Species of Dickeya and Pectobacterium Isolated during an Outbreak of Blackleg and Soft Rot of Potato in Northeastern and North Central United States

An outbreak of bacterial soft rot and blackleg of potato has occurred since 2014 with the epicenter being in the northeastern region of the United States. Multiple species of Pectobacterium and Dickeya are causal agents, resulting in losses to commercial and seed potato production over the past decade in the Northeastern and North Central United States. To clarify the pathogen present at the outset of the epidemic in 2015 and 2016, a phylogenetic study was made of 121 pectolytic soft rot bacteria isolated from symptomatic potato; also included were 27 type strains of Dickeya and Pectobacterium species, and 47 historic reference strains. Phylogenetic trees constructed based on multilocus sequence alignments of concatenated dnaJ, dnaX and gyrB fragments revealed the epidemic isolates to cluster with type strains of D. chrysanthemi, D. dianthicola, D. dadantii, P. atrosepticum, P. brasiliense, P. carotovorum, P. parmentieri, P. polaris, P. punjabense, and P. versatile. Genetic diversity within D. dianthicola strains was low, with one sequence type (ST1) identified in 17 of 19 strains. Pectobacterium parmentieri was more diverse, with ten sequence types detected among 37 of the 2015–2016 strains. This study can aid in monitoring future shifts in potato soft rot pathogens within the U.S. and inform strategies for disease management.


Introduction
In 2020, potatoes were, by weight, the fifth most produced food crop globally, behind sugar cane, maize, wheat, and rice [1]. The United States was the fifth top country, producing 41.5 billion pounds of potatoes valued at $3.9 billion in 2020 [1,2]. Pectolytic soft rot diseases cause annual field and storage losses in potato production [3][4][5]. Severe outbreaks of soft rot in 2014 in particular led to yield and crop losses across the Northeastern and North Central U.S. [6].

DNA Extraction and Amplification
DNA extractions were performed using DNeasy Blood and Tissue kit (Qiagen). Three loci were targeted for analysis: dnaJ, dnaX, and gyrB [56]. PCR mixtures contained 5 µL

Multilocus Sequence Analysis (MLSA)
Nucleotide sequences obtained by direct amplification as described above and from GenBank, ASAP, and collaborators, were aligned, trimmed to a consistent length, and concatenated using CLC Main Workbench (Qiagen, Germantown, MD, USA). Fragment lengths for each gene locus were selected as previously described [56]. Sequence fragments and sequences were concatenated in the following order: dnaJ (672 bp), dnaX (450 bp), gyrB (822 bp for Dickeya spp., 711 bp for Pectobacterium spp.), with a total concatenated length of 1944 bp for Dickeya spp. and 1833 bp for Pectobacterium spp. Evolutionary model testing was run in CLC Main Workbench which determined the general time reversible model (GTR+G+T) to be the best fit model for this data set. Phylogenies were inferred via Bayesian analysis using Bayesian Evolutionary Analysis Sampling Trees (BEAST 1.8.2) assuming a strict molecular clock and 10 million generations [80]. Output from BEAST was analyzed in Tracer v1.6.0 [81] and phylogenetic trees were constructed in FigTree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree, accessed on 11 August 2021).

Multilocus Sequence Typing (MLST)
Sequence types were assigned for 25 strains of D. dianthicola and 39 strains of P. parmentieri to further characterize diversity within these groups. Unique haplotypes for each locus and for concatenated sequences were identified using DnaSP v5 [82]. Minimum spanning trees were constructed in PHYLOViZ 2.0 to infer distance-based relativity of STs and predict founder STs within clonal complexes [83].

Nucleotide Accession Numbers
Sequences were deposited in GenBank and assigned to following accession numbers: MW978791 to MW979234.

Phylogeny of 2015-2016 and Reference Strains of Dickeya spp.
The phylogeny predicted by concatenated dnaJ, dnaX, and gyrB sequences placed several reference and 2015-2016 strains, some previously identified only to genus, within clades containing type strains of D. chrysanthemi, D. dianthicola, D. dadantii, and D. zeae (Figure 1; Tables 1-3). All strains identified previously as D. dianthicola were grouped in the same MLSA clade as the D. dianthicola type strain, LMG 2485 T [13,15,22]. Seven strains in the 2015-2016 collection identified by their providers as Dickeya sp. Were also grouped with LMG 2485 T , as was reference strain Dickeya sp. 600, which was isolated from a sweet potato in Georgia. Five potato strains from Hawaii were most closely related to the type strain of D. dadantii subsp. dadantii, as were four strains from maize. Two strains isolated from pond water adjacent to potato fields, PA24 and ST64, in the 2015-2016 collection were related to D. dianthicola, while two others, 16H2-68-A and 16H2-68-B were most closely related to D. zeae.
Some reference strains, but none of the 2015-2016 strains, were grouped with type strains of D. aquatica, D. chrysanthemi pv. parthenii, D. solani, D. dadantii subsp. dieffenbachiae, D. fangzhongdai, or D. paradisiaca (Figure 1 and Tables 1 and 3). Only D. aquatica DW 0440, a reference strain isolated from river water in Finland, was identified as D. aquatica [54]. In our analyses, one reference strain, Dickeya sp. 699, was identified as D. dadantii subsp. dieffenbachiae, while the subspecies status of D. dadantii strain 655 was ambiguous. Reference strain Dickeya sp. K1015 was most closely related to D. chrysanthemi pv. parthenii. The D. solani clade contained all six reference strains initially identified as D. solani: MK10, MK16, IFB 0099, GBBC 2040, D s0432-1, and RNS08.23.3.1.A. The MLSA phylogeny placed four reference strains recently classified as D. fangzhongdai within the same clade as the type strain of this species DSM 101947 T (Figure 1, Table 3).  (Figure 1 and Tables 1 and 3). Only D. aquatica DW 0440, a reference strain isolated from river water in Finland, was identified as D. aquatica [54]. In our analyses, one reference strain, Dickeya sp. 699, was identified as D. dadantii subsp. dieffenbachiae, while the subspecies status of D. dadantii strain 655 was ambiguous. Reference strain Dickeya sp. K1015 was most closely related to D. chrysanthemi pv. parthenii. The D. solani clade contained all six reference strains initially identified as D. solani: MK10, MK16, IFB 0099, GBBC 2040, D s0432-1, and RNS08.23.3.1.A. The MLSA phylogeny placed four reference strains recently classified as D. fangzhongdai within the same clade as the type strain of this species DSM 101947 T (Figure 1, Table 3).

Phylogeny of 2015-2016 and Reference Strains of Pectobacterium spp.
The phylogenetic tree of Pectobacterium placed several strains from the 2015-2016 and reference collections into well-supported clades containing the type strains of P. atrosepticum, P. brasiliense, P. carotovorum, and P. parmentieri (Figure 2). In addition, strains of the 2015-2016 collection were most closely related to type strains of P. polaris, P. punjabense, and P. versatile. Minnesota strains CIR1140, CIR1152 and CIR1152 from potato were grouped with P. polaris NIBIO 1006 T . Five strains (CIR1015, CIR1026, CIR1028, CIR1163, CIR1164), all from potato originating in Minnesota or North Dakota, were most closely related to type strain P. punjabense SS95 T . P. versatile was the closest relative of three water strains from Maine (16H2-64-A, 16H2-64-B, 16ME-31) and four potato strains isolated in Maine, Minnesota, and North Dakota. Five strains isolated from potato tubers in Minnesota clustered with type strains of either P. versatile, P. parmentieri, or P. carotovorum. The other tuber isolate, CIR1183 was most closely related to P. versatile but was on a separate branch nearly as divergent from P. versatile as that of P. polaris and P. versatile. None of the 2015-2016 strains were found in MLSA clades corresponding to P. actinidae, P. aquatica, P. aroidearum, P. betavasculorum, P. cacticida, P. fontis, P. odoriferum, P. peruviense, P. polonicum, P. wasabiae, or P. zantedeschiae.  Some of the major phylogenetic clades predicted by concatenated dnaJ, dnaX, and gyrB sequence alignments were further subdivided into well-supported sub-lineages. This was especially evident in P. brasiliense and P. carotovorum, which contained sub-branching of greater divergence than detected within other Pectobacterium species. Two major branches of P. brasiliense were evident. One, which was further subdivided into four lineages, contained the type strain, eight strains from the 2015-2016 collection, and reference strains LMG 21370 and LMG 21372. The other was comprised of seven strains collected in 2015 and 2016 from Minnesota or North Dakota. Three groups of P. carotovorum were detected. Two strains, 16H2-LB and CIR 1080 were located on a branch that was distinct from the major clade containing P. carotovorum LMG 2402 T (Figure 2, Table 1).

Diversity within Dickeya dianthicola
Twenty-five Dickeya dianthicola isolates were included in this study, with six STs identified (Table S1). Most (17/19) D. dianthicola strains from the 2015-2016 collection were assigned sequence type 1 (ST1). ST1 strains came from Maine, New York, Pennsylvania, Massachusetts, and Florida. None of the strains from Netherlands, France, Belgium, or the United Kingdom were ST1. ST2 was found in only one strain isolated in 2016 from New York. The sweet potato strain, 600, from R. S. Dickey's collection, previously identified as Dickeya sp. and identified here as D. dianthicola, shared the same sequence type (ST5) as New Jersey strain 16NJ-12 1, and strains RNS0.4.9 and LMG 2485 T from France and from the United Kingdom, respectively ( Figure 3 and Table S1). Some of the major phylogenetic clades predicted by concatenated dnaJ, dnaX, and gyrB sequence alignments were further subdivided into well-supported sub-lineages. This was especially evident in P. brasiliense and P. carotovorum, which contained sub-branching of greater divergence than detected within other Pectobacterium species. Two major branches of P. brasiliense were evident. One, which was further subdivided into four lineages, contained the type strain, eight strains from the 2015-2016 collection, and reference strains LMG 21370 and LMG 21372. The other was comprised of seven strains collected in 2015 and 2016 from Minnesota or North Dakota. Three groups of P. carotovorum were detected. Two strains, 16H2-LB and CIR 1080 were located on a branch that was distinct from the major clade containing P. carotovorum LMG 2402 T (Figure 2, Table 1).

Diversity within Dickeya dianthicola
Twenty-five Dickeya dianthicola isolates were included in this study, with six STs identified (Table S1). Most (17/19) D. dianthicola strains from the 2015-2016 collection were assigned sequence type 1 (ST1). ST1 strains came from Maine, New York, Pennsylvania, Massachusetts, and Florida. None of the strains from Netherlands, France, Belgium, or the United Kingdom were ST1. ST2 was found in only one strain isolated in 2016 from New York. The sweet potato strain, 600, from R. S. Dickey's collection, previously identified as Dickeya sp. and identified here as D. dianthicola, shared the same sequence type (ST5) as New Jersey strain 16NJ-12 1, and strains RNS0.4.9 and LMG 2485 T from France and from the United Kingdom, respectively ( Figure 3 and Table S1).  dnaX, and gyrB). Sequence type numbers are given in black in each circle. Sizes of circles are relative to the number of individuals sharing the same ST. The relatedness between strains is indicated by relative distance as indicated by branch lengths. Geographic origins of strains are depicted by color: New York (blue); Maine (light blue); Pennsylvania (orange); Netherlands (light orange); Georgia (green); Michigan (light green); Massachusetts (red); New Jersey (pink); United Kingdom (purple); Belgium (light purple); France (brown); Florida (light brown).

Diversity within Pectobacterium parmentieri
Several (37) strains in the 2015-2016 collection clustered with type strain, RNS08.42.1A, of P. parmentieri, including strains from Michigan, Minnesota, New York, and North Dakota. Ten sequence types were identified within P. parmentieri (Figure 4 and Table S2). P. parmentieri RNS08.42.1A T from France was assigned ST2. ST2 was also detected in Minnesota, New York, and North Dakota. Strains with ST5 originated in Finland,

Diversity within Pectobacterium parmentieri
Several (37) strains in the 2015-2016 collection clustered with type strain, RNS08.42.1A, of P. parmentieri, including strains from Michigan, Minnesota, New York, and North Dakota. Ten sequence types were identified within P. parmentieri (Figure 4 and Table S2). P. parmentieri RNS08.42.1A T from France was assigned ST2. ST2 was also detected in Minnesota, New York, and North Dakota. Strains with ST5 originated in Finland, Minnesota, New York, and North Dakota. Multiple sequence types were found in some states. For example, strains from Minnesota and New York separated into five and six sequence types, respec-tively. While some STs, such as ST2 and ST5, were found in multiple states, ST8 and ST10 appeared only in Minnesota and North Dakota, and ST1, ST3, ST4, ST6, ST7, and ST9 were unique to particular states.
Minnesota, New York, and North Dakota. Multiple sequence types were found in some states. For example, strains from Minnesota and New York separated into five and six sequence types, respectively. While some STs, such as ST2 and ST5, were found in multiple states, ST8 and ST10 appeared only in Minnesota and North Dakota, and ST1, ST3, ST4, ST6, ST7, and ST9 were unique to particular states.

Discussion
The taxonomy of Pectobacterium and Dickeya has been refined substantially in the past ten years by the widespread adaption of DNA sequencing strategies in soft rot research. Classification and identification of soft rot bacteria by multilocus sequence alignments of concatenated sequences and by whole genome sequence comparisons has added clarification to the complex polyphyletic nature of some earlier-described species of Pectobacterium and Dickeya. In this study, all strains of Pectobacterium and Dickeya previously identified to species level by whole genome sequencing grouped with the corresponding type strains in phylogeny predicted by MLSA of dnaJ, dnaX, and gyrB. The clades and phylogenies predicted in this study using concatenated sequences of dnaJ, dnaX, and gyrB are also similar to those generated by single and multiple sequence alignments [15,35,37,51,56,58,63,[84][85][86]. The consistency among tree topologies generated by single, multiple, or entire genome sequences enables presumptive identification of unknown soft rot bacteria to species and provides insights on the diversity, ecology, and epidemiology of specific taxa [6,54,85,87,88].
The inclusion of several reference and type strains in our MLSA study provided new insights on species diversity of soft rot bacteria prior to the 2014 blackleg outbreak. MLSA was sufficient for identification of several reference strains previously identified only to the genus level.

Discussion
The taxonomy of Pectobacterium and Dickeya has been refined substantially in the past ten years by the widespread adaption of DNA sequencing strategies in soft rot research. Classification and identification of soft rot bacteria by multilocus sequence alignments of concatenated sequences and by whole genome sequence comparisons has added clarification to the complex polyphyletic nature of some earlier-described species of Pectobacterium and Dickeya. In this study, all strains of Pectobacterium and Dickeya previously identified to species level by whole genome sequencing grouped with the corresponding type strains in phylogeny predicted by MLSA of dnaJ, dnaX, and gyrB. The clades and phylogenies predicted in this study using concatenated sequences of dnaJ, dnaX, and gyrB are also similar to those generated by single and multiple sequence alignments [15,35,37,51,56,58,63,[84][85][86]. The consistency among tree topologies generated by single, multiple, or entire genome sequences enables presumptive identification of unknown soft rot bacteria to species and provides insights on the diversity, ecology, and epidemiology of specific taxa [6,54,85,87,88].
The inclusion of several reference and type strains in our MLSA study provided new insights on species diversity of soft rot bacteria prior to the 2014 blackleg outbreak. MLSA was sufficient for identification of several reference strains previously identified only to the genus level. Reference strains identified as Dickeya sp. in Kelman's and Dickey's collections were assigned to MLSA clades D. chrysanthemi, D. dadantii subsp. dadantii, D. dadantii subsp. dieffenbachia, D. dianthicola, and D. zeae. The placement of the sweet potato strain Dickeya sp. 600 within D. dianthicola is evidence that D. dianthicola was present in the USA prior to the 2014 blackleg outbreak. The four strains identified as D. fangzhongdai in the reference collection all originated outside the U.S. D. fangzhongdai has very recently been identified in New York as the cause of soft rot of onion [89]. Historic isolates of Pectobacterium versatile have been reported from the U.S. In a retrospective study of soft rot bacteria, Portier et al. identified four historic strains from the USA as P. versatile: three from potato isolated in 2001 and one from Iris sp. isolated in 1973 [37]. Reference strain Pectobacterium sp. Ecc71 was recently reclassified as P. versatile [30]. In our study, Ecc71 grouped with other P. versatile strains. The taxon includes soft rot strains isolated in Russia from potato and cabbage unofficially named "Candidatus Pectobacterium maceratum" [30,90].
MLSA and MLST of the 2015-2016 collection provides a broad view of the species of pectolytic soft rot bacteria present in the Northeastern and North Central U.S. immediately following the 2014 blackleg outbreak. The identification of P. parmentieri, P. brasiliense, P. carotovorum, and P. atrosepticum is consistent with prior reports of the occurrences of these species in the U.S. [15,20,21,23,24,26]. We did not expect to find nine strains of P. versatile in the 2015-2016 collection. P. versatile was recently reported in potato from New York and appears to represent a shift in 2017 from the P. parmentieri that predominated in 2016 [91]. Less expected was finding P. polaris and P. punjabense, as these have not been reported in the U.S. P. polaris was validly reported in 2017 and is found in Norway [52] and Morocco [85]. P. punjabense was reported in Pakistan [36]. No strains of P. polaris or P. punjabense originating from the U.S. were noted in the original species description or in recent examinations of other species of Pectobacterium [30,31,35,37,52,84]. Further polyphasic analyses are warranted to validate the MLSA identification of P. polaris or P. punjabense within the 2015-2016 collection.
The 2015-2016 collection contained D. dianthicola strains originating from multiple states. D. dianthicola has been reported previously in New York, Maine, Michigan, and New Jersey [6,13,15,16,22]. Our findings confirmed that D. dianthicola is also present in Florida, Massachusetts, and Pennsylvania. The majority (17/19) of D. dianthicola isolates in the 2015-2016 collection from Maine, New York, Pennsylvania, Massachusetts, and Florida are sequence type 1 (ST1). This result might be expected if soft rot outbreaks in the Northeastern U.S. were caused by a single (or limited) introduction of one strain of D. dianthicola. However, isolates of ST2 and ST5, from New York and New Jersey, respectively, were also identified within the 2015-2016 collection, which suggests the outbreak was not only clonal. A genetic diversity analysis of 256 isolates of D. dianthicola collected over 5 years concluded that the blackleg outbreak in Northeastern U.S. was caused by multiple strains [41]. Ge et al., 2021 identified three genotypes within D. dianthicola, the frequencies of which varied by year and by state [41]. Some D. dianthicola isolates reported by Ge et al. were also included in our study. Based on overlapping results, ST1 in our study corresponds well to Type I of Ge et al., 2021, while ST 2-6 strains are represented in Type II and III [41].
The 2015-2016 collection is not comprehensive, in that the number of strains from the contributing states varied and no systematic sampling strategy was applied for obtaining isolates over states. The variation in coverage by state resulted in different findings. For example, the deep coverage of strains from Minnesota and North Dakota led to first reports of some species and the likely identification of P. polaris and P. punjabense in the U.S. The abundance of strains of D. dianthicola and P. parmentieri in the collection enabled sequence type analyses, from which we can conclude that the genetic diversity within P. parmentieri was greater than that of D. dianthicola during those years and across states.
Most strains in the 2015-2016 collection originated from plants. The eight strains from water were enough to show that water can serve as a source of D. dianthicola, D. zeae, P. actinidae, and P. versatile. In a large comprehensive study of Dickeya in temperate regions of Central Europe, the diversity of Dickeya species recovered from water was different than that obtained from potato [45]. In that study, D. dianthicola was not recovered from water. Future studies involving larger numbers of aquatic strains could improve our understanding of the relative diversity of soft rot species in water and plant sources in the U.S. [6,13,15,16,22,41].
Four newly described species of Dickeya were validly published as we completed our analyses [27,29,86,92]. Future studies with two of these, D. oryzae and D. poaceiphila, would aid in resolution of isolates within the MLSA clade D. zeae. Since D. oryzae ZYY5 T was not included in our study, we did not delineate a clade of D. oryzae. NCPPB 3531, CSL RW192, DZ2Q, ZJU 1202, EC1, and the reference strain K1030 from maize clustered together in the MLSA phylogeny presented here. All these strains are currently identified as D. oryzae [86]. We also note that two water strains from Maine, 16H2-68-A and 16H2-68-B, and the reference strain, Ech586, are divergent of other strains within the D. zeae clade. This might reflect the natural diversity within D. zeae or an unresolved species assignment. In a phylogenetic study that included NCPPB 569, the type strain of D. poaceiphila [92] and the reference strain Ech586, Ech586 was placed in D. zeae [31,93]. It is likely that the water strains are also correctly placed within D. zeae; however, because D. zeae, D. poaceiphila, and D. chrysanthemi are closely related species within Dickeya, future studies of water strains 16H2-68-A and 16H2-68-B including the type strain of D. poaceiphila could verify their species assignment.
P. parvum is the only validly published and currently used species name of Pectobacterium not included in our study [84]. The inclusion of P. parvum might have provided insights on the identity of Pectobacterium sp. CIR1183, the 2016 isolate from Minnesota, which by dnaJ, dnaX, and gyrB phylogeny represented a separate branch within the P. polaris/P. versatile clade. The closest relatives of P. parvum are P. polaris and P. versatile. Isolates of P. parvum were described as a group of atypical, less virulent strains closely related to, but distinguished from P. polaris [84]. In the future, whole genomic sequence comparisons, evaluation of differentiating biochemical traits, toxicity tests in insects, and soft rot virulence assays that include the type strain of P. parvum might enable more accurate placement of Pectobacterium sp. CIR1183 within the P. versatile/P. polaris/P. parvum super clade.
While much progress has been made toward defining a stable, monophyletic taxonomy for species of Dickeya and Pectobacterium, some taxa remain notably complex and certain strains of soft rot bacteria do not align well with known species [37,63,84]. Two strains, classified here as P. carotovorum, CIR1080 and 16H2-LB, clustered together on a separate branch of P. carotovorum. These might belong to other species related to P. carotovorum. Similarly, the diverging branches of P. brasiliense strains within the 2015-2016 collection and reference strains is consistent with the conclusion that P. brasiliense is less homogenous than other species of Pectobacterium [87].
Phylogenetic relationships predicted by the MLSA schema of dnaJ, dnaX, and gyrB support and extend recent evidence that several newly described species of Dickeya and Pectobacterium cause soft rot diseases of potato in the U.S. The importance of specific species on disease occurrence, symptomology, and severity remains unclear. MLSA/MLST studies could be extended to study the contributions of within-and between-location genetic variability, multiple species and genotype combinations, and cultivar on disease severity. Comprehensive genomic comparisons of large numbers of representative strains within a given species will continue to advance stable, well-delineated species definitions with the goal of improving pathogen detection and disease management [6,41,87].
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.