Department DEMETRA, Laboratory of Acarology Eliahu Swirski, University of Palermo, Viale delle Scienze 13, Palermo, Italy 2

bs_bs_banner Zoological Journal of the Linnean Society, 2012, 165, 253 273. With 8 figures The concept of genus within the family Phytoseiidae (Acari: Parasitiformes): historical review and phylogenetic analyses of the genus Neoseiulus Hughes HARALABOS TSOLAKIS 1 *, MARIE STEPHANE TIXIER 2, SERGE KREITER 2 and SALVATORE RAGUSA 1 1 Department DEMETRA, Laboratory of Acarology Eliahu Swirski, University of Palermo, Viale delle Scienze 13, 90128 Palermo, Italy 2 Montpellier SupAgro, Unité Mixte de Recherche Centre de Biologie pour la Gestion des Populations INRA/IRD/CIRAD/ Montpellier SupAgro, Campus International de Baillarguet, CS 30016, 34988 Montferrier-sur-Lez cedex, France Received 1 April 2011; revised 21 November 2011; accepted for publication 1 December 2011 Systematic studies on the family Phytoseiidae were first conducted at the beginning of the 20 th century but increased greatly after the Second World War. Various classifications have been proposed based on different characters such as: dorsal, ventral, and leg chaetotaxy; the shape of ventrianal and sternal shields; the shape of the insemination apparatus (spermatheca) and spermatodactylus; the number of teeth on the movable digit of chelicera; and dorsal and ventral adenotaxy. The genus concepts developed over the last five decades can be divided into two main categories or hypotheses. The first, supported mainly by Chant and McMurtry, focuses on dorsal and ventral chaetotaxy, and the genera so defined usually include a great number of species. The second category, proposed by Athias-Henriot, considers the shape of the insemination apparatus as the key character, and the genera so defined usually include a limited number of species. From a diagnostic point of view, both classifications have a valid structure, but the question investigated herein was: which of the two classifications or hypotheses fits phylogenetic evolution? To answer this, we conducted molecular phylogenetic analyses (using the genes ITS and 12S rrna) on the genus Neoseiulus, which has been subjected to classification based on the two main genus concepts. The results showed that the first hypothesis (Chant and McMurtry) leads to polyphyly of the genus Neoseiulus, while the second (Athias-Henriot) leads to paraphyly of the genus. The results show that acarologists who first decided that the insemination apparatus was of evolutive importance could be correct as the shape of the insemination apparatus seems to better fit evolutive clades than dorsal and ventral chaetotaxy. The morphology of this organ, however, must be more accurately studied to better define homologies. The present paper investigates the two main hypotheses proposed until now for classification of Phytoseiidae and thereby opens the way for improved classification.. doi: 10.1111/j.1096-3642.2011.00809.x ADDITIONAL KEYWORDS: 12S rrna chaetotaxy insemination apparatus ITS phylogeny. INTRODUCTION The family Phytoseiidae has great economic importance because many species are natural enemies of *Corresponding author. E-mail: tsolakis@unipa.it phytophagous mites and small insects (McMurtry & Croft, 1997). Their role as biocontrol agents of phytophagous mites was emphasized in the early 20 th century (Parrott, Hodgkiss & Schoene, 1906), and they have received increasing attention from scientists during the last 50 years (Chant, 1961; Oatman & McMurtry, 1966; McMurtry, 1982; McMurtry, Morse 253

254 H. TSOLAKIS ET AL. & Johnson, 1992). The number of described phytoseiid species increased from 61 in 1951 to 165 a few years later (Chant, 1959a) and to 1565 by the end of the 1980s (Chant & Yoshida-Shaul, 1989). In the last world catalogue, de Moraes et al. (2004) reported 2250 phytoseiid species. The systematics of the family Phytoseiidae has a tumultuous and confused history due to a notable lack of agreement among taxonomists on the diagnostic characters that delimitate taxonomic entities (Chant & McMurtry, 2007). Different Phytoseiidae classifications were proposed during the last half century and the characters considered were as follows: the dorsal, ventral, and leg chaetotaxy, the shape of ventrianal and sternal shields, the shape of the insemination apparatus (spermatheca) and spermatodactylus, the number of teeth on the movable digit of chelicera, and the dorsal and ventral adenotaxy (Athias-Henriot, 1977, 1978; Ragusa & Athias-Henriot, 1983; Chant & Yoshida- Shaul, 1989; Takahashi & Chant, 1993; Chant & McMurtry, 2003a, b, 2004a, b, 2005a, b, c, 2006a, b, 2007). The systematic concepts of genus in the Phytoseiidae that were developed can be divided into two main hypotheses. The first emphasizes similarities in dorsal and ventral chaetotaxy; the genera so defined usually include a great number of species (Wainstein, 1962; Muma & Denmark, 1968; Chant & McMurtry, 2003a, b). The second hypothesis emphasizes similarities in the shape of the insemination apparatus; the genera so defined usually include a limited number of species (Athias-Henriot, 1977, 1978; Ragusa & Athias-Henriot, 1983; Ragusa Di Chiara & Tsolakis, 1994). However, none of these hypotheses has been rigorously tested with phylogenetic analyses (likelihood maximum, Bayesian, or parsimony analyses); no bootstrap values were calculated and no outgroup was used in the cladograms sometimes provided (Ragusa & Athias- Henriot, 1983; Takahashi & Chant, 1993; Ragusa Di Chiara & Tsolakis, 1994; Chant & McMurtry, 2003a, b, 2004a, b, 2005a, b, c, 2006a, b, 2007). These studies thus clearly fail to satisfy the requirements for defining evolutionary lines within the family. In 1971, Athias-Henriot stated that because systematists had failed to consider the phenomena of parallel and convergent evolution, which are so common in the Gamasida (Mesostigmata), their classifications seemed more numerical than systematic. Nearly 40 years later, the situation has not changed. The first objective of this paper is to provide a historical review of the characters that have been used to discriminate between Phytoseiidae genera. The second objective is to perform a molecular phylogenetic analysis to test the two approaches, focusing on the example of the genus Neoseiulus Hughes 1948. HISTORICAL REVIEW OF THE FAMILY PHYTOSEIIDAE AND ORGANIZATION OF THE GENERA For many years, species and genera within the Mesostigmata were defined based on only a small number of vague characters, mainly because of the low magnification and poor resolution of optical instruments but also because these characters were sufficient to distinguish the small number of known species. However, as the number of new species increased, these poorly documented descriptions did not provide more accurate assignment of species to genera. Table 1 provides examples of some characters considered in historical definitions for some Phytoseiidae genera. From the first attempts at classification of Phytoseiidae (Berlese, 1913) until the 1970s, the characters used for defining genera increased (Table 1). The first characters to be considered for genus descriptions were the lengths and serrations of some long setae on the dorsal shield, the shape of the ventral shields and the epistoma, and the presence of macrosetae on legs. To base the classification on homologies, Sellnick (1944) and Garman (1948) were the first to propose a nomenclature for dorsal setae. Nesbitt (1951) used Garman s nomenclature and reported in the genera descriptions the number of setae present on dorsal and ventral shields, as well as the serration of some dorsal and ventral setae, the shape of ventral shields, the length of peritrematal plate, and the number of teeth on chelicera. Based on an idea of Bernhard (Bernhard, 1955 in Chant, 1957), Chant (1957) was the first to use the dorsal anterior (notocephalic) chaetotaxy to delineate supra-specific taxa (sub-genera). At the same time, Athias-Henriot (1957) developed the idea that the sub-family Phytoseiinae showed a moderate hypotrichy ( dépilation accusée ) in comparison with other Gamasida and like Chant (1957) she thought that the idiosoma chaetotactic patterns could allow a clearer and a more useful delimitation of taxonomic entities but at suprageneric levels (Athias-Henriot, 1958). Moreover, she suggested a new chaetotactic nomenclature system (Athias-Henriot, 1957, 1958) that merged Garman s (1948) denomination of the vertical series of setae (D, M, L) and Sellnick s (1944) proposal for division of the dorsal scutum into horizontal ranks. Hirschmann (1957) suggested a similar nomenclature system, and named the longitudinal rows i-i, z-z, s-s, and r-r (for dorso-central, medio-lateral, lateral, and interscutal membrane setae, respectively); the lower-case letters correspond to the setae on the podosoma, and the upper-case letters correspond to those on the

PHYLOGENETIC GENUS DEFINITION 255 Table 1. Genera descriptions by various historical acarologists Berlese, 1915: 143 Genus Amblyseius new genus Parvuli, ovati, convexi, nitidissimi; dorso scuto unico latiori protecto. Adsunt setae dorsuales utrinque sex, caeteris multo longiores et dimidium corporis latitudinem aequantes vel superantes, sive: humeralis; lateralis (post quartos pedes); postica, in angulo margini lateralis et postici, haec caeteris longior. Seta quoque adanalis, ventralis est, ad angulos posticos ad latera scuti analis a quo scuto sat remota, caeteris curtior. Setae longiores sunt una in quoque articulo pedum posticorum genu, tibia, praetarso. Caetera ut gen. Ameroseius. Species typica: Zercon obtusus K. Small, oval, convex, clear; dorsal scutum undivided and lateraly protected. Dorsal setae present in both sexes, some of them very long, reach beyond the half of body or exceed it: humerus setae and lateral setae (after fourth leg); at the back of the body in the postero-lateral margin other long setae. Also the adanal setae (may be JV 5) is situated in the ventral part, in the posterior lateral corner and distant to the ventrianal shield, other (setae) shorter (are present). Long setae are present one in each article, genu, tibia, pretarsus of the posterior leg. Others (setae) as in the genua Ameroseius. Type species: Zercon obtusus K. Nesbitt, 1951: 52 53 Genus Kampimodromus Nesbitt, 1951 The dorsum is very slightly if at all imbricate and bears 7 to 9 pairs of distinctly pectinate lateral setae and 5 or 6 pairs of median setae which may or may not show slight pectinations; D 1 and VL 1 are pectinate. The epistome is decidedly rounded. The chelicerae are denticulate with 3 to 5 teeth on the movable digit and an equal number on the fixed. The peritremal plate is small yet extends posteral of coxa IV. The sternal scutum bears three pairs of setae, and the ventrianal, which is equal in width to the genital (has) three pairs of preanal. The fourth pair of legs may not bear conspicuous sensory setae. Type species: Typhlodromus elongatus Oudemans, 1930 Chant, 1959a: 48 49 Genus Typhlodromus Scheuten, 1857 Dorsal shield with 13 to 20 pairs of setae, six to 12 in the lateral rows. If some setae on dorsal shield long, not flattened, club-shaped, greatly thickened, or serrated, but slender and smooth. Setae S 1 and S 2, if present, on interscutal membrane. Ventrianal shield of various shapes, with one to four pairs of preanal setae. Type species by monotypy: Typhlodromus pyri Scheuten, 1857 Key to subgenera of Typhlodromus 1. Four pairs of anterior lateral setae on dorsal shield; never more than nine pairs of lateral setae......amblyseius Berlese 2. More than four pairs of anterior lateral setae on dorsal shield; eight to 12 pairs of lateral setae......typhlodromus Scheuten Wainstein, 1962: 12 Genus Amblyseius Berlese, 1904 Usually the number of D is 6, seldom 5, only in one case 7; AL-3; ML-1, only in one case 0; PL-1 3; AM-2; PM-2; S-2; V-2; MV-2, and only in one case 0. Dorsal setae are slender, lisse, or more or less serrated. Ventral plate usually has three pairs of setae, rarely 1 2. Type species: Zercon obtusus Koch 1839. More than 100 species are known. More than 90 of them are related to the sub genus Amblyseius s. str., the remaining are included in other six subgenera (1 3 species in each) Muma & Denmark, 1968: 232 Genus Chelaseius new genus Setal characters are four pairs of dorsal setae; three pairs of median setae; eight pairs of lateral setae, some of which are elongate and weakly plumose; two pairs of sulateral setae; three pairs of sternal setae; three pairs of ventrianal setae; and three pairs of ventrolateral setae excluding caudal setae. Scutal characters are one dorsal scutum, sternal scutum wider than long, ventrianal scutum pentagonal and wider than genital scutum, and primary metapodal scutum normal in size and elongate in outline. Peritremal and stigmatal scuta indistinguishably fused and with a reinforced secondary pore; peritreme extended forward between vertical setae. Spermathecal cervix saccular with an undifferentiated to nodular atrium. Spermatodactyl with foot terminal, heel obscure, and lateral process distinct. Chelicerae very large with no denticules on the movable finger and two to four denticules and basal pilus dentilis on the fixed finger. Legs ranked in size 1423, respectively, with leg I only slightly longer than others; leg I with one macroseta and an erect seta on the tarsus; leg II with one macroseta, leg III with two macrosetae; and leg IV with three macrosetae, that on genu longest. Type species: Amblyseiopsis floridanus Muma, 1955

256 H. TSOLAKIS ET AL. Table 1. Continued Athias-Henriot, 1977: 64 66 Genus Cydnodromus, Muma, 1961 Adults are recognised for the ennomic of dorsal chaetotaxy, for the presence of J2, for the isotrichy, for the dorsal polygonal reticulation, for the four setae on genu II, for the evolved condition of gv3, for the entire or slightly shorter peritreme, and for the shape of the insemination apparatus. The spermatodactyl of males is caudatum, and five setae are present of which V are hoplochorous. Dorsal shield of adults is meriadenic: pluri or monodeficient. Insemination apparatus: receptaculum not or weakly differentiated. Adductor channel at least the same length of the calyx, fin, cylindrical. Atrium oblong, straight, free, and separated from the calyx by a constriction. Accessus is straight, its length is almost that of the trivium or a little longer. Calyx sub-hemispheric or troncoconique, as long as large or 1.5 times longer than large. Situation mediopeltidial e opisthodorsal. Lateral setae long; Z 5 and often Z 4 serrated; gd1 equidistant from j1 and z3; id2 close to z4 than to z3, antiaxial to the segment z3-z4, dm1 more close to S 2 than to Z 1; idl3 closer to S 4 than S 2; idl4 almost equidistant from S 4 and S 5; gd4 posterior to s4 or marginal, closer to id4 than to the seta, gd2 between saiii and z4 or slightly posterior at this segment; gd8 between idx and Z 4, closer to the sensillum than to the seta; gd9 very close to S 5; idm2 antiaxial or anteroantiaxial to J 2. Sternogenital region (female): sternal shield almost as long as large; v3 hoplochorous; endopodal sclerite III simple with a tail; posterior margin of genital shield straight. Opistogastre (female): 8 setae V. Sclerite of the sigillum sgpa free. Subpentagonal sclerification; its lateral marginconvex at the level of perianal sigilla. V 1,V 2,V 3 situated in a triangle; preanal sigillum posteroparaxial to V 3; gv3 evoluted: it is located between preanal and V 3 closer to the first; solenostome situated transversally, crescentic, forms a small tectum. Gnatosoma and legs: Fixed digit with 2 7 teeth; mobile digit without teeth or with a maximum of 3 teeth. Genu II with 7 setae. Simple macroseta on the basitarsus IV. Evolution of gv3: in its primitive condition, in protoadenic gamasides, gv3 (very strong organ) is located posteroantiaxial to the peranal sigilla; its solenostome is small and round (simple). The more common evolution of the gland observed is a location towards the preanal sigillum. At this stage, the solenostome is usually simple. In a second step, gv3 is located between this sigillum and V 3 and it is transformed (gv3 evolved). Type species: Typhlodromus californicus McGregor, 1954 opisthosoma. Hirschmann named also the ventral setae and afterwards used this character to define some phytoseiid genera (Hirschmann, 1962). Muma (1961) modified Garman s system to define the genera, but he also considered other characters (the form and setation of the sternum, the number and position of the setae in the scapular region, and the development of elongate or otherwise modified setae on the genu, tibia, and tarsus of leg IV). Wainstein (1962) modified Athias-Henriot s chaetotactic nomenclature system and used the shape of the ventrianal shield to redefine some existing genera and subgenera and to describe some new ones. After considering the ideas of Sellnick (1944), Athias-Henriot (1957), and Hirschmann (1957, 1962) concerning chaetotactic systems, Lindquist & Evans (1965) proposed a chaetotactic nomenclature for dorsal setae in the Gamasina. Rowell, Chant & Hansell (1978) applied the same system to various genera of Phytoseiidae in order to establish the dorsal setal homologies, as had Hirschmann (1957) and Athias-Henriot (1957) previously. This system, which is the most commonly used at the time of writing, is simply the combination of ideas of all the above-mentioned authors. Until the early 1960s, generic descriptions were thus mainly based on dorsal chaetotaxy, the shape of ventral shields, and the presence of macrosetae. Oudemans (1928: 285) first noticed the insemination apparatus in a phytoseiid species:... under the dorsal shield, between coxa III and IV (the position can vary little), there are two pear shaped bladders; the very thin steel extends to the ventral side, between coxae III and its scutum parapodicum externum, on the height of the stigma. The bladders sometimes are big and thin-walled, and other times they are small and with thick walls. Therefore, I believe that we are talking about a gland. Subsequently, Oudemans (1930a) drew these structures in the description of Typhlodromus tiliarum Oudemans 1930, without mentioning them as a discriminant taxonomic character. In their redescription of Phytoseiulus macropilis (Banks, 1904), Smith & Summers (1949) also noticed a pair of coxal glands annulated tubes in female only, which is clearly a reference to the insemination apparatus, but they did not consider it as a diagnostic character. The first systematist to use the shape of the insemination apparatus (spermatheca) for taxonomic purposes was

PHYLOGENETIC GENUS DEFINITION 257 Dosse (1957) in his description of Amblyseius zwoelferi Dosse 1957. One year later, Dosse (1958) emphasized the structure s function, described and drew the structure for 18 species of the genera Typhlodromus, Phytoseiulus, and Phytoseius, and stated that it had utility for species delimitation. Chant (1959b: 18) also considered the shape of the insemination apparatus as an important character for species diagnosis within a same genus. Schuster & Smith (1960: 181) considered that the shape of the insemination apparatus, especially the calyx (cervix according to some American acarologists), was of... definite value in the identification of species having similar setal arrangements and lengths. There is also an indication that the shape of the spermathecae will best demonstrate the relation of groups of species. Then, Athias-Henriot (1960a, b) and De Leon (1962) also reported the shape of insemination apparatus as a useful taxonomic character for new species descriptions. A short time later, Athias-Henriot (1966) grouped the various supra-specific ensembles based not only on dorsal setal pattern but also on the shape of the insemination apparatus. In redescribing and renaming some genera, Muma & Denmark (1968) took into account the insemination apparatus as a feature of the genus together with other characters such as the dorsal and ventral chaetotaxy, the shape of ventral shields and metapodal plates [re-named after Athias- Henriot, (1971) as inguinal sigilla], the shape of spermatodactyl, the number of teeth on the movable digit, and the presence of macrosetae on all four legs; these authors, however, did not give any priority to any of these characters. Hansell (1970) pointed out the weight of each character for Phytoseiidae taxonomy, and Hansell & Chant (1971) then considered the classical taxonomic method as a self-correcting process through time. Beside chaeotaxy and shape of the insemination apparatus, genus descriptions by Athias-Henriot (1977) considered the position and the shape of some solenostomes, especially solenostome gv3 (on the ventrianal shield). From this bibliographic synthesis, it is clear that in the history of Phytoseiidae classification, taxonomists paid attention to the discriminant values of characters/states of characters but not to their evolutive importance. As a consequence we can assume that the characters considered could be homoplastic and the genera today defined could be not sustained by apomorphies. Some authors have considered that apormorphies supporting genera are the occurrence of setae bearing on the dorsal shield (Wainstein, 1962; Muma & Denmark, 1968; Chant & McMurtry, 2003a, b, 2007) whereas others thought that genera are defined by apomorphic states of characters related to the insemination apparatus (Athias-Henriot, 1977, 1978; Ragusa & Athias-Henriot, 1983; Ragusa Di Chiara & Tsolakis, 1994). But in reality, neither of these two views has been tested, and neither could be considered right or wrong. CURRENT SITUATION AND THE GENUS NEOSEIULUS HUGHES, 1948 The previously mentioned literature led to the formulation of two theories supporting supra-specific classification within the family Phytoseiidae. The genus Neoseiulus is useful for illustrating how these two theories differ and have been applied (Ragusa & Athias-Henriot, 1983; Chant & McMurtry, 2003a). Each theory has produced a different definition of this genus, and Table 2 details the characters used for each definition. The first definition considers the shape of the insemination apparatus as a key character (apomorphy), as proposed by Athias-Henriot (1977, 1978) in the revision of the genera Cydnodromus Muma 1971 and Dictydionotus Athias-Henriot 1979 (= Dictyonotus Athias-Henriot, 1978), and as subsequently proposed by Ragusa & Athias-Henriot (1983) in the revision of the genus Neoseiulus. In the second definition, which is currently the most commonly used and which was reported by Chant & Yoshida-Shaul (1989), the key characters (apomorphies) are the presence/absence of idiosomal setae. These latter authors report (l.c. 223): However, it is important for the reader to understand that in the present paper we are not proposing a classification system nor are we suggesting that the setal patterns reported in it represent taxa in any formal sense. This paper is intended simply to report all known dorsal setal patterns observed in the family. Nevertheless, these patterns were subsequently used in all taxonomic revisions carried out by Chant & McMurtry (1994, 2003a, b, 2004a, b, 2005a, b, c, 2006a, b, 2007). Chant & McMurtry (2003a: 3) considered 268 species in the genus Neoseiulus, which they placed in the tribe Neoseiulini. Beyond the dorsal and ventral setal pattern, generic diagnosis is also based on ratios of length/width of the shields, the length of some dorsal setae, the length of peritreme, and the presence of macrosetae on legs (Table 2). They also considered the shape of the insemination apparatus for distinguishing between species sub-groups (in the taxonomic keys provided). Ragusa & Athias-Henriot (1983), on the other hand, included only 17 species in the genus Neoseiulus. Their generic diagnosis is based essentially on the shape of the insemination apparatus, and then considers other associated characters, i.e. idiosomal chaetotaxy, dorsal adenotaxy, and position of solenostome gv3. All the above-mentioned studies underline the difficulty in identifying valid supra-specific groups (i.e.

258 H. TSOLAKIS ET AL. Table 2. Description of the genus Neoseiulus according to Ragusa & Athias-Henriot (1983) and to Chant & McMurtry (2003a) Ragusa & Athias-Henriot (1983) Insemination apparatus*: Adductor duct about as long as calyx, broad, soft. Accessus large, strongly indentated, thick walled, not separated from trivium by a diameter modification. Trivium prominent, globular to oviform, as wide as or slightly wider or narrower than calyx base, fused to this base but never projecting on calyx bottom. Calyx basically tubular, from 2 to 6 times longer than its average diameter. Related characters: Dorsal adenotaxy bideficient (gd5 and gd8 absent); gd2 weak (= metatactic). Isotrichy. On genu II, 7 hairs. Solenostome gd4 simple (female) or altered (male). Dorsal scutum smooth or anterolaterally with a few ridgelets, sometimes slightly ornamented between Z1-J2-Z4-S5 (ridgelets or grooves). Dorsal hairs short, simple, rather thin; s4 equal or sometimes longer than z4. Z4 longer than S2; gd1 posteroantiaxial to j3, equidistant from j3 and z3 or closer to the former. Segments j3-gd1-z3, ids-z4-z5, Z1-idl1-gd5 and s4-gd4-id4 nearly straight or obtuse angled; idl1 anterior and hardly paraxial to Z1 or anteroparaxial to this hair; z3-id2-z4, gd2-z4-id2, idm1-j2-idm2 not very far from right angle: id2 much closer to z4 than to z3; gd6 equidistant from Z1 and gd6 or much closer to the gland; idm1 antiaxial to J2 and anterior or posterior to the level of hair. Peritreme: Peritreme entire. Sternogenital shield: Sternogenital sclerocuticle smooth, with few lateral ridgelets on the sternal scutum. This shield almost as long as wide; its posterior margin weakly concave between the slightly prominent lobes on which v3 and v 3 are inserted. Front tip of endopodal IV cochleariform. Epigynium with hind margin shorter than width of opisthogastral shield, enlarged posterior to v5 where the margins are straight or somewhat convex; these are connected with hind margin in a sharp or slightly rounded angle. Large inguinal sigillum elongate, narrow, cuneiform, neither egg- nor regularly spindle-shaped. Genital sigilla of 4 th and 5 th pairs thin, linear, hardly individualized. Genital sigillum of 6 th pair (= sgpa) mostly located on opisthogastral shield, frequently projecting marginally, or les often tylochore. The hoplochorous condition of sgpa is more primitive than the tylochorous one found in numerous amblyseiine genera. Opisthogastral shield: Opisthogastral shield at least as long as wide, with transverse striae and lateral margin straight or slightly concave; paranal sigilla submarginal. The solenostome gv3 is either punctiform, removed from preanal sigillum and posteroparaxial to V3 (primitive condition) or crescentic, close to preanal sigillum and about paraxial to V3 (evoluted condition). Legs: On basitarsus IV, StIV usually, though not always, elongated. Chelicera: Chelicera paucidentated: movable jaw unidentated, sometimes unilaterally bidentate (vertition); fixed jaw with two to four teeth. Chant & McMurtry (2003a) Female idiosomal dorsal pattern: 10A:9B/JV-3, ± JV4, ± ZV3 Dorsum: Dorsal shield generally significantly longer than wide (L/W ratio 1.6 3.0:1.0) but with some species more rotund (L/W ratio 1.1:1.0). Ornamentation variable, usually reticulate, with or without waist at level of seta R1. Posterior margin rounded. Dorsal setae short to medium in length approximately subequal except that Z5, and sometimes Z4, are longer on some species; setiform, occasionally strongly serrated. Setae s4, Z4, and Z5 never markedly longer than other setae. Seta R1 inserted on dorsal shield of the females of few species. There are important setal ratios that distinguish most species in Neoseiulus from those in the tribe Amblyseiini: s4:z4 = 1.0 2.0:1.0; s4:z1 = 0.6 3.0:1.0, usually < 2.5; s4:j6 = < 3.0:1.0 (the species in the canadensis group are an exception). The ratios of the lengths of setae s4, Z4, and Z5 to the length of the dorsal shield are other important distinguishing features: length dorsal shield:z5 = 3.0 14.3:1.0; average = 7.0 8.0:1.0; :s4 = 5.0 30.0:1.0, average = 12.0 15.0:1.0; :Z4 = 4.0 19.0:1.0, average = 10.0 11.0:1.0; :j6 = 1.0 35.0:1.0, average = 1.0 3.0:1.0. There is an even cline in the length of dorsal setae from uniformly very short to uniformly quite long. There are no significant discontinuities in this cline, and hence lengths of dorsal setae cannot be used as the basis for separating the species in this genus into groups. Sternal shield: The L/W ratio of the sternal shield usually approximates 1:1, but on some species the shield is much longer than wide. Seta St3 usually inserted on sternal shield but on some species this seta is inserted on a separate platelet. Posterior margin straight or concave. Many species have the sternal shield reticulated or striated. Genital shield: Genital shield narrow or of medium width, truncate posteriorly, with genital setae, sometimes reticulated. Ventrianal shield: Female ventrianal shield variable in shape: on most species pentagonal and as wide or somewhat wider than genital shield (ratio 1.0 2.2:1.0), but on some species long and narrow, or reduced. Without conspicuous waist on most species, never vase shaped. Preanal setae JV1, JV2, and ZV2 arranged in broad triangular pattern, with seta JV2 well removed from anterior margin of shield. Setae JV4, ZV1, and ZV3 variable in occurrence, present in most species. Usually with, but sometimes apparently without, a pairs of pores. Of the species for which this feature has been described, 58% have the pores narrowly separated and 42% have the pores widely separated. Many species have the ventrianal shield reticulated or striated. Metapodal plates normal. Peritreme: Peritreme usually extending to level of seta j1, but shorter on some species. Peritremal shield narrow, fused anteriorly with dorsal shield. Spermatheca*: There is considerable variation in the morphology of the spermatheca in this genus. These variations are used as a basis for distinguishing the species subgroups in the species groups that make up the genus Neoseiulus. Chelicera: Cheliceral dentition variable. Most species have 3 4 teeth on the distal portion of the fixed digit but a few are multidentate, with from 8 10 teeth evenly distributed along the digit. Movable digit with 0 3 teeth. Legs: Leg I always and legs II III usually without macrosetae, at most with a slightly adapted seta on Ge III, more rarely on Ge II. Leg IV with 0 3 macrosetae, most commonly 1. *Insemination apparatus = Spermatheca. Solenostome = Pore. Inguinal sigillum = Metapodal plate.

PHYLOGENETIC GENUS DEFINITION 259 monophyletic groups) and the morphological innovations that support them, especially without phylogenetic analyses and consequently without knowledge of homoplasy rates, which may be high in the Phytoseiidae. Each of the two theories concerning supraspecific classification within the family Phytoseiidae adopts a monistic approach:... one perspective is given priority and used as framework for organizing the evidence (Knox, 1998: 4). So, from a strict view based only on classification, both theories have a valid structure. The question asked in the present paper is: which of the two theories better fits phylogenetic evolution? MATERIAL AND METHODS TAXA CONSIDERED The study includes 21 species of the sub-family Amblyseiinae and two outgroup species belonging to the sub-family Typhlodrominae, genus Typhlodromus: Typhlodromus (Typhlodromus) exhilaratus Ragusa 1977 and Typhlodromus (Typhlodromus) phialatus Athias-Henriot 1960. The source of the species, the number of DNA sequences analysed, and their GenBank accession numbers are given in Table 3. The species of Neoseiulus considered were included in four different species sub-groups by Chant & McMurtry (2003a): (1) Neoseiulus cucumeris (Oudemans, 1930b), Neoseiulus fallacis (Garman, 1948), Neoseiulus californicus (McGregor 1954), Neoseiulus picanus (Ragusa, 2000), Neoseiulus reductus (Wainstein, 1962), and Neoseiulus longilaterus (Athias-Henriot, 1957) were included in the cucumeris species group and cucumeris species sub-group; (2) Neoseiulus tunus (De Leon 1967) was included in the species group cucumeris and species sub-group tunus; (3) Neoseiulus barkeri Hughes 1948 and Neoseiulus agrestis (Karg 1960) were included in the barkeri species group and barkeri species sub-group; and (4) Neoseiulus womersleyi (Schicha 1975) and Neoseiulus alpinus (Schweizer 1922) were included in the barkeri species group and womersleyi species sub-group. According to Ragusa & Athias-Henriot (1983), N. barkeri and N. agrestis belong to the genus Neoseiulus, while N. californicus, N. fallacis, and N. picanus belong to the genus Cydnodromus, as reviewed by Athias-Henriot (1977) (Table 4). Although they do not belong to the genus Neoseiulus according to both Chant & McMurtry (2003a) and Ragusa & Athias-Henriot (1983), Amblyseius andersoni (Chant, 1957), Amblyseius swirskii Athias-Henriot 1962, and T. montdorensis were included in the current study because the shape of their insemination apparatus is similar to that of species included in the genus Cydnodromus [as defined by Athias-Henriot, (1977)]. Finally, to test the monophyly of the genus Neoseiulus as defined by Chant & McMurtry (2003a) or by Ragusa & Athias-Henriot (1983) and of the genus Cydnodromus as defined by Athias-Henriot (1977), we included other genera of the sub-family Amblyseiinae such as Kampimodromus, Euseius Wainstein 1962, Iphiseiodes De Leon 1966, and Iphiseius Berlese 1916. MORPHOLOGICAL FEATURES To determine the morphological characters that could support the molecular phylogeny, we considered the morphology of the above-mentioned species belonging to the genus Neoseiulus as defined by Chant & McMurtry (2003a). Morphological observations were carried out on various specimens of each species (from eight to 40). Specimens were obtained from the same series of specimens (same population, same locality, and same date of collection) that were DNA-extracted. Apart from commercial breedings, the majority of the species were collected in the field and reared in the laboratory on plexiglas arenas (Swirski, Amitai & Dorzia, 1970) on Carpobrotus edulis and/or Oxalis corniculata pollen and/or on various stages of Tetranychus urticae Koch. Some of the collected mites was conserved in alcohol 100% for genetic studies whereas the remainder were cleared in Nesbitt liquid and mounted on slides in Hoyer medium for identification and morphological observations. Comparative analyses were also carried out for each species, using material from different localities or countries, present in our or in other collections; also type material, when avalaible (Table 3), was taken into account. When dealing with such small mites, it was impossible to extract DNA from part of a single specimen and then use the remainder as a voucher. However, the mite carcasses were retrieved on the Qiagen column after DNA extraction according to the protocol described in Tixier et al. (2010) and then mounted on slides. Even if this technique allows for correct identification, the mites were sometimes not in very good condition for drawings and accurate observations. For this reason we used the mounted specimens from the same locality for morphological analyses and observations. The internal transcribed spacer (ITS) DNA sequence used for N. reductus was deposited at GenBank (GU966582) by Pham & Linden (unpublished data). The specimens of this latter species used in the current study for morphological analyses were kindly provided to us by A. van der Linden, so that we could work on the same specimens as those that had been subjected to DNA analysis. Drawings were made using a differential interference contrast microscope. For dorsal and ventral chaetotaxy, the nomenclature for Phytoseiidae commonly

260 H. TSOLAKIS ET AL. Table 3. Sources of the species of genera considered for morphological observations and molecular experiments and accession numbers in GenBank database Species Origin of specimens considered for genetic analyses Host plant Accession numbers of ITS sequences Sequences deposited by: Accession numbers of 12S sequences Sequences deposited by: Origin of specimens considered for morphological analyses Taxonomists N. cucumeris Rearing units (SyngentaBioline) AY121985 Jeyaprakash & Hoy (2002) N. fallacis GenBank source Y18271 Navajas et al. (1999) N. californicus Rearing units (Koppert) and rearing units in Authors labs N. picanus Chile and rearing units in authors labs N. longilaterus Israel, rearing units in authors labs HQ404847 Rearing units (SyngentaBioline), Italy, Israel AY099364 Jeyaprakash & Hoy (2002) Identified by authors Slides from USA Identified by J. A. McMurtry and checked by the authors* HQ404802 HQ404836 USA, Chile, Italy, France Identified by authors HQ404803 HQ404837 HQ404804 HQ404838 HQ404805 HQ404839 HQ404840 HQ404806 HQ404841 HQ404807 HQ404842 HQ404808 HQ404843 Citrus sp. HQ404809 HQ404844 Chile (Holotype, Paratypes) and HQ404810 HQ404845 rearing units HQ404811 HQ404846 Unknown host plant HQ404827 HQ404828 N. reductus GenBank source GU966582 Pham & Van der Linden unpublished N. tunus Brazil Unknown host plant N. alpinus Genbank source FJ515685 Pham et al. unpublished N. barkeri GenBank source FJ515686 Pham et al. unpublished N. barkeri Tunisia and rearing units in authors labs Unknown host plant N. womersleyi Japan Unknown host plant N. agrestis France Unknown host plant A. andersoni France and rearing units in authors labs A. swirskii Rearing units (Koppert), Israel Unknown host plant HQ404826 Israel, rearing units in authors labs Identified by authors Identified by authors Slides from the Netherlands Identified by A. van der Linden and checked by the authors* HQ404812 HQ404848 Brazil Identified by authors Slides from France Identified by authors* France, Italy HQ404813 HQ404849 Rearing Units in authors labs Identified by authors HQ404820 Japan, Australia (Holotype and paratypes) Identified by authors HQ404814 HQ404850 France Identified by authors Apple tree HQ404822 HQ404857 Canada (Holotype), rearing units HQ404823 HQ404858 in authors labs Identified by authors HQ404821 HQ404856 Rearing units (Koppert), Israel Identified by authors T. montdorensis Rearing units (SyngentaBioline) HQ404824 HQ404859 Rearing units Identified by authors HQ404825 HQ404860 E. ovalis Rearing units (SyngentaBioline) HQ404801 HQ404835 I. degenerans Italy and rearing units in authors labs K. aberrans France Celtis australis K. corylosus France Corylus avellana Ph. longipes Argentina and rearing units in authors labs Ph. persimilis France and rearing units in authors labs Citrus spp. HQ404800 HQ404834 Italy, rearing units in authors labs Identified by authors HQ404798 HQ404832 France, Italy Identified by authors HQ404799 HQ404833 Identified by authors Solanaceae HQ404815 HQ404851 Argentina, Chile, rearing units HQ404816 HQ404852 in authors labs Unknown host plant A. urquharti Guadeloupe Unknown host plant T. phialatus France Viburnum tinus T. exhilaratus France Vitis vinifera HQ404817 HQ404853 France, Italy, rearing units in HQ404818 HQ404854 authors labs Identified by authors Identified by authors HQ404819 HQ404855 Identified by authors HQ404829 HQ404861 Identified by authors HQ404830 HQ404831 Identified by authors *Identification is referred to material considered only for morphological observations. Where none is specified, sequences were deposited in GenBank by the present authors.

PHYLOGENETIC GENUS DEFINITION 261 Table 4. Species grouped to supraspecific taxa according to the two different concepts of the genus Neoseiulus Sensu Athias-Henriot (1977) and Ragusa & Athias-Henriot (1983) Habitat Sensu Chant & McMurtry (2003a) Genus Neoseiulus Genus Neoseiulus Species group: barkeri Species group: cucumeris Neoseiulus barkeri Hughes (1948) on germinating barley Species sub-group: cucumeris Species group: marinellus Neoseiulus cucumeris Neoseiulus agrestis Karg (1960) in soil Neoseiulus reductus Genus Cydnodromus Neoseiulus fallacis Cydnodromus fallacis Garman (1948) on apple Neoseiulus californicus Cydnodromus californicus McGregor (1954) on Citrus (lemon) Neoseiulus picanus Cydnodromus picanus Ragusa (2000) on Citrus (orange) Neoseiulus longilaterus According to the Ragusa & Athias-Henriot concept, the following species do not belong to the above-mentioned genera: Species sub-group: tunus Neoseiulus tunus Species group: barkeri Typhlodromus cucumeris Oudemans (1930) on Cucumis melo (melon) Species sub-group: barkeri Amblyseius reductus Wainstein (1962) on blackberry Neoseiulus barkeri Typhlodromus longilaterus Athias-Henriot (1957) on Cynodon dactylon Neoseiulus agrestis (Bermuda grass) Typhlodromips tunus De Leon (1967) on Psidium guajava (guava) Species sub-group: womersleyi Amblyseius alpinus Schweizer (1922) on moss Neoseiulus alpinus Amblyseius womersleyi Schicha (1975) on strawberry Neoseiulus womersleyi attributed to Rowell et al. (1978) was used. For the insemination apparatus, the terminology followed that of Athias-Henriot (1975) with some additions and changes. MOLECULAR ANALYSES Total genomic DNA was individually extracted from several females per species with a Qiagen DNeasy tissue kit (Qiagen, Hilden, Germany), according to the DNA extraction protocol Purification of Total DNA from Animal Blood or Cells (Spin-Column Protocol) adapted for extracting total DNA from mites (Kanouh et al., 2010). Two markers, one mitochondrial and one nuclear DNA fragment, were used: 12S rrna and ITS1-ITS2-5.8S (ITSS). The primers used to amplify the 12S rrna and ITSS DNA fragments were: 5 - TACTATGTTACGACTTAT-3 and 3 -AAACTAGGAT TAGATACCC-5 (Jeyaprakash & Hoy, 2002), and 5 -AGAGGAAGTAAAAGTCGTAACAAG-3 and 3 -AT ATGCTTAAATTCAGGGGG-5 (Navajas et al., 1999), respectively. PCR was performed in a total volume of 25 ml containing 2 ml of mite DNA, 2.5 ml (1mM) of buffer 10, 1mL (1.5 mm) of MgCl 2, 0.5 ml (0.05 mm for each) of dntp, 0.175 ml (0.7 mm) of each primer, 0.125 ml (0.625 U) of Taq Qiagen, and 18.525 ml of water. Thermal cycling conditions for the 12S rrna marker were: 95 C for 1 min; followed by 35 cycles of 94 C for 30 s, 40 C for 30 s, and 72 C for 1 min; and an additional 5 min at 72 C. Thermal cycling conditions for the ITS marker were: 92 C for 1 min; followed by 30 cycles of 92 C for 15 s, 50 C for 45 s, and 72 C for 1 min; and an additional 7 min at 72 C. Electrophoresis was carried out on a 1.5% agarose gel in 0.5 TBE buffer for 30 min at 100 V. PCR products were sequenced with the Dynamic ET Terminator Cycle Sequencing Kit (Amersham). DNA was purified with Exosap-IT (Amersham). All DNA fragments were sequenced along both strands with a MegaBACE 1000 apparatus. Sequences were analysed and aligned with Geneious v. 3.5.4 (Drummond et al., 2007). The new sequences generated in this study were deposited in GenBank (see Table 3 for accession numbers). Data were analysed using both parsimony and Bayesian approaches. Each approach has its drawbacks: parsimony analysis is sensitive to long-branch attraction (Alfaro, Zoller & Lutzoni, 2003; Simmons, Pickett & Miya, 2004), while Bayesian methods may produce bias in clade support due to the standard use of flat topological priors (Pickett & Randle, 2005) and may often overestimate confidence levels for especially short nodes (Suzuki, Glazko & Nei, 2002; Alfaro et al., 2003). As these sensitivities are not (fully) overlapping, our confidence in the results was increased when the results of the two methods were congruent. PARSIMONY ANALYSIS Separate analyses were conducted for each data set (two loci) using parsimony in PAUP*, v. 4.0b.10

262 H. TSOLAKIS ET AL. (Swofford, 2002). Incongruence Length Divergence (ILD) (Mickevich & Farris, 1981) was calculated to examine possible conflicting signals among the two components of the data set. Although several sequences were missing for some specimens (only the ITS data set was complete), we have analysed the combined data set following Wiens (1998), who demonstrated, by performing multiple analyses of simulated data sets, that the addition of an incomplete data set to a complete data set is more likely to increase than decrease the phylogenetic accuracy. To reduce misleading effects of homoplasious characters, a posteriori reweighting was applied according to the rescaled consistency index (RC) after each tree search, until the number of trees was stabilized (Farris, 1969, 1989). A heuristic search procedure repeated 100 times was applied, with randomized taxon additions and branch-swapping algorithm (TBR). Tree number was limited to 10 000 for reasons of computation speed. The tree topologies obtained with and without reweighted characters were similar, only the bootstrap values and thus the relevant phylogenetic support of the nodes being changed. The Wilcoxon signed rank test (Templeton, 1983), as implemented in PAUP*, was conducted to test the possible paraphyly of the genus Neoseiulus under parsimony. This test was used to determine whether the difference between an optimal tree (resulting from an unconstrained analysis under likelihood) and a constrained tree (e.g. a tree in which the species of the genus Neoseiulus were constrained to be monophyletic) was significant. BAYESIAN INFERENCE We performed partitioned Bayesian analyses (Jordal & Hewitt, 2004; Nylander et al., 2004) on the combined data set. Under this approach, the computational efficiency of the Bayesian Markov chain Monte Carlo (MCMC) method allows the use of more realistic and complex evolutionary models for each data partition (Nylander et al., 2004). For each defined partition (one partition per gene sequenced), the bestfit substitution model was determined by Modeltest 3.06 (Posada & Crandall, 1998) through hierarchical likelihood-ratio tests (LRTs). The model of evolution GTR was selected by the LRTs with a proportion of invariable sites and a gamma distribution. The GTR model was implemented in MrBayes 3.1 (Ronquist & Huelsenbeck, 2003). The number of categories used to approximate the gamma distribution was set at 4, and four Markov chains were run for 1000 000 generations. Stabilization of model parameters (burn-in) occurred around 250 generations. The results are presented in the form of a 50% majority-rule consensus tree (in which trees corresponding to the burn-in period were discarded), and the support for the nodes of this tree was given by posterior probability estimates for each clade. To avoid the confusion resulting from the assignment of different generic names to the same species by different authors, the Results refer to the species name but not to the genus name. RESULTS AND DISCUSSION CHARACTER INTERACTION ILD values (Mickevich & Farris, 1981) were calculated to check the level of conflict between data sets. The interactions of two loci were examined (ILD = 0.07). This value suggests that the two data sets are compatible. PHYLOGENETIC ANALYSES After exclusion of uninformative characters and unalignable regions, the 12S rrna and ITS matrices retained, respectively, 244 and 258 informative characters, with a combined total of 502. The ITS analysis yielded 10 000 equally most-parsimonious trees [Length = 485, consistency index (CI) = 0.798, retention index (RI) = 0.859]. The 12S rrna analysis yielded 10 000 equally most-parsimonious trees (Length = 334, CI = 0.687, RI = 0.81). Results of the parsimony and Bayesian analyses on the combined data are shown in Figure 1. The parsimony analysis resulted in 6105 equally most-parsimonious trees (Length = 819, CI = 0.75, RI = 0.83). Phylogenetic analyses strongly support the polyphyly of the genus Neoseiulus and also of the species groups and species sub-groups of this genus, as they are defined by Chant & McMurtry (2003a). The Wilcoxon signed rank test (carried out on the combined data analysis) significantly rejected the monophyly of the genus Neoseiulus (P < 0.0001). The characters used by Chant & McMurtry (2003a) to define this genus and species sub-groups within the genus thus seem to be unreliable for defining monophyletic clades. The ten species considered within the genus Neoseiulus by Chant & McMurtry (2003a) are distributed in five clades (Figs 1 3): (i) A first monophyletic clade comprises barkeri, agrestis, womersleyi, and alpinus (Figs 1, 2). According to Chant & McMurtry (2003a), all these species belong to the species group barkeri but are included in two species sub-groups. According to Ragusa & Athias-Henriot (1983), both barkeri and agrestis are included in the genus Neoseiulus (sub-group barkeri), whereas womersleyi and alpinus are not because of the presence of a neck at the base of the calyx (Fig. 2). As a consequence, the genus Neoseiulus defined by Ragusa & Athias-Henriot (1983) seems to be

PHYLOGENETIC GENUS DEFINITION 263 Figure 1. Strict consensus trees obtained after Parsimony analysis (A) and Bayesian analysis (B) carried out on combined molecular data sets (ITS and 12S) including 11 species of the genus Neoseiulus defined by Chant & McMurtry (2003a), ten species belonging to seven other genera of the sub-family Amblyseiinae, and two outgroup species belonging to the sub-family Typhlodrominae: Typhlodromus (Typhlodromus) phialatus and T. (T.) exhilaratus. Values below branches indicate bootstrap support (A) and posterior probabilities (B). Roman numerals indicate the well-supported clades: I, genus Neoseiulus; II, genus Cydnodromus; III, genus A; IV, genus B; V, genus C. paraphyletic whereas the species group defined by Chant & McMurtry (2003a) seems to be monophyletic. However, we herein consider this monophyletic group to be a genus (we called it Neoseiulus) and not a species sub-group because of its position in the tree relative to the other genera and because of the high morphological similarity among the species belonging to this group (Figs 1, 2). The main visible morphological feature common to species of this group is the shape of the entire insemination apparatus: major (adductor) duct membranous, as long as the calyx; atrium differentiated, cylindrical or sub-conical, as wide as the insertion of calyx in some species or slightly larger in others, strongly indentated at the base; minor duct membranous, clearly visible, and large; and calyx tubular, flaring distally, with or without a neck at the base. However, these species also share the following other morphological similarities: smoothness or slight reticulation of the dorsal shield, isotrichy (= dorsal setae have similar length), ventrianal shield sub-pentagonal with few tranverse striae, genital sigillum of 6th pair (sgpa) located on the shield (hoplochorous) or less often on a sclerite (tylochorous), and a clearly visible solenostome gv3, between setae JV2 and near the preanal sigilla (Fig. 2). (ii) A second, well-supported clade includes the species californicus, fallacis, and picanus (Figs 1, 3). According to Chant & McMurtry (2003a), these species belong to the cucumeris species group and cucumeris species sub-group (Table 4), together with reductus, longilaterus, and cucumeris. The present phylogenetic analyses show that the species group cucumeris as well as the species sub-group cucumeris are polyphyletic. On the other hand, because of a similar shape of insemination apparatus and other