Defining species boundaries in the genus Phytophthora: the case of Phytophthora andina A response to ‘Phytophthora andina sp. nov., a newly identified heterothallic pathogen of solanaceous hosts in the Andean highlands’ (Oliva et al., 2010)



Phytophthora is a genus comprising upwards of 100 species. The most recent phylogeny of the genus was inferred from seven nuclear genes (Blair et al., 2008) and was subdivided into 10 well-defined clades that are concordant with the morphology-based Waterhouse criteria (Waterhouse, 1963; Kroon, 2010). The description of new species in this genus has traditionally been based on morphological characters, ecology, geography and host specialization. Although these traits are useful in the nomenclatural recognition of different species, they are not necessarily correlated to their evolutionary history and thus should not be used alone as characters to define new species. It is now routine that new Phytophthora species descriptions also include a detailed phylogenetic analysis typically based on several loci. In an evolutionary context, reproductive isolation of a species from other close relatives is a fundamental requirement of the biological species concept (Mayr, 1942). The use of molecular markers, traditionally the nuclear internal transcribed spacer (ITS) region, has allowed recognition of species complexes such as P. citricola and P. megasperma. In species complexes the definition of species boundaries is difficult. When is a species in a lineage differentiated enough to be considered as a separate one (Mishler, 1985; Avise & Ball, 1990)? We believe that careful analysis of several nuclear and mitochondrial loci is needed to support the definition of species boundaries. Here we examine problems of defining species boundaries using the example of the recently proposed new species, Phytophthora andina.

The newly proposed species P. andina, first found in Ecuador, is a relative of the potato late blight pathogen P. infestans (Oliva et al., 2010). The proposed species has a narrower host range and does not infect tomato or potato in the field. In contrast, P. infestans infects potato (Solanum tuberosum) and other members of the Solanaceae family (i.e. S. phureja, S. lycopersicum, S. betaceum, S. melongena, S. quitoense and Physalis peruviana). To date, four clonal lineages have been described for P. infestans in South America: US-1, EC-1, EC-2 and EC-3 (Forbes et al., 1997, 1998; Ordoñez et al., 2000; Adler et al., 2004). Based on the phylogenetic analyses of the CoxI gene and the intron 1 of the Ras gene, Gómez-Alpizar et al. (2008) proposed that isolates with the Ic mitochondrial haplotype from the EC-2 lineage corresponded to a new species, P. andina. In the same study, the P. andina EC-3 isolates were found to be different from the P. infestans EC-1 lineage but were described as P. infestans EC-3 isolates without assuming a new species.

More recently, Oliva et al. (2010) formally describe P. andina, lineages EC-2 and EC-3, as a distinct taxonomic species based on three types of evidence. First was the fact that these Ecuadorian isolates were found causing disease on wild Solanum spp. different from those that are hosts for P. infestans. Secondly, an AFLP analysis clearly separated both species into different clades by means of neighbour joining with 100% bootstrap support. However, the third criterion was a phylogenetic analysis that, as we argue here, was less compelling than the first two lines of evidence mentioned above. We argue in this letter that the available data are not yet sufficient for description of a new species.

We believe that the CoxII phylogeny presented in Figure 5 of Oliva et al. (2010) does not support P. andina as a monophyletic species. In fact, the proposed P. andina is placed into two separate clades with significant bootstrap support. As a result, we argue that the species description for the proposed P. andina is incomplete and more sequence-based work is required to conclusively describe P. andina as a unique taxon. It may be that the two clades are each a new species, so that the current description is premature. The phylogenetic species concept requires a monophyletic grouping and the congruence among different gene genealogies to avoid inferring conclusions from a gene tree instead of a species tree (Cracraft, 1989; Taylor et al., 2000). Furthermore, a novel Phytophthora species description should be based ideally on more than one sequenced locus and should include both the mitochondrial and nuclear genomes. Including nuclear genes is important because recombination events subsequent to speciation can only be detected in nuclear data, as mitochondrial regions are maternally inherited (Nichols, 2001).

Seventy-seven P. infestans isolates from Colombia, Ecuador and Venezuela were used (Vargas et al., 2009; Cardenas et al., 2011), and four proposed P. andina isolates from Ecuador (isolate E74 collected in 2001 and also used by Oliva et al. (2010); E77 in 2003; E82 in 2003; E83 in 2004), to sequence two nuclear and one mitochondrial locus in order to determine the phylogenetic relationships between P. infestans and the newly proposed species P. andina. Phytophthora andina isolates were kindly provided by M. Coffey from the World Oomycete Genetic Resource Collection Database. Sequences for each locus were obtained from databases for P. phaseoli, P. citricola, P. brassicae, P. mirabilis and P. ipomoeae to find out if the proposed P. andina is really a different species under the phylogenetic species concept (Table 1). Additionally, the sequences were used to establish if there is potential for cytonuclear discordance between the nuclear and mitochondrial genomes (Seehausen, 2004).

Table 1.   GenBank accession numbers, strains and locations of Phytophtora infestans and the new proposed species Phytophthora andina used in this study
Strain/GB accession numbersSpeciesCountryLocationHost
E74/P13400P. andinaEcuadorTungurahuaDatura bicolor
E77/P13660P. andinaEcuadorTungurahuaBrugmansia sanguinea
E82/P13648P. andinaEcuadorTungurahuaAnarrhichomenum
E83/P13803P. andinaEcuadorPinchinchaSolanum betaceum
COASB25P. infestansColombiaRionegro–AntioquiaS. betaceum
COASB27P. infestansColombiaRionegro–AntioquiaS. betaceum
COASQ21P. infestansColombiaRionegro–AntioquiaS. quitoense
COASQ22P. infestansColombiaRionegro–AntioquiaS. quitoense
COASQ23P. infestansColombiaRionegro–AntioquiaS. quitoense
COASQ24P. infestansColombiaRionegro–AntioquiaS. quitoense
COAST51P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST52P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST54P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST55P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST57P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST58P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST59P. infestansColombiaRionegro–AntioquiaS. tuberosum
COAST60P. infestansColombiaRionegro-AntioquiaS. tuberosum
COAST61P. infestansColombiaRionegro–AntioquiaS. tuberosum
COBST62P. infestansColombiaToca–BoyacáS. tuberosum
COBST63P. infestansColombiaToca–BoyacáS. tuberosum
COBST64P. infestansColombiaToca–BoyacáS. tuberosum
COBST65P. infestansColombiaToca–BoyacáS. tuberosum
COBST66P. infestansColombiaToca–BoyacáS. tuberosum
COBST67P. infestansColombiaToca–BoyacáS. tuberosum
COBST68P. infestansColombiaToca–BoyacáS. tuberosum
COBST70P. infestansColombiaToca–BoyacáS. tuberosum
COBST71P. infestansColombiaToca–BoyacáS. tuberosum
COBST72P. infestansColombiaToca–BoyacáS. tuberosum
COBST73P. infestansColombiaToca–BoyacáS. tuberosum
COCPP16P. infestansColombiaSan Francisco–CundinamarcaPhysalis peruviana
COCPP17P. infestansColombiaSan Francisco–CundinamarcaP. peruviana
COCPP18P. infestansColombiaGranada–CundinamarcaP. peruviana
COCPP19P. infestansColombiaGranada–CundinamarcaP. peruviana
COCPP20P. infestansColombiaGranada–CundinamarcaP. peruviana
COCPP28P. infestansColombiaGranada–CundinamarcaP. peruviana
COCSB26P. infestansColombiaGranada–CundinamarcaS. betaceum
COCSB29P. infestansColombiaGranada–CundinamarcaS. betaceum
COCSL11P. infestansColombiaVillapinzón–CundinamarcaS. lycopersicum
COCSL12P. infestansColombiaFusagasugá–CundinamarcaS. lycopersicum
COCSL13P. infestansColombiaTibacuy–CundinamarcaS. lycopersicum
COCSL14P. infestansColombiaBogotá–CundinamarcaS. lycopersicum
COCSL15P. infestansColombiaTibacuy–CundinamarcaS. lycopersicum
COCSP10P. infestansColombiaCogua–CundinamarcaS. phureja
COCSP7P. infestansColombiaCogua–CundinamarcaS. phureja
COCSP8P. infestansColombiaCogua–CundinamarcaS. phureja
COCSP9P. infestansColombiaCogua–CundinamarcaS. phureja
COCST1P. infestansColombiaZipaquirá–CundinamarcaS. tuberosum
COCST2P. infestansColombiaCogua–CundinamarcaS. tuberosum
COCST3P. infestansColombiaZipaquirá–CundinamarcaS. tuberosum
COCST4P. infestansColombiaZipaquirá–CundinamarcaS. tuberosum
COCST5P. infestansColombiaCogua–CundinamarcaS. tuberosum
COCST6P. infestansColombiaCogua–CundinamarcaS. tuberosum
CONSB31P. infestansColombiaGenoy–NariñoS. betaceum
CONSB45P. infestansColombiaGenoy–NariñoS. betaceum
CONSB48P. infestansColombiaBuesaco–NariñoS. betaceum
CONSB49P. infestansColombiaSan Bernardo–NariñoS. betaceum
CONSB50P. infestansColombiaObonuco–NariñoS. betaceum
CONSP39P. infestansColombiaPasto–NariñoS. phureja
CONST34P. infestansColombiaPasto–NariñoS. tuberosum
CONST38P. infestansColombiaPasto–NariñoS. tuberosum
CONST40P. infestansColombiaPasto–NariñoS. tuberosum
CONST44P. infestansColombiaPasto–NariñoS. tuberosum
ESC80P. infestansEcuadorNapoS. colombianum
ESM79P. infestansEcuadorPinchinchaS. muricatum
ESM85P. infestansEcuadorAzuayS. muricatum
ESO84P. infestansEcuadorCarchiS. ochranthum
ESQ75P. infestansEcuadorTungurahuaS. quitoenese
EST76P. infestansEcuadorPinchinchaS. tuberosum
EST81P. infestansEcuadorTungurahuaS. tuberosum
V86STP. infestansVenezuelaSanta RosaS. tuberosum
V87STP. infestansVenezuelaLas PorquerasS. tuberosum
V88STP. infestansVenezuelaEl ValleS. tuberosum
V89STP. infestansVenezuelaPueblo LlanoS. tuberosum
V90STP. infestansVenezuelaPueblo LlanoS. tuberosum
V91STP. infestansVenezuelaPueblo LlanoS. tuberosum
V92STP. infestansVenezuelaEl CobreS. tuberosum
V94STP. infestansVenezuelaValle del MocotíesS. tuberosum
V95STP. infestansVenezuelaValle del TuñameS. tuberosum
V97STP. infestansVenezuelaSanto DomingoS. tuberosum
VST93P. infestansVenezuelaValle del ChamaS. tuberosum
AF266777.1P. mirabilisMexicoTexcocoMirabilis jalapa
AY564041.1P. mirabilisMexicoNAM. jalapa
AY129171.1P. mirabilisMexicoNAM. jalapa
AF266778.1P. phaseoli
AY129168.1P. phaseoliUSAMarylandPhaseolus lunatus
AY564044.1P. phaseoli
AY564158P. ipomoeaeMexicoIpomoea longipedunculata
AY564170P. citricolaSouth AfricaMedicago sativa
AY564198P. brassicaethe NetherlandsBrassica oleracea
AY564161P. tropicalisthe NetherlandsRosa sp.

Sequence data were subjected to phylogenetic analysis by maximum parsimony (MP), maximum likelihood (ML) and Bayesian inference (BI) for each gene (Figs 1 & 2). For MP, the software paup* 4b10 (Swofford, 1998) was used, and 10 000 bootstrap replicates were made to assess support values to the branches. RAxML was used for ML assessment, using 1000 bootstrap replicates (Stamatakis et al., 2008). MrBayes 3·1·2 (Huelsenbeck & Ronquist, 2001) was used for BI using the models obtained by MrModeltest 2·3 (Nylander, 2004) for each gene and partition schemes. MCMC generations (10 000 000) were used in two runs of four chains each (one cold chain and three heated chains) with a 250 000 generations burn-in (Huelsenbeck & Ronquist, 2001, 2005). Nucleotide substitution priors were used in each matrix to improve the quality of the runs.

Figure 1.

 Phylogenetic trees obtained from the ITS2 region (a) and the β-tubulin region (b). The obtained substitution models for each region are shown. Samples in grey highlighting (red font online) have been previously defined as Phytophthora andina. Support values are above branches, and represent bootstrap values for MP and ML and posterior probabilities for BI: MP/ML/BI.

Figure 2.

 Phylogenetic reconstruction obtained from the CoxI region. The obtained substitution models are shown. Samples in grey highlighting (red font online) have previously been defined as the new proposed species Phytophthora andina. Support values are provided above branches and represent values for MP/ML/BI.

The results (Figs 1 & 2) show that the proposed species P. andina and P. infestans fall into the same clade with significant support values differentiating them from close relatives such as P. mirabilis and P. ipomoeae, previously characterized as taxa distinct from P. infestans (Flier et al., 2002; Grünwald & Flier, 2005). Both locus-by-locus and concatenated analyses revealed that the newly proposed species, P. andina cannot be distinguished by limited sequence data from P. infestans. Thus, the phylogenetic species concept could not be applied. The evidence presented here and in Oliva et al. (2010) places the proposed P. andina within the P. infestans population. Given this analysis, the current description of the proposed P. andina cannot yet be considered as accurate.

This analysis shows that further work is needed to understand the evolutionary history of P. andina. AFLP analysis suggests that P. andina may be a new species but this result is clearly in conflict with the phylogenetic analysis of nuclear and mitochondrial sequence data. In addition, we have found that P. infestans could be isolated from Solanum betaceum where the proposed P. andina is found. Further work might include analysis of reproductive isolation, gene flow and population structure (Goss et al., 2009). One mechanism that could explain the distinct nature of P. andina could be hybridization. In fact, Oliva et al. (2010) argue that P. andina might have evolved from a hybridization event between P. infestans and another as yet unknown Phytophthora species. Further studies should investigate this possibility in a substantive manner.