Classical biological control of the African citrus psyllid Trioza erytreae, a major threat to the European citrus industry

Citrus greening or huanglongbing (HLB) is the main threat to the European citrus industry since one of its vectors, the African citrus psyllid, Trioza erytreae, has recently become established in mainland Europe. In this context, classical biological control programmes should be implemented to reduce the spread of the psyllid. The aims of this study were to: i) disentangle the parasitoid complex of T. erytreae combining morphological and molecular characterization; and ii) to study the biology of its main parasitoids in its area of origin in South Africa for their future importation into Europe. The main citrus producing areas of South Africa were surveyed during 2017. In contrast to previous studies, the parasitoid complex of T. erytreae included three species of primary parasitoids: Tamarixia dryi, Psyllaephagus pulvinatus and another parasitoid of the genus Tamarixia. Molecular analysis showed that it is a new species closely related to T. dryi. Tamarixia dryi was the most abundant parasitoid but its relative abundance varied among sampling sites. The sex ratio (males/females) of T. dryi and Tamarixia sp. decreased with T. erytreae size and became female biased when psyllid nymphs were larger than 0.6 and 1.2 mm2, respectively. These parasitoids were attacked by three species of hyperparasitoids, Aphidencyrtus cassatus, Marietta javensis and a species of the genus Aphanogmus. Aphidencyrtus cassatus, the most abundant hyperparasitoid, tended to emerge from large nymphs, and adult females lived as long as those of T. dryi. The implications of these results are discussed within the framework of the introduction of T. dryi into Europe.

In Europe, T. erytreae was first detected in 1994 in Madeira (Portugal) 22 and later on in the Canary Islands (Spain) in 2002 23,24 . Until then, it seemed to be restricted to non-continental areas, but in 2014 it first appeared in the north-western Iberian Peninsula 13 . Despite the initial insecticide treatments to eradicate it, T. erytreae is now spreading from the north-west to the south-west of the Iberian Peninsula 15 . Parasitoids of the genus Tamarixia are amongst the most effective natural enemies of HLB vectors 5,[25][26][27] . However, no native parasitoids have been recorded from T. erytreae in the Atlantic islands or in the Iberian Peninsula 15,28 . In this context, classical biological control seems to be the most feasible measure for preventing T. erytreae to spread further in the Mediterranean citrus growing areas.
The complex of parasitoids in South Africa and Swaziland was analyzed in detail during the 1960's and 70 s 29,30 and in Cameroon twenty years ago 31 . According to these studies, the two main primary parasitoids of T. erytreae in Southern Africa are Tamarixia dryi (synonym Tetrastichus dryi) (Waterston) (Hymenoptera: Eulophidae) and Psyllaephagus pulvinatus (Waterston) (Hymenoptera: Encyrtidae). Both are solitary koinobiont parasitoids. The former is an ectoparasitoid, whereas the encyrtid is an endoparasitoid. These primary parasitoids are frequently attacked by a complex of hyperparasitoids 26,32 that, accordingly to Van Der Berg and Greenland 26 , severely decrease the impact of the primary parasitoids. Tamarixia dryi was used in a classical biological control programme in Reunion Island when T. erytreae was detected in 1974 25,33 . In less than five years, T. dryi became established and controlled T. erytreae 25,26 . Similarly, in 1982, T. dryi was imported into Mauritius 34 . In these islands both T. erytreae and D. citri coexisted as well as the African and Asian forms of HLB. However, only the classical biological control of T. erytreae with T. dryi was successful 34,35 . The lack of alternative hosts for T. erytreae, the presence of alternative hosts for T. dryi and the absence of hyperparasitoids were considered key aspects for the establishment of T. dryi and the successful control of T. erytreae populations [36][37][38][39] . Whether T. dryi would find these conditions in mainland Europe its unknown. However, T. dryi is highly specific. The parasitoid did not parasitize and develop in any of the eleven alternative host species that were offered in host-specificity tests 40 .
In this study, we propose a classical biological control programme to introduce the main parasitoids of T. erytreae from its area of origin into Europe. We first identified the parasitoid complex of T. erytreae in several areas of South Africa using morphological and molecular characterisation. We then determined several aspects of the biology of the main parasitoids: sex ratio, longevity and hyperparasitism. Implications of using T. dryi in classical biological control programmes of T. erytreae are discussed.

Results
Insect survey. The Western Cape was the only province in South Africa where neither T. erytreae nor its symptoms were recorded (see supplementary material Table S1). Leaves of trees from 17 out of the 65 sampled sites in the other provinces had the characteristic open gall-like structures, indicating the presence of T. erytreae. Live nymphs were collected from five out of the 17 sites and parasitoids were recorded from four sites. parasitoid emergence and species abundance. A total of 580 parasitized T. erytreae individuals were collected during the survey. From these samples 334 parasitoids belonging to five species emerged. The parasitoids in the remaining 246 psyllids failed to develop. The parasitoid complex was composed of at least five species whose relative abundance varied with sampling site (Fig. 1). Tamarixia dryi was the most abundant primary parasitoid in Pretoria and Nelspruit (>95% of the emerged parasitoids). This parasitoid species was also present in Tzaneen. On the other hand, the primary parasitoids P. pulvinatus (79%) and Tamarixia sp. (65%) were the most abundant species in Nelspruit and Tzaneen, respectively. The most abundant hyperparasitoid was Aphidencyrtus cassatus Annecke (Hymenoptera: Encyrtidae), which was recovered in Nelspruit, Tzaneen and Pretoria. One specimen of an Aphanogmus sp. (Hymenoptera: Ceraphronidae) was recorded in Tzaneen.
The phylogenetic tree of these Tamarixia species was paraphyletic with two distinct branches (Fig. 3). The first branch separated into two clusters. The first cluster grouped together the COI barcode sequences obtained in this work from the T. dryi specimens, while the second cluster hosted the T. radiata, T. drukyulensis and T. triozae GenBank accessions included in the analysis. The COI barcode sequence from the new Tamarixia sp. branched separately from the rest of the Tamarixia species included in this analysis. seasonal trend of the parasitoid complex of T. erytreae. The most abundant parasitoid was T. dryi, followed by its hyperparasitoid A. cassatus, but their relative abundance showed different trends (Fig. 4). While the relative abundance of T. dryi decreased over the five weeks, the number of hyperparasitoids increased during  www.nature.com/scientificreports www.nature.com/scientificreports/ the same period. The primary parasitoid Tamarixia sp. was recorded during the first and third sampling periods, and the hyperparasitoid Marietta javensis (Howard) (Hymenoptera: Aphelinidae) was found only at the second sampling date.

Discussion
Trioza erytreae was highly parasitized by several species of hymenopteran parasitoids. Parasitism rates were high in all the sampled areas in spring and ranged between 0.40 and 0.70. These rates are similar to those reported by van der Merwe 42 (1923) and Catling 43 in South Africa and Swaziland, respectively, and Tamesse et al. 44 in Cameroon. Therefore, as demonstrated by Catling in the 1960s, parasitoids are important biotic regulators of T. erytreae in those areas where insecticides are not sprayed in South Africa, e.g. abandoned and experimental orchards and public gardens 29,45 . This result reinforces the suggestion of introducing exotic parasitoids to those areas where T. erytreae has arrived and where effective native parasitoids are absent. This is the case of Maderia (Portugal), the Canary Islands (Spain) and, more recently, mainland Europe 15 . Among the three species of primary parasitoids of T. erytreae, T. dryi was the most effective and abundant, as has been previously demonstrated by other studies in South Africa and Swaziland 30,43 . Its relative abundance was higher than 90% in two sites. Similar values were obtained by Catling 43 in South Africa. Another primary parasitoid of T. erytreae, P. pulvinatus, was only present in Nelspruit-ARC, where it was the most abundant species (80%). The high abundance of P. pulvinatus at this site may be due to the absence of T. dryi. Although P. pulvinatus parasitizes younger nymphs than T. dryi, its developmental time is longer than that of T. dryi 30 . This reason might partially explain why T. dryi tends to be more abundant where T. erytreae is present at low population densities. www.nature.com/scientificreports www.nature.com/scientificreports/ A new parasitoid species from the genus Tamarixia was recorded in Tzaneen and Pretoria. It was the most abundant species in Tzaneen (70%), coexisting with T. dryi and the hyperparasitoid A. cassatus. In Pretoria, it was recorded sporadically. This new species could be the same species named as "Tetrastichus sp. n" in Western and Eastern Africa and classified as a primary parasitoid (Aubert 1986). In other studies in Cameroon, Zimbabwe and Malawi, an unknown "Tetrastichus sp. " was also found, but it was recorded as a hyperparasitoid 30,34,44 . The high abundance of Tamarixia sp. and the fact that we never observed the pupae or larvae of any Tamarixia species attacked by other larvae suggest that it is a primary parasitoid. The molecular analysis confirmed that this new species has not been reported yet in the database Genbank and it is closely related to T. dryi.
The sex ratio of T. dryi and Tamarixia sp. turned female biased when the size of T. erytreae was greater than 0.40 and 0.90 mm 2 , respectively. Many species of solitary parasitoids lay male eggs on small hosts and female eggs on large hosts 46,47 . This host-size-dependent sex ratio is presumed to be advantageous because females gain a greater benefit from the resulting increase in body size than do males 46,48 . According to Waage 49 , this occurs mostly in idiobionts, which paralyze the host, because in koinobiont parasitoids, hosts continue growing and its size at oviposition is a less reliable predictor of the resources that offspring will have available for development 50,51 . In our case, although T. dryi is a koinobiont, host size increases only slightly after parasitism (personal observations). Therefore, host size could be a reliable predictor of the resources when T. dryi recognizes and assesses the size of T. erytreae individuals. These results are important for mass-rearing the parasitoids and maximizing the production of females 52 .
Hyperparasitism rates were low and ranged between zero and 0.1 at the four sites. In Pretoria, where the seasonal trend of the parasitoids was determined in spring, hyperparasitism also reached a maximum of 0.1. These values were higher than the ones obtained by Catling and Anneke in the Letaba District [Limpopo (then Transvaal province) South Africa] from 1965 to 1967, although hyperparasitism was thought to be of little apparent significance, later on Mc Daniel and Moran 30 and Anneke and Moran 53 determined that it was the main factor in the increase of T. erytreae population levels in citrus. Among the hyperparasitoids, A. cassatus was the most abundant and widely distributed. Aphidencyrtus cassatus is considered the most abundant hyperparasitoid of T. erytreae and it attacks, at least, the two primary parasitoids T. dryi and P. pulvinatus 43,44,54 . Moreover, A. cassatus has been observed host feeding in the thoracic region of the primary parasitoid host, which sometimes killed the host 30 . This mortality and the high temperatures reached at the end of spring could have caused the high mortality rates of the primary parasitoids observed in Pretoria.
Two traits of the biology of A. cassatus could explain the negative impact that this hyperparasitoid has on T. dryi. Firstly, the probability that the hyperparasitoid A. cassatus emerged from T. erytreae nymphs increased with the size of the nymph. Since T. dryi females develop in larger nymphs than males, A. cassatus may affect the secondary sex ratio of T. dryi and more males will emerge. Secondly, females of the hyperparasitoid A. cassatus lived for more than 30 days when they had access to carbohydrates, and their longevity was similar to that of T. dryi. Therefore, the hyperparasitoid is capable of surviving long periods of host scarcity or, at least, as long as the primary parasitoid. Although hyperaparasitoids can regulate herbivore populations by stabilising host-parasitoid interactions [55][56][57] , from the point of view of classical biological control, the accidental introduction of A. cassatus could impair the establishment of T. dryi 58 . Therefore, great care should be taken to exclude this parasitoid when importing T. dryi to Europe, for example introducing only adult parasitoids, establishing isolines and following quarantine procedures before the release 59,60 .

Materials and Methods
Insect survey. Sampling took place in citrus producing areas in four provinces, Gauteng, Limpopo, Mpumalanga and Western Cape, of South Africa. A total of 65 citrus orchards, 5 public parks and 60 private properties were examined from September 21st to December 9th of 2017. At all sites, the sampling date, the number of trees sampled, the variety of the trees and the presence of T. erytreae or its visual symptoms and parasitoids were recorded (see supplementary material Table S1). Visual symptoms of T. erytreae on citrus, in contrast to those of Diaphorina citri, consist of pit galls protruding from the upper surface of the leaves, chlorosis and leave twisting 15,61 .

Parasitoid identification, relative abundance and parasitism rates. From those areas and trees
where T. erytreae was collected, the psyllids were transported to the laboratory to identify potential parasitoids, determine their relative abundance and parasitism rates in each location. From each tree, 3 to 20 leaves infested by T. erytreae were collected and transported in enclosed individualized plastic bags to the laboratory. Due to the scarcity of T. erytreae in some of the orchards, the number of sampled trees and leaves was variable (see supplementary material Table S1). Once in the laboratory, the number of live psyllids and psyllids suitable for parasitism (2 nd -to the 5 th -instar nymph) and parasitized psyllids was recorded using a stereomicroscope. In order to identify all the parasitoids and calculate the rate of emerging parasitoids, psyllid nymphs were placed individually in 1 ml microtubes closed with cotton wool. Afterwards, the microtubes tubes were kept in an incubator (Labcon TM 2 LTGC 20, Laboratory Marketing Services cc, South Africa) under controlled conditions (12 L: 12 Dh, 25 °C, 60-70% RH) and checked daily until parasitoids emerged. Once emerged, their sex was determined and the identification at species level was carried out using the key of Tamesse 62 . The morphological identifications of T. erytreae and Tamarixia parasitoids were confirmed by David Ouvrard (Natural History Museum, London) and Roger Burks (University of California, Riverside), respectively. Moreover, voucher specimens of all the parasitoid species collected during this study were deposited at the University of California, Riverside and labeled with the database number UCRC_ENT from 00517324 to 00517336.
In order to calculate the parasitism rate, each tree was used as a sampling unit because T. erytreae, as other psyllids, has an aggregative distribution pattern [63][64][65] . (2019) 9:9440 | https://doi.org/10.1038/s41598-019-45294-w www.nature.com/scientificreports www.nature.com/scientificreports/ DNA extraction, pCR and sequencing of barcode fragment. The morphological identifications were verified with molecular identifications using cytochrome c oxidase subunit 1 (COI) sequences of T. erytreae and of the parasitoids T. dryi and Tamarixia sp. The insects were collected at three different locations, Nelspruit, Pretoria, and Tzaneen, in South Africa. DNA was extracted from individual insects using a salting out method 66 adapted from Monzó et al. 67 . Polymerase chain reaction (PCR) was performed to amplify the mitochondrial COI gene. Standard primers LCO1490 (5′-GGTCAACAAATCATAAAGATATTGG-3′) and HCO2198 (5′-TAAACTTCA GGGTGACCAAAAAATCA-3′) 68 were used for the amplification of a 708-bp fragment of the COI gene of Tamarixia species, while Te-6U30 (5′-ATTTTTAAGCACTAATCATAAAATTATTGG-3′) and Te-720L26 (5′-TATACTTCAGGATGTCCAAAAAATCA-3′) specific primers were used to amplify a COI barcode fragment of a 714-bp fragment of the COI of T. erytreae. PCR was carried out in a total reaction volume of 20 µl containing 1X Reaction Buffer (Mg free), 1625 µM MgCl2, 250 µM of all four dNTPs, 0.25 µM of each primer, 1 U of DNA polymerase (1 U/µl, Biotools), and 1 µl of DNA in a thermal cycler (Eppendorf Mastercycler). Reactions were cycled as follow: initial denaturation at 95 °C for 120 seconds (s); 40 cycles of 94 °C for 60 s, 45 °C for Tamarixia or 54 °C for T. erytreae for 60 s and 72 °C for 90 s; and a final extension at 72 °C for 600 s. Amplified PCR products were resolved in a 1.2% agarose gel and successfully amplified barcode fragments were cleaned up using the UltraClean PCR Clean-up DNA Purification Kit (MO BIO Laboratories Inc., Carlsbad -California, USA). Amplified barcode fragments were bidirectional sequenced using both the forward and reverse PCR primers. Sanger-sequencing was performed by capillary electrophoresis using a 3130XL Genetic Analyzer (Applied Biosystems, Carlsbad -California, USA), at the Sequencing Service of the IBMCP (Valencia, Spain). Sequences were analysed and trimmed to remove primer sequences, using Sequencer DNA Sequence Analysis Software (Gene Codes Corporation, Ann Arbor -Michigan, USA). Forward and reverse high-quality reads obtained for each individual were assembled into consensus sequences and submitted to the GeneBank public sequence repository.
sequence data and phylogenetic analysis. The phylogenetic analysis of the two species of Tamarixia was carried out using the COI barcode sequences obtained in this work together with those already included in the GenBank: T. radiata (Waterston), T. drukyulensis Yefremova and Yegorenkov and T. trioza (Burks). Consensus sequences corresponding to the amplified COI barcode fragment of the Tamarixia parasitoids were first used as query to BLASTN against the non-redundant nucleotide collection of the GenBank database (http:// blast.ncbi.nlm.nih.gov/Blast.cgi), optimising for highly similar sequences. The barcode based taxonomic assignment at the genus level was set at 94% identity over a 90% sequence overlap. Multiple alignments of consensus sequences and closest sequence identities were done using Clustal Omega software 69 . The phylogenetic analysis was carried out in MEGA X 70 , using the Neighbor-Joining method 71 . The reliability of the tree pattern was evaluated using a bootstrap test with 10000 replicates 72 , and the evolution distances, given as units of the number of base substitution per sites, were computed using the Maximum Composite Likelihood method 73 . The rate variation among sites was modelled with a gamma distribution (shape parameter = 1). seasonal trend of the parasitoid complex of T. erytreae. The seasonal trend of T. erytreae and its parasitoids was studied in an infested lemon orchard located at the University of Pretoria Experimental Farm (25°44′51.1″S 28°15′31.2″E). The orchard was ~10 years old, and the trees sampled were not treated with pesticidesduring the sampling period. From October to December, five infested leaves from five trees were sampled every two weeks. Samples were processed following the same methodology described above until parasitoids emerged and were identified.
Effect of host size on secondary sex ratio and hyperparasitism of parasitoids. The effects of host size on the secondary sex ratio of T. dryi and Tamarixia sp. and on hyperparasitism were analyzed. After parasitoids emerged they were identified and their sex was determined. The psyllid nymphs have an oval shape and host size was determined by calculating the area of an ellipse by multiplying r 1 × r 2 × π (r 1 : major radius, r 2 : minor radius). Both sex and hyperparasitism ratios were analyzed using generalized linear models assuming binomial errors. The assumed error structure was assessed by a heterogeneity factor equal to the residual deviance divided by the residual degrees of freedom. If an over-or an under dispersion was detected, we re-evaluated the significance of the explanatory variables using an F-test after rescaling the statistical model by a Pearson's χ2 divided by the residual degrees of freedom. All data analyses were performed with the R freeware statistical package (Version 1.0.143) 74 .

Longevity of Tamarixia dryi and its hyperparasitoid Aphidencyrtus cassatus. Tamarixia dryi
and A. cassatus longevity was recorded. From the individual nymph sampled at the University of Pretoria Experimental Farm, a total of 20 females and 12 males of T. dryi as well as 14 females and 7 males of A. cassatus were selected. Individual parasitoids were placed singly in 1 ml microtubes tubes closed with cotton wool and 1 M sucrose drop renewed every two days. The microtubes were kept under controlled conditions in an incubator (Labcon 2 LTGC 20, Laboratory Marketing Services CC, Roodepoort, South Africa) at 12L: 12D, 25 ± 2 °C and 60-70% RH, and the survival of the insects was checked daily. The Cox regression model was used to determine differences between sexes within the same species and between species. Analyses were carried out using the R freeware statistical package (Version 1.0.143) 74 . ethical approval. This article does not contain any studies with human participants performed by any of the authors. All applicable international, national, and/or institutional guidelines for the care and use of animals were followed.