match his song types with those of hisneighbours, even when he settles into a newpopulation, whereas the female-typical sen-sitive phase preserves her natal dialect. Songmatching may be important for male cardi-nals in the context of establishing territo-ries, as in other species9, but evidently is notfor females, for reasons yet to be identified.These findings parallel subtle but consistentgender differences in human speech acqui-sition10 and offer an opportunity to studythe biological basis of distinctions in maleand female learning.Ayako YamaguchiAnimal Communication Laboratory, University ofCalifornia, Davis, California 95616, USAe-mail: ay64@columbia.eduPresent address: Department of Pharmacology, YaleUniversity School of Medicine, 333 Cedar Street,New Haven, Connecticut 06520, USA

1. Marler, P. & Tamura M. Science 146, 1483–1486 (1964).2. Marler, P. J. Comp. Physiol. Psychol. Monogr. 71, 1–25 (1970).3. Nelson, D. A., Marler, P. & Palleroni, A. P. Anim. Behav. 50,

83–97 (1995).4. Eales, L. A. Anim. Behav. 35, 1336–1365 (1987).5. Yamaguchi, A. Condor 100, 504–511 (1998).6. Nelson, D. A. Anim. Behav. 56, 487–491 (1998).7. Lemon, R. E. & Scott, D. M. Can. J. Zool. 44, 191–197 (1965).8. Smith, P. C. Survival and Dispersal of Juvenile Cardinals.

Thesis, Univ. Western Ontario (1969).9. Beecher, M. D., Campbell, S. E. & Stoddard, P. K. Proc. Natl

Acad. Sci. USA 91, 1450–1454 (1994).10.Fenson, L. et al. Monogr. Soc. Res. Child Dev. 59, 1–185 (1994).

mass, because microbes can be killed byosmotic shock and cell rupture when rapidrehydration follows a period of drying1.

Phosphorous released from microbesafter drying and rewetting the soil couldcontribute to the pollution of water bodies,because soil drying coincides with cracking,which enhances preferential movement ofwater and associated phosphorus throughthe soil. This may explain the increasedconcentrations of phosphorus found indrainage waters after periods of soil dryingand rewetting2. The impact of this effect onwater quality may be increased by irrigationpractices and in regions where climatechange causes longer dry periods or morefrequent cycles of wetting and drying3.

Our results also have implications forthe usefulness of environmental soil-phosphorus tests that measure only inor-ganic phosphorus, rely on aqueous extrac-tion, and take no account of soil moisture4.The effect shown here may also apply toother soil nutrients and could partiallyexplain a similar phenomenon observed fornitrogen mineralization in tropical soils5.Benjamin L. Turner*†, Philip M. Haygarth**Soil Science Group, Institute of Grassland andEnvironmental Research, North Wyke,Okehampton, Devon EX20 2SB, UKe-mail: phil.haygarth@bbsrc.ac.uk†Present address: USDA–ARS, Northwest Irrigationand Soils Research Laboratory, Kimberly, Idaho83341, USA1. Salema, M. P., Parker, C. A., Kidby, D. K., Chatel, D. L. &

Armitage, T. M. Soil Biol. Biochem. 14, 15–22 (1982).

2. Turner, B. L. & Haygarth, P. M. Soil Sci. Soc. Am. J. 64,

1090–1097 (2000).

3. Houghton, J. T., Jenkins, G. J. & Ephraums, J. J. Climate Change.

The IPCC Scientific Assessment (Cambridge Univ. Press,

Cambridge, 1992).

4. Pote, D. H. et al. J. Environ. Qual. 28, 170–175 (1999).

5. Birch, H. F. Plant Soil 12, 81–96 (1960).

6. Brookes, P. C., Powlson, D. S. & Jenkinson, D. S. Soil Biol.

Biochem. 14, 319–329 (1982).

Epidemiology

Foot-and-mouth diseaseunder control in the UK

Following the first reported case on 20February this year, foot-and-mouthdisease spread to over 1,500 livestock

farms in the United Kingdom by the end ofApril1–3. From late March, the Ministry ofAgriculture, Fisheries and Food (MAFF)required livestock on infected farms to beculled within 24 hours of the disease beingreported and those on neighbouring farmswithin 48 hours. Here we investigatewhether progress towards meeting thesetargets3 has had a detectable impact on thecourse of the epidemic in the United King-dom. We conclude that it has now beenbrought under control, but it will be impor-tant to contain rapidly any new outbreaks

258 NATURE | VOL 411 | 17 MAY 2001 | www.nature.com

in life may be necessary to produce songs infemales, but not in males. In summary,male and female cardinals differed marked-ly in when they acquired songs, but weresimilar in what they acquired. The sensitivephase estimated by using tape tutoring gen-erally agrees with that measured using livetutoring6 and is likely to guide song acquisi-tion in wild birds; the sensitive phase ofwild male cardinals is entirely consistentwith my results7.

The different sensitive phases for thesexes may be an adaptation to meet thefunctional requirements of songs. Juvenilecardinals disperse from their parents’ terri-tory to establish their own territory betweenthe time they become independent fromtheir parents (about 45 days of age) and thestart of the first breeding season (when theyare about 8.5 to 11 months old; Fig. 1b)8.My results show that, although both sexesstart learning songs in their natal territory,the females lose the ability to acquire newsongs soon after independence and malescontinue to learn songs until after dispersalis complete.

Songs of adult cardinals form ‘dialects’in different geographical regions, so dis-persing juveniles may settle into a popula-tion with a foreign dialect. The male-typicalsensitive phase therefore allows him to

brief communications

Biogeochemistry

Phosphorus solubilizationin rewetted soils

Biogeochemical cycles are shaped byevents that follow soil drying andrewetting. Here we show that the

process of drying and rapidly rewetting soilincreases the amount of water-solublephosphorus present and that this is pre-dominantly in organic form after havingbeen released from the soil microbial biomass. This effect could not only signifi-cantly affect phosphorus pollution of water-bodies but might also corrupt results fromanalyses involving water extraction of driedsoils.

We investigated the effect of drying andrewetting soil on its content of solublephosphorus by determining the amount ofwater-soluble phosphorus in 29 lowlandpermanent grassland soils from Englandand Wales. The total amounts of water-soluble phosphorus in moist soils weresmall, but increased after drying by185–1,900%. Comparable amounts ofphosphorus were solubilized when soilswere dried at different temperatures, butsolubilization occurred more rapidly at 30 7C (maximum phosphorus solubiliza-tion after 1–3 days) than at 15 7C (maxi-mum solubilization after 7–10 days).

Most of the soluble phosphorus was pre-

sent in an organic form, which accountedfor 56–100% of the total phosphorusreleased by drying at 30 °C. The amount ofsolubilized organic phosphorus was posi-tively related to the soil microbial phospho-rus concentration (P*0.0001; Fig. 1). Weattributed this effect to direct release ofphosphorus from the soil microbial bio-

Figure 1 Increase in water-soluble organic phosphorus after soil

drying as a function of soil microbial phosphorus. Data are from

29 permanent grassland soils in England and Wales containing

various amounts of phosphorus (NaHCO3-extractable inorganic

phosphorus was 9–48 mg kg11), total carbon (29–80 g kg11)

and clay (22–68%). Increase in water-soluble organic phosphorus

4(0.9550.33)&(0.01850.003) (microbial phosphorus);

R 240.58; n429; P41.621016. The amount of water-soluble

phosphorus was determined by extracting soils at field-moisture

capacity (after drainage from saturation for 48 h) with water in a

4:1 water/soil ratio for 1 h. Subsamples were air-dried for 7 days

at 30 7C and extracted in the same way. Extracts were filtered

(pore size*0.45 mm) and analysed for inorganic and organic

phosphorus2. Phosphorus levels in soil microbial biomass were

determined by measuring the phosphorus released by chloroform

fumigation6.

0 125 2500

3

6

Incr

ease

in w

ater

-sol

uble

orga

nic

P (µ

g p

er g

dry

soi

l)

Microbial P (µg per g dry soil)

© 2001 Macmillan Magazines Ltd

in previously unaffected areas.A key parameter describing the state of

an epidemic is the case-reproduction ratio4,R, where R is the average number of newcases generated by each case (a case isdefined here as the diagnosis of foot-and-mouth disease on a single farm). Only if Ris less than 1 can the epidemic be regardedas ‘under control’5. We estimated R directlyusing a combination of demographic andcontact tracing methods6–8. The key inputsare MAFF data giving the estimated date offirst infection (based on clinical assessment— age of oldest lesions and known incuba-tion periods in different livestock species —and epidemiological information — prox-imity and date of infection of potentialsources) and the exact location of each casediagnosed and reported1. For some cases,MAFF identified the most probable sourceof infection using a combination of report-ed contacts with infected farms and (rarely)predicted patterns of aerial spread1. Weused the nearest infectious farm as the mostprobable source of infection for the remain-ing cases. We corrected R values for censor-ing (that is, new cases not yet reported) andallowed for the culling of at-risk farms (the‘extended cull’; Fig. 1 legend).

The results indicate that R was initiallyhigh (3.3), reflecting extensive movementsof infected livestock around the UnitedKingdom before restrictions were imposedon 23 February, after which R fell, but not

to below 1 (Fig. 1). Reduced average report-ing-to-slaughter intervals (from 2.9 days inlate February to 1.1 days in early April —together with reporting occurring an aver-age of 0.7 days earlier), and intensificationof the extended cull (from less than oneadditional farm culled per reported casebefore April to more than four duringApril) should have reduced R, but theeffects would not have been immediate.However, the extended cull also had animmediate effect on reported incidence byremoving incubating cases before theycould be diagnosed. The true impact ofincreased culling effort was therefore notseen until cases generated by farms infectedin late March were reported. Since thattime, R has been below 1 for all Britain andfor the three most affected regions (Fig. 1),implying that the epidemic is in decline.The expected rate of decline in incidence isrelated to the mean generation time (theaverage interval between infection of sourceand new cases, estimated as 7.6 days): forexample, R=0.65 (the overall value in lateMarch) corresponds to incidence halvingevery 12 days, consistent with subsequentlyreported trends.

The reduction in R and correspondingfall in the incidence of new cases was con-siderably greater than that recently predict-ed using a mathematical model9. Otherfactors that may have contributed includelocal ‘burn out’ of susceptible farms, more

brief communications

NATURE | VOL 411 | 17 MAY 2001 | www.nature.com 259

susceptible farms being affected earlier thanless susceptible farms8, and changes inunderlying patterns of transmission.

Long-term projections are very sensitiveto any further changes in R, and any inten-sification or relaxation of control effortscould greatly affect the final scale and dura-tion of the epidemic. We note that theexpected course of the epidemic is also verysensitive to the introduction of disease intopreviously unaffected areas and it is impor-tant that new outbreaks, especially in areaswith high livestock densities, are containedrapidly and effectively.Mark Woolhouse*, Margo Chase-Topping*,Daniel Haydon†, John Friar‡, Louise Matthews*, Gareth Hughes*§,Darren Shaw*, John Wilesmith||¶, Alex Donaldson§, Stephen Cornell#,Matthew Keeling#, Bryan Grenfell#*Centre for Tropical Veterinary Medicine,University of Edinburgh, Easter Bush, Roslin,Midlothian EH25 9RG, UK†Department of Zoology, University of Guelph,Guelph, Ontario N1G 2W1, Canada‡Eoscene Corporation, 700 5th Avenue, Seattle,Washington 98104, USA§Institute for Animal Health Pirbright Laboratory,Ash Road, Pirbright, Surrey GU24 0NF, UK||Veterinary Laboratories Agency, New Haw,Addlestone, Surrey KT15 3NB, UK¶Department of Infectious and Tropical Diseases,London School of Hygiene and Tropical Medicine,Keppel Street, London WC1E 7HT, UK#Department of Zoology, University of Cambridge,Downing Street, Cambridge CB2 3EJ, UK1. Anon. Vet. Rec. 148, 286–288 (2001).

2. Anon. Vet. Rec. 148, 390–393 (2001).

3. http://www.maff.gov.uk

4. Anderson, R. M. & May, R. M. Infectious Diseases of Humans:

Dynamics and Control (Oxford Science, Oxford, 1991).

5. Woolhouse, M. & Donaldson, A. Nature 410, 515–516 (2001).

6. Patil, G. P., Pielou, E. C. & Waters, W. E. (eds) Statistical Ecology

Vol. II (Pennsylvania State Univ. Press, 1971).

7. Sellers, R. F., Garland, A. J. M., Donaldson, A. I. & Gloster, J.

Br. Vet. J. 137, 608–620 (1981).

8. Haydon, D. T., Woolhouse, M. E. J. & Kitching, R. P. IMA J.

Math. Appl. Med. Biol. 14, 1–9 (1997).

9. Ferguson, N., Donnelly, C. & Anderson, R. Science 13 April

2001 (http://www.scienceexpress.org).

Genomics

Annotation of theDrosophila genome

The publication of the genome of thefruitfly Drosophila by Celera Genomicslast year generated considerable excite-

ment1. The initial release predicted 14,226proteins, including about 550 that arise asthe result of alternative splicing1. Here wecompare the annotation of the CeleraDrosophila genome with known proteinsequences lodged in a databank and findthat although there is substantial agreementfor many gene sequences, there are alsonumerous and significant discrepancies.These uncertainties call for caution in

Figure 1 Changes through time in the case-reproduction ratio, R, for all of Britain and for the three most affected regions based on case

report data up to 29 April. R values are for populations not samples. We estimated numbers of new cases not yet reported (censored)

using the distribution of the interval between infection of a source case and reporting of cases infected from it, using negative binomially

distributed variances to generate 95% confidence limits for R values corrected for censoring (vertical lines). We also adjusted R values to

allow for the effect of culling incubating cases using the estimated proportion of total cases represented by at-risk farms culled (hatching).

Incorrect infection dates – these could be earlier (rarely later) than estimated if clinical signs do not develop or are not reported – would

lead to overestimation of recent R values. Incorrect infection sources would not affect average R values but could affect confidence limits.

2.0

1.5

1.0

0.5

0.0

24 Feb – 2 Mar 3 – 9 Mar 10 – 16 Mar 17 – 23 Mar 24 – 30 Mar 31 Mar – 6 Apr 7 – 13 Apr

All Britain Devon Cumbria Dumfries and Galloway

Infection date

Cas

e re

pro

duc

tion

ratio

, R

© 2001 Macmillan Magazines Ltd