Upload
sarah-j
View
212
Download
0
Embed Size (px)
Citation preview
LETTERSPUBLISHED ONLINE: 1 SEPTEMBER 2013 | DOI: 10.1038/NCLIMATE1990
Crop pests and pathogens move polewards in awarming worldDaniel P. Bebber1, Mark A. T. Ramotowski2 and Sarah J. Gurr1*
Global food security is threatened by the emergence and spreadof crop pests and pathogens. Spread is facilitated primarilyby human transportation, but there is increasing concern thatclimate change allows establishment in hitherto unsuitableregions. However, interactions between climate change, cropsand pests are complex, and the extent to which crop pestsand pathogens have altered their latitudinal ranges in responseto global warming is largely unknown. Here, we demonstratean average poleward shift of 2.7±0.8 km yr−1 since 1960, inobservations of hundreds of pests and pathogens, but withsignificant variation in trends among taxonomic groups. Ob-servational bias, where developed countries at high latitudesdetect pests earlier than developing countries at low latitudes,would result in an apparent shift towards the Equator. Theobserved positive latitudinal trends in many taxa support thehypothesis of global warming-driven pest movement.
Since crop domestication 10,000 years ago, farmers have beenplagued by multitudes of pests and pathogens (hereafter termedpests) causing starvation and social upheaval1–4. Classic examplesinclude the 1840s Irish potato famine caused by the oomycetePhytophthora infestans and the 1943 Great Bengal Famine due to thefungus Helminthosporium oryzae3. The threat persists. Between 10and 16% of crop production is lost to pests, with similar losses post-harvest1,4,5. Indeed, losses of major crops to fungi and oomycetesalone amount to enough to feed 8.5% of today’s population2. Thediversity of crop pests is daunting (fungi, bacteria, viruses, viroids,oomycetes, insects and nematodes) and continues to expandthrough evolution and dissemination of new pathotypes2,6–8.Recently emerged strains of the rusts Puccinia graminis andP. striiformis are among the most virulent and rapidly spreadingpathogens ever seen9,10, and a new and invasive lineage ofP. infestanshas rapidly displaced other late blight genotypes11.
Dissemination occurs through both natural and anthropogenicprocesses, facilitated by the increasing interconnectedness of theglobal food chain. More than half of all emerging diseases ofplants are spread by introduction6. Weather is the second mostimportant factor6. For example, fusarium head blight of wheathas re-emerged in the USA, favoured by warm, wet weatherat anthesis5. Insect pests are also influenced by weather, withchewing insects responding negatively to drought and borerspositively12. Warming generally stimulates insect herbivory athigher latitudes, primarily through increased winter survival13,as seen in mountain pine beetle (Dendroctonus ponderosae)outbreaks in the US Pacific Northwest14. The effects of weatherare dependent on both host and pest responses. For example,drought stress can decrease plant resistance15, but infectionprobability is lower in dry conditions16. Although pests are
1Department of Biosciences, University of Exeter, Stocker Road, Exeter EX4 4QD, UK, 2Christ Church College, University of Oxford, St Aldates, Oxford OX11DP, UK. *e-mail: [email protected]
spread by human activities and aerial dispersal6,8, prevailingclimatic conditions are likely to determine their subsequentestablishment and growth.
The influence of weather on crop disease has led to speculationabout the effects of anthropogenic climate change on globalfood security5,6,17,18. Projections are complicated by the interactinginfluences of increasing atmospheric CO2 concentrations, changingclimatic regimes, altered frequency/intensity of extreme weatherevents, and differing responses of the plant and its enemies17–19.However, a general pattern of increasing latitudinal range withmean global temperature is anticipated6, either through directeffects of climate change on the pests, or on the availability ofhost crops. Latitudinal shifts in species distributions, as organismstrack temperature optima, have been detected in thousands of wildpopulations20–22. However, a comprehensive analysis of latitudinalrange shifts of crop pests has not hitherto been attempted. Here,we undertake this analysis using published observations of 612crop pests and pathogens (Supplementary Table S1 and Fig. S1).The data were investigated for the presence of observational biases,caused by latitudinal gradients in the abilities of countries to detect,identify and report pests, and latitudinal trends in observations forindividual pest species.
Identification of reporting biases is central to the analysis oflatitudinal trends in pest observation. The earliest observation fora particular pest in a particular region is equal to the true dateof arrival plus a delay due to observation, identification, reportingand selection of a reliable record for inclusion in the database. In aregression of the latitude of observation against observation date,the regression coefficient will be positive if there is an observedincrease over time, negative if there is a decrease, and statisticallyundifferentiated from zero if no trend is detected. A bias will ariseif the delay period is related to latitude. Scientific and technicalcapacity are greater for countries at high latitudes23, and thesecountries also report more pests (see Supplementary Information).Therefore, countries at high latitudes should report earlier thanlow latitudes, and the regression coefficient of latitude on year ofobservation should be negative in the absence of any real latitudinaltrend in observations.
Two-thirds of pests were observed either solely in the NorthernHemisphere (restricted above 23.4◦N) or northern and tropical(between 23.4◦ S and 3.4◦N) zones for the first decade ofobservations (Supplementary Table S2). Around one-tenth of thepests were found solely outside the tropics, and another tenthwithin the tropics, with the remainder global (in both tropical andextra-tropical zones). Only two pests were restricted to the south(below 23.4◦ S). By the end of the observation period more thanhalf were global in distribution, a third were either northern or
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange 1
© 2013 Macmillan Publishers Limited. All rights reserved.
LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1990
120
100
80
60
40
20
0
0 100 200 300
Longitudinal range (°)
Latit
udin
al r
ange
(°)
Figure 1 | Latitudinal range versus longitudinal range. Grey points showvalues for individual pests in each year. The curves show fitted values (solidline) (±s.e.m., shown by dashed curves) from generalized additive mixedmodels, with random slopes and intercepts for individual pest species. Thestraight dashed line shows the 1:1 relationship for scale.
northern and tropical, and less than one-tenth of the pests wererestricted to the tropical or tropical and southern zones. Of thoseoriginally restricted to the tropics, more than two-thirds spread out,most becoming global. Of those originally found outside the tropics,half were later found in the tropics. The latitudinal range (thedifference between the most-northerly and most-southerly knownlatitudes) for a pest in a given year increased roughly linearly withthe longitudinal range (Fig. 1).However, this occurredmore rapidlyover smaller ranges, such that, on average, the rate of increase wasapproximately equal for latitude and longitude.
Latitudinal trends in observations varied greatly among in-dividual pest species (Supplementary Fig. S2), but taking allspecies together, significant positive latitudinal trendswere detected(Fig. 2). For Northern Hemisphere observations, the Acari, Bacte-ria, Coleoptera, Diptera, Fungi, Hemiptera, Isoptera, Lepidopteraand Oomycota show increased detection towards the north since1960 (Fig. 3). In contrast, Nematoda and viruses show the oppositetrend, towards the Equator. Randomization tests showed that notrend should be detected, if no temporal pattern were present (seeSupplementary Information). Linearmixed-effectsmodels of coun-try (or region) latitude against year of first observation showed anaverage poleward shift in recorded incidences of 2.7±0.8 km yr−1(t -test versus zero, t = 3.3, df = 22,387, p = 0.0009) since 1960for both hemispheres combined, 2.2±0.8 km yr−1 in the NorthernHemisphere (t = 2.7, df = 18,769, p= 0.007) and 1.7±1.7 km yr−1in the SouthernHemisphere (t =1.0, df =3,222, p=0.3).
Linear mixed-effects models were also fitted to detect averagetrends within pest species or pathotypes. For all pests combined,the mean latitudinal shifts were not significant (SupplementaryTable S3), but this seemed to be due to large variability amongpest groups (Supplementary Tables S4 and S5). For all years,observations of Coleoptera and Lepidoptera shifted north in theNorthern Hemisphere, whereas Nematoda and viruses shiftedsouth (Fig. 4). From 1960 onwards, Acari, Coleoptera, Fungi,Hemiptera and Lepidoptera shifted north and Nematoda andviruses shifted south, towards the Equator (Fig. 4). Taking multiplecomparisons into account, significant trends were found in a fewpests (Supplementary Table S6). From 1960 onwards, 12 pests(of which ten were fungi) showed significant trends towards theEquator, and 17 pests (of which six were nematodes) away.
The results indicate significant positive latitudinal shifts formany important groups of crop pests and pathogens. Overall, therehas been a significant trend of increasing numbers of pest and
North
South
32
33
34
35
¬24
1900 1920 1940 1960
Year
1980 2000
¬22
¬20
¬18
Latit
ude
(°)
Figure 2 | Latitude versus year of earliest observation for all pests, in theNorthern and Southern hemispheres. Fitted values (solid line) andstandard errors (dashed lines) are derived from generalized additive mixedmodels of latitude against year of observation.
pathogen observations at higher latitudes, globally and in both theNorthern and Southern hemispheres. The mean shift in detectionsince 1960 (26.6 km per decade) is more rapid than that reportedfor many wild species (17.6 km per decade; ref. 22), but is nearlyidentical to that expected by temperature changes (27.3 km perdecade; ref. 21). Latitudinal variation in countries’ abilities to reportpests would probably bias the data towards earlier detection athigher latitudes. Therefore, the positive trends cannot be explainedby likely latitudinal variation in the ability to detect pathogens.
Overall trends in new observations could include increaseddetection probabilities at high latitudes unrelated to predictors suchas gross domestic product (GDP), or result from real shifts in peststhat have not yet been detected at lower latitudes. Therefore,modelsfor mean shifts within species were also fitted. Within-species shiftswere significant for some groups, particularly ‘mobile invertebratepests such as Lepidoptera, Coleoptera and Hemiptera, but alsoFungi. The viruses and Nematoda showed clear observational shiftstowards the Equator. Both viruses and Nematoda lack the meansfor airborne dispersal, and the trend could therefore be due totrade alone, whereas the aerially dispersed groups exhibit polewardshifts. Other possibilities are that viruses andNematoda are difficultto identify in the field, being soil-borne, and their symptomspotentially misidentifiable as abiotic stresses. Therefore, reportingbias due to latitudinal variation scientific and technical capacitycould explain these negative trends.
It is likely that movements of wild species are hampered byhabitat fragmentation, dispersal limitation, and some by longgeneration times. A climatic debt can be incurred, whereby speciesdo notmove as rapidly as expected given shifting climatic regimes24.In contrast, pathogens have evolved to disperse and grow rapidly,and their spread is facilitated by the global trade in seeds andagricultural produce. It is likely that anthropogenic6 and aerial8dispersal continuously introduce pathogens to new areas, and in
2 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange
© 2013 Macmillan Publishers Limited. All rights reserved.
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1990 LETTERS
1,000
500
0
¬500
¬1,000
1,000
500
0
¬500
¬1,000
1,000
500
0
¬500
¬1,000
Dis
tanc
e fr
om E
quat
or (
km)
All Acari Bacteria Coleoptera Diptera
Fungi Hemiptera Hymenoptera Isoptera Lepidoptera
Nematoda Oomycota Protozoa Thysanoptera Viruses
1960 1980 2000 1960 1980 2000 1960 1980 2000
Year
1960 1980 2000 1960 1980 2000
Figure 3 | Latitude versus year of observation for pest taxonomic groups in the Northern Hemisphere from 1960 onwards. Fits for all pests combined areshown for comparison. Fitted values (solid line) and standard errors (dashed lines) are derived from generalized additive mixed models.
many cases only inclement weather prevents their establishmentin a new habitat. As such, an unwanted assisted colonizationprogramme is taking place for plant pests and pathogens25.
Observed changes in pest distributions accord with observationsof wild species20,22, direct responses of pests to warming14,and with expectations for expanding pest ranges under climatechange6. Although recent climate change is implicated as animportant driver of these observations, other factors could biasthe results. New crop varieties and agricultural technologieshave extended the agricultural margin northward in the USA26,and deforestation has increased production in the tropics, thusproviding new opportunities for pest invasions at high and lowlatitudes. Correlations between land use change and climate changecan obscure analyses based on species temperature ranges20,24,27.Range expansions could be biased in one direction if equatorialbarriers, such as the Sahara desert, were more restrictive to pestmovement than poleward barriers such as permafrost. However,randomization tests demonstrated that no latitudinal shift would beexpected in the absence of a directional temporal trend. Althoughfactors such as land use change do influence species distributions,the influence of such confounding factors decreases in large-scalestudies, and detecting climate signals in noisy data is unlikely in theabsence of real climate drivers20.
Global food security is dependent on numerous physical,agronomic and socioeconomic factors. There is little doubt,however, that climate change and its effects on plant healthwill increasingly threaten human populations, particularly thoseliving in poorer regions1,18,28,29. We have shown that reportedobservations of hundreds of pests and pathogens are consistentwith the hypothesis of climate change drivers, and contrary to thehypothesis of greater detection capability in developed countries.Although countries at higher latitudes are better able to monitorand manage emerging pests and diseases, these countries alsotend to have the greatest productivity per unit land area, andthe threat to food security is troubling. If climate change willmake it easier for crop-destroying organisms to spread, renewedefforts to monitor the occurrence of pests and diseases and control
All 1960 onwards
Lep.
Pro.
Col.
Hem.
Fun.
Hym.
Aca.
Thy.
All.
Vir.
Nem.
Bac.
Iso.
Oom.
Dip.
¬40 ¬20 0 20 40 60 ¬40 ¬20 0 20 40 60Rate (km yr¬1)
Figure 4 |Mean latitudinal shift (km yr−1) for pest taxonomic groups inthe Northern Hemisphere for all years, and for 1960 onwards. Estimatesare from linear mixed-effects models of latitude against observation yearfor centred species-level data. Positive values denote a poleward shift,negative values a shift towards the Equator. Error bars show 95%confidence intervals of the mean. Taxonomic groups are abbreviated, andcombined observations (All) included for comparison. Groups are orderedby the mean of the coefficients.
their transport will be critical in controlling this growing threat toglobal food security1,2,6.
MethodsThe latitudes and dates of the earliest record of 612 crop pests and pathogens wereabstracted from two exhaustive historical databases—the CABI Distribution Mapsof Plant Pests, and of Plant Diseases30 (Supplementary Table S1). The maps areavailable from CABI (www.cabi.org). Pest observations were at country level, andregional for some large countries (USA, Brazil, India, China, Japan, Russia andAustralia); therefore, latitudes of country or region centroids were used in analyses,
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange 3
© 2013 Macmillan Publishers Limited. All rights reserved.
LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1990
to determine whether the latitude of new observations has changed significantlyover time, and whether any shift was consistent either with any observational biasor with the expected effect of climate change.
The presence of a record for a particular geographical region in a given yeardepends on numerous factors, including the presence of the pest, occurrence at adetectable and economically significant level, and scientific and technical capacityto sample, identify and publish a report in a source abstracted by CABI. The dataare therefore likely to suffer from strong observational bias. When considering thepotential impact of climate change on crop pest distributions, any observationalbias linked to latitude must be investigated.
Let L be the latitude of earliest observation, and Do be the year of earliestobservation in the map. Do is not the date of arrival, but includes both the delayin reporting and selection of a record for the map, that is, Do =Da+Dd, whereDa is the true date of arrival, and Dd is a random variable describing the delaybetween arrival and reporting in the map. If there has been a real latitudinal shiftin pest distributions, we propose a relationship E(L)∼ aDa, where the coefficient ais positive. Estimation of a will be biased if delay in observation varies with latitudeE(Dd)∼ bL, such that E(L)∼ aDa+bL, where b is non-zero. If b is positive, thenregression of L on Do will overestimate a, and if b is negative then the regressionwill underestimate a. In other words, if countries at higher latitudes delay reportinglonger than those at low latitudes, it will seem as though pests arrived at higherlatitudes later, and a poleward latitudinal shift could be erroneously inferred. Ifcountries at low latitudes report later, the situation is reversed, and a latitudinalshift towards the Equator would be inferred. On the basis of known biases inspecies observational capacity towards higher latitudes, correlation between percapita GDP and scientific capacity23, increases in per capita GDP with latitude,and increase in pest detection number with latitude, we infer that countries athigher latitudes are likely to have better pest detection capacity, meaning that Dd
is smaller and b< 0. Therefore, the likely bias in observational capacity meansthat a positive latitudinal shift in observations is unlikely in the absence of a realclimate change signal.
The data were tested for the presence of non-Gaussian errors and spatialautocorrelation before linear mixed-effects models and generalized additive mixedmodels were applied, to estimate latitudinal shifts in observations of the entire dataset, taxonomic groups and individual species. Linear mixed-effects models on 1,000randomizations of year against latitude, with pest as a random effect, were usedto remove any temporal trend in the pest observations and thereby determine thelatitudinal shift expected in the absence of a global trend such as climate change.The latitudes of the centroids of countries or regions were used in the analysis.Randomization tests gave an expected latitudinal shift of −0.011±0.017 km yr−1,that is, no significant shift under the null hypothesis of no temporal trendaffecting pest observations.
Full methods are described in the Supplementary Information.
Received 15 December 2012; accepted 31 July 2013;published online 1 September 2013
References1. Flood, J. The importance of plant health to food security. Food Secur. 2,
215–231 (2010).2. Fisher, M. C. et al. Emerging fungal threats to animal, plant and ecosystem
health. Nature 484, 186–194 (2012).3. Strange, R. N. & Scott, P. R. Plant disease: A threat to global food security.
Annu. Rev. Phytopathol. 43, 83–116 (2005).4. Oerke, E-C. Crop losses to pests. J. Agric. Sci. 144, 31–43 (2006).5. Chakraborty, S. & Newton, A. C. Climate change, plant diseases and food
security: An overview. Plant Pathol. 60, 2–14 (2011).6. Anderson, P. K. et al. Emerging infectious diseases of plants: Pathogen
pollution, climate change and agrotechnology drivers. Trends Ecol. Evol. 19,535–544 (2004).
7. Stukenbrock, E. H. & McDonald, B. A. The origins of plant pathogens inagro-ecosystems. Annu. Rev. Phytopathol. 46, 75–100 (2008).
8. Brown, J. K. M. & Hovmøller, M. S. Aerial dispersal of pathogens on theglobal and continental scales and its impact on plant disease. Science 297,537–541 (2002).
9. Singh, R. P. et al. The Emergence of Ug99 races of the stem rust fungus is athreat to world wheat production. Annu. Rev. Phytopathol. 49, 465–481 (2011).
10. Hovmøller, M. S., Yahyaoui, A. H., Milus, E. A. & Justesen, A. F. Rapidglobal spread of two aggressive strains of a wheat rust fungus. Mol. Ecol. 17,3818–3826 (2008).
11. Cooke, D. E. L. et al. Genome analyses of an aggressive and invasive lineage ofthe Irish potato famine pathogen. PLoS Pathog. 8, e1002940 (2012).
12. Koricheva, J., Larsson, S. & Haukioja, E. Insect performance onexperimentally stressed woody plants: a meta-analysis. Annu. Rev. Entomol. 43,195–216 (1998).
13. Bale, J. S. et al. Herbivory in global climate change research: direct effects ofrising temperature on insect herbivores. Glob. Change Biol. 8, 1–16 (2002).
14. Woods, A. Is the health of British Columbia’s forests being influencedby climate change? If so, was this predictable? Can. J. Plant Pathol. 33,117–126 (2011).
15. Mauch-Mani, B. & Mauch, F. The role of abscisic acid in plant–pathogeninteractions. Curr. Opin. Plant Biol. 8, 409–414 (2005).
16. Huber, L. & Gillespie, T. J. Modeling leaf wetness in relation to plant diseaseepidemiology. Annu. Rev. Phytopathol. 30, 553–577 (1992).
17. Garrett, K. A., Dendy, S. P., Frank, E. E., Rouse, M. N. & Travers, S. E. Climatechange effects on plant disease: genomes to ecosystems.Annu. Rev. Phytopathol.44, 489–509 (2006).
18. Gregory, P. J., Johnson, S. N., Newton, A. C. & Ingram, J. S. I. Integrating pestsand pathogens into the climate change/food security debate. J. Exp. Bot. 60,2827–2838 (2009).
19. Shaw, M. W. & Osborne, T. M. Geographic distribution of plant pathogens inresponse to climate change. Plant Pathol. 60, 31–43 (2011).
20. Parmesan, C. & Yohe, G. A globally coherent fingerprint of climate changeimpacts across natural systems. Nature 421, 37–42 (2003).
21. Burrows, M. T. et al. The pace of shifting climate in marine and terrestrialecosystems. Science 334, 652–655 (2011).
22. Chen, I-C., Hill, J. K., Ohlemüller, R., Roy, D. B. & Thomas, C. D. Rapid rangeshifts of species associated with high levels of climate warming. Science 333,1024–1026 (2011).
23. Furman, J. L., Porter, M. E. & Stern, S. The determinants of national innovativecapacity. Res. Policy 31, 899–933 (2002).
24. Devictor, V. et al. Differences in the climatic debts of birds and butterflies at acontinental scale. Nature Clim. Change 2, 121–124 (2012).
25. Hoegh-Guldberg, O. et al. Assisted colonization and rapid climate change.Science 321, 345–346 (2008).
26. Reilly, J. et al. US agriculture and climate change: New results. Climatic Change57, 43–67 (2003).
27. Clavero, M., Villero, D. & Brotons, L. Climate change or land use dynamics: Dowe know what climate change indicators indicate? PLoS ONE 6, e18581 (2011).
28. Schmidhuber, J. & Tubiello, F. N. Global food security under climate change.Proc. Natl Acad. Sci. USA 104, 19703–19708 (2007).
29. Lobell, D. B. et al. Prioritizing climate change adaptation needs for foodsecurity in 2030. Science 319, 607–610 (2008).
30. Pasiecznik, N. M. et al. CABI/EPPO distribution maps of plant pests andplant diseases and their important role in plant quarantine. Eppo Bull. 35,1–7 (2005).
AcknowledgementsThe authors thank Earthwatch and the HSBC Climate Partnership for financiallysupporting D.P.B.
Author contributionsM.A.T.R. collected the data, D.P.B. analysed the data and following discussion withS.J.G., D.P.B. and S.J.G. wrote the paper.
Additional informationSupplementary information is available in the online version of the paper. Reprints andpermissions information is available online at www.nature.com/reprints. Correspondenceand requests for materials should be addressed to S.J.G.
Competing financial interestsThe authors declare no competing financial interests.
4 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange
© 2013 Macmillan Publishers Limited. All rights reserved.