REVIEW RESEARCH BOX I population size to the point at which the species is vulnerable owing to catastrophic collapses as a result of stochastic3z or Allee effects40 Modelling host extinctions caused by pathogenic fungi Long-lived environmental stages Fungi have remarkably resilient dispersal stages (a feature that they share with some spore-forming bacteria,such as Bacillus anthracis).The ability A simple susceptible-infected model shows that the presence of a to survive independently outside of their host,as a free-living saprophyte threshold host population size for disease persistence does not or as durable spores in the environment,is probably the most important prevent host extinction during a disease outbreak,especially in cases feature in driving the emergence of pathogenic fungi,owing to an increased in which a lethal pathogen invades a large host population.In a large risk of transporting the inocula to naive hosts(Fig.2b)".Furthermore, host population transmission is rapid and all hosts can become infected before the host population is suppressed below the threshold. pathogenic fungi with a saprophytic stage(called sapronoses;Fig.2c)can lead to host extirpation because their growth rate is decoupled from host The model follows the dynamics of susceptible (S)and infected ( densities and many fungal diseases threatening natural populations are hosts during the short time duration of an epidemic(deaths that are caused by opportunistic fungi with long-lived environmental stages.Many not due to disease,and births,are ignored):dS/dt=-BS/;dl/ fungi in the phylum Ascomycota are common soil organisms and are dt=BSI-al,where B is the pathogen transmission rate,and a is the tolerant of salinity with the consequence that,when they enter the marine disease-induced death rate.For the parameters shown in Fig.2a,the system through freshwater drainage,they are able to infect susceptible threshold population size below which the pathogen has a negative hosts such as corals (A.sydowii),sea otters (Coccidioides immitis 3)and growth rate is N=20 individuals.Figure 2a shows that large host the nests of loggerhead turtles (Fusarium solani).In terrestrial environ- populations are rapidly driven extinct,but only a fraction of individuals ments,potentially lethal fungi are ubiquitous,such as the causative agent of are killed in small host populations. aspergillosis,Aspergillus fumigatus,and soil surveys have shown that Pathogens with a long-lived infectious stage have an increased potential to cause host extinction.In this model.the disease is Geomyces spp.are common soil organisms.Viable G.destructans has been transmitted through contact between susceptible hosts and free-living recovered from the soil ofinfected bat caves,showing that the pathogen is able to survive and persist in infected roosts when the bats are absent. infectious spores(Z),resulting in infected hosts,I:dS/dt=-BSZ:dl/ dt BSZ-al:dz/dt=-uZ-BNZ,where B is the transmission rate Likewise,long-term persistence of fungal inoculum in the agricultural is the pathogen-induced death rate and is the rate of release of landscape is achieved by quiescent survival on plant debris,such as the spores from infected hosts.Figure 2b shows that fraction of hosts spores of wheat stem rust(Puccinia graminis),which overwinter on straw killed in a disease outbreak increases with the duration of the free- stubble before infecting a secondary host. living infectious spore stage (1/g where u is the spore mortality rate). Saprophytic growth by a pathogen can lead to extinction of the host. Generalist pathogens and opportunistic pathogens and even allow the pathogen to persist in the absence of its host.In this Although many fungi demonstrate extreme host specialization,exem- model,free-living infectious spores are released from infected hosts plified by the gene-for-gene interactions between biotrophic fungi and their plant hosts,broad host ranges twinned with high virulence can be a (with rate )and can increase in abundance through saprophytic growth,with rate a.To illustrate the effects of saprophytic pathogen lethal combination.Fungi exhibit the broadest spectrum of host ranges growth on host and pathogen equilibria(Fig.2c),density-independent for any group of pathogens,and B.dendrobatidis and the oomycete Phytophthora ramorum (the cause of sudden oak death and ramorum host reproduction (with rate b),density-dependent host mortality(with rate do +diN,where N=S+D,and density-dependent spore blight)are known to infect 508 (ref.16)and 109 (ref.3)host species, mortalities (at rate o+Zwere included:dS/dt=bN-(do+dN)S respectively.Different host species vary in their susceptibility to infec- BSZ;dl/dt=BSZ-al-(do +dN)l;dZ/dt=+Z-(uo+u tion and these differences create the potential for parasite-mediated Z-BNZ. competition when the pathogens concerned are generalists.Host The presence of a tolerant host species can lead to the extinction of a species that can tolerate high infection loads while serving as a source susceptible host species.In this model,species A is the tolerant host of infectious stages (known as pathogen spill-over)act as community species,which can become infected and shed infectious spores but 'super spreaders'by maintaining persistent infectious stages in the does not die as a result of the disease,whereas the susceptible host system(Fig.2d).Invasive North American signal crayfish,which tol- species (species B)has a disease-induced per-capita mortality rate of erate infection by the oomycete A.astaci,force the infection into more aB.Figure 2d shows that species B is driven extinct at high densities of susceptible European species that then decline2,and similarly,although species A dSA/dt=bANA-(dAo +dAINA)SA-BASAZ;dlA/ p.ramorum is deadly to Notholithocarpus densiflorus (tanoak)and dt=BASAZ-(dao+dANA)IA:dSa/dt baNA-(dao+dBiNs) several Quercus species,many of its other hosts survive infection but SB-BBSBZ;dlB/dt BBSBZ-(dso+d8NB)8-BlB:dZ/ generate inoculum themselves for new infections.Furthermore,disease- dt=lA+-uZ-BANAZ-BaNBZ,where all parameters are as tolerant life-history stages of otherwise susceptible species can maintain previously defined,but with the subscripts A or B referring to host high pathogen levels leading to extinction dynamics.In chytridiomycosis, species A or B,respectively the long-lived multi-year tadpole stages of amphibians such as the mountain yellow-legged frog Rana muscosa and the midwife toad Alytes obstetricans are not killed by chytrid infections,but they can B.dendrobatidis in amphibians,G.destructans in bats and Ophiostoma build up high loads of B.dendrobatidis that can infect and overwhelm ulmi in elm trees).Virulence is a measure of the relative capacity of a juvenile metamorphs of the same species,leading to rapid population microbe to cause damage to a host,and high virulence is associated with loss2.Ultimately,when host-generalist pathogens manifest long-lived rapid intra-host growth rates,ultimately leading to rapidinter-host trans- environmental stages,conditions may occur that lead to long-distance mission"Fungi have a high reproductive potential and in a large host dispersal and infection of naive hosts and environments3. population this effect can result in all individuals becoming infected before the population is driven to the low densities at which the pathogen Trade and transport promotes globalization of fungi can no longer spread(Fig.2a).Thus,host extirpation can occur before Fungi comprise most of the viable biomass in the air,with an average density dependence limits the rate of transmission,a feature that has human breath containing between one and ten fungal spores4.This contributed to the mass extirpations seen in frog populations across ability of fungi to disperse results in some species with cosmopolitan the US Sierra Nevada mountains".Similarly,even if the pathogen does distributions However,these species are in the minority and it is not drive the host to complete extinction,it may severely reduce the noticeable that few fungi exhibit truly globally distributions;instead they 12 APRIL 2012 VOL 484 NATURE 189 2012 Macmillan Publishers Limited.All rights reservedB. dendrobatidis in amphibians, G. destructans in bats and Ophiostoma ulmi in elm trees). Virulence is a measure of the relative capacity of a microbe to cause damage to a host36, and high virulence is associated with rapid intra-host growth rates, ultimately leading to rapid inter-host trans￾mission37,38. Fungi have a high reproductive potential and in a large host population this effect can result in all individuals becoming infected before the population is driven to the low densities at which the pathogen can no longer spread (Fig. 2a). Thus, host extirpation can occur before density dependence limits the rate of transmission, a feature that has contributed to the mass extirpations seen in frog populations across the US Sierra Nevada mountains39. Similarly, even if the pathogen does not drive the host to complete extinction, it may severely reduce the population size to the point at which the species is vulnerable owing to catastrophic collapses as a result of stochastic32 or Allee effects40. Long-lived environmental stages Fungi have remarkably resilient dispersal stages (a feature that they share with some spore-forming bacteria, such as Bacillus anthracis). The ability to survive independently outside of their host, as a free-living saprophyte or as durable spores in the environment, is probably the most important featurein driving the emergence of pathogenicfungi, owing to an increased risk of transporting the inocula to naive hosts (Fig. 2b)41. Furthermore, pathogenic fungi with a saprophytic stage (called sapronoses; Fig. 2c) can lead to host extirpation because their growth rate is decoupled from host densities and many fungal diseases threatening natural populations are caused by opportunistic fungi with long-lived environmental stages. Many fungi in the phylum Ascomycota are common soil organisms and are tolerant of salinity with the consequence that, when they enter the marine system through freshwater drainage, they are able to infect susceptible hosts such as corals (A. sydowii42), sea otters (Coccidioides immitis43) and the nests of loggerhead turtles (Fusarium solani44). In terrestrial environ￾ments, potentially lethalfungi are ubiquitous, such as the causative agent of aspergillosis, Aspergillus fumigatus, and soil surveys have shown that Geomycesspp. are common soil organisms. Viable G. destructans has been recoveredfrom the soil of infected bat caves45, showing that the pathogen is able to survive and persist in infected roosts when the bats are absent. Likewise, long-term persistence of fungal inoculum in the agricultural landscape is achieved by quiescent survival on plant debris, such as the spores of wheat stem rust (Puccinia graminis), which overwinter on straw stubble before infecting a secondary host. Generalist pathogens and opportunistic pathogens Although many fungi demonstrate extreme host specialization, exem￾plified by the gene-for-gene interactions between biotrophic fungi and their plant hosts, broad host ranges twinned with high virulence can be a lethal combination. Fungi exhibit the broadest spectrum of host ranges for any group of pathogens, and B. dendrobatidis and the oomycete Phytophthora ramorum (the cause of sudden oak death and ramorum blight) are known to infect 508 (ref. 16) and 109 (ref. 3) host species, respectively. Different host species vary in their susceptibility to infec￾tion and these differences create the potential for parasite-mediated competition when the pathogens concerned are generalists46. Host species that can tolerate high infection loads while serving as a source of infectious stages (known as pathogen spill-over) act as community ‘super spreaders’ by maintaining persistent infectious stages in the system (Fig. 2d). Invasive North American signal crayfish, which tol￾erate infection by the oomycete A. astaci, force the infection into more susceptible European species that then decline25, and similarly, although P. ramorum is deadly to Notholithocarpus densiflorus (tanoak) and several Quercus species47, many of its other hosts survive infection but generate inoculum themselves for new infections. Furthermore, disease￾tolerant life-history stages of otherwise susceptible species can maintain high pathogen levels leading to extinction dynamics. In chytridiomycosis, the long-lived multi-year tadpole stages of amphibians such as the mountain yellow-legged frog Rana muscosa and the midwife toad Alytes obstetricans are not killed by chytrid infections, but they can build up high loads of B. dendrobatidis that can infect and overwhelm juvenile metamorphs of the same species, leading to rapid population loss39. Ultimately, when host-generalist pathogens manifest long-lived environmental stages, conditions may occur that lead to long-distance dispersal and infection of naive hosts and environments5 . Trade and transport promotes globalization of fungi Fungi comprise most of the viable biomass in the air, with an average human breath containing between one and ten fungal spores48. This ability of fungi to disperse results in some species with cosmopolitan distributions5,49,50. However, these species are in the minority and it is noticeable that few fungi exhibit truly globally distributions; instead they BOX 1 Modelling host extinctions caused by pathogenic fungi A simple susceptible–infected model shows that the presence of a threshold host population size for disease persistence does not prevent host extinction during a disease outbreak, especially in cases in which a lethal pathogen invades a large host population. In a large host population transmission is rapid and all hosts can become infected before the host population is suppressed below the threshold. The model follows the dynamics of susceptible (S) and infected (I) hosts during the short time duration of an epidemic (deaths that are not due to disease, and births, are ignored): dS/dt 5 2 bSI; dI/ dt 5 bSI 2 aI, where b is the pathogen transmission rate, and a is the disease-induced death rate. For the parameters shown in Fig. 2a, the threshold population size below which the pathogen has a negative growth rate is NT5 20 individuals. Figure 2a shows that large host populations are rapidly driven extinct, but only a fraction of individuals are killed in small host populations. Pathogens with a long-lived infectious stage have an increased potential to cause host extinction. In this model, the disease is transmitted through contact between susceptible hosts and free-living infectious spores (Z), resulting in infected hosts, I: dS/dt 5 2bSZ; dI/ dt 5 bSZ 2 aI; dZ/dt 5 wI 2 mZ 2 bNZ, where b is the transmission rate, a is the pathogen-induced death rate and w is the rate of release of spores from infected hosts. Figure 2b shows that fraction of hosts killed in a disease outbreak increases with the duration of the free￾living infectious spore stage (1/m, where m is the spore mortality rate). Saprophytic growth by a pathogen can lead to extinction of the host, and even allow the pathogen to persist in the absence of its host. In this model, free-living infectious spores are released from infected hosts (with rate w), and can increase in abundance through saprophytic growth, with rate s. To illustrate the effects of saprophytic pathogen growth on host and pathogen equilibria (Fig. 2c), density-independent host reproduction (with rate b), density-dependent hostmortality (with rate d01 d1N, where N 5 S 1 I), and density-dependent spore mortalities (at rate m01 m1Z) were included: dS/dt 5 bN 2 (d0 1 d1N) S 2 bSZ; dI/dt 5 bSZ 2 aI 2 (d01 d1N)I; dZ/dt 5 wI 1 sZ 2 (m0 1 m1Z) Z 2 bNZ. The presence of a tolerant host species can lead to the extinction of a susceptible host species. In this model, species A is the tolerant host species, which can become infected and shed infectious spores but does not die as a result of the disease, whereas the susceptible host species (species B) has a disease-induced per-capita mortality rate of aB. Figure 2d shows that species B is driven extinct at high densities of species A. dSA/dt 5 bANA2 (dA01 dA1NA)SA2 bASAZ; dIA/ dt 5 bASAZ 2 (dA0 1 dA1NA)IA; dSB/dt 5 bBNA 2 (dB01dB1NB) SB 2 bBSBZ; dIB/dt 5 bBSBZ 2(dB01 dB1NB)IB 2 aBIB; dZ/ dt 5 wAIA1 wBIB 2 mZ 2 bANAZ 2 bBNBZ, where all parameters are as previously defined, but with the subscripts A or B referring to host species A or B, respectively. REVIEW RESEARCH 12 APRIL 2012 | VOL 484 | NATURE | 189 ©2012 Macmillan Publishers Limited. All rights reserved