BioControl (2019) 64:457–468

https://doi.org/10.1007/s10526-019-09946-0 (0123456789().,-volV)( 0123456789().,-volV)

Post-release evaluation of a combination of biocontrol agents on Crofton weed: testing extrapolation of greenhouse results to field conditions

Lisa Buccellato . Marcus J. Byrne . Jolene T. Fisher . Ed T. F. Witkowski

Received: 17 January 2019 / Accepted: 24 June 2019 / Published online: 3 July 2019 Ó International Organization for Biological Control (IOBC) 2019

Abstract Two biocontrol agents, a leaf-spot patho- growth). The greenhouse trails were therefore not gen, Passalora ageratinae, and a stem gall fly, predictive of field conditions. utilis, have been released against Crofton weed, Ageratina adenophora (syn. Eupato- Keywords Ageratina adenophora Á Crofton weed Á rium adenophorum) (Asteraceae), in South Africa. -plant-pathogen interaction Á Multiple This work reports the first post-release evaluation of biocontrol agents Á Passalora ageratinae Á the effect of both agents acting together in the field. A Procecidochares utilis greenhouse trial using both agents had predicted an additive (beneficial) interaction between the agents. This study investigated if the additive interaction was Introduction present in the field. Four month old stems were exposed to one of the following three treatments Classical biocontrol of invasive alien plants involves (n = 20 plants per treatment): pathogen-only, patho- the deliberate introduction of agents, such as gen plus single fly-galled, and pathogen plus double and pathogens, to negatively influence the invasive fly-galled, for 11 months. The interaction between the plant’s growth parameters and population dynamics agents was equivalent to both agents acting indepen- (Mu¨ller-Scha¨rer and Schaffner 2008; Morin et al. dently (i.e. there was no additive effect on the weed’s 2006; Zachariades et al. 2017). Often, more than one biocontrol agent is released onto a specific target weed, based on the theory that the combined effects of multiple biocontrol agents will increase control of the Handling Editor: S. Raghu weed (Denoth et al. 2002; Julien and Griffiths 1998; L. Buccellato Á M. J. Byrne Á J. T. Fisher (&) Á Myers 1985, 2008; Stephens et al. 2013; Stiling and E. T. F. Witkowski Cornelissen 2005). Synergistic relationships between School of Plant and Environmental Sciences, insect and pathogen agents are considered responsible University of the Witwatersrand, PO Wits, Johannesburg 2050, South Africa for some of the first successes in biocontrol, of e-mail: [email protected] Opuntia stricta and of Hypericum perforatum for example (Caesar 2000, 2003). M. J. Byrne However, there is an increasing call not to release DST-NRF Centre of Excellence for Invasion Biology, School of Animal, Plant and Environmental Sciences, agents which are unlikely to be very effective (Paynter University of the Witwatersrand, et al. 2018; Raghu et al. 2006) because of the risks PO Wits, Johannesburg 2050, South Africa 123 458 L. Buccellato et al. associated with non-target effects. Therefore in addi- natural vegetation (Erasmus et al. 1992; Henderson tion to host specificity, the effectiveness of biocontrol 2001; Niu et al. 2007; Poudel et al. 2019). Crofton agents is often assessed under controlled conditions in weed also reduces crop yield, the carrying capacity of laboratory or greenhouse studies (Cowie et al. 2017; grazing land, is unpalatable to cattle, and causes a fatal Kumaran et al. 2018; Morin et al. 2009). However, lung disease in horses (Land Protection 2001; Page under such controlled conditions the impact of the and Lacey 2006; Plant Protection News 1988). agents can be either under- or over-estimated, as these In 1984 a biocontrol programme for the control of are not always representative of conditions in the field Crofton weed was initiated in South Africa (Plant (Morin et al. 2006, 2009; Rosskopf et al. 1999). For Protection News 1987; Kluge 1991). Two biocontrol example, the Eucalyptus weevil, Gonipterus ‘‘scutel- agents, a stem gall fly, Procecidochares utilis Stone latus’’ Gyllenhal (Curculionidae), was more selective (), and a leaf-spot pathogen, Passalora in its feeding and oviposition behaviour on Eucalyptus ageratinae Crous and A.R. Wood (Mycosphaerel- species in the field in comparison to laboratory trials laceae) (previously named Cercospora eupatorii Peck (Newete et al. 2011) highlighting potential differences or Phaeoramularia sp.) (Crous et al. 2009), were in the predicted versus realised host ranges of released in South Africa in 1984 and 1987, respec- biocontrol agents. Biocontrol risk assessment gener- tively (Kluge 1991). The biology and impact of the fly ally centres on the dangers of potential non-target in have been described by Haseler (1965) feeding of the putative agent (Blossey et al. 2018). and in South Africa by Bennett (1986). The pathogen’s However, post-release evaluations of released agents host specificity was explored by Morris (1989) and are critical for the future of biocontrol (Morin et al. Wang et al. (1997) who measured its impact in 2009; Raghu et al. 2006), not least to continuously Australia. Heystek et al. (2011) reviewed the Crofton reassure the public that it is a safe and cost-effective weed biocontrol programme in South Africa and pest control method (Messing and Brodeur 2018), but concluded that additional agents should be sought. also to inform politicians that government investment However, this is the first post-release evaluation of in the technique is money well spent (Ivey et al. 2019; both agents’ impact when released together in the field Schwarzla¨nder et al. 2018a). In light of this, this study in South Africa. assesses the impact of two biocontrol agents in the A controlled greenhouse trial using both the fungal field, released against Ageratina adenophora, follow- pathogen and the gall fly, individually and in combi- ing on from greenhouse trials of efficacy of the two nation, showed that the fly reduced plant height and biocontrol agents acting together. growth of Crofton weed, and repeated galling by the Originating from Mexico, Ageratina adenophora fly was most detrimental to the weed (Buccellato et al. (Sprengel) King and Robinson (syn. Eupatorium 2012). The pathogen did, however, reduce the number adenophorum Spreng.) (Asteraceae), also known as of sideshoots, supressing vegetative reproduction of Crofton weed or the Mexican Devil, is an invasive the weed, which was unfortunately promoted by weed in several countries worldwide, including South galling (Buccellato et al. 2012). Nevertheless, the Africa, Australia, New Zealand, Hawaii, , China, combination of the two agents resulted in an additive and recently in Italy (Julien and Griffiths 1998; Poudel effect on control of Crofton weed in the greenhouse. et al. 2019). Crofton weed is a perennial herb, with a The categorisation of combined impacts of the agents woody rootstock, and many stems reaching up to 2 m follows those proposed by Hatcher and Paul (2001), in height (Bess and Haramoto 1959; Henderson 2001; and Turner et al. (2010). Page and Lacey 2006). Trailing Crofton weed stems In South Africa, anecdotal field observations sug- produce roots when in contact with the soil, resulting gested that neither the fly nor fungal pathogen are in dense infestations (Bess and Haramoto 1959; successfully controlling Crofton weed (Heystek et al. Morris 1991). The plant invades steep slopes and 2011). However, the impact of the two agents together wet areas along streams, roadsides, forests, and has not been formally assessed in the field in South plantations (Dodd 1961; Henderson 2001; Page and Africa. The aim was therefore to evaluate whether the Lacey 2006; Trounce 2003). Conservation areas are pathogen and fly in combination will control the adversely affected by this weed, as it is allelopathic, vegetative growth of Crofton weed under field alters soil microbial communities, and displaces 123 Post-release evaluation of a combination of biocontrol agents on Crofton weed 459 conditions, in South Africa, as predicted by the In June 2007, 20 plants were randomly allocated to greenhouse studies. each of the following three treatments: (1) plants infected with the fungal pathogen, P. ageratinae (pathogen-only); Materials and methods (2) plants infected with the fungal pathogen, P. ageratinae, and exposed to one release of the Field site fly, P. utilis (hereafter referred to as pathogen- single galled); A field trial was conducted on the 150 ha Kloofwaters (3) plants infected with the fungal pathogen, P. farm (1380 m a.s.l.), in the Magaliesberg, North West ageratinae, and exposed to two releases of the Province, South Africa (25°49045.100S, 27°26026.000E). fly, P. utilis (hereafter referred to as pathogen- The farm is privately owned, and predominantly used double galled. Pathogen-double galled 1st gen- for school camps. However, a low stocking level of eration refers to the first fly release, and 2nd cattle is present. The field site was located within generation refers second fly release). Crofton weed infestations along a flat stream embank- ment, facing eastwards. Plants were in full sunshine, Cages were used to restrict flies from ovipositing on with late afternoon shading and only the fungal non-galled treatment plants. The cages were con- pathogen, P. ageratinae, was present. The site is on structed from fine white netting, attached to a 1 m Moot Plains Bushveld (central bushveld, Savanna central pole with a steel ring at the top, and pegged into Biome), with thorny savanna, dominated by Acacia the soil. All plants in all treatments were covered with species and woodlands (Mucina et al. 2006; Ruther- cages. Cages were removed two weeks after flies were ford et al. 2006). The Magaliesberg region experiences released, as the average life span of the adult flies is summer rainfall and very dry winters with frost ten days. No galling on pathogen-only stems was (Mucina et al. 2006). In winter, temperatures ranged observed during the trial period. from 2.6 to 18.3 °C (mean = 9.9 °C), and from 14.0 to Adult P. utilis were collected from a colony on 29.5 °C (mean = 20.5 °C) during summer. The mean galled plants at the University of the Witwatersrand, RH was 55% (range 10.2–100%) during the winter Johannesburg (26°11020.9700S, 28°01055.3700E). The months (June 2007 to August 2007) and 87.5% (range first release of flies was made in August 2007 when 45.0–100%) during the summer months (December one pair of flies per plant was released into the cages 2007 to February 2008). with plants of the pathogen-double galled treatment (treatment 3). emerged from galls 12 weeks later Field trials (November 2007), and were collected from the cages with an aspirator, over a five week period, to prevent In February 2007 all Crofton weed plants within a further galling of the treatment plants, allowing 50 m2 area within the field site were cut back to soil comparison with the laboratory trials (Buccellato level and marked with a plastic labelled peg in the et al. 2012). In January 2008, one pair of flies per ground. Plants were cut back to promote growth of plant was released onto the pathogen-single galled and new stems of known age from the woody rootstock. pathogen-double galled plants (treatments 2 and 3). The fungicide AMISTAR Ò (Syngenta, South Four stems per plant were randomly selected for Africa), used in greenhouse trials to restrict P. measurements of vegetative growth. Monthly mea- ageratinae infection, resulted in phytotoxic symptoms surements of stem height, number of living, dead and (chlorotic spots) on Crofton weed seedling leaves abscised leaves per stem, and the number of sideshoots (Buccellato et al. 2012). Therefore, it was not possible per stem, were recorded from June 2007 to April 2008. to chemically exclude the pathogen from the new The number of leaves per stem infected with the stems, as they had already been infected when they pathogen was also recorded monthly. On the day of the were at the seedling stage, from surrounding plants. fly release the oviposition site (apical leaf bud) was Fungal leaf spots were visible on new stems by April marked on all the treatment stems with a piece of 2007. sewing thread. Monthly measurements of stem growth above this oviposition site were then taken for all 123 460 L. Buccellato et al. treatment stems. These measurements were only taken 10% of pathogen-double galled stems died. There was after the second fly release, as stems were flowering no statistical association between stem death and during the first fly release and no apical leaf buds were treatment (v2 = 3.3, df = 2, P = 0.192). present. Gall length and diameter (at the longest and broadest point respectively) were measured and used Agent establishment and infection severity to calculate a gall size index (Bennett 1986). The number of emergence holes per gall was used as an The pathogen infected 10–40% of living leaves from index of the number of adult flies successfully June to October 2007 (Fig. 1a). The number of emerging per gall. The position of galls was classified infected living leaves increased to 70–80% by into two categories, stem (galls formed on the main November 2007, and thereafter 50% of living leaves stem) or side (galls formed on the leaf petiole or leaf were infected with the pathogen. There was no node). At the end of the trial the treatment stems were statistical difference in the percentage of leaves cut, collected and separated into bare stems, live infected with the pathogen on galled and ungalled leaves, dead leaves and sideshoots. The bare stems, stems over the trial period (F2,45 = 1.14, P = 0.331). leaves and sideshoots were then dried in a forced- draught oven for seven days at 60 °C and weighed. Stem height The total dry mass of the treatment stems was calculated. In addition the allocation of biomass to The stem height under all the treatments increased live leaves, bare stems and sideshoots per stem was until the flowering period, but decreased after flower- calculated as a percentage of stem total biomass. ing, and thereafter increased again for the remainder of the trial period (Fig. 1b). A repeated measures Data analysis ANOVA showed no overall significant difference

between the treatments (F2,41 = 0.92, P = 0.405). Repeated measures ANOVA with nesting (stems However, in September (F2,41 = 5.63, P = 0.007) nested within plants) were used to assess the and October 2007 (F2,41 = 6.41, P = 0.004), the influence of the biocontrol agents on the vegetative pathogen-double galled treatment stems were signif- growth of Crofton weed over the trial period (GLM icantly shorter than the pathogen-only and the procedure, StatSoft 2007). One-way ANOVA with pathogen-single galled stems. nesting and LSD tests were used to assess differ- ences in stem measurements between treatments for Percentage of live leaves per stem each month. One-way ANOVAs were used to assess differences in the gall size index and number of The percentage of live leaves decreased until the end emergence holes per gall between the single and of the flowering period, after which stems maintained double galling treatments. A v2 2 9 3 contingency 30–40% live leaves (Fig. 1c). There was no statistical table was used to compare the association between difference between the percentage of live leaves on 2 stem death and treatments. A v 2 9 2 contingency any of the treatments (F2,39 = 0.11, P = 0.893). table with Yates correction, was used to compare The gall size index differed significantly between the association between the position of galls and the release 1 (August 2007) and release 2 (January 2008) incidence of flowering. The difference in biomass (F2,93 = 3.05, P = 0.0522; refer to host-hoc LSD tests between the three treatments was assessed with a in Fig. 2a). Galls that formed in August 2007, during one-way ANOVA with nesting. the flowering period, were smaller than galls formed in January 2008. There was an association between the position of galls and flowering period (v2 = 4.09, Results df = 1, P = 0.043), with more galls formed on stems, in comparison to the side stems, when Crofton weed is Stem death not flowering. There was no statistical difference between the sizes of galls formed on stems for either Over the trial 15% of the pathogen-only stems died, the first or second time, in January 2008. The number while 6% of pathogen-single galled stems died and of adult flies emerging did not differ statistically 123 tm,and stems, P at treatments, two other the to different significantly is galled Pathogen-double where 2007 October and September of months the for differences significant *Indicate leaves. live (Mean otrlaeeauto facmiaino icnrlaet nCotnwe 461 pathogen, fungal the with infected leaves live 2008. April utilis fly the by double) or (single galling ageratinae fungus (by infection pathogen of treatments, three to plants adenophora Ageratina weed 1 Crofton Fig. on agents biocontrol of combination a of evaluation Post-release \ .5(LSD) 0.05 rmJn 07to 2007 June from , Procecidochares ± Passalora epneof Response SE) iho without or with ) c a ecnaeof Percentage ecnaeof Percentage b egtof Height (c) (b) (a) Stem height (mm) % Live leaves/stem % Pathogen Infected leaves/stem 100 200 300 400 500 600 700 800 100 10 20 30 40 50 60 70 80 90 10 20 30 40 50 60 70 80 90 0 0 0 u 0 u 0 u 0 e 0 c 0 o 0 e 0 a 0 e 0 a 0 Apr'08 Mar'08 Feb '08 Jan'08 Dec'07 Nov'07 Oct'07 Sep'07 Aug'07 Jul'07 Jun '07 u 0 u 0 u 0 e 0 c 0 o 0 e 0 a 0 e 0 a 0 Apr'08 Mar'08 Feb'08 Jan'08 Dec'07 Nov'07 Oct'07 Sep'07 Aug'07 Jul'07 Jun '07 u 0 u 0 u 0 e 0 c 0 o 0 e 0 a 0 e 0 a 0 Apr'08 Mar'08 Feb'08 Jan'08 Dec '07 Nov'07 Oct'07 Sep '07 Aug'07 Jul'07 Jun '07 Pathogen-doublegalled Pathogen-singlegalled Pathogen-only Release 1 Release 1 Release 1 Flowering period Flowering period Flowering period Fly Fly * * Month Release 2 Release 2 Release 2 Fly Fly Fly 123 462 L. Buccellato et al.

(a) 2.2 Fly a Pathogen-double galled (1st release) Release 2 2.0 Pathogen-single galled Pathogen-double galled (2nd release) a 1.8

1.6 Fly Release 1 1.4 b 1.2

1.0

Gall size index 0.8

0.6

0.4

0.2

0.0 Aug '07 Jan '08

(b) 5.0 Fly Release 2 a 4.5

4.0 a Fly 3.5 Release 1

3.0 b

2.5

2.0

1.5

1.0 Number of emergence holes/gall

0.5

0.0 Aug '07 Jan '08 Month

Fig. 2 a Gall size index (gall diameter 9 length) of Proceci- emerging from galls on pathogen infected stems, from the dochares utilis on Ageratina adenophora stems infected with release of P. utilis in August 2007 and January 2008. Means the fungal pathogen Passalora ageratinae, and b the number of followed by different letters are significantly different, at P. utilis adults (represented by the number of emergence holes) P \ 0.05 (LSD) (Mean ± SE)

123 Post-release evaluation of a combination of biocontrol agents on Crofton weed 463 between release 1 and 2, or between single and double 5 galled stems (F2,93 = 2.30, P = 0.106) (Fig. 2b). a 4 Stem growth above the oviposition site ab Growth above the oviposition site differed signifi- 3 cantly in February and April between the three b treatments after the second fly release in January 2 2008 (F2,41 = 4.70, P = 0.015) (Fig. 3). Pathogen- galled stems showed significantly less growth above the oviposition site compared to pathogen-only stems. Number of sideshoots/stem 1

Number of sideshoots per stem 0 Pathogen-only Pathogen-single Pathogen-double galled galled At the end of the trial the number of sideshoots Treatment differed significantly between treatments Fig. 4 Number of sideshoots on Ageratina adenophora stems (F2,39 = 3.26, P = 0.049) (Fig. 4). Repeated galling of stems promoted sideshoot growth in comparison to in response to three different treatments, of pathogen infection with or without galling (single or double), at the end of the trial pathogen-only stems. in April 2008. Means followed by different letters are significantly different, at P \ 0.05 (LSD) (Mean ± SE) Biomass (F2,39 = 3.33, P = 0.046) (Fig. 5). Although the three The galled stems allocated significantly more biomass treatments allocated their biomass to different areas of to sideshoots than pathogen-only stems (F2,39 = 4.74, the stem, there was no significant difference in the P = 0.014) (Fig. 5). The pathogen-only stems had a total biomass of the stems (F2,39 = 0.14, P = 0.873) higher percentage of their biomass allocated to live (Fig. 5). leaves (F2,39 = 4.12, P = 0.024), and bare stems

Fig. 3 Growth of Ageratina adenophora stems above the fly’s oviposition site, in response to three different treatments, of pathogen infection with (single or double) or without galling, after the second fly release in January 2008 until April 2008. For each month, means followed by different letters are significantly different, at P \ 0.05 (LSD) (Mean ± SE)

123 464 L. Buccellato et al.

Fig. 5 The percentage 80 10 biomass allocation of Ageratina adenophora stems to live leaves, 70 sideshoots and bare stems 8 (left axis), and the total 60 biomass (g) (right axis) of stems in response to three different treatment 50 conditions, of pathogen 6 infection with (single or double) or without galling, 40 at the end of the trial in April 2008. Means of biomass 4 Biomass (g) 30 allocation (%) followed by

different letters are Biomass allocation (%) significantly different, at 20 P \ 0.05 (LSD) 2 (Mean ± SE) 10

0 0 Pathogen-only Pathogen-single galled Pathogen-double galled Treatment

Discussion defoliation of pathogen infected stems has been recorded on Crofton weed in Australia (Page and Following the greenhouse trial, a substantial impact on Lacey 2006; Wang et al. 1997). The percentage of Crofton weed by the fly was expected in the field, with pathogen infected leaves reached 80%, when leaf loss repeated galling reducing stem growth, and the fungal was high after flowering. Thereafter only half the pathogen reducing sideshoot growth (Buccellato et al. leaves were infected at any one time. While death of 2012). However, in the field the fungal pathogen alone seedlings was observed in the field here, most of the was the more successful biocontrol agent, and no mature stems on the test plants were able to limit additional cumulative control was achieved with the pathogen infection to leaves on the lower half of the addition of the gall fly. Rather, the plant’s allocation of stem, leaving sites higher up on the stems for biomass was different when both biocontrol agents oviposition by the fly. All treatment stems maintained were used in the field. While the quality of stems and at least approximately 30% of their living leaves. The live leaves was reduced, which might constrain plant pathogen alone, and the combination of pathogen and growth, the biomass of sideshoots, which might galling, was not sufficient to completely defoliate promote asexual generation of new plants, was higher Crofton weed as has been seen in Australia (Dodd when both agents were present in the field. This study 1961). As Crofton weed naturally sheds its lower highlighted that caution should be used when extrap- leaves, this impact may not be sufficient to reduce olating greenhouse trial results, to predict the effec- Crofton weed growth. tiveness of multiple biocontrol agents in the field. The pathogen with single or double galling slowed In this insect-pathogen system, the agents partition stem growth above the oviposition site, in comparison the resources of Crofton weed plants in space because to pathogen-only stems. Weeds are often able to the fly galls the shoot tips, and the pathogen infects the compensate for the pressure of reduced leaf area by lower leaves (Milbrath and Nechols 2014). These field increasing vegetative growth (Charudattan results show that galling does not influence the 2005, 2010) or upregulating photosynthesis in adja- severity of pathogen infection of Crofton weed, which cent leaves (Cowie et al. 2018a). Reduced growth of is a positive result. The pathogen is a valuable agent Crofton weed above the oviposition site highlights the because death of seedlings and severe or complete ability of the fly galls to potentially act as nutrient

123 Post-release evaluation of a combination of biocontrol agents on Crofton weed 465 sinks, as seen in other systems (Dorchin et al. 2006; exposure of the plants to the flies was limited for strict Florentine et al. 2005; Moseley et al. 2009). Galls in comparison with the greenhouse trials. This may not the field were significantly larger in size (1.54 ± 0.12, completely represent true field conditions and as a mean ± SE) compared to those observed in the multivoltine agent, the fly’s impact might be greater greenhouse (1.16 ± 0.06) (t364 = 2.47, P \ 0.001), than measured here (Kumaran et al. 2018). In Hawaii, possibly indicating a bigger nutrient sink effect of Crofton weed has been successfully controlled by the galls in the field. gall fly alone, where Crofton weed has been elimi- Galling influences apical dominance, and Crofton nated over large tracks of land, in low rainfall areas weed compensates by producing more sideshoots on (Bess and Haramoto 1959, 1972; Muniappan et al. double-galled stems (Buccellato et al. 2012; Erasmus 2009). Heavy galling in Hawaii, up to seven galls per et al. 1992). Tall Crofton weed stems eventually bend stem, resulted in stunted plants, weakened stems and over and trail along the soil, where the sideshoots can the death of some plants (Bess and Haramoto 1972), then produce roots and eventually grow into new but the fly was not so successful in high rainfall areas plants (Muniappan et al. 2009). Pathogen-only stems (Bess and Haramoto 1959, 1972). However, flies had significantly fewer sideshoots. Although we do generally have a low impact as biocontrol agents (only not know what sideshoot growth would be like on 7% of releases have a major impact on their target) galled stems without pathogen-infection in the field, compared to 19% of fungal biocontrol releases there is an antagonistic effect between these biocontrol (Schwarzla¨nder et al. 2018b). agents on sideshoot growth, which again highlights the Across South Africa the impact of these agents may difficulty of extrapolating from the laboratory to the differ depending on the severity and prevalence of field. agent attack, parasitism of the fly, and abiotic envi- There was no cumulative effect of the biocontrol ronmental factors, and as such future work should agents on the total biomass of Crofton weed stems. assess agent impacts across a broader range of Stems infected only with the fungal pathogen allo- locations. Factors which may affect the success or cated less biomass to sideshoots. However, galling failure of biocontrol agents include parasitism (King reduced the biomass of bare stems and live leaves, et al. 2011), climate (Byrne et al. 2002; Zachariades suggesting that although galled stems have similar et al. 2011) and acquired natural enemies (Muniappan heights to pathogen-only stems, galled stems may be and McFadyen 2005), which are unlikely to be present weaker and thinner than the pathogen-only stems. in the controlled conditions of a greenhouse trial. The Therefore, the fly may be effective over a longer time gall fly and fungal pathogen together have slowed the period if growth above the oviposition site is weak. encroachment of Crofton weed and thinned infesta- This effect was seen in Hawaii where galled Crofton tions along the east coast of Australia (Dodd 1961; weed stems were weaker and thinner than ungalled Page and Lacey 2006). There, the pathogen has led to stems (Bess and Haramoto 1959, 1972). death of seedlings and major or complete defoliation The results of the greenhouse trials on agent of stems in the drier months (Page and Lacey 2006), interactions on Crofton weed were not predictive of and high levels of galling have reduced plant vigour the field results. However, given time, further field and killed plants (Dodd 1961). Differences in plant trials might show more beneficial effects of the agents. varieties may be responsible for the differential The field trials were only able to assess the impact of success of agents both internationally, and potentially the fly and pathogen on vegetative growth within one within South Africa (Mukwevho et al. 2017). How- area of South Africa, where the fly had not been ever, Asteraceae are more difficult targets for biocon- previously released and therefore no parasitoids were trol, yielding a 61% success rate [compared for found in this study. Up to 100% of P. utilis are instance with 100% for Cactaceae (Schwarzla¨nder parasitized in New Zealand, and Bennett (1986) found et al. 2018b)]. This study only assessed the vegetative three parasitoid wasp species in KwaZulu-Natal growth of Crofton weed. However, the efficacy of Province, parasitizing 76% of flies in South Africa. agents to control reproductive output of the target So parasitism is likely to reduce the fly’s impact in the weed is also important in biocontrol (Pyseˇk and field and influence its relationship with the fungus. By Richardson 2007; Morin et al. 2009; Cowie et al. removing the flies after one galling event per release, 2017). 123 466 L. Buccellato et al.

Morin et al. (2009) recommended that field trials in Procecidochares utilis Stone. MSc Thesis, University of the introduced range are necessary to understand the Natal, South Africa Bess HA, Haramoto FH (1959) Biological control of pamakani, response of an agent to new environmental conditions, Eupatorium adenophorum, in Hawaii by a tephritid gall fly, as well as gauging the response of the weed to the Procecidochares utilis. 2. Population studies of the weed, agent. Inevitably, research and funding is focused on the fly, and the parasites of the fly. Ecology 40:244–249 the initial host-specificity stages of a biocontrol Bess HA, Haramoto FH (1972) Biological control of pamakani, Eupatorium adenophorum, in Hawaii by a tephritid gall fly, programme. However it is essential to continue that Procecidochares utilis. 3. Status of the weed, fly and par- research after field releases to fully understand the asites of the fly in 1966-71 versus 1950-57. Proc Hawaii dynamics of any biocontrol programme (Aigbedion- Entomol Soc 21:165–178 Atalor et al. 2019; Briese 2006; Cowie et al. 2018b; Blossey B, Da´valos A, Simmons W, Ding J (2018) A proposal to use plant demographic data to assess potential weed bio- Morin et al. 2009; Raghu et al. 2006) and to support logical control agents impacts on non-target plant popu- funding for future projects (Messing and Brodeur lations. BioControl 63:461–473 2018) in which South Africa generally performs above Briese DT (2006) Can an a priori strategy be developed for average in terms of agent impact, compared to the biological control? The case of Onopordum spp. thistles in Australia. Aust J Entomol 45:317–323 other four leading biocontrol nations (Schwarzla¨nder Buccellato L, Byrne MJ, Witkowski ETF (2012) Interactions et al. 2018b). This study has highlighted that caution between a stem gall fly and a leaf-spot pathogen in the should be used when using only greenhouse trials, biological control of Ageratina adenophora. Biol Control with controlled conditions, to predict the effectiveness 61:222–229 Byrne MJ, Currin S, Hill MP (2002) The influence of climate on of multiple biocontrol agents and supports the pro- the establishment and success of the biocontrol agent posal for open field testing in the agent’s home range Gratiana spadicea, released on Solanum sisymbriifolium in (Schaffner et al. 2018). South Africa. Biol Control 24:128–134 Caesar AJ (2000) Insect–pathogen synergisms are the founda- Acknowledgements The Working for Water Capacity tion of weed biocontrol. In: Spencer NR (ed) Proceedings Building Programme of DWAF, the National Research of the X international symposium on biological control of Foundation (NRF2069152) and the University of the weeds, July 4–14, 1999. Montana State University, Boze- Witwatersrand are thanked for funding. man, pp 793–798 Caesar AJ (2003) Synergistic interaction of soilborne plant pathogens and root-attacking insects in classical biological Funding Research funding for the work (University of control of an exotic rangeland weed. Biol Control Witwatersrand post-graduate merit award; Working for Water 28:144–153 bursary) were independent of the research conducted and Charudattan R (2005) Ecological, practical, and political inputs therefore did not influence any outcomes. No authors received into selection of weed targets: what makes a good biolog- any financial support to for attending symposia or to conduct ical control target? Biol Control 35:183–196 education programs related to the research. Charudattan R (2010) A reflection on my research in weed biological control: using what we have learned for future Compliance with ethical standards applications. Weed Technol 24:208–217 Cowie BW, Venturi G, Witkowski ETF, Byrne MJ (2017) A Conflict of interest The authors declare that they have no villainous hero: does the biological control agent, An- conflict of interest. thonomus santacruzi, pollinate its target weed, Solanum mauritianum? Biol Control 105:79–85 Research involving in human and animal rights No humans Cowie BW, Witkowski ET, Byrne MJ, Strathie LW, Goodall or were used in this research. JM, Venter N (2018a) Physiological response of Parthe- nium hysterophorus to defoliation by the leaf-feeding beetle Zygogramma bicolorata. Biol Control 117:35–42 References Cowie BW, Venter N, Witkowski ETF, Byrne MJ, Olckers T (2018b) A review of Solanum mauritianum biocontrol: prospects, promise and problems: a way forward for South Aigbedion-Atalor PO, Day MD, Idemudia I, Wilson DD, Africa and globally. BioControl 63:475–491 Paterson ID (2019) With or without you: stem-galling of a Crous PW, Schoch CL, Hyde KD, Wood AR, Gueidan C, de tephritid fly reduces the vegetative and reproductive per- Hoog GS, Groenewald JZ (2009) Phylogenetic lineages in formance of the invasive plant Chromolaena odorata the Capnodiales. Stud Mycol 64:17–47 (Asteraceae) both alone and in combination with another Denoth M, Frid L, Myers JH (2002) Multiple agents in biolog- agent. BioControl 64:103–114 ical control: improving the odds? Biol Control 24:20–30 Bennett PH (1986) An investigation into the biological control Dodd AP (1961) Biological control of Eupatorium adenopho- of crofton weed, Ageratina adenophora (K. & R.) (Syn. rum in Queensland. Aust J Sci 23:356–365 Eupatorium adenophorum Spreng.) by the gall fly

123 Post-release evaluation of a combination of biocontrol agents on Crofton weed 467

Dorchin N, Cramer MD, Hoffmann JH (2006) Photosynthesis Moseley CT, Cramer MD, Kleinjan CA, Hoffmann JH (2009) and sink activity of wasp-induced galls in Acacia pyc- Why does Dasineura dielsi-induced galling of Acacia nantha. Ecology 87:1781–1791 Cyclops not impede vegetative growth? J Appl Ecol Erasmus DJ, Bennett PH, Van Staden J (1992) The effect of 46:214–222 galls induced by the gall fly Procecidochares utilis on Mucina L, Rutherford MC, Powrie LW (2006) Vegetation atlas vegetative growth and reproductive potential of crofton of South Africa, Lesotho and Swaziland. In: Mucina L, weed, Ageratina adenophora. Ann Appl Biol 120:173–181 Rutherford MC (eds) The vegetation of South Africa, Florentine SK, Raman A, Dhileepan K (2005) Effects of gall Lesotho and Swaziland. South African National Biodi- induction by Epiblema strenuana on gas exchange, nutri- versity Institute, Pretoria, pp 748–790 ents, and energetic in Parthenium hysterophorus. Mukwevho L, Simelane D, Olckers T (2017) Host-plant variety BioControl 50:787–801 and not climate determines the establishment and perfor- Haseler BW (1965) Life-history and behaviour of the crofton mance of Aceria lantanae (Eriophyidae), a biological weed gall fly Procecidochares utilis Stone (Diptera: control agent of Lantana camara in South Africa. Exp Appl Trypetidae). Aust J Entomol 4:27–32 Acarol 71:103–113 Hatcher PE, Paul ND (2001) Plant pathogen-herbivore interac- Mu¨ller-Scha¨rer H, Schaffner U (2008) Classical biological tions and their effects on weeds. In: Jeger MJ, Spence NJ control: exploiting enemy escape to manage plant inva- (eds) Biotic interactions in plant-pathogen associations. sions. Biol Invasions 10:859–874 CAB International, Wallingford, pp 193–225 Muniappan R, McFadyen RE (2005) Gall-inducing Henderson L (2001) Alien weeds and invasive plants: a com- in the biological control of weeds. In: Raman A, Schaefer plete guide to declared weeds and invaders in South Africa. CW, Withers TM (eds) Biology, ecology, and evolution of ARC-PPRI, PPRI Handbook No. 12: Pretoria gall-inducing arthropods, vol 1. Science Publishers Inc, Heystek F, Wood AR, Neser S, Kistensamy Y (2011) Biological New Hampshire, pp 709–730 control of two Ageratina species (Asteraceae: Eupatorieae) Muniappan R, Raman A, Reddy GVP (2009) Ageratina ade- in South Africa. Afr Entomol 19:208–216 nophora (Sprengel) King and Robinson (Asteraceae). In: Ivey PJ, Hill MP, Coetzee JA (2019) South Africa tackles Muniappan R, Reddy GVP, Raman A (eds) Biological invasive plants. Nature 565:161. https://doi.org/10.1038/ control of tropical weeds using arthropods. Cambridge d41586-019-00037-9 University Press, Cambridge, pp 63–73 Julien MH, Griffiths MW (1998) Biological control of weeds: a Myers JH (1985) How many insect species are necessary for world catalogue of agents and their target weeds, 4th edn. successful biocontrol of weeds? In: Delfosse ES (ed) Pro- CSIRO Entomology CABI Publishing, Oxon ceedings of the VI international symposium on biological King AM, Brudvig R, Byrne MJ (2011) Biological control of control of weeds, August 19–25, 1984, Vancouver. Agri- dense-thorned bitter apple, Solanum sisymbriifolium Lam. culture Canada, Ottawa, pp 77–82 (Solanaceae), in South Africa. Afr Entomol 19:427–433 Myers JH (2008) One agent is usually sufficient for successful Kluge RL (1991) Biological control of crofton weed, Ageratina biological control of weeds. In: Julien MH, Sforza R, Bon adenophora (Asteraceae), in South Africa. Agric Ecosyst MC, Hatcher PE, Hinz HL, Rector BG (eds) Proceedings of Environ 37:187–191 the XII international symposium on biological control of Kumaran N, Lockett C, Dhileepan K (2018) Effect of simulated weeds. CAB International Wallingford, UK, pp 601–606 herbivory on bellyache bush (Jatropha gossypiifolia L.) Newete SW, Oberprieler RG, Byrne MJ (2011) The host range growth and implications for biological control. Weed Biol of the Eucalyptus Weevil, Gonipterus ‘‘scutellatus’’ Gyl- Manage 18:151–159 lenhal (Coleoptera: Curculionidae), in South Africa. Ann Messing R, Brodeur J (2018) Current challenges to the imple- For Sci 68:1005–1013 mentation of classical biological control. BioControl Niu H, Liu W, Wan F, Liu B (2007) An invasive aster (Ager- 63:1–9 atina adenophora) invades and dominates forest under- Milbrath LR, Nechols JR (2014) Plant-mediated interactions: stories in China: altered soil microbial communities considerations for agent selection in weed biological con- facilitate the invader and inhibit natives. Plant Soil trol programs. Biol Control 72:80–90 294:73–85 Morin L, Evans KJ, Sheppard AW (2006) Selection of pathogen Page AR, Lacey KL (2006) Economic impact assessment of agents in weed biological control: critical issues and Australian weed biological control. CRC for Australian peculiarities in relation to agents. Aust J Ento- weed management Technical Series #10, pp. 39 mol 45:349–365 Paynter Q, Fowler SV, Groenteman R (2018) Making weed Morin L, Reid AM, Sims-Chilton NM, Buckley YM, Dhileepan biological control predictable, safer and more effective: K, Hastwell GT, Nordblom TL, Raghu S (2009) Review of perspectives from New Zealand. BioControl 63:427–436 approaches to evaluate the effectiveness of weed biological Plant Protection News (1987) Crofton weed. Plant Protection control agents. Biol Control 51:1–15 News, 7. The Plant Protection Research Institute, Pretoria Morris MJ (1989) Host specificity studies of a leaf spot fungus, Plant Protection News (1988) Fungus released against crofton Phaeoramularia sp., for the biological control of crofton weed. Plant Protection News, 11. The Plant Protection weed (Ageratina adenophora) in South Africa. Phytophy- Research Institute, Pretoria lactica 21:281–283 Poudel AS, Jha PK, Shrestha BB, Muniappan R (2019) Biology Morris MJ (1991) The use of plant pathogens for biological and management of the invasive weed Ageratina adeno- control in South Africa. Agric Ecosyst Environ phora (Asteraceae): current state of knowledge and future 37:239–255 research needs. Weed Res 59:79–92 123 468 L. Buccellato et al.

Land Protection (2001) NRM facts: Ageratina adenophora. plant performance. Proc R Soc B 280:20122756. https:// Queensland Government, Natural Resources and Mines, doi.org/10.1098/rspb.2012.2756 Queensland Stiling P, Cornelissen T (2005) What makes a successful bio- Pyseˇk P, Richardson DM (2007) Traits associated with inva- control agent? A meta-analysis of biological control agent siveness in alien plants: where do we stand? In: Nentwig W performance. Biol Control 34:236–246 (ed) Biological invasions. Springer-Verlag, Berlin, Trounce R (2003) Crofton weed. In: Gosper H (ed) Agfact pp 97–125 P7.6.36, 2nd edn. New South Wales Agriculture, Australia, Raghu S, Wilson JR, Dhileepan K (2006) Refining the process of pp 1–4 agent selection through understanding plant demography Turner PJ, Morin L, Williams DG, Kriticos DJ (2010) Interac- and plant response to herbivory. Aust J Entomol tions between a leafhopper and rust fungus on the invasive 45:308–316 plant Asparagus asparagoides in Australia: a case study of Rosskopf EN, Charudattan R, Kadir JB (1999) Use of plant two agents being better than one for biological control. Biol pathogens in weed control. In: Bellows TS, Fisher TW Control 54:322–330 (eds) Handbook of biological control. Academic Press, Wang F, Summerell BA, Marshall D, Auld BA (1997) Biology London, pp 891–916 and pathology of a species of Phaeoramularia causing a Rutherford MC, Mucina L, Lo¨tter MC, Bredenkamp GJ, Smit leaf spot of crofton weed. Aust Plant Pathol 26:165–172 JHL, Scott-Shaw CR, Hoare DB, Goodman PS, Zachariades C, Strathie LW, Retief E, Dube N (2011) Progress Bezuidenhout H, Scott L, Ellis F, Powrie LW, Siebert F, towards the biological control of Chromolaena odorata (l.) Mostert TH, Henning BJ, Venter CE, Camp KGT, Siebert R.M. King and H. Rob. (Asteraceae) in South Africa. Afr SJ, Matthews WS, Burrows JE, Dobson L, van Rooyen N, Entomol 19:282–302 Schmidt E, Winter PJD, du Preez PJ, Ward RA, Wil- Zachariades C, Paterson ID, Strathie LW, Hill MP, van Wilgen liamson S, Hurter PJH (2006) Savanna biome. In: Mucina BW (2017) Assessing the status of biological control as a L, Rutherford MC (eds) The vegetation of South Africa, management tool for suppression of invasive alien plants in Lesotho and Swaziland. South African National Biodi- South Africa. Bothalia 47(2):a2142. https://doi.org/10. versity Institute, Pretoria, pp 438–539 4102/abc.v47i2.2142 Schaffner U, Smith L, Cristofaro M (2018) A review of open- field host range testing to evaluate non-target use by her- bivorous biological control candidates. BioControl Lisa Buccellato This research was part of her PhD project 63:405–416 investigated the interactions between the stem-gall fly, and Schwarzla¨nder M, Moran VC, Raghu S (2018a) Constraints in leaf-spot pathogen, released against Crofton weed, and their weed biological control: contrasting responses by imple- effect on the plant. menting nations. BioControl 63:313–317 Schwarzla¨nder M, Hinz HL, Winston RL, Day MD (2018b) Marcus Byrne studies various aspects of biocontrol including Biological control of weeds: an analysis of introductions, agent effects on invasive alien plants, and post-release rates of establishment and estimates of success, worldwide. evaluation. BioControl 63:319–331 StatSoft, Inc. (2007) STATISTICA (data analysis software Jolene Fisher is involved with post-release monitoring of system), version 8.0. http://www.statsoft.com/Products/ biocontrol programs. STATISTICA-Features. Accessed 23 May 2019 Stephens AEA, Srivastava DS, Myers JH (2013) Strength in Ed Witkowski is a plant ecologist working on the ecology and numbers? Effects of multiple natural enemy species on control of invasive alien plants.

123