patterns ofmeteorological variables atNortham, Western Australia, overJune (circles) and September(triangles), 2007. Units used areWm-2 (Q), ms-1 (u and u*),°C (T)

582 D. Savage et al.

particular spore will undergo long-distance dispersal. Inparticular, adaptations relating to the shape, size and surfaceof spores can influence their terminal velocity [9, 11], and aretherefore believed to influence an individual spores ability toundergo long-distance dispersal. Particular mechanisms ofrelease affect the height and initial velocity of released spores,and therefore also affect dispersal capabilities ([2, 4], see also[38]). While small changes to the terminal velocity of fungalspores [23], or its initial velocity, may not necessarilytranslate to a large increase in the distances travelled, thecombined effect of spore characteristics and the timing andmechanism of release may contribute significantly to theoverall likelihood that a particular spore will undergo long-distance dispersal.

For our case study, we simulated the wind-assisteddispersal of fungal spores at Merredin and Northam in Juneand September 2007. However, we suggest that conclusionsdrawn from our results would also apply across differentmonths and years, and across different geographic areas,whenever seasonal and diurnal patterns in the relevantmeteorological variables occur. We also speculate that ourconclusions can be generalised to a range of organisms thatundergo wind-assisted dispersal, have a comparable termi-nal velocity, and also release their propagules according toseasonal or diurnal patterns. This may include the wind-assisted dispersal of some bacteria, pollen, seeds and smallwingless invertebrates.

Results presented in this paper suggest that accuratemodelling of wind-assisted dispersal, not only of fungal

spores but any propagule of similar size, may require thespecific timing of propagule release to be taken intoaccount. In particular, the use of dispersal sub-models inlarger simulations of epidemiology or population dynamicsmay need to be undertaken with parameter sets thataccurately reflect patterns of propagule release, as over- orunder-estimating the tail of dispersal kernels can havesignificant effects on simulation results [7]. Where modelsconsist of dispersal kernels fitted to real dispersal data, theinfluence of release patterns is of course implicitly takeninto account. However, mechanistically parameterisedkernels and models that directly represent mechanisticprocesses may need to be explicitly parameterised torepresent dispersal at a particular time. This could be easilyachieved by simply filtering the input or weighting thecontribution of hourly meteorological variables accordingto the seasonal and diurnal pattern of release.

Concluding Notes

The simulations presented in this paper indicate that thetime at which fungal spores are released can significantlyaffect the proportion of released spores expected to undergolong-distance dispersal. For real populations of fungalpathogens, this effect may have important implications forthe evolution of populations whose survival depends on thetiming of release, or their ability to promote long-distancedispersal. The effect of release time on long-distancedispersal also has important implications for modelling of

Figure 4 Diurnal patterns ofmeteorological variables at Mer-redin, Western Australia, overJune (circles) and September(triangles) 2007. Units used areWm-2 (Q), ms-1 (u and u*),°C (T)

Timing of Spore Release Affects Long-Distance Dispersal 583

dispersal, particularly when dispersal sub-models areincluded in larger simulations of landscape-scale processes.Such models may need to consider the seasonal and diurnalpatterns of propagule release if they are to provide anaccurate description of dispersal and the effect of dispersalcapabilities on disease transmission, biological invasionsand meta-population survival and genetic structure.

Acknowledgements This work was supported by iVEC through theuse of advanced computing resources provided by the AustralianResources Research Centre located in the Western Australia TechnologyPark. The authors acknowledge the support of the Cooperative ResearchCentre for National Plant Biosecurity, established and supported underthe Australian Government’s Cooperative Research Centres Programme,and also the Department of Agriculture and Food Western Australia forhalf the salary of Martin Barbetti.

References

1. Asai GN (1960) Intra- and inter-regional movement of uredo-spores of black stem rust in the Upper Mississippi Valley.Phytopathology 50:535–541

2. Aylor DE (1990) The role of intermittent wind in the dispersal offungal pathogens. Annu Rev Phytopathol 28:73–92

3. Aylor DE, Fry WE, Mayton H, Andrade-Piedra JL (2001) Quanti-fying the rate of release and escape of Phytophthora infestanssporangia from a potato canopy. Phytopathology 91:1189–1196

4. Aylor DE, McCartney HA, Bainbridge A (1981) Deposition ofparticles liberated in gusts of wind. Agr Meteorol 20:1212–1221

5. Aylor DE, Parlange J (1975) Ventilation required to entrain smallparticles from leaves. Plant Pysiol 56:97–99

6. Barrès B, Halkett F, Dutech C, Andrieux A (2008) Geneticstructure of the poplar rust fungus Melampsora larici-populina:evidence for isolation by distance in Europe and recent foundereffects overseas. Infect Genet Evol 8:577–587

7. Bohrer G, Nathan R, Volis S (2005) Effects of long-distancedispersal for metapopulation survival and genetic structure atecological time and spatial scales. J Ecol 93:1029–1040

8. Brown JKM, Hovmøller MS (2002) Aerial dispersal of pathogenson the global and continental scales and its impact on plantdisease. Science 297:537–541

9. Doi M, Makino M, (2005) Sedimentation of particles of generalshape. Phys Fluids 17:043601-043601-7

10. Ferrandino FJ, Aylor DE (1984) Settling speed of clusters ofspores. Phytopathology 74:969–972

11. Gregory PH (1945) The dispersion of air-borne spores. T BritMycol Soc 28:26–72

12. Gregory P (1978) Distribution of airborne pollen and sporesand their long distance transport. Pure Appl Geophys 116:309–315

13. Gregory PH, Stedman OJ (1953) Deposition of air-bornelycopodium spores on plane surfaces. Ann Appl Biol 40:651–674

14. Guo XW, Fernando W (2005) Seasonal and diurnal patternsof spore dispersal by Leptosphaeria maculans from canolastubble in relation to environmental conditions. Plant Dis 89:97–104

15. Hurley PJ, Hill J, Blockley A (2005) An evaluation and inter-comparison of AUSPLUME, CALPUFF and TAPM - part 2:Angelsea and Kwinana annual datasets. Clean Air Environ Qual39:46–51

16. Hurley PJ, Luhar AK (2005) An evaluation and inter-comparison of AUSPLUME, CALPUFF and TAPM - part 1:the Kincaid and Indianapolis field datasets. Clean Air EnvironQual 39:40–45

17. Hurley PJ, Physick WL, Luhar AK (2005) TAPM: A practicalapproach to prognostic meteorological and air pollution model-ling. Environ Modell Softw 20:737–752

18. de Jong MD, Bourdôt GW, Powell J, Goudriaan J (2002) A modelof the escape of Sclerotinia sclerotiorum ascospores from pasture.Ecol Model 150:83–105

19. Knudsen GR (1989) Model to predict aerial dispersal of bacteriaduring environmental release. Appl Environ Microb 55:2641–2647

20. Kot M, Lewis MA, van den Driessche P (1996) Dispersaldata and the spread of invading organisms. Ecology 77:2027–2042

21. Kuparinen A (2006) Mechanistic models for wind dispersal.Trends Plant Sci 11:296–301

22. Kuparinen A, Katul G, Nathan R, Schurr FM (2009) Increases inair temperature can promote wind-driven dispersal and spread ofplants. Proc Biol Sci 276:3081–3087

23. Kuparinen A, Markkanen T, Riikonen H, Vesala T (2007)Modeling air-mediated dispersal of spores, pollen and seeds inforested areas. Ecol Model 208:177–188

24. Li X, Yang X, Mo J, Guo T (2009) Estimation of soybean rusturedospore terminal velocity, dry deposition, and the wetdeposition associated with rainfall. Eur J Plant Pathol 123:377–386

25. Mondal SN, Gottwald TR, Timmer LW (2003) Environmentalfactors affecting the release and dispersal of ascospores ofMycosphaerella citri. Phytopathology 93:1031–1036

26. Nathan R, Katul GG, Horn HS, Thomas SM, Oren R (2002)Mechanisms of long-distance dispersal of seeds by wind. Nature418:409–413

27. Nathan R, Sapir N, Trakhtenbrot A, Katul GG (2005) Long-distancebiological transport processes through the air: can nature’s complex-ity be unfolded in silico. Divers Distrib 11:131–137

28. Pady SM, Kramer CL (1971) Spore discharge in Glomerella. TBrit Mycol Soc 56:81–87

29. Paulitz TC (1996) Diurnal release of ascospores by Gibberellazeae in inoculated wheat plots. Plant Dis 80:674–678

30. Roelfs AP, Martell LB (1984) Uredospore dispersal from apoint source within a wheat canopy. Phytopathology 74:1262–1267

31. Rossi V, Ponti I, Marinelli M, Giosuė S, Bugiani R (2001)Environmental factors influencing the dispersal of Venturiainaequalis ascospores in the orchard air. J Phytopathol 149:11–19

32. Salam MU, Fitt BDL, Aubertot JN, Diggle AJ, Huang YJ,Barbetti MJ, Gladders P, Jedryczka M, Khangura RK, Wratten N,Fernando WGD, Penaud A, Pinochet X, Sivasithamparam K(2007) Two weather-based models for predicting the onset ofseasonal release of ascospores of Leptosphaeria maculans or L.biglobosa. Plant Pathol 56:412–423

33. Salam MU, Khangura RK, Diggle AJ, Barbetti MJ (2003)Blackleg Sporacle: a model for predicting onset of pseudotheciamaturity and seasonal ascospore showers in relation to blackleg ofcanola. Phytopathology 93:1073–1081

34. Savage D, Barbetti MJ, MacLeod WJ, Salam MU, Renton M(2010) Timing of propagule release significantly alters thedeposition area of resulting aerial dispersal. Divers Distrib16:288–299

35. Singh AB, Babu CR (1980) Diurnal periodicity of commonallergenic pollen. Allergy 35:311–317

36. Skelsey P, Kessel GJT, Rossing WAH, van der Werf W(2009) Parameterization and evaluation of a spatiotemporal

584 D. Savage et al.

model of the potato late blight pathosystem. Phytopathology99:290–300

37. Soons MB, Heil GW, Nathan R, Katul GG (2004) Determinants oflong-distance dispersal by wind in grasslands. Ecology 85:3056–3068

38. Sundberg S (2010) Size matters for violent discharge height andsettling speed of Sphagnum spores: Important attributes fordispersal potential. Ann Bot 105:291–300

39. Tackenberg O, Poschlod P, Kahmen S (2003) Dandelion seeddispersal: the horizontal wind speed does not matter for long-distance dispersal - it is updraft! Plant Biol 5:451–454

40. Xu X, Butt D, Ridout U (1995) Temporal patterns of airborneconidia of Podosphaera leucotricha causal agent of applepowdery mildew. Plant Pathol 44:944–955

41. Yarwood C, Hazen W (1942) Vertical orientation of powderymildew conidia during fall. Science 96:316–317

Timing of Spore Release Affects Long-Distance Dispersal 585