An update on the identity of Longitarsus (Coleoptera: Chrysomelidae) species introduced or relocated in Canada for the biological control of tansy ragwort (Asteraceae)

Abstract

Starting in the 1960s, the flea beetle Longitarsus jacobaeae (Waterhouse) (Coleoptera: Chrysomelidae) has been imported and relocated to North America for the control of the weed tansy ragwort, Jacobaea vulgaris (Asteraceae). Some Longitarsus species are morphologically almost indistinguishable, leading to questions regarding the taxonomic identity of the beetles released. Here, we present the results of a molecular study that addresses these questions and updates a list of the releases of Longitarsus within Canada. Our findings confirm the suspected release of the cryptic species L. flavicornis concurrent with releases of L. jacobaeae at sites in British Columbia, Canada, in the 1970s. However, we find no evidence for the continued presence of L. flavicornis at historical release sites in British Columbia nor in Nova Scotia, Canada, where it might potentially have been redistributed. We confirm the presence of L. gracilis Kutschera in Nova Scotia and British Columbia, the latter population likely established from a 2005 release of Longitarsus from Nova Scotia. To our knowledge, this study provides the first published confirmation of L. gracilis in Canada. Finally, we provide evidence that what were recognised in earlier work as three separate taxa (L. succineus (Foudras), L. near noricus, and Longitarsus sp.) are likely all L. succineus.

Introduction

Tansy ragwort, Jacobaea vulgaris (synonym Senecio jacobaea, Asteraceae), is a biennial or short-lived perennial weed of Eurasian origin. In North America, it is widespread along the Pacific Coast, from California in the United States of America to British Columbia, Canada, and along the Atlantic Coast in the northeastern United States of America and the Atlantic provinces of Canada (i.e., New Brunswick, Newfoundland and Labrador, Nova Scotia, and Prince Edward Island), but it also has made inland incursions from both coasts since its arrival in the 1800s (Bain 1991; Winston et al. 2011). The weed currently occurs in eight provinces and 14 states (Natural Resources Conservation Service 2023), where it invades disturbed sites, including roadsides, recently logged or burned forest stands, and agricultural settings. Agriculturally, it impacts the yields of forage species of perennial pastures and native rangelands used in livestock production (Winston et al. 2011). However, the plant also produces highly toxic pyrrolizidine alkaloids that can cause livestock death (Wiedenfeld 2011). Animals normally avoid grazing on live plants but can ingest toxic levels of alkaloids from foliage incorporated in hay or silage (Pethick 1983; de Lanux-Van Gorder 2000; Walsh and Dingwell 2007; Stegelmeier 2011).

Research on Longitarsus jacobaeae (Waterhouse) (Coleoptera: Chrysomelidae) as a candidate biocontrol agent for tansy ragwort began in the 1960s (Harris et al. 1984), with early studies identifying Italian and Swiss biotypes (Frick 1970). The Italian biotype has proven to be more adapted to the coastal climates invaded by the weed (i.e., hot, dry summers and mild, wet winters), with a summer aestivation and late-autumn emergence/reproductive period. The Swiss biotype was identified as being more adapted to colder, continental climates where tansy ragwort also can be invasive (e.g., Montana, the United States of America, and the southern interior of British Columbia). It lacks the summer aestivation of the Italian biotype and breeds in early summer. Both biotypes were introduced in California in 1969 (Turner and McEvoy 1995). Only the Italian biotype was determined to have established in coastal regions of the state to provide successful control of J. vulgaris (McEvoy et al. 1991; McEvoy and Coombs 1999), as later supported by molecular studies (Puliafico 2003; Szűcs et al. 2011). Beetles from California were subsequently redistributed to sites throughout the Pacific Northwest, including coastal British Columbia (Harris et al. 1984), where they also have proven effective (Winston et al. 2011).

In Canada, multiple introductions and notable relocations of L. jacobaeae into new ecoclimatic areas were made up until 2013, using beetles from different source populations (Table 1). Initial introductions were made in the south–coastal region of British Columbia, in Vancouver – University of British Columbia rearing plots (1971, 1973), Abbotsford (1971, 1972, 1974), and Nanaimo (1974, 1976), using beetles from England, Switzerland, and Italy (Harris et al. 1984). Those beetles originating from Italy were obtained from established populations of the Italian biotype in California and Oregon, United States of America (Harris et al. 1984). First established populations of Longitarsus from the Abbotsford and Nanaimo releases then provided source material for redistributions on tansy ragwort within British Columbia (1978–2001), with the majority released in the south–coastal area, where they continued to perform well (Turner and Cesselli 2013). However, subsequent redistributions of coastal beetles to the British Columbian Interior (i.e., Naramata, south Okanagan Valley, 1992, 1994, 1997, 2001; Turner and Cesselli 2013) failed to establish, perhaps due to a climate mismatch involving the Italian biotype. The more cold-adapted Swiss biotype was introduced (via Montana) into the same area in 2011–2013, where it successfully established (Turner and Cesselli 2013; Winston et al. 2023). Currently, the biotype of the L. jacobaeae introduced from England to Canada in the 1970s is unknown, nor is it known why L. jacobaeae were sourced from England at that time.

Table 1. Historically significant introductions and relocations of Longitarsus jacobaeae in Canada, 1971–2013

* Harris et al. (1984).

^† UBC, University of British Columbia.

^‡ At least some of the beetles released were subsequently determined to be L. flavicornis (from voucher specimen labels and personal communication, P. Harris).

^§ From original release records held at the Lethbridge Research and Development Centre, Agriculture and Agri-Food Canada, Lethbridge Alberta, and Agriculture Canada (Ottawa) internal reports, Insect Liberations in Canada, Parasites and Predators (1971, 1973, 1974, 1976, 1978, 1981, 1984, 1986, 1988, 1989, 1990, 1991).

^|| Turner and Cesselli (2013).

An effort also was made between 1978 and 1991 to redistribute L. jacobaeae from British Columbia into Atlantic Canada, that is, Prince Edward Island, New Brunswick, and Nova Scotia (Harris et al. 1984; Winston et al. 2023). During this period, a few releases also were made in southern Ontario (Klimaszewski et al. 2020). Due to low beetle numbers initially found at the Atlantic sites after release, a poor biotype–climate match is suspected (Harris et al. 1984; Julien and Griffiths 1998). However, which biotype(s) may have been involved at the time is unclear because the beetles were redistributed from established populations that had received release material from three different sources (Italy, England, and Switzerland; Harris et al. 1984). Although establishment is currently confirmed for all Atlantic provinces where L. jacobaeae was released (Winston et al. 2023), some uncertainty remains on the occurrence of L. jacobaeae within these provinces (LeSage 1988; Hoebeke and Wheeler 2005; Klimaszewski et al. 2020).

The presence of other Longitarsus species further complicates the identification of what has been established or was relocated from previous releases. These other species either were adventive at sites at the time of release of L. jacobaeae or were introduced unintentionally with the biocontrol agent. For beetles from England introduced in south–coastal British Columbia (a key source for subsequent Canadian redistributions), morphological examination of voucher material identified the occurrence of L. flavicornis Stephens (Shute 1975). This latter species is morphologically almost indistinguishable from L. jacobaeae (Shute 1975; Government of British Columbia 2015) and therefore may have been redistributed with L. jacobaeae into Atlantic Canada. Two other species of Longitarsus associated with Asteraceae have been reported to occur in Canada (LeSage 1988; Klimaszewski et al. 2020). For L. ganglbaueri Heikertinger, which feeds on species of Senecio, Westcott et al. (1985) reported specimens from multiple sites in Oregon and a single specimen from Glenlea, Manitoba, Canada, concluding that the species is likely native to North America. Conversely, LeSage (1988) suggested that L. ganglbaueri was probably present in shipments of L. jacobaeae from Italy that were released in Oregon. Populations of L. ganglbaueri have since been identified in Québec, Nova Scotia, and Prince Edward Island (Hoebeke and Wheeler 2005; Bousquet et al. 2013). Of European origin, the second species L. succineus (Foudras) feeds on species of several Asteraceae genera, including Jacobaea (LeSage 1988). It has been reported only in Newfoundland and Labrador (LeSage 1988; Bousquet et al. 2013).

Yet another species of uncertain presence or history in Canada, Longitarsus gracilis Kutschera is reported to feed on tansy ragwort and other Asteraceae in Europe (Kevan 1967). Although first mentioned in a provincial government report as being released on tansy ragwort in the central Okanagan Valley (Kelowna) in 2005 using beetles collected in Nova Scotia (Turner and Cesselli 2013), the species has not been formally reported in the scientific literature as present in Canada (Bousquet et al. 2013; Klimaszewski et al. 2020). Thought at the time of release in Kelowna to be L. jacobaeae, subsequent morphological examination by Chrysomelidae taxonomists (L. LeSage, Agriculture and Agri-Food Canada, Ottawa, Ontario, and R. Beenen, the Netherlands) of specimens collected from the Kelowna site in 2008 revealed only L. gracilis; that is, no L. jacobaeae were present within the sample of 44 beetles examined (De Clerck-Floate, unpublished data).

The current molecular study was undertaken with a twofold purpose: (1) to provide an update, with confirmations on the history of Longitarsus introductions and relocations in Canada, of Harris et al.’s (1984) summary of releases to 1981 and of key releases that occurred from 1981 to 2013, and (2) to clarify the identity of Longitarsus species at sites of biological control releases made on tansy ragwort in British Columbia and in Nova Scotia. Previous studies have documented the utility of CO1 gene sequencing as a tool for species determinations of Longitarsus species (Baselga et al. 2013; Magoga et al. 2018; Salvi et al. 2020; Schwarz et al. 2021).

Materials and methods

Source material

We sequenced DNA from pinned voucher specimens and from specimens stored in 70 or 95% ethanol (Table 2). The pinned voucher material comprised specimens in the main insect collection of Agriculture and Agri-Food Canada’s Lethbridge Research and Development Centre, (Lethbridge, Alberta, Canada). Based on specimen label information and personal communications with P. Harris (Agriculture and Agri-Food Canada) in the 1990s by R.D.-F., vouchers were collected in 1983 from an established Longitarsus release site at Nanaimo, British Columbia (listed in Harris et al. 1984 as a release made in 1974 using material from England) and identified as L. flavicornis (also some as L. jacobaeae) in 1985 by S.L. Shute (British Museum of Natural History, London, United Kingdom) using morphological characters. Other pinned vouchers were of the Swiss biotype of L. jacobaeae collected in 2003 at St. Imier, Switzerland, and from a field population of this biotype in Montana established using beetles originating from the St. Imier collection. The identity of both of these specimen sets was confirmed, based on morphology by L. LeSage (Agriculture and Agri-Food Canada, Ottawa) before releases of the Swiss biotype in British Columbia in 2011–2013 (R. De Clerck-Floate, unpublished data). Also used were pinned voucher specimens morphologically determined by L. LeSage to be L. gracilis that were collected in 2008 from the established release of beetles from Nova Scotia in Kelowna, British Columbia.

Table 2. Specimens sequenced for the present study. Sequences used for the similarity analyses are deposited in GenBank and assigned accession numbers. Full details of all specimens are provided in Supplementary material, Table S1.

Specimens stored in ethanol were collected from tansy ragwort at or near to sites in British Columbia and Nova Scotia, where Longitarsus were previously released for biological control of the weed. In 2010, specimens from British Columbia were collected at the aforementioned Kelowna site, where a release was made in 2005 using material relocated from Nova Scotia, and in 2013, they were collected at sites where Longitarsus were released in the 1970s (i.e., the first introductions to Canada). Specimens from Nova Scotia were collected in 2014 at locations as near as possible to sites where beetles from British Columbia had been released between 1984 and 1991 (Table 2).

DNA extraction and amplification

Genomic DNA was extracted from either the leg or the whole body of individual specimens using DNeasy^® Blood & Tissue Kits (Qiagen Group, Hilden, Germany). A portion of the mitochondrial DNA cytochrome c oxidase subunit 1 (CO1) gene was amplified from the extracted DNA, using the universal primers HCO 2198 (5′-TAA ACT TCA GGG TGA CCA AAA AAT CA-3′) and LCO 1490 (5′- GGT CAA CAA ATC ATA AAG ATA TTG G – 3′; Folmer et al. 1994) and the general methods described in Hebert et al. (2003). Amplicon sequencing was done at the University of Calgary (University Core DNA Services, Calgary, Alberta, Canada).

Similarity analysis

The CO1 sequences obtained for the present study were pooled with CO1 sequences from GenBank to assist species identification (Table 2). We also included sequences for beetles recovered during an earlier unpublished European study that assessed the host specificity of L. succineus as a potential biological control agent for common tansy (Tanacetum vulgare Linnaeus) (Asteraceae) (Gassmann et al. 2012). In doing so, we hoped to resolve the identities of what the earlier study described as three taxa that included L. succineus, which we anticipated might be one of the species recovered in our samples from Nova Scotia.

Sequences were trimmed and aligned using the ClustalW method, resulting in sequences of 591 base pairs. The sequences were used to obtain a “best fit” similarity tree, derived from 1000 iterations of a distance matrix using the neighbour-joining method. For this method, we used a nucleotide substitution type, the Kimura two-parameter model, d: transitions + transversions for substitutions, pairwise deletion of gaps/missing data, and uniform rates among sites. Mean genetic distance within species was calculated using the maximum composite likelihood method with a bootstrap test of 1000 repetitions. Sequence analysis and tree construction were done in Mega 7 (Kumar et al. 2016). When multiple identical sequences were available from the same locality, only one or two of the sequences were used in the analysis. Sequences used for analyses were deposited in GenBank under accession numbers PP061620–PP061639 (Table 2).

The similarity analysis organised sequences into groups or clades. Genetic variation within these clades was measured as the number of base-pair substitutions per site (± standard error), averaged for all pairs of sequences within the clade. These analyses were done using a maximum composite likelihood model (Tamura et al. 2004) using MEGA11 (Tamura et al. 2021).

Results and discussion

We obtained 154 sequences for the present study, including 32 unpublished sequences from the aforementioned unpublished European study (Table 2; Supplementary material, Table S1). A preliminary analysis identified many identical sequences for specimens within populations (Supplementary material, Fig. S1). In combination with sequences in GenBank, similarity analyses identified three main clades that we identify here as (1) L. gracilis, (2) L. flavicornis and L. jacobaeae, and (3) L. succineus (Fig. 1).

Figure 1. Similarity analysis showing the separation of Longitarsus CO1 sequences into three clades: (1) L. succineus, (2) L. jacobaeae and L. flavicornis, and (3) L. gracilis. Sequences obtained in the present study are noted in bold. Bootstrap values lower than 70 are not reported.

Sequences for the first clade comprising L. gracilis derived from specimens collected in British Columbia and Nova Scotia (Fig. 2). At the British Columbia (Kelowna) site, Longitarsus beetles from Nova Scotia were released in 2005 and reportedly included L. gracilis (Turner and Cesselli 2013). No evidence was provided in the report to support this assertion, but it may have been based on a personal communication to the report authors by R.D.-F. She learned that L. gracilis was present at the site from L. LeSage, who examined the morphology of beetles collected from the site in 2008. Our molecular findings from beetles recovered from the site in 2008 and 2010 confirm the presence of L. gracilis in British Columbia. We also obtained 30 sequences corresponding to L. gracilis for specimens recovered from four locations in Nova Scotia in 2014 (Table 2; Supplementary material, Table S1), which suggests well-established populations of this species in that province. These combined findings support the earlier report of beetles from Nova Scotia providing the source L. gracilis population in British Columbia (Turner and Cesselli 2013) and provide the first published record of L. gracilis in Canada. We note, however, the presence of an unpublished L. gracilis DNA sequence in GenBank (MG057846) for a beetle collected in 2016 by C. Rumbolt at St. John’s, Newfoundland and Labrador.

Figure 2. Sites of historically significant introductions and relocations of Longitarsus jacobaeae in Canada, 1971–2013. Source populations for releases are identified in Table 1. Colours indicate which species were detected in the present study at or near release sites. No specimens were examined for the present study from sites marked as “unknown.” Map modified from https://www.printablemaps.net.

Sequences for the second clade originate from the almost morphologically indistinguishable species L. flavicornis and L. jacobaeae (Shute 1975). Sequences for L. flavicornis derive from two pinned voucher specimens collected at Nanaimo, British Columbia, in 1983, which are associated with some of the first introductions of L. jacobaeae to Canada from Europe, including England, in 1974 (Table 1). The genetic separation between sequences for L. flavicornis and L. jacobaeae supports the interpretation of two taxa in this clade (but see the section, Use of barcodes in Longitarsus taxonomy, in this paper). This finding also supports the earlier finding by taxonomist S.L. Shute that both L. flavicornis and L. jacobaeae were present at the site in England from which beetles were collected for use in the initial Canadian biological control releases, consistent with the hypothesis that L. flavicornis was introduced to Canada at that time (Shute 1975). Despite the morphological confirmation that L. flavicornis was present (with L. jacobaeae) at the Nanaimo site nine years after release, no sequences corresponding to L. flavicornis were recovered for beetles collected at the site in 2013, suggesting that the species is no longer present. Similarly, we did not recover any sequences corresponding to L. flavicornis from sites in Nova Scotia, where specimens originating from Nanaimo were released in the 1970s to 1990s. Hoebeke and Wheeler (2005) previously reported the collection of numerous Longitarsus specimens from tansy ragwort in Nova Scotia over a 10-year period that, based on examination of male genitalia, were all L. jacobaeae. Thus, if L. flavicornis were among the Longitarsus collected at Nanaimo and released into Atlantic Canada, it does not appear to have been established. Shute (1975) noted that L. flavicornis “proved” to be less successful than L. jacobaeae in the Canadian climate but did not provide context for this statement.

Sequences in the second clade for L. jacobaeae derive from specimens collected at multiple locations (Fig. 2). These included sites at or near original sites of release at Nanaimo and Abbotsford, British Columbia, and from five sites in Nova Scotia (Table 2; Supplementary material, Table S1). Sequences in this clade also were obtained using processed pinned voucher specimens from a lineage originating from field collections at St. Imier, Switzerland, and at a subsequent 2003 field release in Montana using beetles collected from the St. Imier site (i.e., the Swiss biotype of L. jacobaeae). The latter established population then provided the source material for introductory releases of the lineage near Naramata, British Columbia, in 2011–2013 (Table 1).

All of the remaining sequences obtained in the present study represent a third clade that we identify as L. succineus (Table 2; Supplementary material, Table S1). Sequences for this clade derived from material collected in the earlier mentioned unpublished European study on L. succineus host specificity (Gassmann et al. 2012). For that work, Longitarsus beetles were collected from T. vulgare in St. Petersburg, Russia. Newly hatched larvae of these beetles were subsequently transferred onto T. vulgare and species of Asteraceae for a common garden study in Delémont, Switzerland. The sequences represent both specimens collected from St. Petersburg and those reared from plants in the common garden. The latter presumably include the F₁ offspring of the beetles from St. Petersburg but also, possibly, the Longitarsus species native to the Jura region.

The interpretation of the unpublished study was that these sequences represented three taxa corresponding to L. succineus (represented in Figure 1 by PP061622, PP061623, and PP061624), Longitarsus near noricus (represented in Figure 1 by PP061620 and PP061621), and an unidentified Longitarsus species (represented in Figure 1 by PP061625, PP061626, PP061627, and PP061628; Gassmann et al. 2012). However, Gassmann et al. (2012) noted that further phylogenetic study was needed to confirm their interpretation. Voucher sequences for L. succineus have since become available in public databases (Baselga et al. 2013; Magoga et al. 2018; Salvi et al. 2020) that better characterise genetic diversity within the species. We now have greater confidence that what was recognised as three taxa are likely all L. succineus (but see the section, Use of barcodes in Longitarsus taxonomy, in the present paper).

Use of barcodes in Longitarsus taxonomy

Longitarsus is a taxonomically diverse genus represented by more than 700 species worldwide (Salvi et al. 2020). The genus requires taxonomic revision that will likely identify new undescribed species (Riley et al. 2002). Longitarsus species can be difficult to identify by external morphology (Shute 1975; LeSage 1988), such that accurate identification may require comparison of male genitalia (LeSage 1988) or molecular sequences (Dobler 2001; Puliafico 2003; Salvi et al. 2020; Roslin et al. 2022) against suitable voucher material.

Two studies that assess the use of molecular sequences for Longitarsus species determination identify challenges that temper our conclusions. The first challenge is the availability of suitable voucher sequences linked to specimens for which species determinations have been made by qualified taxonomists (Salvi et al. 2020). We were unable to locate suitable voucher sequences for comparison to sequences that we obtained in the present study for L. flavicornis. The only sequence we could locate for this taxon (GenBank accession no. HQ164779) contained anomalies that negated its use. We also were unable to locate voucher sequences for L. noricus. Doing so would have helped clarify the status of the clade identified in the unpublished European study as Longitarsus near noricus (Gassmann et al. 2012).

The second challenge is variation in the expected congruence of CO1 barcodes with morphologically discernable species. Although this congruence exists for the majority of Longitarsus species that have been assessed, there are exceptions (Baselga et al. 2013; Salvi et al. 2020) that may include the results of the present study. Genetic variation in the clade that we identify as containing only L. succineus is 0.017 (± 0.003) and in the clade containing only L. gracilis is 0.012 (± 0.002). In contrast, genetic variation within the clade that we identify as containing the two species L. jacobaeae and L. flavicornis is 0.014 (± 0.003). This finding suggests that there can be more variation in CO1 gene sequences within a species than may be present between two closely related species. Further study is required to clarify this matter.

In summary, we report the following three findings based on the evidence available. First, we find no evidence for the continued presence of Longitarsus flavicornis at sites of historical release in British Columbia nor in Nova Scotia where it might potentially have been redistributed. This negative finding does not exclude the possibility that the species is at these sites. However, if it is present at the sites (or elsewhere in Canada), further study is needed to provide confirmation. Second, we confirm the presence of L. gracilis in both British Columbia and Nova Scotia, and we now believe it likely that the population established in British Columbia came from a 2005 release of Longitarsus originating from Nova Scotia. Third, we identify as L. succineus the specimens originating from St. Petersburg that were recognised in earlier work as three separate taxa; that is, L. succineus, L. near noricus, and Longitarsus sp.

The initial releases of Longitarsus into North America for control of tansy ragwort occurred in the 1970s and 1980s when molecular barcoding was not available for species determination. The present study highlights the value of this method, particularly for use with species that are morphologically difficult to distinguish. As such, the use of DNA barcoding is a recommended practice for the import and release of arthropod biological control agents in Canada. Doing so increases confidence in morphological determinations of agents and facilitates the detection of possible cryptic species, thereby allowing greater certainty in the safety (i.e., host specificity) and efficacy of species introduced or relocated for biological control (Mason et al. 2017).