Insect Herbivores Associated with Lycium Ferocissimum (Solanaceae) in South Africa and Their Potential As Biological Control Agents in Australia

Insect herbivores associated with Lycium ferocissimum (Solanaceae) in South Africa and their potential as biological control agents in Australia

§ L.D. Chari1* , E.V. Mauda2 , G.D. Martin1 & S. Raghu2 1Centre for Biological Control, Department of Zoology and Entomology, Rhodes University, Makhanda, 6400 South Africa 2CSIRO, Ecosciences Precinct, GPO Box 2583, Brisbane, Queensland, 4001 Australia

Lycium ferocissimum Miers (Solanaceae) is an indigenous shrub in South Africa but has become invasive in several countries including Australia, where chemical and mechanical control methods have proved costly and unsustainable. In Australia, biological control is being considered as a management option, but the herbivorous insects associated with the plant in its native range are not well known. The aim of this study was to survey the phytophagous insects associated with L. ferocissimum in South Africa and prioritise promis- ing biological control agents. In South Africa, the plant occurs in two geographically distinct areas, the Eastern and Western Cape provinces. Surveys for phytophagous insects on L. ferocissimum were carried out repeatedly over a two-year period in these two regions. The number of insect species found in the Eastern Cape Province (55) was higher than that in the Western Cape Province (41), but insect diversity based on Shannon indices was highest in the Western Cape Province. Indicator species analysis revealed eight insect herbivore species driving the differences in the herbivore communities between the two provinces. Based on insect distribution, abundance, feeding preference and available literature, three species were prioritised as potential biological control agents. These include the leaf-chewing beetles Cassida distinguenda Spaeth (Chrysomelidae) and Cleta eckloni Mulsant (Coccinel- lidae) and the leaf-mining weevil Neoplatygaster serietuberculata Gyllenhal (Curculionidae). Key words: invasive plants, weed biocontrol, native range surveys, phytophagous insects, agent prioritisation.

INTRODUCTION

Many weed biological control programmes have boxthorn, is a woody plant belonging to the been based on limited and ad hoc surveys that Solanaceae family that includes several important considered only the most damaging and abun- agricultural and environmental species (Haegi dant insect species, often resulting in mixed out- 1976; Arnold & Wet 1993). The genus Lycium L. comes (Goolsby et al. 2006). Biological control consists of small to large shrubs with a wide distri- programmes can be greatly facilitated by a com- bution in arid to semi-arid and temperate to sub- prehensive understanding of the natural enemies tropical regions of the world (Minne et al. 1994). associated with the target plant in its region of Although L. ferocissimum is indigenous to the origin (Syrett et al. 2000; Goolsby et al. 2006). Western and Eastern Cape provinces of South Having an extensive suite of potential agents to Africa (Arnold & Wet 1993; Venter 2000), the choose from allows researchers to make informed species comprises two distinct populations that decisions about the most appropriate agent for the are separated by a 200 km geographic barrier, target weed, ultimately reducing the risk of intro- comprising a combination of the Cape Fold Moun- ducing unsuitable, ineffective agents (Balciunas tains and the Knysna Forest (Fig. 1; Venter 2000). 2004; Goolsby et al. 2006). Such a focus on agent As a result of the plant’s ability to grow in diverse prioritisation/selection is increasingly being advo- environments, it has become invasive in some cated and practised in weed biological control countries. (Morin et al. 2009). In the 19th century, L. ferocissimum was used in Lycium ferocissimum Miers, known as African Australia as a hedge plant and wind-break but has

*Author for correspondence. E-mail: [email protected]

Received 25 November 2019. Accepted 2 March 2020

ISSN 1021-3589 [Print]; 2224-8854 [Online] African Entomology 28(2): 00–00 (2020) DOI: https://doi.org/10.4001/003.028.0000 ©Entomological Society of Southern Africa 2 African Entomology Vol. 28, No. 2, 2020

Fig. 1. Surveyed Lycium ferocissimum sites (l) in the geographically distinct Eastern (EC) and Western (WC) Cape provinces of South Africa (SANBI; Venter 2000). The two populations are separated by a 200 km geographic barrier consisting of a combination of the Cape Fold Mountains and the Knysna Forest. The hypothesised distribution of the two populations was projected (red and blue circles) using the furthest apart points (sites) to determine the diameter. See supplementary figure (Fig. S1) for all known localities of the plant from herbarium records. since spread outside its propagated range, where nation of management strategies (Noble & Rose it outcompetes local plant species (Noble & Adair 2013; Noble & Adair 2016). To facilitate the weed’s 2016). Lycium ferocissimum is widespread in several management, a biological control programme Australian states, including New South Wales, against L. ferocissimum was initiated in Australia in South Australia, Tasmania and Victoria, with 2016. An Australian project, funded by the Austra- restricted distributions in Queensland and West- lian Government’s Department of Agriculture and ern Australia, and extremely restricted distribu- Water Resources, sought the services of Rhodes tions in the Northern Territory. Due to its exten- University, Centre for Biological Control, to pro- sive environmental and economic damage, it vide assistance to The Commonwealth Scientific was declared a ‘Weed of National Significance’ in and Industrial Research Organisation in Australia Australia in 2012 (Australian Government 2013). by conducting surveys for natural enemies of Lycium ferocissimum has also become an invasive L. ferocissimum in its native range, South Africa. weed in New Zealand, where it is restricted to This study aimed to identify the herbivorous coastal areas of the South and North Islands insect fauna associated with the plant’s two popu- (Haegi 1976; Webb et al. 1988; Kriticos et al. 2010; lations in the Eastern and Western Cape provinces Adair 2013). In Australia, despite chemical and of South Africa. In addition to cataloguing the mechanical control efforts, the weed continues to insect herbivores, comparisons were made between expand its distribution into new areas (Julien 2006; the Eastern and Western Cape populations of Noble & Rose 2013). Due to its capacity to regener- L. ferocissimum to test for any differences in insect ate from root stocks, stems and seeds, successful herbivore species richness, diversity and commu- control of the weed will probably require a combi- nity composition. These analyses, combined with Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 3 information on host-specificity, were used to to beat the branches above the tray to knock off all prioritise species that warrant further investiga- insects. Beating was carried out for 5 min at each site. tion as potential biological control agents. All insects were collected and placed in killing bottles charged with ethyl acetate. Insects were METHODS then preserved in 70 % alcohol or pinned. Using identification guides (e.g. Scholtz & Holm 1985; Site determination Picker et al. 2004), insects were identified to the The coordinates for localities known to have had lowest possible taxonomic level and then sent to L. ferocissimum (its natural distribution) were the Biosystematics Division of the Plant Health & obtained from South African herbarium records, Protection Research Institute (Agricultural Research the South African National Biodiversity Institute Council, South Africa) for confirmation of identity. online database (SANBI 2016) and available literature. Sample sites within the two distributio- Data analysis nal ranges were then located during roadside To establish sampling representativeness and surveys, from which a total of 53 sites were sur- adequacy, a sample-based rarefaction (accumula- veyed. At each sampling event, an individual of tion) curve was compiled from the abundances of the L. ferocissimum population was tagged and its insect species collected during each sampling GPS coordinates recorded (see supplementary occasion, using the software EstimateS 9 (Colwell Tables S1–S2). 2005). The rarefaction curve was created using the analytically calculated Sest (number of species Insect surveys expected; Mao Tau). The non-parametric Michae- Native range surveys for phytophagous insects lis-Menten (MMMean) estimator and the inci- on L. ferocissimum were carried out over a two-year dence-based coverage estimator (ICE) were used period, between 2017 and 2018, in both provinces. to evaluate sample size adequacy (Chazdon et al. To account for seasonal variations, collections 1998; Toti et al. 2006). When the observed rarefac- were made during different times of the year, at tion curve (Sest) and the estimators (MMMean and least twice during both the winter (June–August) ICE) converge closely at the highest observed rich- and summer (October–February) months, and ness, the richness estimates can be considered once during the autumn (March–May) and spring representative (Longino et al. 2002). Insect species (October) months. Searches were timed to ensure that were encountered only once were not uniformity across all sites, with three researchers included in the analyses. searching for 5 min each. Leaves were scrutinised To investigate the differences in herbivore for both ectophages and endophages, whilst species richness between the Eastern and Western stems were dissected to find borers. Stems were Cape provinces, rarefaction curves of the two dissected in the field, using secateurs, a dissection provinces were plotted with their 95 % confidence knife and a saw. Leaves with leaf-mining larvae intervals. Any overlap in the confidence intervals were collected and stored in well aerated plastic was taken to reflect no significant differences at containers until they developed into adults. Simi- P > 0.05 (Colwell 2005). To assess the insect species larly, fruits and flowers with any signs of feeding diversity between the two regions, the Shannon were kept in the dark, well aerated chambers to diversity index (H’) was calculated using Esti- allow any developing larvae to complete their mateS 9 (Colwell 2005). The Shannon diversity development. The fruits were kept in the cham- index combines species richness with the equit- bers for 2–3 weeks, as it was assumed that three ability (evenness) with which individuals are dis- weeks would be sufficient time for most larvae to tributed among sites and is less sensitive to rare develop into pupae/adults or for the fruit to species (Shannon & Weaver 1949). As a result of become uninhabitable. Predators, pollinators and unequal variances in diversity between the two parasites were also collected but are not reported provinces, the Mann-Whitney non-parametric in this paper. After active searching, a beating test was used to determine the significance of method was used to sample species that were differences in Shannon diversity indices between difficult to detect or those that were difficult to the Eastern and Western Cape provinces. capture. A single person held a beating tray (50 × Analysis of similarity (ANOSIM), conducted in 50 cm), whilst another person using a beating-stick the PRIMER software package (Clarke & Warwick 4 African Entomology Vol. 28, No. 2, 2020

2001), was used to infer differences in insect RESULTS assemblages between the two regions. A Bray- Curtis similarity measure was used to calculate the A total of 1315 individuals belonging to 96 similarity matrix and non-metric multidimen- morphospecies, were collected from 53 sites across sional scaling (nMDS) was used to visualise differ- the Eastern and Western Cape provinces of South ences in assemblages between the two regions Africa (see Table 1). The analytically calculated (Clarke & Warwick 2001). To ensure that the repre- number of species (Sest) did not converge with the sentation of the data in two dimensions was two richness estimators (ICE and MMMean), appropriate (i.e. not misleading), the stress score suggesting that herbivorous insects were under- was interpreted following recommendations by sampled (Fig. 2). However, since the cumulative Clarke (1993). Values below 0.20 are considered species sampling curve approached an asymptote, acceptable stress values, but the lower the values, sampling can be considered adequate although the more acceptable the outcome. Solutions with not exhaustive (Fig. 2). The overlapping confi- stress values above 0.30 require cautious interpre- dence bands of the rarefaction curves suggested tation (Clarke 1993). Additionally, a scree plot, no significant difference in projected species rich- with stress versus number of dimensions, was ness between the two provinces (Fig. 3). However, plotted to ensure that the reduction of the dimen- diversity in the form of Shannon indices was sions (to two) in the insect community data was significantly higher in the Western Cape Province appropriate. (Fig. 4; Mann-Whitney, U = 451.50, z = –2.59, P = Indicator Species Analysis was used to identify 0.01). insect species which drove the differences in There were differences in insect species compo- clustering into the two assemblages (Eastern and sition between the Eastern and Western Cape Western Cape provinces); i.e. those characterising provinces; but there was also some overlap (Fig. 5). the assemblages of the two provinces. This Of the 96 insect herbivore species collected across method combines relative frequency and relative all sites, 18 (18.8 %) were collected in both prov- abundance of species between groups (provinces; inces. The ANOSIM analysis revealed distinct Dufrêne & Legendre 1997). The resulting indicator herbivorous assemblages between the two prov- values were tested for significance (P < 0.05) using inces, but there was overlap among the assem- a randomisation test (1000 iterations). Indicator blages (Global R = 0.27, P < 0.01). A total of eight Species Analysis was conducted using the ‘labdsv’ indicator species were identified across the two package (Roberts 2010) in the statistical software R provinces. Cassida distinguenda Spaeth (Coleop- (v 3.0.1) (R Core Team 2017). tera: Chrysomelidae), Cleta eckloni Mulsant To identify insects that warrant further investi- (Coleptera: Coccinellidae) and Epilachna sp. 1 gation as potential biological control agents, they (Coleoptera: Coccinellidae) were associated with were prioritised on the following attributes; the Eastern Cape Province sites (Indicator val- prevalence, occurrence, distribution, host speci- ues = 0.37, P < 0.01; 0.34, P < 0.01 and 0.27, P = ficity and available literature. Prevalence of insect 0.03, respectively), whilst Apalochrus sp. 1 (Coleop- taxa was estimated as the percentage of total tera: Malachiidae), Ceratitis spp. (Diptera: Tephri- abundance of each species (at each site), per tidae), Epilachna sp. 2 (Coleptera: Coccinellidae), province. The distribution of each taxon was Homoptera spp. (Hemiptera) and Pachycnema sp. 1 described as the number of sites where it was (Coleoptera, Melolonthidae) were significantly recorded. Historical host range information of associated with the Western Cape Province sites each taxon was obtained from the literature, (Indicator values = 0.13, P = 0.05; 0.34, P < 0.01; including anecdotal evidence and museum 0.14, P = 0.05; 0.18, P = 0.04 and 0.23, P < 0.01, reports. This information was used to estimate the respectively). potential level of specificity of each taxon, i.e. Some 1315 individual insect herbivores were monophagous, polyphagous or oligophagous. collected from both provinces (Eastern Cape = Furthermore, comparisons of prevalence and 995, Western Cape = 320). A greater number of distribution were made between the two popula- insect species was collected from the Eastern Cape tions. Mann-Whitney non-parametric tests were Province (55 species), compared to the Western used to investigate the significance of differences Cape Province (41 species). There was generally in prevalence between the two populations. a lower insect prevalence in the Western Cape Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 5 sed on pub- Continued on p. 00 bol represents un- Miers (Solanaceae) in

Lycium ferocissimum prev. sites prev. sites X 0.2 1 0 0 specificity Monophagous 0 0 1.1 1 Oligophagous 13.7 19 1.8 3 ska ½ toja

/ wi Ð

Gyllenhal Monophagous/ 1.8 6 4.1 4 Suffrian X 1.1 5 0 0 Boheman Schuh Menard Gmelin (fl)(fr) Polyphagous 2.7 4 1.5 2 Spinola Oligophagous 0.6 2 0 0 Borowiec & Spaeth Monophagous 5.7 15 1.1 3 Distant X 0.6 4 0 0

liturellus Fabricius Polyphagous 0.2 1 0 0 nr Mulsant Monophagous 11.4 12 0.4 1 sp. Polyphagous 10.6 10 5.5 4 sp. (rt) X 0.5 2 0 0 sp. 1 X 1.8 1 0 0 sp. 1 X 0 0 0.7 1 sp. 1sp. 2 X X 0.4 0.4 2 4 0 1.5 0 2 sp. X 0.2 2 0 0 sp. (rt)sp. 2 Oligophagous Oligophagous 0.9 1 3 2 0 0 0 0

Cassida distinguenda Cassida melanophthalma Cassida reticulipennis Clytrini Cryptocephalus Epitrix Epitrix Macetes Peploptera Chnootriba Cleta eckloni Epilachna Epilachna Acanthocoris spinosus Neoplatygaster serietuberculata Cicadidae Coccoidea Beaufortiana cornuta Pseudambonea capeneri Schuhistes lekkersingia Antestiopsis thunbergii Cenaeus carnifex Taxa Diet Eastern Cape Western Cape . Prevalence (% of total abundance; prev.) and distribution (number of sites) of herbivorous insects associated with HemipteraHemiptera Pentatomidae Pyrrhocoridae Pentatomidae sp. 1 X 0.3 2 0 0 Hemiptera Pentatomidae HemipteraHemiptera Membracidae Miridae Membracidae X 3 4 0.7 1 lished literature, consultations with taxonomists and anecdotalknown evidence. diet Species encountered specificity. only The once were letters excluded in from the brackets analysis. represent The X foliage sym feeders that are also flower feeders (fl), fruit feeders (fr) or root feeders (rt). the Eastern Cape (EC) and Western Cape (WC) provinces of South Africa. Diet specificity (monophagous, oligophagous and polyphagous) was assessed ba Coleoptera Coccinellidae ColeopteraHemiptera Curculionidae Cicadellidae Curculionidae X 0.5 4 11.4 5 Coleoptera Chrysomelidae Coleoptera Coccinellidae Coleoptera Chrysomelidae Coleoptera Chrysomelidae Coleoptera Chrysomelidae Hemiptera Membracidae Homoptera Homoptera X 0.4 2 3.3 4 Hemiptera Coccoidea ColeopteraColeoptera Coccinellidae Coreidae ColeopteraColeopteraColeoptera Chrysomelidae Chrysomelidae Chrysomelidae Lamprosomatinae X 0.9 7 0 0 ColeopteraColeoptera Curculionidae Curculionidae Lixini sp. Polyphagous 0.6 1 0 0 FOLIAGE FEEDERS Coleoptera Chrysomelidae Table 1 Order Family Species ColeopteraColeoptera Chrysomelidae Coccinellidae Chrysomelidae X 1.4 9 0.4 1 Coleoptera Chrysomelidae 6 African Entomology Vol. 28, No. 2, 2020 prev. sites prev. sites Diet Eastern Cape Western Cape specificity Spinola Polyphagous 0,6 2 0 0 Trybom (fl) Stål X 0,2 1 2,6 1 sp. X 32,4 5 0 0 sp. 1 X 0 0 3 2 Morrison/ Polyphagous 0,3 2 1,1 2 sp. 1 X 0 0 6,6 4 sp. 1 X 0 0 1,5 3 sp. 1 X 0,2 2 0 0 sp. 1 X 0 0 4,8 2 sp. Polyphagous 0,2 1 6,6 8

Thrips simplex Frankliniella schultzei Aculus/Aculops Ceratitis Acanthocoris spinosus Cenaeus pectoralis Promeces Micraspis Apalochrus Pachycnema Amphipsocidae (continued). Coleoptera Coccinellidae DipteraHemiptera Coreidae Tephritidae Tephritidae X 0 0 4,1 3 FLOWER FEEDERS Coleoptera Cerambycidae Trombidiformes Eriophyidae Coleoptera Malachiidae ColeopteraColeopteraUNDETERMINED Coleoptera Melyridae Melyridae Malachiinae Scarabaeidae sp. Melyridae X X 4,1 1 12 6 1,8 1 1,1 2 FRUIT FEEDERS Diptera Tephritidae SEED FEEDERS Hemiptera Pyrrhocoridae Thysanoptera Thripidae Table 1 Taxa Order Family Species Psocoptera Amphipsocidae Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 7

Fig. 2.Sample-based rarefaction curves indicating the observed number of species (Sest;Mao Tau), incidence-based coverage estimator (ICE) and Michaelis-Menten mean (MMMean) richness estimators of all insects collected on Lycium ferocissimum across the surveyed sites (Eastern and Western Cape provinces combined; 76 samples from 53 sites).

Fig. 3. Sample-based rarefaction curves indicating the observed number of insect species (Sest) associated with Lycium ferocissimum in the Eastern and Western Cape provinces of South Africa. Species richness should be compared when the number of samples is equal in all provinces (i.e. approximately 33 samples; vertical dashed line). The dashed lines represent the 95 % confidence intervals. Where confidence intervals overlap, the differences in species richness are not significant (i.e. P > 0.05). 8 African Entomology Vol. 28, No. 2, 2020

Fig. 4. Box plots of Shannon diversity indices (H’) representing the diversity of insects collected on Lycium ferocissimum in the Western and Eastern Cape provinces of South Africa.Asterisk (*) represents a statistically significant difference (Mann-Whitney; P < 0.05) in diversity between the two provinces. The boxes represent 50 % of the data (interquartile range) and the horizontal line inside each box is the median value. The vertical lines represent the maximum and minimum diversity values whilst the dots represent outliers.

Fig. 5. Non-metric multidimensional scaling representation of herbivorous insect species collected on the two Lycium ferocissimum populations in South Africa, in the Eastern (black circles) and Western Cape provinces (open squares). The circles drawn around the points of each province, represent separate insect assemblages identified by ANOSIM (Global R = 0.271, P < 0.001). Data were ordinated using the Bray-Curtis similarity matrix. The stress value of 0.11 indicates that the two dimensional non-metric multidimensional scaling ordination was useful in giving an idea of the main features of the high dimensional structure. Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 9

Province and insects displayed more limited dis- ince, but these differences were not significant tributions compared to the Eastern Cape Province (P > 0.05). (Figs 6 and 7), but only the differences in distribu- Insect herbivore species with the widest distri- tion were statistically significant (Mann-Whitney, butions (at least 10 sites across both provinces) U = 287.50, z = 3.00, P < 0.01; Fig. 7). Insect species included Cassida distinguenda, Cleta eckloni, Neo- in the Eastern Cape Province displayed a substan- platygaster serietuberculata Gyllenhal 1837 (Coleop- tially wider distribution, with at least 12 species tera: Curculionidae), Pseudambonea capeneri Schuh occurring in more than 16 % of the sites when (Hemiptera: Miridae), Chnootriba spp. (Coleop- compared to the Western Cape Province. There tera:, Coccinellidae) and Malachiinae spp. was no difference in overall insect prevalence (Coleoptera: Melyridae) (Table 1). Three of these between the two provinces, probably because species were also amongst the most abundant, there was greater variation in prevalence values in namely C. eckloni, P. capeneri and Chnootriba sp. the Eastern Cape Province (Mann-Whitney, U = The following insect species were the most geo- 454.50, z = 0.77, P = 0.44; Fig. 6). graphically widespread across the native range; When insect species’ prevalence was categorised C. distinguenda, C. eckloni, N. serietuberculata, according to feeding guilds, foliage-feeding P. capeneri, Chnootriba spp. and Malachiinae spp. insects were more prevalent in the Eastern Cape The mite species (Aculus sp. or Aculops sp.; Province, while fruit- and flower-feeding insects Trombidiformes: Eriophyidae) and the insect were more prevalent in the Western Cape Prov- species C. distinguenda, C. eckloni and N. serie- ince (Fig. 8). With regard to the number of sites tuberculata were also amongst the most prevalent where insects were recorded (distribution), herbivores on L. ferocissimum (Table 1). foliage-feeding insects were found at more sites in the Eastern Cape than in the Western Cape Prov- DISCUSSION ince (Mann-Whitney, U = 287.80, z = 3.10, P < 0.01; Fig. 9). Fruit feeders were found at more sites This study revealed a high diversity and abun- in the Western Cape Province, whilst flower feed- dance of herbivorous insects associated with ers were found at more sites in Eastern Cape Prov- L. ferocissimum in South Africa, with considerable

Fig. 6. Insect species prevalence on Lycium ferocissimum in the Western and Eastern Cape provinces of South Africa. There was no statistically significant difference in species prevalence between the two provinces (Mann- Whitney, U = 454.50, z = 0.77, P = 0.44). The boxes represent 50 % of the data (interquartile range) and the horizontal line inside each box is the median value. The vertical lines represent the maximum and minimum values whilst the dots represent outliers. 10 African Entomology Vol. 28, No. 2, 2020

Fig. 7. The number of sites where each insect species occurred (distribution) on Lycium ferocissimum in the Western and Eastern Cape provinces of South Africa.There was a statistically significant difference in number of sites between the two provinces (Mann-Whitney, U = 287.50, z = 3.00, P < 0.01). The boxes represent 50 % of the data (interquartile range) and the horizontal line inside each box is the median value.The vertical lines represent the maximum and minimum values whilst the dots represent outliers. geographic variability. The sampling conducted considerably more than the number of insect was extensive and representative of habitats and species collected over three years on two closely climatic zones of the L. ferocissimum distribution in related and indigenous Solanaceae, Solanum South Africa. Additionally, repeated sampling panduriforme E. Mey (49 species) and Solanum during different times of the year accounted for incanum L. (33 species) (Olckers et al. 1995). seasonal variations. It is unlikely that additional Completely exhaustive surveys are often not sampling activities would yield more than a few achievable except for species that can be identified occasional/rare species. The number of herbivo- easily, such as plants and breeding birds where it is rous insect species collected in this study are possible to obtain a count of all the species present

Fig. 8. Mean (+ S.D.) species prevalence, expressed as the percentage of total abundance of each species, in the Eastern and Western Cape provinces. Prevalence was compared for each insect feeding guild. Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 11

Fig. 9. The mean (+S.D.) distribution expressed as the number of sites where each insect species was observed in the Eastern and Western Cape provinces. Asterisk (*) represents a statistically significant difference (Mann-Whitney, P < 0.05) in number of sites between the two provinces (Colwell & Coddington 1994). For other taxa like Eastern Cape Province generally receives higher arthropods, it is often not possible to identify all rainfall over a more extended period, mostly species, particularly over large spatial scales during summer (Conradie 2012). The Eastern (Ugland et al. 2003). Cape Province biomes and vegetation types are Although sampling effort was extensive, it is also different to those of the Western Cape Prov- possible that some species with cryptic habits may ince and are dominated by Thicket and Nama have been missed. However,surveys for biological Karoo vegetation types (Mucina & Rutherford control agents are inherently different to compre- 2006). hensive insect herbivore surveys to record bio- Conspecific plants growing in different regions diversity because there is a tendency to focus on usually experience differing biotic and abiotic specialist taxonomic groups known to produce pressures; hence they can be expected to display biocontrol agents. While the simultaneous use of phenotypic and genotypic variations, as is the case two different insect collection methods, active with L. ferocissimum (McCulloch et al. in review). In searching and beating, increased the chances of turn, the herbivorous insect assemblages associ- capturing a wider suite of insects, the use of ated with the phenotypic and genotypic variants additional methods may have yielded additional may then differ accordingly (e.g. Denno & species. Roderick 1991; Langellotto & Denno 2004). It is The reasons for the differences in insect preva- plausible that the geographic separation between lence, abundance, distribution and composition the Eastern and Western Cape Province popula- between the two sampled provinces were not tions of L. ferocissimum is the cause for the different determined but may be related to habitat and/or insect assemblages seen in South Africa. Further- climatic differences. The Western Cape Province is more, the presence of different indicator species climatically unique in comparison to the rest of associated with each province suggests that some South Africa because of its Mediterranean-type insect species are not suited for the climatic condi- climate characterised by warm, dry summers and tions or phenotypic variants of L. ferocissimum mild, moist winters (Midgley et al. 2005). The prov- associated with the other province. This highlights ince is home to two of the world’s largest bio- the need to survey the entire distribution of plant diversity hotspots, the Cape Floristic Region and species when trying to identify the full comple- the Succulent Karoo (Myers et al. 2000), both of ment of insect associates. This is particularly appli- which are recognised for their unique characteris- cable to biological control as it allows researchers tics and habitat types (Mucina & Rutherford 2006). to highlight species that are likely to be environ- The province also regularly experiences climate mentally suitable to the bioclimatic distribution of extremes such as droughts and floods (Rouault & the weed in its invasive range. Richard 2003; Araujo et al. 2016). Alternatively, the An important objective of this study was to 12 African Entomology Vol. 28, No. 2, 2020 identify insect species that warrant further investi- have the widest potential distribution in Australia. gation as potential biocontrol agents based on Cassida distinguenda, C. eckloni and N. serietuber- their occurrences, abundance and associated culata were highly prevalent across the native damage as well as information from the available/ range (Table 1). Insects feeding on plants rarely published literature. Some of herbivorous insects have an impact when feeding in low numbers associated with L. ferocissimum are likely to be because their effect is dependent on cumulative monophagous or oligophagous, e.g. C. eckloni, damage. Therefore, multivoltine insects are often C. reticulipennis, C. distinguenda, N. serietuberculata selected as biocontrol agents as this allows their and Epitrix sp. (e.g. Heron & Borowiec 1997; GBIF numbers to build up, increasing the amount of 2017; van Noort & Ranwashe 2017; M. Stiller, pers. damage inflicted on the target species and their comm.) and therefore could be considered as potential to spread in search of more food (Hajek candidate agents. However, since alternative host 2004; Kumaran et al. 2018). Additionally, when plants of insect species are seldom known unless both the adults and immature stages of a potential they are of economic importance, our host-range agent damage the plant, the effects of the agent are assessments are tentative and any prioritised increased. All the above-mentioned candidates species would still require careful host-specificity are multivoltine and damaging across their life testing to confirm their host range. Nevertheless, cycles. Cleta eckloni larvae and adults feed by C. distinguenda, C. eckloni and N. serietuberculata, scraping the soft tissue of the leaf, chewing it were never observed on surrounding vegetation and sucking the juices; causing ‘windowing’ (leav- at any of our study sites across South Africa. ing one epidermis and the veins intact) and Hence, it is plausible that these insects have a rendering the leaves skeletonised and desiccated strong host preference for L. ferocissimum and (Tomaszewska & Szawaryn 2016). Both adults and should be prioritised for confirmation of their host larvae of C. distinguenda consume the leaves of the specificity. target plant. While N. serietuberculata adults chew Lycium ferocissimum has already invaded a large shot holes in the leaves, the larvae create mines proportion of the Australian continent. Selecting within the leaf cuticle. Therefore, all three species biocontrol agents that could establish across the show potential to build up large numbers and entire invaded range would ensure that fewer inflict substantial foliar damage if released in the agents are required to manage the species in invaded range. Australia. For example, the confamilial invasive Based on historical successes of biocontrol on species, Solanum mauritianum Scop. (Solana- other weeds, C. distinguenda, C. eckloni and N. serie- ceae) has a broad distribution in South Africa tuberculata should be given preference because (Olckers 2003, 2011). The flowerbud weevil other congeneric/confamilial species have been Anthonomus santacruzi Hustache (Coleoptera: successful. For instance, several species of Cassida Curculionidae) was released in 2008 (Olckers 2008; have been considered as weed biocontrol agents Klein et al. 2011), but has established in only a small (Goeden et al. 1974; Maw 1976; Kleinjan & Scott part of the weed’s geographic range (Olckers 2011; 1996; Winston et al. 2014) with some released (e.g. Cowie et al. 2016). Retrospective climate matching Kok 2001; Downey et al. 2007; Winston et al. 2014). revealed that regions where the weevil had estab- For example, in 1995 the bitou tortoise beetle lished in South Africa were matched to areas (Cassida sp.) was released against Chrysanthemoides where it was sourced in its native range, while monilifera (L.) Norlindh (Asteraceae), an invasive areas that did not support establishment had a less plant in Australia that is native to South Africa than 50 % match to the native range (Cowie et al. (Downey et al. 2007). Although there are no formal 2016). Collecting extensive distribution data on records of Cleta spp. being considered for weed potential biocontrol agents in the native range can biocontrol anywhere, other herbivorous ladybirds assist researchers in predicting the climates that have been tested. Mada polluta Mulsant (Coleop- could support establishment in the invaded range. tera: Coccinellidae), a member of the subfamily Insect species that were the most geographically Epilachninae, has been tested and proposed for widespread across the native range of L. ferocissi- the biocontrol of Tecomastans (L.) Jussex Kunth var. mum included C. distinguenda, C. eckloni, N. stans (Bignoniaceae) in South Africa (Madire 2013). serietuberculata, P. capeneri, Chnootriba spp. and Similarly, weevils in the subfamily Ceutorhyn- Malachiinae spp. and are therefore most likely to chinae are often specialists that develop on a Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 13 narrow range of plants within a single family or these data will provide a reference point for other genus, or on a single species (Anderson 1993; countries considering biological control against Colpetzer et al. 2004). Prior to the current study, not just L. ferocissimum, but other species within N. serietuberculata had only been collected from the Solanaceae. leaves of L. ferocissimum at two sites in the Eastern Cape Province of South Africa (GBIF 2017; van ACKNOWLEDGEMENTS Noort & Ranwashe 2017), suggesting a restricted This project was supported by funding from the host range. Australian Government’s Department of Agricul- Foundational biodiversity lists of insects associ- ture and Water Resources, as part of its Rural R&D ated with plant species contribute to our under- for Profit programme, through AgriFutures standing of the complex relationships that exist Australia (Rural Industries Research and Develop- between plants and their associated insects. These ment Corporation) (PRJ–010527). Part of the fund- types of studies are more common in European ing for work on this paper was provided by the and American literature and are infrequent in South African Research Chairs Initiative of the developing countries (Lewinsohn et al. 2005, Department of Science and Technology and the Goolsby et al. 2006; but see Novotny et al. 1997). National Research Foundation (NRF) of South This study provides valuable information towards Africa. Any opinion, finding, conclusion or recom- understanding the insect herbivore diversity and mendation expressed in this material is that of the structure associated with the genus Lycium in authors and the NRF does not accept any liability South Africa that will prove beneficial to many in this regard. A. Dold (Rhodes University, Depart- disciplines and not just biological control. With ment of Botany) of the Albany Museum is regards to biocontrol, this study emphasises the acknowledged for his assistance in identifying importance of conducting quantitative studies of Lycium ferocissimum and some of its localities. We the insects associated with the target plant. Iden- also thank M.P. Hill for comments on the manu- tifying the available herbivores across the native script and the two reviewers for refining our work. range allows researchers to prioritise agents that are suitable to manage the species within the §ORCID iDs invaded range. Collecting metadata other than L.D. Chari: orcid.org/0000-0000-0000-0000 simply species identifications allows researchers E.V. Mauda: orcid.org/0000-0000-0000-0000 to further prioritise species before they are intro- G.D. Martin: orcid.org/0000-0000-0000-0000 duced into quarantine facilities. Additionally, S. Raghu: orcid.org/0000-0000-0000-0000

REFERENCES ADAIR, R. 2013. Feasibility of biological control of African ily safe? The need for pre-release assessment of prob- boxthorn Lycium ferocissimum. Unpublished report. able impact of candidate biocontrol agents, with 61 pp. Department of Primary Industries, Parks, some examples. In: Cullen, J., Briese, D., Kriticos, D., Water and Environment. Devonport, Tasmania. Lonsdale, W., Morin, L. & Scott, J. (Eds) Proceedings of ANDERSON, R.S. 1993. Weevils and plants: phylogen- the XI International Symposium on Biological Control etic versus ecological mediation of evolution of host of Weeds. 252–257. CSIRO Entomology, Canberra, plant associations in Curculioninae (Coleoptera: Australia. Curculionidae). Memoirs of the Entomological Society of CHAZDON, R.L., COLWELL, R.K., DENSLOW, J.S. & Canada 125: 197–232. GUARIGUATA, M.R. 1998. Statistical methods for ARAUJO, J.A., ABIODUN, B.J. & CRESPO, O. 2016. estimating species richness of woody regeneration in Impacts of drought on grape yields in Western Cape, primary and secondary rain forests of northeastern South Africa. Theoretical and Applied Climatology 123: Costa Rica. In: Dallmeier, F. & Comiskey, J.A. (Eds) 117–130. Forest Biodiversity Research, Monitoring and Modeling: ARNOLD, T.& DE WET,B. 1993. Plants of Southern Africa: Conceptual Background and Old World Case Studies. Names and Distribution. Memoirs of the Botanical 285–309. Parthenon Publishing, Paris, France. Survey of South Africa No. 62. 825 pp. National CLARKE, K.R. 1993. Non-parametric multivariate analy- Botanical Institute, Pretoria, South Africa. sis of changes in community structure. Australian AUSTRALIAN GOVERNMENT. 2013. Weeds of National Journal of Ecology 18: 117–143. Significance. Online at: CLARKE, K.R. & WARWICK, R.M. 2001. Change in Marine https://www.environment.gov.au/biodiversity/inva- Communities: an Approach to Statistical Analysis and sive/weeds/weeds/lists/wons.html Interpretation. PRIMER-E, Plymouth, U.K. (accessed 21 October 2019). COLPETZER, K., HOUGH-GOLDSTEIN, J., DING, J. & BALCIUNAS, J. 2004. Are mono-specific agents necessar- FU, W. 2004. Host specificity of the Asian weevil, 14 African Entomology Vol. 28, No. 2, 2020

Rhinoncomimus latipes Korotyaev (Coleoptera: Curcu- KLEIN, H., HILL, M.P.,ZACHARIADES, C. & ZIMMER- lionidae), a potential biological control agent of mile- MANN, H.G. 2011. Regulation and risk assessment a-minute weed, Polygonum perfoliatum L. (Polygo- for importations and releases of biological control nales: Polygonaceae). Biological Control 30: 511–522. agents against invasive alien plants in South Africa. COLWELL, R.K. 2005. EstimateS: Statistical Estimation of African Entomology 19: 488–497. Species Richness and Shared Species from Samples. Version KLEINJAN, C.A. & SCOTT,J.K. 1996. Selection of Cassida 9. User’s Guide and Application. Published at: spp. from southern Africa for the biological control of http://viceroy.colorado.edu/estimates.html Chrysanthemoides monilifera in Australia. Annals of (accessed October 2019). Applied Biology 128: 373–385. COLWELL, R.K. & CODDINGTON, J.A. 1994. Esti- KOK, L.T. 2001. Classical biological control of nodding mating terrestrial biodiversity through extrapolation. and plumeless thistles. Biological Control 21: 206–213. Philosophical Transactions of the Royal Society of London. KRITICOS, D.J., CROSSMAN, N.D., OTA, N. & SCOTT, Series B, Biological Sciences 345: 101–118. J.K. 2010. Climate change and invasive plants in CONRADIE, D.C.U. 2012. South Africa’s climatic zones: South Australia. Report for the South Australian Depart- today, tomorrow. 1–9 International Green Building ment of Water, Land and Biodiversity Conserva- Conference and Exhibition. CSIR, Sandton, South tion. 97 pp. CSIRO Climate Adaptation Flagship, Africa. Canberra, Australia. COWIE, B.W., VENTURI, G.,WITKOWSKI, E.T.F. & KUMARAN, N., LOCKETT, C. & DHILEEPAN, K. 2018. BYRNE, M.J. 2016. Does climate constrain the spread Effect of simulated herbivory on bellyache bush of Anthonomus santacruzi, a biological control agent (Jatropha gossypiifolia L.) growth and implications for of Solanum mauritianum, in South Africa? Biological biological control. Weed Biology and Management 18: Control 101: 1–7. 151–159. DENNO, R.F. & RODERICK, G.K. 1991. Influence of LANGELLOTTO, G.A. & DENNO, R.F. 2004. Responses patch size, vegetation texture, and host plant archi- of invertebrate natural enemies to complex-struc- tecture on the diversity, abundance, and life history tured habitats: a meta-analytical synthesis. Oecologia styles of sapfeeding herbivores. 169–196. Habitat 139: 1–10. Structure. Springer, Dordrecht, Netherlands. LEWINSOHN, T.M., NOVOTNY, V. & BASSET, Y. 2005. DOWNEY, P.O., HOLTKAMP, R.H., IRESON, J., Insects on plants: diversity of herbivore assemblages KWONG, R.M. & SWIREPIK, A.E. 2007. A review of revisited. Annual Reviews 36: 597–620. the Chrysanthemoides monilifera biological control LONGINO, J.T., CODDINGTON, J. & COLWELL, R.K. program in Australia: 1987–2005. Plant Protection 2002. The ant fauna of a tropical rain forest: estimat- Quarterly 22: 24–32. ing species richness three different ways. Ecology 83: DUFRÊNE, M. & LEGENDRE, P. 1997. Species assem- 689–702. blages and indicator species: the need for a flexible McCULLOCH, G., MAUDA, E.V., CHARI, L.D., asymmetrical approach. Ecological Monographs 67: MARTIN, G., GURDASANI, K., MORIN, L., 345–366. WALTER, G. & RAGHU, S. (In review). Genetic diver- GBIF. 2019. Neoplatygaster serietuberculatus Colonnelli, sity and morphological variation in African boxthorn 1984 in GBIF Secretariat. GBIF Backbone Taxonomy. (Lycium ferocissimum) – characterising the target weed Checklist dataset: for biological control. Biological Control. https://doi.org/10.15468/39omei MADIRE, L.G. 2013. Biology and host range of Mada (accessed 21 October 2019). polluta, a potential biological control agent of Tecoma GOEDEN, R.D., ANDRES, L.A., FREEMAN, T.E., stans in South Africa. Biocontrol Science and Technology HARRIS, P., PIENKOWSKI, R.L. & WALKER, C.R. 23: 944–955. 1974. Present status of projects on the biological con- MAW, M.G. 1976. Biology of the tortoise beetle, Cassida trol of weeds with insect and plant pathogens in the hemisphaerica (Coleoptera: Chrysomelidae), a possi- United States of America and Canada. Weed Science ble biological control agent for bladder campion, 22: 490–495. Silene cucubalus (Caryophyllaceae), in Canada. Cana- GOOLSBY, J.A.,VAN KLINKEN, R.D. & PALMER, W.A. dian Entomologist 108: 945–754. 2006. Maximising the contribution of native-range MIDGLEY, G.F., CHAPMAN, R.A, HEWITSON, B., studies towards the identification and prioritisation JOHNSTON, P., DE WIT, M., ZIERVOGEL, G., of weed biocontrol agents. Australian Journal of Ento- MUKHEIBIR, P., VAN NIEKERK, L., TADROSS, M., mology 45: 276–286. VAN WILGEN, B.W., KGOPE, B., MORANT, P.D., HAEGI, L. 1976. Taxonomic account of Lycium (Solana- THERON, A.K., SCHOLES, R.J. & FORSYTH, G.G. ceae) in Australia. Australian Journal of Botany 24: 669. 2005. A status quo, vulnerability and adaptation HAJEK, A.E. 2004. Natural Enemies?: an Introduction to assessment of the physical and socio-economic Biological Control. 378 pp. Cambridge University effects of climate change in the Western Cape. 171 pp. Press, Cambidge, U.K. Report to the Western Cape Government, Cape HERON, H. & BOROWIEC, L. 1997. Host plants and Town, South Africa. CSIR Report No. ENV-S-C feeding patterns of some South African tortoise 2005-073. CSIR, Stellenbosch, South Africa. beetles (Coleoptera: Chrysomelidae: cassidoid MINNE, L., SPIE, J.J., VENTER, H.J.T. & VENTER, A.M. Hispinae). Genus 8: 625–658. 1994. Breeding systems in some representatives of JULIEN, M.H. 2006 Biological control of rangeland the genus Lycium (Solanaceae). Bothalia 241: 107–110. weeds in Australia. The Rangeland Journal 28: 47–54. MORIN, L., REID, A.M., SIMS-CHILTON, N.M., Chari et al.: Insect herbivores as biological control agents of Lycium ferocissimum in Australia 15

BUCKLEY, Y.M., DHILEEPAN, K., HASTWELL, ROBERTS, D. 2010. Labdsv: Ordination and Multivariate G.T., NORDBLOM, T.L. & RAGHU, S. 2009. Review Analysis for Ecology. R Package Version1.4-1. Online at: of approaches to evaluate the effectiveness of http://CRAN.R-project.org/package=labdsv weed biological control agents. Biological Control 51: (accessed 1 November 2019) 1–15. ROUAULT,M. & RICHARD, Y.2003. Intensity and spatial MUCINA, L. & RUTHERFORD, M.C. 2006. The Vegetation extension of drought in South Africa at different time of South Africa, Lesotho and Swaziland. 617–657. Strelit- scales. Water SA 29: 489–500. zia. South African National Biodiversity Institute, SOUTH AFRICAN NATIONAL BIODIVERSITY INSTI- Pretoria. South Africa. TUTE. 2016. Botanical Database of Southern Africa MYERS, N., MITTERMEIER, R.A., MITTERMEIER, C.G., (BODATSA) [dataset]. Online at: DA FONSECA, G.A.B. & KENT, J. 2000. Biodiversity http://newposa.sanbi.org/.html hotspots for conservation priorities. Nature 403: (accessed 29 October 2019) 853–858. SCHOLTZ, C.H. & HOLM, E. 1985. Insects of Southern NOBLE, M. & ROSE, M. 2013. Weeds of National Signifi- Africa. 502 pp. Protea Book House, Pretoria, South cance: African Boxthorn: National Best Practice Manual. Africa. 66 pp. Australian Government Department of Agri- SHANNON, C.E. & WEAVER, W.1949. The Mathematical culture Fisheries and Forestry, State of Tasmania, Theory of Communication. 1st Edition. 117 pp. Univer- Australia. sity of Illinois Press, Champaign, Urbana, IL, U.S.A. NOBLE, M.R. & ADAIR, R.J. 2016. The biology of Austra- SYRETT, P., BRIESE, D.T. & HOFFMANN, J.H. 2000. lian weeds 66. Lycium ferocissimum Miers. Plant Protec- Success in biological control of terrestrial weeds by tion Quarterly 31: 70–79. arthropods. In: Gurr, G. & Wratten, S. (Eds) Biological NOVOTNY, V.,BASSET, Y., MILLER, S., ALLISON, A. & Control: Measures of Success. 189–230. Springer, SAMUELSON, G.A. 1997. The diversity of tropical Dordrecht, Netherlands. insect herbivores: an approach to collaborative inter- TOMASZEWSKA, W. & SZAWARYN, K. 2016. Epila- national research in Papua New Guinea. 1–11. Pro- chnini (Coleoptera: Coccinellidae): a revision of the ceedings of The International Conference on Taxonomy and world genera. Journal of Insect Science 16: 1–91. Biodiversity Conservation in East Asia. Korean Institute TOTI, D.S., COYLE, F.A. & MILLER, J.A. 2006. A struc- for Biodiversity Research of Chonbuk National tured inventory of Appalachian grass bald and heath University, Jeonju, Korea. bald spider assemblages and a test of species richness OLCKERS, T. 2003. Assessing the risks associated with estimator performance. Journal of Arachnology 28: the release of a flowerbud weevil, Anthonomus santa- 329–345. cruzi, against the invasive tree Solanum mauritianum UGLAND, K.I., GRAY,J.S. & ELLINGSEN, K.E. 2003. The in South Africa. Biological Control 28: 302–312. species-accumulation curve and estimation of OLCKERS, T. 2008. Anthonomus santacruzi Hustache species richness. Journal of Animal Ecology 72(5): (Curculionidae), a new biological control agent for 888–897. bugweed, Solanum mauritianum Scopoli, in South VAN NOORT, D.S. & RANWASHE, F. 2017. Iziko South Africa, poses no risks to cotton production. African Africa Museum Collection (1800–2013). Version 1.3. Entomology 16: 137–139. Occurrence Dataset. South African National OLCKERS, T. 2011. Biological control of Solanum Biodiversity Institute, Pretoria, South Africa. mauritianum Scop. (Solanaceae) in South Africa: will VENTER, A.M. 2000. Taxonomy of the genus Lycium L. perseverance pay off? African Entomology 19: 416–426. (Solanaceae) in Africa. Ph.D. thesis, University of the OLCKERS, T., HULLEY, P.E. & HILL, M.P. 1995. Insect Orange Free State, Bloemfontein, South Africa. herbivores associated with indigenous species of WEBB, C.J., SYKES, W.R. & GARNOCK-JONES, P.J.1988. Solanum (Solanaceae) in the Transvaal, South Africa, Flora of New Zealand. Volume IV. Naturalized Pterido- and in Namibia. African Entomology 3: 49–57. phytes, Gymnosperms, Dicotyledons. 1365 pp. Botany PICKER, M., GRIFFITHS, C. & WEAVING, A. 2004. Field Division, D.S.I.R, Christchurch, New Zealand. Guide to Insects of South Africa. 2nd Edition. 444 pp. WINSTON, R.L., SCHWARZLÄNDER, M., HINZ, H.L., Struik Publishers, Cape Town, South Africa. DAY, M.D., COCK, M.J.W.& JULIEN, M.H. 2014. Bio- R CORE TEAM. 2017. R: A Language and Environment for logical Control of Weeds: a World Catalogue of Agents and Statistical Computing. R Foundation for Statistical Their Target Weeds. 838 pp. USDA Forest Service, Computing, Vienna, Austria. Online at: Forest Health Technology Enterprise Team. 5th Edi- https://www.R-project.org/ tion. Morgantown, West Virginia, U.S.A. Supplementary material to: L.D. Chari, E.V. Mauda, G.D. Martin & S. Raghu,

Insect herbivores associated with Lycium ferocissimum (Solanaceae) in South Africa and their potential as biological control agents in Australia

African Entomology 28(2): 00–00 (2020).

SUPPLEMENTARY TABLES AND FIGS

Table S1. GPS locations of all Lycium ferocissimum sites surveyed in the Eastern Cape Province of South Africa. Site name Latitude Longitude EC1 –33.3108 26.5169 EC2 –33.2023 26.6190 EC3 –33.1474 26.6415 EC4 –33.1474 26.6415 EC5 –33.1474 26.6415 EC6 –33.1474 26.6415 EC7 –32.8292 26.8917 EC8 –33.2962 26.1474 EC9 –33.3160 26.5207 EC10 –33.2872 26.5309 EC11 –33.2767 26.5417 EC12 –33.2668 26.5547 EC13 –33.4719 26.4846 EC14 –33.6785 26.6772 EC15 –33.6817 26.6797 EC16 –33.5966 26.8870 EC17 –33.5966 26.8887 EC18 –33.3916 26.7075 EC19 –33.3118 26.5141 EC20 –33.5746 25.8704 EC21 –33.6945 25.7999 EC22 –33.7488 25.6886 EC23 –33.7194 24.7202 EC24 –34.1239 24.8038 EC25 –34.0569 24.9238 EC26 –33.9330 25.0106 EC27 –33.4571 26.6893 EC28 –33.3196 26.5381 EC29 –33.7410 24.6154 EC30 –33.8429 24.8660

Table S2. GPS locations of all Lycium ferocissimum sites surveyed in the Western Cape Province of South Africa. Site Name Latitude Longitude WC1 –33.6938 19.5959 WC2 –33.4893 18.7234 WC3 –33.4888 18.7231 WC4 –33.1945 18.7059 WC5 –32.1149 18.8554 WC6 –32.1149 18.8554 WC7 –32.1149 18.8554 WC8 –33.3017 18.6862 WC9 –34.2353 18.4749 WC10 –34.2318 18.4744 WC11 –34.4180 19.1729 WC12 –34.4707 19.8884 WC13 –34.2762 19.5351 WC14 –33.9712 18.3715 WC15 –34.0411 18.3595 WC16 –34.1303 18.4490 WC17 –34.1303 18.4490 WC18 –33.724 18.5006 WC19 –33.726 18.74028 WC20 –33.729 23.26174 WC21 –32.319 18.40577 WC22 –34.42 19.17765 WC23 –34.452 19.88289

Table S3. Geographic occurrences of herbivorous insect taxa collected on Lycium ferocissimum in the Eastern and Western Cape provinces of South Africa. The x denotes that the species was not recorded in the respective province. Foliage feeders Eastern Cape Province Western Cape Province Acanthocoris spinosus Spinola EC1, EC23 x Aculus sp. or Aculops sp. EC9, EC15, EC16, EC21, EC26 x Thrips simplex (Morison)/ EC4, EC20 WC4, WC20 Frankliniella schultzei Trybom Antestiopsis thunbergii EC1, EC15, EC19, EC23 WC12, WC18 Cassida distinguenda Spaeth EC1, EC6, EC10, EC12, EC13, WC13, WC14, WC19 EC14, EC15, EC16, EC18, EC19, EC20, EC22, EC23, EC25, EC26 Cassida melanophthalma x WC3 Boheman/Cassida reticulipennis Borowiec & ĝwiĊtojaĔska Cenaeus carnifex EC18 x Chnootriba sp. EC1, EC10, EC11, EC13, EC15, WC2, WC4, WC14, EC18, EC19, EC20, EC21, EC22 WC21 Cicadidae sp. 1 x WC13 Cleta eckloni Mulsant EC1, EC7, EC10, EC12, EC13, WC21 EC18, EC19, EC20, EC21, EC22, EC26, EC30 Clytrini EC2 x Cryptocephalus nr liturellus EC7, EC13, EC16, EC19, EC23 x Suffrian Curculionidae EC13, EC17, EC21, EC22 WC1, WC2, WC4, WC13, WC19 Epilachna sp. 2 (Coleoptera, EC12, EC15, EC19, EC26 WC15, WC16 Coccinellidae) Epilachna sp. x EC2, EC20 x Epitrix sp. EC2, EC12, EC23 x Epitrix sp. 2 EC2, EC12, EC23 x

Lamprosomatinae EC2, EC3, EC11, EC13, EC18, WC4, WC5 EC20, EC25 Lixini sp. EC22 x Membracidae EC9, EC18, EC25, EC30 WC19 Neoplatygaster EC4, EC21, EC22, EC23, EC26, WC1, WC14, WC18, serietuberculata (Gyllenhal) EC30 WC21 Pseudambonea capeneri EC1, EC2, EC3, EC4, EC5, EC6, WC2, WC13, WC19 Schuh/Schuhistes lekkersingia EC9, EC10, EC11, EC12, EC13, Menard EC14, EC15, EC18, EC19, EC20, EC21, EC23, EC28

Peploptera sp. EC2, EC5 x Fruit feeders Acanthocoris spinosus Spinola EC1, EC23 x Antestiopsis thunbergii EC1, EC15, EC19, EC23 WC12, WC18 Ceratitis sp. EC12 WC9, WC10, WC12, WC13, WC15, WC18, WC19, WC22 Tephritidae EC10, EC13 WC3 Seed feeders Cenaeus pectoralis Stål EC18 WC10 Flower feeders Thrips simplex (Morison)/ EC4, EC20 WC4, WC20 Frankliniella schultzei Trybom Antestiopsis thunbergii EC1, EC15, EC19, EC23 WC12, WC18 Melyridae EC2, EC3, EC9, EC14, EC16, WC5, WC21 EC20 Micraspis sp. 1 x WC2, WC3 Promeces sp. 1 EC3, EC4 x Root feeders Epitrix sp. 2 EC2, EC12, EC23 x Peploptera EC2, EC5 x Undetermined Amphipsocidae sp. 1 x WC14, WC15 Apalochrus sp. 1 x WC13, WC20, WC21 Beaufortiana cornuta Distant EC8, EC13, EC17, EC18 Chrysomelidae EC10, EC11, EC14, EC15, EC18, WC19 EC20, EC22, EC23, EC29 Coccoidea sp. 1 EC8, EC12 WC16 Homoptera EC17, EC19 WC1, WC4, WC9, WC13 Hymenoptera EC13, EC15, EC18, EC22, EC25 WC10, WC11, WC12, WC16, WC18, WC19, WC21, WC22 Macetes sp. EC11, EC23 x Malachiinae sp. EC2, EC3, EC4, EC5, EC6, EC14, WC13 EC15, EC20, EC23, EC25, EC26, EC28 Pachycnema sp. 1 x WC4, WC7, WC12, WC13, WC22 Pentatomidae sp. EC19, EC22 x

Fig. S1. All known localities of Lycium ferocissimum in South Africa. Locality data were collected from herbarium specimens and the South Africa Biodiversity Institute database (SANBI 2016).