Genotyping of Phytophthora infestans in Eastern Africa Reveals a Dominating Invasive European Lineage

The oomycete plant pathogen Phytophthora infestans (Mont.) de Bary causes devastating late blight, mainly on potato (Solanum tuberosum L.) and tomato (S. lycopersicum L.); however, the pathogen has a wide host range within the Solanaceae family. P. infestans is a heterothallic organism requiring both mating types A1 and A2 for sexual reproduction. In areas where sexual reproduction is absent, asexual clonal populations (lineages) dominate. The disease emerged in the mid-1840s (Berkeley 1846), when it contributed to the Great Famine in Ireland (Zadoks 2008). In eastern Africa, the epidemic which led to the discovery of the pathogen in the region was documented by Dr. Nattrass, who found the disease in Kenya in 1941 on the potato cultivar Kerr’s Pink, in a crop that had been planted with imported seed (Natrass 1944). A year after the original outbreak in Kenya, the disease was discovered in eastern Uganda, which borders Kenya. The disease then spread, moving west into the Democratic Republic of Congo and south into Tanzania (Natrass 1944).

The complexity of controlling late blight arises from factors associated with pathogen movement between continents. This has resulted in the introduction of strains with novel virulence, increased fitness, and fungicide resistance and of the opposite mating type to the one already present. At the time of the original introduction of P. infestans in eastern Africa, a large shipment of potato cultivar Kerr’s Pink was imported from the United Kingdom for the 1941 cropping season (Cox and Large 1960). Therefore, it is likely that the original strains of P. infestans in eastern Africa resembled those in the United Kingdom at the time.

Strains of P. infestans have been monitored using one or more standardized markers, including mating type, fungicide (metalaxyl) resistance (Cardenas et al. 2012), mitochondrial (mt)DNA haplotype, restriction fragment length polymorphisms (RFLPs), analyses with the probe RG-57 (Goodwin et al. 1994), isozymes (glucose-6-phosphatase isomerase [Gpi] and peptidase) (Shattock et al. 1986a; Tooley et al. 1985), amplified fragment length polymorphisms (Cooke et al. 2003; Flier et al. 2003), and, more recently, simple sequence repeats (SSRs) (Knapova and Gisi 2002; Lees et al. 2006; Li et al. 2010). The phenotypic and genotypic data obtained from these markers is subsequently used to describe P. infestans multilocus genotypes (MLGs) according to a globally used nomenclature system. The first strain to be isolated and described in the United States was named US-1 (Goodwin et al. 1994) and its designation was maintained wherever it was found in the world (Forbes et al. 1998).

The US-1 lineage, characterized primarily by a Ib mt-DNA haplotype, dominated worldwide population during the mid-20th century and was originally suspected to be responsible for the famine-era disease outbreaks (Goodwin et al. 1994). However, it is now clear that these were actually due to a lineage with a Ia mtDNA haplotype (Birch and Cooke 2013; May and Ristaino 2004). The famine-era MLG, FAM-1, was itself displaced by the US-1 (Saville et al. 2016), which was, in turn, replaced by new genotypes carried over in a shipment of potato from Mexico in 1976 (Birch and Cooke 2013). The precise timing of the displacement of FAM-1 by US-1 is not known but it is believed to have taken place before World War II (Saville et al. 2016). Thus, US-1 was probably dominant in Europe at the time of the introduction of P. infestans in eastern Africa in 1941, and is assumed to be the only lineage introduced into the region then.

Migration of P. infestans has had a big influence on shaping its population structure and dynamics worldwide. The pathogen is thought to have originated in central Mexico, owing to populations characteristics in this area such as high genetic diversity, presence of A1 and A2 mating types, and the presence of two close relatives (P. mirabilis and P. ipomoeae) in that region (Goss et al. 2014; Grünwald and Flier 2005). However, other reports point to a South American origin because the Andes are the center of origin of cultivated potato and variations of Andean lineages of P. infestans were found there (Gómez-Alpizar et al. 2007). For the latter hypothesis, P. infestans is assumed to have migrated to Mexico from South America (Niederhauser 1991). The pathogen reached Europe in 1845 (Bourke 1964) and, from there, it spread to other potato-growing regions of the world, including Africa (Cox and Large 1960). The first introduction of P. infestans into Europe only brought the A1 mating type but, in a second migration event in 1976, the A2 mating type was introduced (Birch and Cooke 2013). The A2 mating type has also been reported in North Africa, and this could be attributed to the high volume of seed potato tubers and, consequently, new pathogen strains imported into Algeria, Egypt, Morocco, and Tunisia from European countries (van Loon 2007). In eastern Africa, only the A1 mating type has been detected thus far (Njoroge et al. 2016; Pule et al. 2013; Were et al. 2013), signifying the persistence of a clonal population.

Population shifts of either clonal or sexual populations have similar epidemiological consequences because they are the result of emergence of fitter pathogen genotypes (Day and Shattock 1997; Hohl 1998). Sexual populations of P. infestans give rise to long-lived oospores in the soil (Mayton et al. 2000), as well as greater genotypic variation via sexual recombination (Shattock et al. 1986b). Clonal populations create intralineage variations via accumulation of mutations or mitotic recombination (Goodwin 1997; Grünwald 2012). Knowledge of P. infestans population structure within a region helps in understanding whether different strains have arisen and increased in frequency over time, hence providing a better insight into pathogen migrations. Disease management strategies can then be refined by knowledge of the phenotypic characteristics of the prevailing pathogen strains such as host range or fungicide sensitivity. Additionally, P. infestans populations can differ dramatically among countries and locations, and predictions concerning their phenotypic behavior need to be based on the correct population structure in a region.

In the eastern-Africa Region (EAR), US-1 had been the only P. infestans lineage reported, apart from appearances of Ia mtDNA haplotypes in Rwanda in the mid-1980s that were designated as RW-1 and RW-2 (Forbes et al. 1998; Goodwin et al. 1994). These authors reported that the two Rwandan genotypes (RW-1 and RW-2) were co-occurring with US-1 in that country. However, the RW-1 and RW-2 genotypes were not detected in a later study in 2007 that reported all P. infestans isolates from Rwanda to be US-1 (Pule et al. 2013). Pule et al. (2013) also reported that US-1 was still the only lineage in central and eastern Africa apart from Kenya, where US-1 and a new lineage were found. The new lineage, designated as KE-1, was reported in Kenya in 2007 (Pule et al. 2013) and in Uganda in 2011 (Njoroge et al. 2016). KE-1 was later found to be the only lineage on potato in Kenya (Njoroge et al. 2016), having earlier been found in only two fields in 2007 (Pule et al. 2013). Njoroge et al. (2016) also found that the KE-1 lineage was, in fact, the European 2_A1 lineage, which comprised about half of Great Britain’s population in the mid-1990s (Day et al. 2004).

The suspicion that KE-1 and 2_A1 were the same lineage was raised by Were et al. (2013) but was not confirmed at the time. In contrast, Pule et al. (2013) supposed that the KE-1 lineage could be the RW genotypes reported in Rwanda in the mid-1980s. In neighboring Ethiopia, a study in 2002 revealed a population of only Ia mtDNA isolates (Schiessendoppler and Molnar 2002). More recently, a subsequent study revealed several lineages coexisting in the country, including a novel lineage with an unknown mtDNA haplotype (Shimelash et al. 2016). Subsequently, a comparison of the SSR fingerprints of the Ia populations in Ethiopia and Kenya indicated that they are identical (Njoroge et al. 2016). Although it has never been clarified whether Ethiopia is the source of the KE-1 population in Kenya, it is evident that the shared population is of European origin. Both countries import seed potato from Europe.

In Kenya, late blight on tomato was first recorded in 1944 and, by 1945, the disease was widespread on tomato in Tanzania (Wallace and Wallace 1945) and later found in Uganda in late 1954 (Baldwin 1956). From population studies conducted between the late 1990s and 2012, it appears that a single US-1 clonal lineage was found on tomato and potato in eastern Africa (Njoroge et al. 2016; Vega-Sanchez et al. 2000). The US-1 isolates from tomato and potato were shown to be host specific in detached-leaf tests and the two pathogen subpopulations also differed for one allele of Gpi, an allozyme previously used to characterize populations of P. infestans. Furthermore, a recent pathogen effector gene study looking at the IpiO (Avr-blb1) gene variants has revealed effector profile differences between the potato and tomato-adapted US-1, further confirming their difference (A. W. Njoroge, M. Ghislain, B. Andersson, E. Magembe, C. Mutai, R. Pelle, J. Yuen, and G. Forbes, unpublished data). It is possible that the difference in time of the first appearances of late blight on potato and tomato in Kenya might mean that the P. infestans strains have been host specific ever since their initial introduction, as indicated by early studies on the disease in the region (Cox and Large 1960). The selective displacement of US-1 by KE-1 only on potato in some countries could be due to adaptation to tomato in US-1 that has not (or not yet) occurred in KE-1.

Although current populations of US-1 in the EAR exhibit features that could be considered beneficial to the pathogen, such as a high level of metalaxyl resistance (Mukalazi et al. 2001), the dominance of KE-1 lineage suggests that it has some attributes that result in a higher fitness compared with the previously established lineage. The date of introduction into Kenya of KE-1 as the European 2_A1 lineage is unknown, although the route of introduction is believed to be seed potato importation from Europe. In this study, we tested the hypothesis that the European strain 2_A1 has recently dominated the P. infestans population in the EAR. We present an analysis of the population structure of P. infestans sampled on potato and tomato between December 2013 and June 2016 in five eastern Africa countries (namely, Kenya, Uganda, Tanzania, Burundi, and Rwanda) using SSR markers.

MATERIALS AND METHODS

Sampling of late-blight-infected plant material.

Samples of diseased potato leaflets were collected from growers’ fields (Supplementary Fig. S1) in 2013 (Uganda), 2014 (Uganda and Kenya), 2015 (Uganda, Kenya, Rwanda, Burundi, and Tanzania), and 2016 (Uganda, Kenya, Burundi, and Tanzania). Additionally, samples were taken from farmers’ fields containing different potato cultivars being evaluated for susceptibility to late blight in southwestern Uganda (2013–14); from field trials on the experimental station of Uganda’s National Agricultural Research Organization at the Kachwekano zonal agricultural research and development institute (2015), where Dutch potato cultivars were under evaluation; and from a confined field trial (2015) with transgenic potato lines. The greatest number of samples came from Uganda, (447 potato, 110 tomato), followed by Kenya (271 potato, 43 tomato), Rwanda (180 potato, 8 tomato), Burundi (82 potato), and Tanzania (76 potato, 4 tomato). Single lesions from diseased leaflets were individually pressed onto Whatman FTA cards (Whatman FTA Classic Card, catalog number WB120205: GE Healthcare UK Ltd.) in the field; the cards were subsequently air dried before storage until they were moved to the laboratory for SSR analysis. In total, 22 reference DNA samples (FTA discs) of the 2_A1 lineage, provided by Dr. Alison Lees (James Hutton Institute, Dundee, United Kingdom), were included in the analysis. The samples from Dr. Lees had been previously described by Were et al. (2013), who identified the lineage as 2_A1.

Eluting DNA from the FTA cards.

To elute the DNA from the FTA cards, a method by Njoroge et al. (2016) was used but modified to include a pH-treatment step. A single 6-mm disc was punched out from the applied dry sample spot and transferred into a 1.5-ml microfuge tube. The disc was washed twice with 400 µl of FTA purification reagent (Whatman Inc., USA) followed by two rinses with 400 µl of modified Tris-EDTA buffer (10 mM Tris and 0.1 mM EDTA). The pH treatment process involved incubating the washed FTA disc in 70 µl of alkaline buffer (solution 1: 0.1 N NaOH and 0.3 mM EDTA, pH 13.0) for 5 min at room temperature. Thereafter, 130 µl of neutralization buffer (solution 2: 0.1 M Tris-HCl, pH 7.0) was added, vortexed five times to mix, then incubated for 10 min at room temperature. The FTA discs were left in the final solution and kept at −20°C until use.

Confirmation of P. infestans DNA presence in FTA elutes using internal transcribed spacer region primers.

The eluted solution from the FTA cards was tested for the presence of P. infestans DNA using a primer-pair, internal transcribed spacer (ITS)5 (GGAAGTAAAAGTCGTAACAAGG) as the forward primer and a species-specific primer Pinf2 (CTCGCTACAATAGCAGCGTC) as the reverse primer in a polymerase chain reaction (PCR). The PCR was performed in a 10-µl volume containing 2 µl of the FTA eluate, 0.25 µM each primer, and 5 µl of AccuPower Taq PCR 2× master mix (Bioneer). Amplifications were performed in Eppendorf AG 22331 Hamburg thermocycler under the following conditions: initial denaturation at 94°C for 2 min; followed by 35 cycles of 15 s at 94°C, 15 s at 55°C, and 15 s at 72°C; and a final extension of 5 min at 72°C. Visualization of the PCR amplicons was done on agarose gel electrophoresis.

SSR multiplex reactions and uniplex PCR.

Determination of the P. infestans SSR alleles was done in two laboratories. The samples collected in Uganda and Kenya from 2013 to mid-2014 were analyzed in single primer reactions using nine SSR markers chosen from previously published markers (Table 1) at the Swedish University of Agricultural Sciences (SLU-Uppsala) in 2014. The other set of samples collected between December 2014 and June 2016 was analyzed at the Biosciences central and eastern Africa–International Livestock Research Institute (BecA-ILRI) hub in Nairobi, Kenya, in 2017, using a modified version of the 12-plex single SSR protocol previously described by Li et al. (2013). Additionally, for the 2017 analysis at the BecA-ILRI hub, a uniplex reaction was set up with an additional 13th marker, Pi56 (Lees et al. 2006). The primer concentrations in the multiplex reaction were adjusted for our laboratory conditions to between 0.16 and 0.3 µM (Table 1). The Qiagen Type-It Microsatellite PCR kit (Qiagen Corporation) was used for PCR and sample volumes were modified to run a 12.5-µl reaction using 6.25 µl of 2× Type-It Master Mix, 1.3 µl of a 10× multiplex primer mix, 1.95 µl of water, and 3 µl of FTA eluate. For the single primer reactions, AccuPower Taq PCR 2× master mix (Bioneer) was used and all primers were used at a final concentration of 0.25 µM. The amplification conditions for the multiplex PCR were as follows: 95°C for 5 min; followed by 33 cycles of 95°C for 30 s, 58°C for 90 s, and 72°C for 30 s; and a final extension at 60°C for 30 min. For the single primer reactions, the cycling conditions were as follows: initial denaturation at 94°C for 3 min; followed by 33 cycles of 30 s at 94°C, 30 s at 62°C (Pi04 and PiG11), 60°C (Pi16), 58°C (Pi56, Pi70, Pi89, and Pi4B), or 50°C (D13), and 45 s at 72°C; and completed with a final extension step of 25 min at 72°C.

TABLE 1. Phytophthora infestans simple sequence repeat (SSR) primers used in this study for the multiplex and single primer reactions

Primer	Reaction type	Primer sequence + dyesa	Final Con (µM)b
PiG11c	Multiplex, uniplex	Fwd_NED-TGCTATTTATCAAGCGTGGG	0.15
	…	Rev_GTTTCAATCTGCAGCCGTAAGA	0.15
Pi02/PinfSSR3d,e	Multiplex	Fwd_NED-ACTTGCAGAACTACCGCCC	0.15
	…	Rev_GTTTGACCACTTTCCTCGGTTC	0.15
PinfSSR11e	Multiplex	Fwd_NED-TTAAGCCACGACATGAGCTG	0.15
	…	Rev_GTTTAGACAATTGTTTTGTGGTCGC	0.15
D13d	Multiplex, uniplex	Fwd_FAM-TGCCCCCTGCTCACTC	0.16
	…	Rev_GCTCGAATTCATTTTACAGACTTG	0.16
PinfSSR8e	Multiplex	Fwd_FAM-AATCTGATCGCAACTGAGGG	0.3
	…	Rev_GTTTACAAGATACACACGTCGCTCC	0.3
PinfSSR4e	Multiplex	Fwd_FAM-TCTTGTTCGAGTATGCGACG	0.15
	…	Rev_GTTTCACTTCGGGAGAAAGGCTTC	0.15
Pi04d	Multiplex, uniplex	Fwd_VIC-AGCGGCTTACCGATGG	0.15
	…	Rev_GTTTCAGCGGCTGTTTCGAC	0.15
Pi70d	Multiplex, uniplex	Fwd_VIC-ATGAAAATACGTCAATGCTCG	0.15
	…	Rev_CGTTGGATATTTCTATTTCTTCG	0.15
PinfSSR6e	Multiplex	Fwd_VIC-GTTTTGGTGGGGCTGAAGTTTT	0.15
	…	Rev_TCGCCACAAGATTTATTCCG	0.15
Pi63d	Multiplex, uniplex	Fwd_VIC-ATGACGAAGATGAAAGTGAGG	0.15
	…	Rev_CGTATTTTCCTGTTTATCTAACACC	0.15
PinfSSR2e	Multiplex	Fwd_PET-CGACTTCTACATCAACCGGC	0.15
	…	Rev_GTTTGCTTGGACTGCGTCTTTAGC	0.15
Pi4Bc	Multiplex, uniplex	Fwd_PET-AAAATAAAGCCTTTGGTTCA	0.3
	…	Rev_GCAAGCGAGGTTTGTAGATT	0.3
Pi56d	Uniplex	Fwd_FAM-ACAACTATCTATCGGCGTGC	0.25
	…	Rev_AGTAGGCTTCACGACCAGC	0.25
Pi16d	Uniplex	Fwd_FAM-CACAGCACGCGGAATC	0.25
	…	Rev_ACGCCGAGTGTCCTGA	0.25
Pi89d	Uniplex	Fwd_FAM-GAGAACGCACAATGTAAGGC	0.25
	…	Rev_ACATAAATACACGCTGAACGG	0.25

^aPrimer sequences and the different fluorescent dyes (NED, FAM, VIC, or PET) added to the different forward primers.

^bPrimer concentrations used for our laboratory conditions.

^cPrimers previously reported by Knapova and Gisi (2002).

^dPrimers previously reported by Lees et al. (2006).

^ePrimers previously reported by Li et al. (2010).

View as image HTML

SSR fragment analysis and scoring.

To confirm successful amplification and band intensity, 5 µl of PCR product was mixed with 1 µl of 6× gel loading dye and separated on 2.5% (wt/vol) agarose gel. The band intensities were used to estimate the amount of PCR amplicons present in each sample. Dilutions were made using 1 to 1.5 µl of the PCR products by mixing them with 9 µl of distilled deionized water. To prepare samples for fragment analysis, 1 µl of the diluted PCR product was mixed with a 9-µl mix of HiDi (400 µl; Hi-Di Formamide part number 4311320; Applied Biosystems) and GeneScan-500LIZ standard (6 µl; part number 4322682). The samples were submitted to the Segoli Laboratory (BecA-ILRI hub, Nairobi, Kenya) while those in Sweden were submitted to the Uppsala Genome Center at Rudbeck Laboratory (Uppsala University, Sweden). Fragment analysis of the samples was done on an automated ABI 3730 capillary sequencer according to the manufacturer’s instruction at both locations. In Sweden, allele sizes were scored using the software GENEMARKER (Soft Genetics) while, at the BecA-ILRI hub, allele sizes were determined using GeneMapper v3.7 software (Applied Biosystems).

mtDNA haplotype analysis.

mtDNA haplotypes were determined for 62 samples selected from the 223 samples analyzed in Sweden. The 62 samples represented all of the SSR MLGs generated from subpopulations defined according to host origin and clonal lineage. Haplotyping was done using the RFLP method of Griffith and Shaw (1998). Two mtDNA haplotypes were tested using the primer pair P2F (TTCCCTTTGTCCTCTACCGAT) and P2R (TTACGGCGGTTTAGCACATACA) and primer pair P4F (TGGTCATCCAGAGGTTTATGTT) and P4R (CCGATACCGATACCAGCACCAA), specific for the P2 and P4 mtDNA regions, respectively. The PCR consisted of 1× Green buffer, 0.1 mM dNTP each, 1.5 mM MgCl₂, 0.32 µM each forward and reverse primer, and 1.5 U of Dream Taq (Fermentas) in a 25-µl final volume. The cycling conditions were as follows: 94°C for 3 min; followed by 30 cycles of 94°C for 30 s, 62°C for 30 s, and 72°C for 60 s; and a final extension at 72°C for 5 min. Amplified products (10 µl) of the P2 and P4 regions were digested with MspI and EcoRI restriction enzymes, respectively. The resulting products were visualized by agarose gel electrophoresis to distinguish the resulting mtDNA haplotypes.

Datasets.

Three datasets were used for analyses in this study as follows:

1. The 2014 dataset comprised 223 samples collected on potato and tomato between 2013 and mid-2014 from Kenya and Uganda. This dataset was generated using nine SSR markers in single PCR at SLU-Uppsala. The dataset could not be combined with the 2017 dataset (see below) due to dissimilarity in the type of SSR markers and laboratory protocols used; hence, it was analyzed separately.

2. The 2017 dataset with 1,013 samples was collected between December 2014 to June 2016 in Kenya, Uganda, Tanzania, Rwanda, and Burundi on potato and tomato. This dataset was created using 13 SSR markers in a single 12-plex and uniplex (Pi56) reaction at the BecA-ILRI hub in Kenya. The dataset was analyzed both separately and combined with the reference samples (see below), depending on the different population parameters that were analyzed.

3. The 2017 full dataset comprised all the samples in the 2017 dataset and an additional 22 reference samples. The reference samples had been previously collected in Kenya in 2009. They were genotyped using a similar genotyping protocol used for the 2017 dataset and, henceforth, assigned the name KenyaRef, for purposes of comparisons in the subsequent analyses.

Assignment of clonal lineages.

The 2014 dataset and 2017 dataset were used for comparison against the reference samples to establish whether samples belonged to the US-1 or KE-1 lineage.

MLG analyses.

Samples with the same combination of alleles (i.e., similar MLGs) were considered to be clones. The number of MLGs in each population and the expected number of MLGs after rarefaction (eMLG) (Hurlbert 1971) were obtained using the R package Poppr (Kamvar et al. 2014).

Genetic diversity indices.

Basic indices of genetic variation for each population were calculated for populations defined by lineages (KE-1 or US-1) and subpopulation host (tomato or potato) for the 2014 dataset and 2017 full dataset. Clonal diversity within the populations was described using the Simpson diversity (Simpson 1949) and the Shannon-Wiener index (Shannon 2001); genetic richness (R), which, in this case, is equivalent to the number of eMLG at the largest shared sample size; and the Hendrick’s standardized Index (G′_ST), which determines within-population genetic diversity for multiallelic markers (Hedrick 1999; Hedrick 2005). An index of G′_ST = 0 indicates that populations are fixed for a single allele while G′_ST = 1 indicates that populations have no overlapping sets of alleles (Meirmans and Hedrick 2011).

Population structure and differentiation.

Estimation of genetic distance between individuals was based on the method of Bruvo et al. (2004), which uses a stepwise mutation model to calculate genetic distance of individuals in the population using the R package POLYSAT (Clark and Jasieniuk 2011). Significance was assessed using 1,000 bootstrap replicates. Genetic groupings of individuals within a clonal lineage were viewed using a discriminant analysis of principal components (DAPC) (Jombart et al. 2010) using the R package ADEGENET (Jombart and Ahmed 2011).

Minimum-spanning networks to test evolutionary relationships among individuals for each subpopulation (KE-1 or US-1), were constructed and visualized for different regions and hosts, using the R package Poppr (Kamvar et al. 2014).

The proportion of genetic variability between and within clonal genotypes was determined by an analysis of molecular variance (AMOVA) using (Weir and Cockerham 1984) an unbiased estimator of fixation index in the ade4 R package (Dray and Dufour 2007).

RESULTS

The findings from this study confirm that the P. infestans population infecting potato in the EAR, represented here by five countries, is dominated by KE-1 lineage, a clonal lineage of European origin. KE-1 and the European 2_A1 had similar SSR fingerprints for the SSR loci that identify this lineage, as well as being of the Ia mtDNA haplotype, and, thus, are assumed to be the same. The KE-1 lineage will hereafter be referred to as 2_A1. The US-1 lineage was found to have diminished significantly but was still present on potato in Uganda, Rwanda, Burundi, and Tanzania. Additionally, a tomato-adapted US-1 subpopulation was also still present in all of the countries.

A rapid displacement of the old US-1 lineage by 2_A1 was observable in all countries, particularly on potato. The 2_A1 was the only lineage on potato in Kenya. However, in Uganda, Rwanda, Burundi, and Tanzania, both 2_A1 and US-1 were found on this host. In some instances, the two lineages were found coexisting in the same field. A notable example was one field in Uganda’s Kabale district where, out of eight potato plants sampled during the dry season in February 2016, five were 2_A1 and three were US-1.

The population change over time was particularly evident in Uganda, where samples collected every cropping season showed a rapid increase of the 2_A1 lineage between December 2013 (23%) and December 2015 (91.5%). In southwestern Uganda, 2_A1 was first observed in December 2014 but made up three-quarters of the samples collected in April 2016. A similar, rapid increase in 2_A1 frequency was observed in Tanzania, where it increased to 85% frequency within a year. 2_A1 also appeared and spread in 2015–16 in Rwanda and Burundi, where it had not been observed before. Finally, 2_A1 spread to tomato in Kenya, a host from which it was previously absent (Fig. 1).

Download as PowerPoint

The P. infestans population across the countries had varying diversity, as demonstrated by the eMLGs. Tanzania had the most diverse P. infestans population while Rwanda displayed the least diversity (Table 2). Burundi, Uganda, and Kenya had almost similar diversity, ranking in that order in total eMLGs (Table 2). Notable, though, was the presence of unique alleles on the US-1 subpopulation in Tanzania in one of the SSR markers that identifies the US-1 lineage. The P. infestans samples from Tanzania had private alleles on the Pi02 locus, 264/270 and 266/272, in place of 266/266, which generally identifies the US-1 lineage. Dissimilarity of the Tanzanian US-1 from the US-1 populations in the other countries was also evident in the cluster analyses, where the two Tanzanian subpopulations on potato and tomato formed a distinct cluster in the DAPC (Fig. 2). Multiple occurrences of similar MLGs was noted across the countries, with Kenya and Burundi sharing the highest number of such genotypes. Consequently, these P. infestans populations in the two countries were more aggregated than the subpopulations in the other countries in the DAPC (Fig. 3).

TABLE 2. Summary of multilocus genotype diversity found in the Phytophthora infestans subpopulations found in different countries and on different hosts for the 2017 full dataset

		Potato		Tomato		All samplesa
Country	Lineageb	Nc	eMLGd	Nc	eMLGd	Ne	All eMLGf
Burundi	2_A1	68	8.11	…	…	…	…
	US-1	11	10	…	…	…	…
	All	…	…	…	…	79	49
Kenya	2_A1	232	7.19	31	9.61	…	…
	US-1	…	…	12	9.32	…	…
	2_A1REF	22	7.68	…	…	…	…
	All	…	…	…	…	275	44.5
Rwanda	2_A1	151	7.35	…	…	…	…
	US-1	29	9.12	8	5	…	…
	All	…	…	…	…	188	41.1
Uganda	2_A1	260	6.57	…	…	…	…
	US-1	75	9.73	56	9.77	…	…
	All	…	…	…	…	391	46.9
Tanzania	2_A1	30	8.27	…	…	…	…
	US-1	46	9.79	4	3	…	…
	All	…	…	…	…	80	64.3

^aNumber of all Phytophthora infestans samples in the respective countries.

^bDenotes the P. infestans clonal lineage; 2_A1REF represents the reference samples.

^cNumber of samples from each clonal lineage in each country on either potato or tomato.

^dExpected number of multilocus genotypes (eMLGs) for each clonal lineage after rarefaction (or genetic richness at the largest shared sample size) on either potato or tomato.

^eTotal number of samples from each country.

^feMLGs for all samples after rarefaction for each country.

View as image HTML

Download as PowerPoint

Download as PowerPoint

MLG variability was found for both 2_A1 and the diminishing resident US-1 subpopulations on potato and tomato. Diverse MLGs of the 2_A1 lineage were found even at the field level and no field had a P. infestans population consisting of a single MLG. Although both lineages had high diversity, there was a higher MLG diversity in US-1 (85%) than in 2_A1 (36%). Greater MLG diversity was also evident on tomato (84.7%) than on potato (43.8%) US-1 subpopulations. Across the countries, Burundi’s US-1 subpopulation on potato had a higher genetic richness at the largest shared sample size (eMLG) than did the US-1 potato subpopulations in the other countries (Table 2). Even though the 2_A1 subpopulation on tomato was only found in Kenya, it was equally diverse, with nearly all of the samples representing a distinct MLG. Within the subpopulations of each lineage on the two hosts, the US-1 lineage and, particularly, the subpopulation on potato exhibited higher genetic diversity, as indicated by three diversity indices (Nei’s index, Simpsons’s index, and Shannon’s index) (Table 3). The intralineage variability observed within the clonal lineages was attributable to some but not all the 13 SSR markers used in this study. The SSR markers D13, PiG11, and PinfSSR4 had the highest polymorphism (number of alleles), hence resulting in a higher number of MLGs within the two clonal lineages (Table 4). High genetic similarity existed between the 2_A1 subpopulations on potato and tomato, as shown by shared nodes on the minimum spanning network (Fig. 4) and overlapping of clones on the DAPC (Fig. 5). Some genetic dissimilarity as shown by the AMOVA was found, however, in the 2_A1 lineage across the countries (Table 5), despite the lineage forming a cluster of its own in the DAPC, with many overlapping samples from different countries.

TABLE 3. Summary of diversity indices conducted on clonal genotypes of Phytophthora infestans (2_A1 and US-1) coming from both potato and tomato in the eastern-Africa region

	2014 dataset			2017 dataset
Diversity indices	US-1 Toma	US-1 Potb	2_A1 Potc	US-1 Toma	US-1 Potb	2_A1 Potc	2_A1 Tomd
Samples, Ne	64	105	64	80	161	763	31
MLGsf	24	33	15	67	138	269	27
eMLGg	24.8	22.1	14.0	28.8	29.9	21.9	27.0
HexPh	0.429	0.443	0.416	0.497	0.527	0.495	0.508
Simpsons Ii	0.850	0.915	0.781	0.983	0.991	0.963	0.959
Shannon Ij	2.57	2.90	2.03	4.14	4.86	4.43	3.26

^aUS-1 lineage on tomato.

^bUS-1 lineage on potato.

^c2_A1 lineage on potato.

^d2_A1 lineage on tomato.

^eNumber of Phytophthora infestans samples of each clonal genotype from each host.

^fNumber of distinct multilocus genotypes (MLGs) discriminated.

^gExpected number of MLGs (eMLG) after rarefaction (or genetic richness at the largest shared sample size).

^hNei’s index of gene diversity defined as the probability that two randomly chosen alleles are different.

ⁱSimpsons’s index, showing the probability estimation that two randomly selected genotypes are different.

^jShannon’s index, providing genotypic abundance and evenness of the genotypes present.

View as image HTML

TABLE 4. Summary of number of alleles, Simpson’s diversity, Nei’s gene diversity, evenness of allele distribution, and the standardized Hedricks’s G′_ST in the Phytophthora infestans populations found in different countries on different hosts for the 2017 dataset only

Simple sequence repeat locus	Allelesa	Simpson’s index	Nei’s gene diversity	Evenness	G′ST _Hedrick
Pi56	2	0.47	0.47	0.95	0.36
D13	22	0.71	0.71	0.52	0.75
Pi02	7	0.67	0.67	0.83	0.30
Pi04	5	0.51	0.51	0.95	0.02
Pi4B	3	0.22	0.22	0.58	0.35
Pi63	8	0.66	0.66	0.98	0.50
Pi70	6	0.60	0.60	0.83	0.31
PiG11	22	0.70	0.70	0.61	0.30
PinfSSR11	3	0.52	0.52	0.91	0.12
PinfSSR2	4	0.27	0.27	0.55	0.30
PinfSSR4	14	0.74	0.74	0.72	0.77
PinfSSR6	6	0.57	0.57	0.83	0.47
PinfSSR8	4	0.66	0.66	0.94	0.46

^aNumber of microsatellite alleles per locus. Locus D13 and PiG11 have the highest number of alleles while locus PinfSSR4 has the highest Simpson diversity (0.74) and Pi63 has the most evenly distributed alleles (0.98). PinfSSR4 locus reveals the highest ability for population segregation, with a Hendrick’s standardized index of 0.77.

View as image HTML

Download as PowerPoint

Download as PowerPoint

TABLE 5. Summary of results of analysis of molecular variance conducted on clone-corrected datasets of Phytophthora infestans, indicating the percentage of genetic variation attributable to population level differences of 2_A1 clonal lineage for regions (2014 dataset) and countries (2017 full dataset) and, between potato-and tomato-adapted US-1 clonal lineages

		2014 dataset		2017 dataset
Lineage	Source of variation	Variance of total (%)	ϕ*a	Variance of total (%)	ϕ*a
2_A1	Between regions or countries	4	0.04b	23	0.24b
	Within countries	96	…	77	…
	Total	100	…	100	…
US-1	Between hosts	7	0.07c	13	0.13c
	Within hosts	93	…	87	…
	Total	100	…	100	…

^aPopulation differentiation statistics where higher values represent a higher amount of differentiation.

^bThe ϕ* includes contribution of the potato 2_A1KenyaRef (reference samples) to the overall variance.

^cThe ϕ* does not include the 2_A1KenyaRef reference samples.

View as image HTML

Host specificity of the US-1 lineage on tomato was evident in all countries. In every country other than Kenya, the tomato-adapted US-1 was the only lineage found on tomato. In Kenya, however, both 2_A1 and the tomato-adapted US-1 were found to co-occur. The lack of shared SSR MLGs between potato and tomato within the US-1 subpopulations was noteworthy. The genetic dissimilarity between the two US-1 subpopulations, although somewhat minimal (13%) as revealed by AMOVA (Table 5), was evident in the cluster analyses. The tomato-adapted US-1 clustered away from the US-1 on potato (Fig. 5). However, within the 2_A1 lineage, some of the genotypes appearing on tomato were also found on potato in Kenya, Burundi, and Uganda.

DISCUSSION

The evidence presented here supports the hypothesis of the occurrence and dominance of the European 2_A1 lineage in EAR. This study also demonstrated a pathogen population with relatively high genotype variation. Although this occurs and can be expected in parts of northern Europe, where sexual populations of P. infestans are dominant (Brurberg et al. 2011; Sjöholm et al. 2013), results from previous studies of subclonal variation in clonal pathogen populations have not been consistent. For example, Delgado et al. (2013) found high subclonal variation within the EC-1 lineage in Ecuador, whereas Blandón-Díaz et al. (2012) found very little diversity in the lineage in Nicaragua. Li et al. (2013) showed higher variation for D13 (seven allele combinations) and G11 (two allele combinations) with the clone identified as 13_A2.

It is not clear why there was higher variation in the Tanzanian US-1 subpopulation. The results here suggest that this subpopulation could be a variant of the US-1 population. This variation may be due to different potato cultivars grown there, specific microclimate, or even a separate introduction of US-1, although the latter would be inconsistent with the chronology of pathogen spread reported by (Natrass 1944).

Factors that may have contributed to subclonal diversity in the present study include a relatively large sample size and sampling of the population over time, on two hosts, and across a relatively large distance. The latter factor, however, is inconsistent with the fact that populations from the two most distant countries (Burundi and Kenya) were most similar. Delgado et al. (2013) discussed other factors that may lead to high subclonal diversity and reviewed the literature on this topic. However, it is evident from the current study that variation within lineages has arisen predominantly from polymorphisms at the three most variable SSR loci. The other loci that discriminate between the P. infestans lineages, mainly 2_A1 and US-1 in the eastern-Africa population, remain largely monomorphic within each lineage, other than the appearance of a few triploids.

Higher diversity in US-1 than in 2_A1 is probably attributable to the longer presence of US-1 in the region, as it is assumed to be the only P. infestans strain existing since the introduction of the disease in EAR in 1940s. The 2_A1 was identified in 1981 in Europe (Day et al. 2004) but was probably introduced into EAR relatively recently, presumably only as one or a few genotypes.

The apparent higher fitness of 2_A1 over US-1 may be due to several factors, but one could be the extended presence of US-1 in the region. One long-term evolutionary consequence of clonality is that deleterious mutations can accumulate and the resulting decline in fitness is the so-called Muller’s ratchet effect in action (Goodwin 1997). This theory has been used to explain why lineages emerging from the sexually reproducing metapopulation described by Yoshida et al. (2013) displaced HERB-1, the lineage responsible for the Irish potato famine. In our case, the US-1 lineage was found to be less fit than 2_A1 for aggressiveness components (Njoroge et al. 2019) and this could explain, at least in part, why 2_A1 has dominated.

The westward migration of 2_A1 from eastern Uganda is likely to be the source of this lineage in Rwanda. Given the pattern of US-1 displacement, it is highly probable that 2_A1 is currently the only lineage on potato in Uganda. From the high number of 2_A1 samples obtained in Rwanda, Burundi, and Tanzania, none of which had initially reported 2_A1, one can speculate on a sequential appearance of this lineage in the aforementioned countries. However, all these countries share land borders and it is also possible that multiple migration events occurred across borders. Human movement across countries is quite common in the region and movement of plant material across borders is heightened when markets for farm produce open in the neighboring countries. For example, trucks heading to Kigali markets in Rwanda with potato tubers freshly harvested in Kabale, Uganda are common sights along the Kabale–Kigali road. Moreover, during conflicts, people move with farm produce across country borders, providing another potential mechanism for human-mediated P. infestans movement.

Notable from this study is the appearance of 2_A1 on tomato in Kenya, which could imply different scenarios. First, it may indicate that this lineage has adapted to both hosts although, with this scenario, there is no way of knowing from our data whether tomato is infected by a tomato-adapted subpopulation or by a dually adapted (potato and tomato) subpopulation. Answering that question would require cross inoculations on both hosts. A second scenario is that the 2_A1 isolates from tomato represent isolates that are not particularly adapted to tomato but were found on tomato because no tomato-adapted isolates were present. In a previous study, isolation of a potato-adapted genotype from tomato in Ecuador was explained by the absence of tomato-adapted genotypes at that location (Oyarzun et al. 1998). A relatively large number of MLGs was found among 2_A1 samples coming from tomato, indicating that this is not a recently adapted subpopulation but, rather, infections by potato-adapted isolates that are weakly aggressive on tomato. However, this high degree of diversity could also be caused by multiple independent adaptation events, or could reflect the fact that 2_A1 has been adapted to tomato for several years; tomato was not sampled in previous studies done since the appearance of 2_A1 in Kenya. No 2_A1 was found on tomato in Uganda, though no samples were collected in eastern Uganda, which borders Kenya. The eastern part of Uganda was the first entry point of 2_A1 into Uganda (Njoroge et al. 2016) and possibly a host shift of 2_A1 from potato to tomato has taken place subsequently.

It will be important in future studies to determine whether 2_A1 occurs on tomato in the other countries. The appearance of the same lineage on both hosts in the region is of epidemiological consequence for either of the scenarios given above because tomato plants are often volunteer plants heavily infested with P. infestans and are neither treated with fungicide nor destroyed when growing in close proximity to potato plants. As an illustration of this, the tomato samples used in this study were collected in different potato-growing areas in the respective countries and mostly from volunteer plants that had no fungicide applied on them. Most of these tomato plants are found in growers’ kitchen gardens, where the environment is conducive for both P. infestans and its hosts, and where these plants are maintained year-round, potentially providing sources of primary inoculum for epidemics in the growers’ fields.

Host specialization within US-1 is still evident in Uganda, Rwanda, and Tanzania, even though the genetic dissimilarity displayed appears minimal. Given the neutrality of the microsatellite markers used in this study, it would appear that host species adaptation has had only a small effect on the overall genetic variability of the two US-1 subpopulations. However, evidence of the differences between the two US-1 subpopulations has been detected using other genetic markers such as effector genes (A. W. Njoroge, M. Ghislain, B. Andersson, E. Magembe, C. Mutai, R. Pelle, J. Yuen, and G. Forbes, unpublished data) and alloenzymes (Vega-Sanchez et al. 2000). Moreover, although these subpopulations appeared to be genetically similar, there were no shared SSR MLGs, further supporting the existence of host specialized subpopulations. No tomato samples from Burundi were available for this study; hence, the lineage of the P. infestans on tomato in that country remains unknown.

Implications of population change.

To effectively manage late blight, understanding how pathogen populations have changed over time can enhance the effectiveness of current management strategies, including host resistance management, fungicide optimization, and the use of integrated pest management tools. In Europe, knowledge of pathogen population changes is used to upgrade highly structured and sophisticated decision support systems from both public and private sectors (Cooke et al. 2011). In EAR and other low income regions, however, such systems do not exist and managing late blight is more of an individual affair, depending on the disease management resources available to a particular farmer.

The results indicating that tomato and potato can serve as alternative hosts in the EAR would argue for a stricter control of garden plants and volunteers, which may serve as sources of initial inoculum. However, these plants are regarded as legitimate sources of food by most small-scale farmers in the region and it is doubtful that they would be willing to remove them.

Population change in P. infestans has been associated with the reemergence of strains that are able to overcome modern control techniques such as fungicides with improved efficacy and host resistance (Fry 2008; Kirk et al. 2009). The EAR has had a US-1 population characterized by isolates with high phenylamide insensitivity (Mukalazi et al. 2001; Vega-Sanchez et al. 2000) and it is highly likely that 2_A1 has also developed insensitivity to these products, or it would not have spread so quickly. Farmers in the region have traditionally been able to treat potato with fewer sprays than predicted by simulation modeling and it was hypothesized that this was due to the reduced pathogenicity of US-1 (Forbes et al. 2009). If this was the case, one could expect that farmers in the EAR will now need to adapt their spray regimes to a new pathogen population that is more difficult to manage. Some growers, particularly in Kenya, recently complained of an increase of severe stem blight, which is difficult to manage because fungicides give poor coverage of stems when the canopy is full. This results in the disease continuing to spread and kill the crop rendering the efforts to use the chemical control ineffective. Although there is no evidence to indicate that this problem is a direct result of the presence of 2_A1, increased stem infections have previously been associated with pathogen population change (Cooke et al. 2011).

When resistant cultivars are released in late-blight-prone areas, pathogen monitoring for factors that could help overcome resistance durability would aid in host resistance deployment. Consequently, the occurrence of the same pathogen strains in different countries within a region means the breeding programs and growers can use similar host cultivars or at least know what resistance genes are appropriate for their region. The International Potato Center and the Ugandan authorities have developed cultivars with cassettes of known resistance genes for release in Uganda and eventually other countries in the region. Monitoring of effector evolution in the pathogen population would be an important tool for increasing the durability of this resistance. Use of appropriate resistance genes and effective fungicides are some of the measures that can lessen the devastating effects of new pathogen strains, despite the fact that new management strategies are difficult to implement in grower communities (Fry 2008; Gachango et al. 2012).

It will also be important to ensure that any population surveillance in the region is coordinated and uses standardized procedures. Microsatellite genotyping provides a fast and reliable tool to track pathogen changes and compare the findings with other networks in Europe, Asia, and the Americas but these meta-analyses can only be done if data are produced using standardized procedures.