A Successful Biocontrol Agent in the USA, Diorhabda Carinulata (Coleoptera: Chrysomelidae) on Tamarix Spp

Biological Control 136 (2019) 104002

Contents lists available at ScienceDirect

Biological Control

journal homepage: www.elsevier.com/locate/ybcon

A successful biocontrol agent in the USA, Diorhabda carinulata (Coleoptera: Chrysomelidae) on Tamarix spp. (Tamaricaceae), rejected in South Africa T due to insuﬃcient host speciﬁcity ⁎ Danica Marlina, , Etienne R. Smita, Marcus J. Byrnea,b a School of Animal, Plant and Environmental Sciences (APES), University of the Witwatersrand, Johannesburg 2050, South Africa b Centre for Invasion Biology, at the School of APES, University of the Witwatersrand, Johannesburg 2050, South Africa

GRAPHICAL ABSTRACT

ARTICLE INFO ABSTRACT

Keywords: Several countries globally, including South Africa, have been invaded by at least one of five species of Tamarix. Diorhadba spp. South Africa therefore considered using one or more species of leaf-feeding beetles in the genus Diorhabda Tamarix spp. (Coleoptera: Chrysomelidae), including Diorhabda carinulata, against invasive T. ramosissima and T. chinensis, fi Host-speci city since the beetles are highly damaging in the USA. The situation in South Africa is possibly more complicated South Africa than that in the USA because there is an indigenous species, Tamarix usneoides, which could potentially serve as a Field tests host for the beetles. To investigate this possibility, a series of field and laboratory host specificity tests were Host range conducted using D. carinulata against invasive target Tamarix species and the indigenous non-target T. usneoides. Field tests showed that D. carinulata had a preference for invasive Tamarix species, but readily settled and laid eggs on T. usneoides. Laboratory paired-choice tests showed that adult beetles preferred T. usneoides over T. chinensis and preferred T. ramosissima over T. usneoides, for both feeding and oviposition. Laboratory no-choice tests showed the egg-to-adult survival rate to be higher for individuals reared on T. usneoides than on T. ramosissima. Furthermore, the fecundity of females reared on T. usneoides was higher than that of females reared on T. ramosissima. Diorhabda carinulata is thus not a suitable biocontrol agent against invasive Tamarix in South Africa. An alternative biocontrol agent is currently being sought, and a short-list of candidate agents has already been compiled.

⁎ Corresponding author. E-mail address: [email protected] (D. Marlin). https://doi.org/10.1016/j.biocontrol.2019.104002 Received 5 December 2018; Received in revised form 27 May 2019; Accepted 6 June 2019 Available online 08 June 2019 1049-9644/ © 2019 Elsevier Inc. All rights reserved. D. Marlin, et al. Biological Control 136 (2019) 104002

1. Introduction usneoides × T. ramosissima (T.u. × T.r.). It was not possible to conduct all experiments on all five of the Tamarix taxa due to the limited In the past century, the number of tree species being moved around numbers of plants available, but several taxa were used within each the world and planted for financial profit has increased rapidly, and this experiment. has led to an increase in importance of trees as invasive species (Van For experiments conducted in South Africa, plants were propagated Wilge and Richardson, 2014). Tamarix L. (Caryophyllales: Tamar- from hardwood cuttings collected from trees of known genetic identity icaceae), commonly known as saltcedars, are riparian woody trees and (Mayonde et al., 2015, 2016) in the field, or provided by the Ecological, shrubs native to Eurasia and Africa. Five Tamarix species (T. aphylla (L.) Engineering and Phytoremediation Programme, University of the Wit- H. Karst, T. chinensis Lour., T. gallica L., T. ramosissima Ledeb. (and watersrand. Cuttings were planted in 4 L pots in a mixture of 2:1 silica hybrids thereof) and T. parviflora DC.) are listed as being invasive in the sand and compost, and fertilized once every two months with a slow Americas, Australia, Indian Ocean Islands and southern Africa release fertilizer (7:1:3 N:P:K). Initially the plants were kept in an in- (Rejmánek and Richardson, 2013). In South Africa, T. aphylla and T. door growth room with natural light and ambient temperature, and gallica have been reported as present (Henderson, 2001; Weiersbye were irrigated by hand twice a week. After one year, the plants were et al., 2006), however, genetic analysis has confirmed the presence of moved to an outside nursery where they were irrigated twice a day by only three ‘pure’ species of Tamarix and hybrids of these: the invasive T. an automated irrigation system. Different parental lineages of each chinensis and T. ramosissima, and the indigenous T. usneoides E.Mey. ex taxon were used to ensure that use of clones did not lead to pseudor- Bunge (Mayonde et al., 2015), with the invasion being dominated by eplication. Plants were used in experiments once they reached a height putative hybrids of T. chinensis and T. ramosissima (Newete et al., 2019; of 0.5 m. Mayonde et al., 2016, 2019). This has led to a search for host specific For field experiments in the USA, hardwood cuttings, as well as biological control agents of the alien species. small rooted plants were transported from South Africa to the USA. The The biological control programme for Tamarix in the United States imported South African cuttings were initially grown in a greenhouse at of America (USA), using four species of leaf-feeding beetles in the genus the Colorado State University and later moved to the Palisade Insectary Diorhabda Weise (Coleoptera: Chrysomelidae), has been very successful in Palisade, Colorado. The plants were irrigated once a day until the (Bean and Dudley, 2018), which has led to the recent initiation of a water dripped through the bottom of the pots. All of the South African similar biocontrol programme for invasive Tamarix in South Africa specimens grown in the USA were kept in pots and their buds were (Marlin et al., 2017). The first candidate to be tested as a biocontrol removed before flower burst to prevent possible cross-pollination and agent for invasive Tamarix spp. in South Africa was Diorhabda carinulata hybridisation with local Colorado plants. All imported plant material (Faldermann), which was introduced into quarantine in South Africa in was destroyed by burning once the experiments were completed. September 2016. In addition to its success in the USA (Bean and Dudley, 2018), D. carinulata was chosen because its native distribution 2.2. Insect culture (western China, southern Russia, Iran through to Mongolia) overlaps with that of T. ramosissima (Tracy and Robbins, 2009). Furthermore, All the Diorhabda carinulata beetles used in laboratory experiments host-specificity tests in the USA have revealed that the Diorhabda bee- in South Africa were collected from Garfield County, Colorado, USA, at tles do not perform as well on T. aphylla as they do on other Tamarix the coordinates 39°22′31.0″N 108°58′52.0″W. The first consignment of spp., having a delayed onset of oviposition, smaller adults and fewer five hundred adult beetles was imported to South Africa in September eggs (Milbrath and DeLoach, 2006a,b). Since the South African in- 2015 and the beetles were initially kept in quarantine at the Plant digenous T. usneoides is most closely related to T. aphylla (Baum, 1978; Protection Research Institute (PPRI) in Tshwane, South Africa. The Mayonde et al., 2015), it was envisaged that D. carinulata would behave culture was securely relocated to a quarantine laboratory at the in a similar manner in South Africa, by attacking T. chinensis, T. ra- University of the Witwatersrand in October 2015 and maintained at mosissima and their hybrids while avoiding T. usneoides. Whilst it may 23.48 ± 0.40 °C, RH of 63.99 ± 1.26%, and a combination of natural be more difficult to find a biocontrol agent that is highly damaging to light and a 16-hour artificial light cycle to prevent diapause induction. only one species and not another within the same genus, such re- Beetles were kept in 5 L cages, separated into generations, and fed on lationships are not unprecedented (for examples see Impson and Moran, bouquets of T. ramosissima × T. chinensis collected from the field (co- 2004; Baker et al., 2003; Githure et al., 1999; Olckers et al., 1995). ordinates 25°40′45.4″S 27°45′26.6″E). The culture reached very low A combination of laboratory-based and field host-specificity tests is numbers during the winter months, and a second consignment, con- ideal for determining a candidate biocontrol agent’s fundamental and sisting of 1000 adult beetles collected from the same source population ecological host range (Briese, 2005; Marohasy, 1998; Clement and in Colorado, was imported in October 2016. The biology of the Cristofaro, 1995). Although an accurate perception of an agent’s eco- Diorhabda beetles is described in Lewis et al. (2003). logical host range may not always be possible to achieve if the range of potential non-target hosts is large, in the present case the primary non- 2.3. Field sites in the USA target species of concern, T. usneoides, is the only indigenous congener, and it was therefore possible to use a combination of laboratory and Field sites in the USA were chosen based on the abundance of D. field tests to determine the host-specificity of D. carinulata, prior to its carinulata adults observed in previous years (D. Bean 2015 pers. potential release in South Africa. comm.). However, because beetle population numbers were extremely This study used laboratory-based choice and no-choice tests, in low in the summers of 2014 and 2015 due to unfavourable weather quarantine in South Africa, and field choice tests in cages, in the USA, conditions and low overwintering success in 2013 (D. Bean 2015 pers. to determine whether D. carinulata was safe for release as a biological comm.) experiments were conducted in large outdoor cages control agent of invasive Tamarix spp. in South Africa. (4.6 m × 4.6 m × 4.9 m) in August/September 2015 and August 2016. All cage experiments were set up outside the insectary in Palisade, 2. Materials and methods Colorado (coordinates 39° 6′ 46.1″ N 108° 21′ 1.4″ W).

2.1. Test plants 2.4. Laboratory trials in South Africa

Tests plants included the following taxa: pure T. ramosissima (T.r.), All replicates of a speciﬁed Tamarix taxon refer to plants from dif- T. chinensis (T.c.) and T. usneoides (T.u.), and hybrids of T. chinensis × T. ferent parental lineages, i.e. four replicates of T. chinensis means that ramosissima (T.c. × T.r.), T. usneoides × T. chinensis (T.u. × T.c.) and T. each of the four T. chinensis plants used in an experiment originated

2 D. Marlin, et al. Biological Control 136 (2019) 104002 from a different T. chinensis parent plant. 108° 21′ 1.4″ W). The following Tamarix taxa (n = 4 for each taxon) were used in both cages set up in August/September 2015: T. usneoides, 2.4.1. No-choice egg-to-adult development T. ramosissima, T. chinensis and T. ramosissima × T. chinensis hybrid. In Bouquets of three different Tamarix taxa were provided as a food August 2016, the following taxa (n = 4 for each taxon) were used: T. source for beetles as follows: four replicates of T. ramosissima, four re- usneoides, T. usneoides × T. ramosissima hybrid, T. usneoides × T. chi- plicates of T. usneoides × T. ramosissima, and five replicates of T. us- nensis hybrid and T. ramosissima × T. chinensis hybrid. In each outdoor neoides. Individual bouquets were placed into 5L cages, each inoculated mesh cage (4.6 m × 4.6 m × 4.9 m), potted Tamarix plants were set up with 20 D. carinulata eggs taken from the F5 to F7 generations of the randomly in a 4 rows × 4 columns Latin square design (Briese et al., first consignment, depending on availability. On hatching, the numbers 2002) in which each taxon occurred only once in each row and each of larvae and their developmental stage i.e. 1st, 2nd or 3rd instar, were column. All plants were of similar size, each plant was placed in a drip noted three times a week. Old bouquets were replaced with fresh tray to retain water, and watered daily. Four hundred field-collected, bouquets twice a week, and all eggs and larvae were moved to the new unsexed D. carinulata adults were placed in the centre of each cage and bouquets with a fine camel-hair brush. Larvae were weighed every allowed to disperse freely inside the cage. second day from 2nd instar until pupation. A 1.5 cm layer of sand was Trials (a) and (b) followed the methods of Briese et al. (2002) and added to the bottom of the buckets to facilitate pupation once the larvae comprised of two phases. In the first phase, the non-target plants (T. reached 3rd instar. Emerged adults were counted and weighed. usneoides and its hybrids) and the target plants were exposed to D. carinulata and the beetle’s host selection was recorded. In the second 2.4.2. No-choice egg-to-egg multi-generational development phase, the target taxa were removed from the system to create a no- Live, potted plants were used: six replicates of T. usneoides and four choice scenario for the beetles. Target taxa were cut down close to the replicates of T. ramosissima, which is the host plant of D. carinulata soil surface and the cuttings left in position. The beetles were thus given (Dalin et al., 2009), and acted as the control against which the non- the opportunity to move from the dead target plants to the non-target target, indigenous T. usneoides was compared. Each plant was kept in- plants (Briese et al., 2002). Trial (c) was not turned into a no-choice dividually in an insect cage (0.5 m × 0.5 m × 1 m). Fifty eggs, taken experiment because there were more D. carinulata adults on the T. us- from the F1 generation of the second beetle consignment, were placed neoides plants than on the target plants, thus eliminating the second onto a piece of paper towel, which was then positioned to be in contact phase of the experiment. The number of adults found on the different with the foliage of the plants. Upon hatching, the number of larvae and Tamarix taxa was counted on alternate days, in each experiment. The their developmental stage was recorded every second day. A 1.5 cm number of eggs found on the different Tamarix taxa was also counted in layer of sand was added to the bottom of the cages to facilitate pupa- the August 2016 experiment. tion. Upon emergence, adults were sexed and weighed and replaced into the same cages for a further two weeks in order to reproduce and 2.6. Statistical analyses lay eggs, after which the experiment was terminated. Newly laid eggs were counted and immediately removed from the plants every second 2.6.1. Laboratory trials in South Africa day for two weeks, so that the same eggs would not be counted twice. Despite various types of transformation, all data violated the as- The average number of eggs laid per female was then used to estimate sumptions of normality and homogeneity, therefore non-parametric the fecundity of females reared on each of the two Tamarix taxa. tests were used for comparisons of means. Data from the no-choice The suitability of T. usneoides and T. ramosissima as hosts of D. experiments were analysed using Kruskal-Wallis tests performed on carinulata was estimated using Maw’s host suitability index (Maw, ranks of data, to compare D. carinulata survival on the different Tamarix 1976). The suitability of the host is evaluated based on the survival rate taxa. Data from the paired-choice tests were analysed using Friedman i.e. percentage pupation, mass and developmental time of the insects tests performed on repeated measures, with Kendall’s Coefficient of utilising the host, and the index is calculated as follows: Concordance, which is used to determine whether individual repeated measures in the Friedman test agree with one another (Field, 2005). A unfed adult mass x% pupation Host suitability index = value of 1 indicates full agreement with repeated measures (in this case, developmental time that Diorhabda carinulata preferred one Tamarix taxon over another) and a value of 0 indicating no agreement (in this case, that D. carinulata 2.4.3. Paired-choice adult oviposition moved frequently among different Tamarix taxa). Two live, potted plants of different Tamarix taxa were placed into an insect cage (0.5 m × 0.5 m × 1 m). Four replicates of T. usneoides were 2.6.2. Field trials in the USA paired with T. ramosissima, and four replicates of T. usneoides were The data on adult counts are presented, but was not analysed due to paired with T. chinensis. Fifty-five adult D. carinulata (20 male: 35 fe- low sample sizes. The difference in egg numbers laid on the different male) from the F1 generation of the second beetle consignment, were Tamarix taxa in August 2016, was compared using a Kruskal-Wallis test released into the center of each cage and allowed to move freely. The performed on ranks of data. number of adults and egg clusters found on each plant were recorded All data analyses were performed using STATISTICA Version 12 every second day. The experiment was terminated once 1st instars (StatSoft 2015). emerged, approximately two weeks following oviposition. To ensure that adult feeding on stock plants would not have an effect on host plant 3. Results selection during the experiment, the stock plants used were a hybrid of T. ramosissima × T. chinensis, therefore, during the experiment the 3.1. Laboratory trials beetles were only able to choose between Tamarix taxa that they had not previously been exposed to. 3.1.1. No-choice egg-to-adult development Diorhabda carinulata had the highest egg-to-adult survival rate on T. 2.5. Field trials in the USA usneoides (36.7%, 18 adults), followed by the T. usneoides × T. ramosissima hybrid (24.1%, 13 adults) (Fig. 1). Survival of D. carinulata on T. 2.5.1. Outdoor cage experiments ramosissima was 0%; no adults emerged from eggs placed on T. ramo- Two outdoor cage experiments (trial (a) and trial (b)) were set up in sissima bouquets. Percentage mortality was high in the 1st instar D. August/September 2015 and a third (trial (c)) in August 2016, outside carinulata on T. ramosissima (66.7%) and the T. usneoides × T. ramo- the Palisade Insectary, Palisade, Colorado (coordinates 39° 6′ 46.1″ N sissima hybrid (55.6%), which was the only taxon on which all the

3 D. Marlin, et al. Biological Control 136 (2019) 104002

T. usneoides was lower (39.17 ± 1.70) than on T. ramosissima (47.33 ± 5.71), but not significantly different (Friedman test; χ2(6,1) = 0.667; p > 0.05). The Coefficient of Concordance was 0.11 indicating that beetles did not consistently select one plant over the other. Fewer D. carinulata eggs were laid per plant on T. usneoides (54 ± 5.06) than on T. ramosissima (97.33 ± 19.70), but not significantly less (Friedman test; χ2(6,1) = 2.667; p > 0.05). The Coefficient of Concordance was 0.44, indicating that the beetles were moderately consistent in their selection for oviposition on T. ramosissima. The mean number of adults per plant was significantly higher on T. usneoides (65.83 ± 10.08) than on T. chinensis (31.67 ± 2.54) (Friedman test; χ2(6,1) = 6.000; p < 0.05) and the Coefficient of Concordance (1.0) indicated that the beetles consistently preferred T. Fig. 1. Percentage of D. carinulata eggs (n = 20 per plant) that survived to usneoides over T. chinensis. Furthermore, the mean number of eggs laid fi adulthood, when reared on bouquets of three different Tamarix taxa (n = 4 for per plant was signi cantly higher on T. usneoides (107.2 ± 17.06) than 2 T. ramosissima and hybrid T. usneoides × T. ramosissima; n = 5 for T. usneoides). on T. chinensis (36.33 ± 7.33) (Friedman test; χ (6,1) = 6.000; P < 0.05). The Coefficient of Concordance was 1.0, indicating that D. carinulata females consistently preferred to oviposit on T. usneoides than pupae emerged as adults (Fig. 1). The mean weight of newly emerged D. on T. chinensis. carinulata adults on T. usneoides was 8.025 mg ± 0.476 mg, and on T. usneoides × T. ramosissima was 8.956 mg ± 0.909 mg, which were not 3.2. Field trials in the USA significantly different from each other (Kruskal-Wallis test on ranks; H = 0.657; p > 0.05). 1,22 3.2.1. Outdoor cage experiments For both experiments conducted in August/September 2015, after 3.1.2. No-choice egg-to-egg multi-generational development the other Tamarix taxa plants were severed, there was an increase in the All life stages of D. carinulata had a higher percentage survival on T. number of adult D. carinulata on T. usneoides (Fig. 3a and b), although usneoides than on T. ramosissima (Fig. 2). The percentage survival to the the difference could not be statistically tested. For the experiment fi 2nd instar was signi cantly higher on T. usneoides (95.7% ± 4.3%) conducted in August 2016, very few adult beetles were available; than on T. ramosissima (72% ± 4.3%) (Kruskal-Wallis test on ranks; nevertheless, the number of adults on T. usneoides remained higher than fi ff H1,8 = 4.288; p < 0.05). There was no signi cant di erence in per- on the other Tamarix taxa for the majority of the experiment (Fig. 4). centage survival to the 3rd instar (Kruskal-Wallis test on ranks; Diorhabda carinulata laid eggs on T. usneoides, T. chinensis × T. ramo- H1,8 = 3.000; p > 0.05) although it was higher on T. usneoides sissima hybrids and T. usneoides × T. chinensis hybrids, but not on the T. (81.8% ± 9.2%) than on T. ramosissima (60.4% ± 4.9%). The per- usneoides × T. ramosissima hybrids (Fig. 5). Differences in the mean fi centage survival to the adult stage was signi cantly greater on T. us- number of eggs laid per plant among the taxa were not significantly neoides (67.6% ± 6.9%) than on T. ramosissima (43.8% ± 7.8%) different (Kruskal-Wallis test on ranks H 3,16 = 4.45p > 0.05). (Kruskal-Wallis test on ranks; H1,8 = 4.136; p < 0.05) (Fig. 2). The number of D. carinulata eggs laid per female on the foliage of each test 4. Discussion plant was higher on T. usneoides (57.66 ± 13.71) than on T. ramosissima (32.48 ± 22.35) (Table 1), but this was not statistically sig- Host specificity tests are a critical and ubiquitous step in classical fi ni cant (Kruskal-Wallis test on ranks; H1,9 = 1.500; p > 0.05). Dior- biological control, conducted before a candidate biocontrol agent is habda carinulata beetles reared on T. usneoides had a higher percentage released into its new target area (Heard, 2002; Schaffner, 2001). These pupation and shorter developmental times that those reared on T. ra- tests usually take place under quarantine laboratory conditions, which ’ mosissima (Table 2). According to Maw s host suitability index, T. us- may result in false positive results as an agent is “forced” to utilise a neoides was 2.33 times a better host than T. ramosissima (Table 2). plant that it would never select under field conditions (Zwölfer and Harris, 1971). Our study had the advantage of using both laboratory- 3.1.3. Paired-choice adult oviposition based and open-field tests, thus increasing our knowledge of D. car- In the paired choice trials, the mean number of adults per plant on inulata’s fundamental and ecological host ranges, respectively. Through the combination of these tests, the host specificity of D. carinulata on various Tamarix taxa was thoroughly examined and results of the field experiments confirmed the results of the laboratory experiments; D. carinulata failed host-specificity requirements in that it both fed on, and completed its life-cycle, on the non-target, indigenous T. usneoides.We consider the failure of any adult D. carinulata to emerge in the no-choice egg-to-adult development test, on T. ramosissima, to be a random effect of low numbers, due to only 20 eggs being used to initiate the experiment. In addition, we suspect that the handling of the larvae and pupae during the experiment may have caused severe damage or death to the larvae and or pupae. Similar observations were made by Lewis et al. (2003) in their study on the development and reproductive parameters of Diorhabda elongata; survival in their laboratory cultures was greatest when the larvae and pupae were not manipulated. However, had D. carinulata fed on T. usneoides without completing Fig. 2. Mean percentage survival of three developmental stages of D. carinulata its lifecycle, there would still be good reason to consider it for release it reared in no-choice tests on potted T. usneoides (n = 5) or T. ramosissima (n = 4) in South Africa (after further host specificity testing with closely related plants. Asterisks (*) above error bars indicate significant difference between plants outside of the Tamaricaceae). This was the case with the water means. Error bars represent the standard error of the mean. hyacinth grasshopper, Cornops aquaticum (Bruner), which fed on 12

4 D. Marlin, et al. Biological Control 136 (2019) 104002

Table 1 The number of eggs laid by D. carinulata females reared from eggs, on two Tamarix taxa. The number of eggs laid per female per day was calculated based on the total number of eggs laid over two weeks and the number of females present.

T. usneoides T. ramosissima

Plant no. 1 2 3 4 5 Mean ( ± S.E.) 1 2 3 4 Mean ( ± S.E.)

No. females 17 10 9 9 10 11.00 ( ± 1.52) 5 2 7 5 4.75 ( ± 1.03) No. eggs 435 924 248 543 824 594.8 ( ± 124.38) 55 199 63 52 92.25 ( ± 35.66) Eggs laid per female in two weeks 25.59 92.4 27.56 60.33 82.4 57.66 ( ± 13.71) 11 99.5 9 10.4 32.48 ( ± 22.35) Eggs laid per female/day 1.83 6.6 1.97 4.31 5.89 4.12 ( ± 0.98) 0.79 7.11 0.64 0.74 2.32 ( ± 1.6)

Table 2 the genus. Thus the Diorhabda beetles have not evolved with T. us- Host suitability of T. usneoides and T. ramosissima for D. carinulata as char- neoides and would therefore not be expected to utilize it as a suitable acterised by the Maw's Host Suitability Index. The index is calculated as: (unfed host. Secondly, in the USA the beetles generally preferred the invasive adult mass * percentage pupation)/development time (Maw, 1976). Relative species over T. aphylla, which is closely related to T. usneoides (Baum, suitability was calculated with the assumption that T. ramosissima represents 1978; Mayonde et al., 2015) and which did experience some herbivory the baseline suitability. but little impact from the beetles (Milbrath and DeLoach, 2006a,b). Tamarix taxa T. usneoides T. ramosissima However, the fact that the beetles were able to oviposit and develop on (mean ± S.E.) (mean ± S.E.) T. aphylla was not of concern for the USA since T. aphylla is an introduced ornamental species, which has become invasive in some areas Unfed adult mass (mg) 9.048 ± 4.738 9.048 ± 4.738 % Pupation 67.56% ± 6.98% 43.78% ± 7.81% such as Lake Powell, Utah (DeLoach et al., 2003). Furthermore, Duration of 22.75 ± 2.99 34.25 ± 6.24 Mayonde et al. (2019) found that T. usneoides is clearly genetically Development (days) differentiated from the invasive T. chinensis and T. ramosissima, which Host suitability 0.02641 ± 0.02687 0.01132 ± 0.01156 could have an effect on plant phenotypic differences (Müller-Schärer Relative suitability 232.3% 100% et al., 2004); these differences could be sufficient that a suitably host specific biocontrol agent can be found. fl non-target plant species, but only completed its development on two of It is possible that the evolution of host use of Diorhabda is in uenced these, Canna indica L. and Pontederia cordata L., and the survival of more by Tamarix chemistry than by its phylogeny (Bernays and nymphs on C. indica and P. cordata was low compared to survival on Chapman, 1994; Becarra, 1997; Erdtman, 1963). Thus, the selection of water hyacinth (Oberholzer and Hill, 2001), such that the grasshopper potential agents would be simpler and more precise if it were possible fi was eventually released as a biocontrol agent of water hyacinth in to match Tamarix species of similar chemistry with a speci c Diorhabda fl South Africa in 2011 (Bownes et al., 2011). species for use in biological control. The in uence of Tamarix chemistry In our study, D. carinulata chose and even preferred the non-target and phylogeny on the evolution of host use in Diorhabda, and other T. usneoides over the invasive targets T. chinensis and T. ramosissima. agents, should therefore be investigated. This was unexpected, firstly because T. usneoides is restricted to Of the 321 insects, mites and pathogens that have been investigated southern Africa (Baum, 1978) and disjunct from the Eurasian species in as potential biocontrol agents of invasive plants in South Africa, only

Fig. 3. The mean number of adult D. carinulata beetles found over time on four diﬀerent Tamarix taxa (n = 4/taxon) in two outdoor cages (trial (a) and trial (b)), in Colorado, USA, during August/ September 2015. “CUT” represents the point where the experiment changes from a multi-choice host- speciﬁcity test to a no-choice test with T. usneoides, by cutting down plants of the other three taxa.

5 D. Marlin, et al. Biological Control 136 (2019) 104002

Fig. 4. The mean number of adult D. carinulata beetles found over time on four diﬀerent Tamarix taxa (n = 4/taxon) in an outdoor cage (trial (c)), in Colorado, USA, during August 2016. Error bars represent the standard error of the mean.

43% have been released while 24% have been rejected, with the re- biological control for Tamarix has been considered in South Africa maining 33% being either shelved, under investigation or pending (Marlin et al., 2017) and Argentina (McKay et al., 2017), following its permission for release (Zachariades, 2018). In many cases, potential success in the USA. agents are rejected by the researchers due to insufficient host-specifi- A list of potential agents has already been compiled with the help of city, yet rarely is such information reported in literature. This is un- colleagues in South Africa, Italy and the USA. Of the 11 potential agents fortunate as it gives the impression that every agent that undergoes selected, the two most promising agents according to field observations host-specificity testing is released. In addition, rejection records are (Cristofaro pers. ob.) are the stem-galling midge Psectrosema noxium important as they ensure that researchers from different countries do (Marikovskij) (Diptera: Cecidomyiidae) and the root-galling weevil not waste resources by repeating experiments with agents that have Liocleonus clathratus (Olivier) (Coleoptera: Cucrulionidae). Psectrosema already been tested, and rejected, in another country. In the case of noxium is of particular interest as it does not oviposit on T. aphylla Diorhabda carinulata in South Africa, we strongly recommend that it not (Cristofaro pers. comm.). It is envisaged that the abovementioned two be introduced or released in southern Africa, as its host-range includes agents will be imported into quarantine in South Africa in the near the indigenous T. usneoides. future and host-specificity tests will follow. In addition, more extensive To prevent, or at the least curtail, the spread of invasive Tamarix in surveys of the arthropod fauna on native and invasive Tamarix in South South Africa, it is vital that a biocontrol agent for the invasive Tamarix Africa are being conducted to further our understanding of the Tamarix be found because the alien species have been shown to be invading arthropod community and to identify niches within the existing com- riparian areas in drier regions of the country (Newete et al., 2019). The munity for effective biocontrol agents. This will also indicate if it is precise area covered by the invasive Tamarix in South Africa is not yet possible for insects to differentiate between the alien and indigenous known, however, given its relatively recent introduction (Mayonde Tamarix taxa. et al., 2019) it is safe to assume that Tamarix is in its early stages of invasion in the country and has the potential to further occupy sub- Acknowledgements stantial areas of riparian zones. In Argentina, Natale et al. (2013) estimated that more than three quarters of the arid and semi-arid regions We thank Andrew Norton for obtaining import permits to the USA of Argentina have the potential to be invaded by the weed. The use of for South African Tamarix and for helping E. Smit with logistics while in

Fig. 5. The mean number of eggs per plant laid by D. carinulata beetles on different Tamarix taxa (n = 4/taxon) in a multi-choice caged experiment conducted in Colorado, USA, during August 2016. No significant difference between means. Error bars represent the standard error of the mean.

6 D. Marlin, et al. Biological Control 136 (2019) 104002 the USA. We thank Dan Bean and Nina Louden for the help and advice 43–48. they provided to E. Smit during his field work, for their advice on Heard, T.A., 2002. Host specificity testing of biocontrol agents of weeds. In: Denslow, J.E., Hight, S.D., Smith, C.W. (Eds.), Proceedings, Hawaii Biological Control rearing the Diorhabda beetles and also sending us more beetles when Workshop. Pacific Cooperative Studies Unit. University of Hawaii, Honolulu, pp. needed. We thank Peter Kgampe for help with Tamarix culture main- 21–28. tenance in South Africa, and the following people for helping with Henderson, L. 2001. Alien weeds and invasive plants: a complete to declared weeds and invaders in South Africa. Agricultural Research Council - Plant Protection Research beetle maintenance: Blair Cowie, Lyriche Drude, Phuluso Mudau, and Institute Handbook No. 12, Pretoria, South Africa. Nic Venter. Impson, F.A.C., Moran, V.C., 2004. Thirty years of exploration for and selection of a succession of Melanterius weevil species for biological control of invasive Australian ff Funding acacias in South Africa: should we have done anything di erently? In: Cullen, J.M., Briese, D.T., Kriticos, D.J., Lonsdale, W.M., Morin, L., Scott, J.K. (Eds.), Proceedings of the XI International Symposium on Biological Control of Weeds pp. 127–134. This research was partially funded by the National Research CSIRO Entomology, Canberra, Australia. Foundation (NRF) of South Africa (Grant 88407 to D. Marlin), the Lewis, P.A., DeLoach, C.J., Knutson, A.E., Tracy, J.L., Robbins, T.O., 2003. Biology of Diorhabda elongata deserticola (Coleoptera: Chrysomelidae), an Asian leaf beetle for University of the Witwatersrand and the South African Department of biological control of saltcedars (Tamarix spp.) in the United States. Biol. Control 27, Environmental Affairs. Funders were not involved in the preparation of 101–116. the manuscript or the decision to publish. Marlin, D., Newete, S.W., Mayonde, S.G., Smit, E.R., Byrne, M.J., 2017. Invasive Tamarix (Tamaricaceae) in South Africa: current research and the potential for biological control. Biol. Invasions 5, 1–24. CRediT authorship contribution statement Marohasy, J., 1998. The design and interpretation of host-specificity tests for weed biological control with particular reference to insect behaviour. Biocontrol News Inf. 19, 13N–20N. Danica Marlin: Data curation, Investigation, Methodology, Project Maw, M., 1976. Biology of the tortoise beetle, Cassida hemisphaerica (Coleoptera: administration, Supervision, Writing - original draft, Writing - review & Chrysomelidae), a possible biological control agent for bladder campion, Silene cu- editing. Etienne R. Smit: Data curation, Formal analysis, Investigation, cubalus (Caryophyllaceae), in Canada. Can. Entomol. 108, 945–954. Mayonde, S.G., Cron, G.V., Gaskin, J.F., Byrne, M.J., 2015. Evidence of Tamarix hybrids Methodology, Writing - review & editing. Marcus J. Byrne: in South Africa, as inferred by nuclear ITS and plastid trnS–trnG DNA sequences. Conceptualization, Funding acquisition, Resources, Supervision, South Afr. J. Botany 96, 122–131. Validation, Writing - review & editing. Mayonde, S.G., Cron, G.V., Gaskin, J.F., Byrne, M.J., 2016. Tamarix (Tamaricaceae) hybrids: the dominant invasive genotype in southern Africa. Biol. Invasions 18, 3575–3594. Declaration of Competing Interest Mayonde, S., Cron, G.V., Glennon, K.L., Byrne, M.J., 2019. Genetic diversity assessment of Tamarix in South Africa – Biocontrol and conservation implications. South Afr. J. – The authors declare that they have no conflict of interests. Botany 121, 54 62. McKay, F., Logarzo, G., Natale, E., Sosa, A., Walsh, G.C., Pratt, P.D., Sodergren, C., 2017. Feasibility assessment for the classical biological control of Tamarix in Argentina. References BioControl 63, 169–184. Milbrath, L.R., DeLoach, C.J., 2006a. Acceptability and suitability of athel, Tamarix aphylla, to the leaf beetle Diorhabda elongata (Coleoptera: Chrysomelidae), a biolo- Baker, J.L., Webber, N.A.P., Johnson, K.K., 2003. Non-target impacts of Aphthona ni- gical control agent of saltcedar (Tamarix spp.). Environ. Entomol. 35, 1379–1389. griscutis, a biological control agent for Euphorbia esula (leafy spurge), on a native plant Milbrath, L.R., DeLoach, C.J., 2006b. Host specificity of different populations of the leaf Euphorbia robusta. In: Cullen, J.M., Briese, D.T., Kriticos, D.J., Lonsdale, W.M., Morin, beetle Diorhabda elongate (Coleoptera: Chrysomelidae), a biological control agent of L., Scott, J.K. (Eds.), Proceedings of the XI International Symposium on Biological saltcedar (Tamarix spp.). Biol. Control 36, 32–48. – – Control of Weeds, pp. 247 251 Canberra, Australia, 27 April 2 May 2003. Müller-Schärer, H., Schaffner, U., Steinger, T., 2004. Evolution in invasive plants: im- Baum, B.R., 1978. The genus Tamarix. Israel Academy of Sciences and Humanities, plications for biological control. Trends Ecol. Evol. 19, 417–422. Jerusalem. Natale, E., Zalba, S.M., Reinoso, H., 2013. Presence–absence versus invasive status data Bean, D., Dudley, T., 2018. A synoptic review of Tamarix biocontrol in North America: for modelling potential distribution of invasive plants: saltcedar in Argentina. – tracking success in the midst of controversy. BioControl 63, 361 376. Ecoscience 20, 161–171. Becarra, J.X., 1997. Insects on plants: macroevolutionary chemical trends in host use. Newete, S.W., Mayonde, S., Byrne, M.J., 2019. Distribution and abundance of invasive – Science 276, 253 256. Tamarix genotypes in South Africa. Weed Res. https://doi.org/10.1111/wre.12356. Bernays, E.A., Chapman, R.F., 1994. Host-plant Selection By Phytophagous Insects. Oberholzer, I.G., Hill, M.P., 2001. How safe is the grasshopper Cornops aquaticum for Chapman and Hall, New York, pp. 312p. release on water hyacinth in South Africa? In: Julien, M.H., Hill, M.P., Center, T.D., Bownes, A., King, A., Nongogo, A., 2011. Pre-release studies and release of the grass- Jianqing, D. (Eds.), Proceedings of the Second Meeting of the Global Working Group – hopper Cornops aquaticum in South Africa a new biological control agent for water for the Biological and Integrated Control of Water Hyacinth. ACIAR, Canberra, hyacinth, Eichhornia crassipes. In: Wu, Y., Johnson, T., Sing, S., Raghu, S., Wheeler, Australia, pp. 82–88. G., Pratt, P., Warner, K., Center, T., Goolsby, J., Reardon, R. (Eds.), Proceedings of the Olckers, T., Zimmermann, H.G., Hoffmann, J.H., 1995. Interpreting ambiguous results of – XIII International Symposium on Biological Control of Weeds, pp. 3 13 September host-specificity tests in biological control of weeds: assessment of two Leptinotarsa – 11 16, 2011, Waikoloa, Hawaii, USA. species (Chrysomelidae) for the control of Solanum elaeagnifolium (Solanaceae) in fi Briese, D., 2005. Translating host-speci city test results into the real world: the need to South Africa. Biol. Control 5, 336–344. – harmonize the yin and yang of current testing procedures. Biol. Control 35, 208 214. Rejmánek, M., Richardson, D.M., 2013. Trees and shrubs as invasive alien species—2013 fi Briese, D.T., Zapater, M., Andorno, A., Perez-Camargo, G., 2002. A two-phase open- eld update of the global database. Div. Distribution 19, 1093–1094. fi test to evaluate the host-speci city of candidate biological control agents for Schaffner, U., 2001. Host range testing of insects for biological weed control: how can it – Heliotropium amplexicaule. Biol. Control 25, 259 272. be better interpreted? BioScience 51, 951–959. fi fi Clement, S.L., Cristofaro, M., 1995. Open- eld tests in host-speci city determination of Tracy, J.L., Robbins, T.O., 2009. Taxonomic revision and biogeography of the Tamarix- – insects for biological control of weeds. Biocontrol Sci. Technol. 5, 395 406. feeding Diorhabda elongate (Brullé, 1832) species group (Coleoptera: Chrysomelidae: ’ Dalin, P., O Neal, M.J., Dudley, T., Bean, D.W., 2009. Host plant quality of Tamarix ra- Galerucinae: Galerucini) and analysis of their potential in biological control of fl mosissima and T. parvi ora for three sibling species of the biocontrol insect Diorhabda Tamarisk. Zootaxa 2101. Magnolia Press, Auckland. – elongate (Coleoptera: Chrysomelidae). Environ. Entomol. 38, 1373 1378. Van Wilge, B.W., Richardson, D.M., 2014. Challenges and trade-offs in the management DeLoach, C.J., Lewis, P.A., Herr, J.C., Carruthers, R.I., Tracy, J.L., Johnson, J., 2003. Host of invasive alien trees. Biol. Invasions 16, 721–734. fi speci city of the leaf beetle, Diorhabda elongata deserticola (Coleoptera: Weiersbye, I., Witkowski, E., Reichardt, M., 2006. Floristic composition of gold and ur- Chrysomelidae) from Asia, a biological control agent for saltcedars (Tamarix: anium tailings dams, and adjacent polluted areas, on South Africa’s deep-level mines. – Tamaricaceae) in the Western United States. Biol. Control 27, 117 147. Bothalia 101–127. – Erdtman, H., 1963. Some aspects of chemotaxonomy. Pure Appl. Chem. 6, 679 708. Zachariades, C. 2018. Biological control of invasive alien plants in South Africa: a list of ’ ffi Field, A.P., 2005. Kendall s coe cient of concordance. Encyclopedia Statist. Behav. Sci. all insects, mites and pathogens considered as biological control agents from 1913- https://doi.org/10.1002/0470013192.bsa327. 2018. http://www.arc.agric.za/arc-ppri/Documents/Table1-NaturalEnemiesAll.pdf. ff fi Githure, C.W., Zimmermann, H.G., Ho mann, J.H., 1999. Host speci city of biotypes of Zwölfer, H., Harris, P., 1971. Host specificity determination of insects for biological Dactylopius opuntiae (Cockerell) (Hemiptera: Oactylopiidae): prospects for biological control of weeds. Ann. Rev. Entomol. 16, 159–178. control of Opuntia stricta (Haworth) Haworth (Cactaceae) in Africa. Afr. Entomol. 7,