A New Cryptic Species Of The Chrysoperla Carnea Group .

382C. S. Henry et al.(2012 and 2013) by P. Duelli; and on the island of Samos,Greece (2013) by P. Duelli. Collecting data are shown inTable 1. All insects were segregated by locality and sex underlong-day photoperiod (17:7 h L:D) and supplied with waterand a Wheast -based diet (Hagen & Tassan, 1970). Becauserecently mated females will not sing, it was necessary to rearprogeny of gravid individuals captured in the field to determinespecies identity. Eggs were therefore collected and the larvaewere reared individually in 2 7 cm vials at 20–25 C onEphestia kuehniella (Zeller) moth eggs plus locally availableaphids, e.g. Brevicoryne brassicae (Linnaeus). In accordancewith the requirements of USDA APHIS-PPQ permit P526P07-06006, newly spun cocoons were shipped to Storrs, CT,USA, where songs were recorded from virgin adults 3–5 daysafter eclosion (see below). Song-verified individuals werepreserved as dried specimens or deep-frozen freshly killedindividuals, or in 95% ethanol for molecular systematic studies(Henry et al., 1999a). Several first-, second- and third-instarlarvae were culled from rearing stocks and preserved in 70%ethanol plus 5% glycerin for morphological study. Voucherspecimens were deposited in (i) the collection of C. S. Henry,Storrs; (ii) the Natural History Museum, London (BMNH); (iii)the collection of Peter Duelli, Zürich, Switzerland; (iv) the collection of Atsushi Mochizuki, Tsukuba City, Japan (NIAES);and (v) the W. F. Barr Museum, Moscow, Idaho (WFBM).Song recording, analysis and comparisonThe substrate-borne songs of 25 individuals of C. heidariisp.n., enumerated by sex in Table 1, were recorded at25 1 C and analysed using methods, software and equipment described in recent publications (e.g. in Henry et al.,2012). Those songs consisted of single volleys of abdominalvibration (Fig. 1), a general song type shared within thecarnea-group by C. plorabunda, C. adamsi , C. agilis Henryet al. and C. zastrowi (Esben-Petersen). In this type of song,the volley may be repeated many times at a regular period by asinging individual, but when a duet is established, the partnersreply to each other volley-for-volley in synchrony (marked byarrows in Fig. 2; also see fig. 3 in Henry et al., 2013). Thus,in species such as C. heidarii and C. plorabunda the volleyis also the ‘shortest repeated unit’ or SRU, whereas in otherspecies including C. carnea (Stephens) and C. pallida a clusterof multiple volleys functions as the SRU (Fig. 2). Also, therate of abdominal vibration – the dominant frequency – canand usually does change over the course of each volley.Consequently, there are five significant song features thatmust be measured in single-volley songs: volley ( SRU)duration, SRU period during a multi-volley solo or duet, andthe dominant frequency at the start, middle and end of thevolley (Fig. 1; Table 2). For C. heidarii , such measurementswere taken from at least ten songs ( volleys or SRUs) perindividual. Therefore, each mean shown in Table 2 is a grandmean. Coefficients of variation (CV s) were calculated toestimate trait variability among individuals, whereas two-tailedt-tests for independent samples (Snedecor & Cochran, 1980)were used to assess the magnitude of sex differences.The same five song features had been measured previouslyat 25 1 C in 133 individuals of C. plorabunda, collected overmany years from populations distributed across the knownTable 1. Collecting sites for living Chrysoperla heidarii , 2002–2013.LocalityAltitudeGPS coordinatesDate collectedSpecimens w/songDamghan / Shahroud, IRAN(colony, from pistachio field)Pandroso (Samos), GREECE(low branches of Pinus)Yeghegnadzor, ARMENIA(at building lights at night)Herher, ARMENIA(bushes, incl. Juniperus)Jermuk, ARMENIA(tall grass field)Sulema Valley, ARMENIA(bushes, incl. Juniperus)1250N36.17, E54.347.vii.20022 females646N37.73, E26.8310.vii.20131 male, 2 females1194N39.76, E45.3326–29.vi.20113 males, 1 female (plus 24)1496N39.76, E45.5423.vi.20112 males, 2 females (plus 10)2017N39.83, E45.6724.vi.20112 males, 1 female (plus 8)1774N39.92, E45.2322.vi.20113 males, 3 females (plus 16)1422N40.28, E44.702.vii.20119.vii.20111 male, 1 female (plus 5)1944N41.26, E43.5917.viii.20131 male (plus 11)1260N41.37, E43.268.vii.20121 male, 1 female (plus 13)449N41.62, E45.5415.viii.201319.viii.20131 male (plus 5)Kotayk, ARMENIA(bushes, mostly Corylus)Ninotsminda, GEORGIA (wildapple)Vardzia, GEORGIA(dry bushes)Sighnaghi, GEORGIA (riverineforest)Altitude is in metres, geographic coordinates are in decimal format, and each specimen with a recorded song is the offspring of a different gravidfemale collected at the specified locality. Also shown in the last column (‘plus’) is the number of additional lab-reared specimens examined for thepresence or absence of a ‘lucasina stripe’ on the second abdominal pleuron. 2014 The Royal Entomological Society, Systematic Entomology, 39, 380–393

Chrysoperla heidarii, a new Eurasian species with an American songAChrysoperla heidariidominant frequency:middleendstart140Frequency (Hz)38312010080604020rumblevolley period0duration0B246812Chrysoperla plorabundadominant frequency:140Frequency (Hz)10120startmiddleend100806040rumble20volley period0duration024681012Time (s)Fig. 1. Oscillograms (time–amplitude plot; lower of each pair of traces) and sonograms (time–frequency plot; upper traces) of (A) a typicalheterosexual duet in Eurasian Chrysoperla heidarii sp.n. and (B) a heterosexual duet in North American C. plorabunda, illustrating convergentsong phenotypes in the two allopatric species. Each duet is drawn to the same timescale of 12 s. In both cases, sexual partners duet by politelyexchanging single-volley SRUs (shortest repeated units). Start, middle, end and rumble sections of volleys are shown by the arrows, whereas volleyduration and period are indicated by dashed lines.range of the species in North America (Henry & Wells,1990). Those data allowed us to compare individual averagesof C. heidarii and C. plorabunda, using two-tailed t-tests toidentify statistically significant differences in species means(Table 2). A graphical comparison of volley period versusvolley duration in the two species is shown in Fig. 3. Allstatistical tests and analyses were performed using Statisticav10 (StatSoft, Inc. 2011).Song phenotypes of six other cryptic song species withwhich C. heidarii co-occurs and might be confused areillustrated in Fig. 2, and have been described and quantifiedin previous papers. These citations – and a summary of theirfindings – are available in a recent review paper (Henry et al.,2013).Behavioural testsIn order to test the responsiveness of C. heidarii andC. plorabunda individuals to one another’s recorded songs(SRUs), each insect was presented with its own (conspecific)song type and the heterospecific song type in a paired design(Wells & Henry, 1992a, 1994; Henry & Wells, 2009). Thecomplete experimental protocol is given in Henry et al. (2012).Responsiveness to C. plorabunda song playbacks was testedin three males and eight females of C. heidarii (total 52individual trials), whereas responsiveness to C. heidarii songplaybacks was determined in two males of C. plorabunda(ten trials). The responsiveness of C. heidarii to the duettingsongs of the six other Eurasian species shown in Fig. 2 was 2014 The Royal Entomological Society, Systematic Entomology, 39, 380–393

384C. S. Henry et al.OscillogramsSRUSpectrograms1401006020C. heidariiSRU1401006020cont.C. carnea1401006020C. mediterranea1401006020cont.C. pallida1401006020C. lucasinaSRUC. agilis1401006020C. z. sillemi1401006020 Hz024681012s024sFig. 2. Oscillograms and spectrograms of species-specific duetting songs of Chrysoperla heidarii sp.n. plus six other Eurasian species of thecarnea-group with which it has been collected in the field. All are drawn to the same time scale. The arrows mark the points at which the partnerin the duet inserts its SRU ( song); single-volley species reply after each of the partner’s volleys, whereas multi-volley species reply when thepartner’s shortest repeated unit (SRU) is finished. Spectrogram vertical axes show frequency in Hertz; cont. SRU is longer than shown, continuingto the right.Table 2. Values at 25 1 C of the song features of Chrysoperla heidarii (Eurasia) and C. plorabunda (pan-North America).Within-volley frequency, HzC. heidariiMales (n 13)C. heidariiFemales (n 12)C. heidariiAll (n 25)C. plorabundaAll (n 133)StartMiddleEndVolley duration, msVolley period, msNumber of volleys per SRU70.47 5.0669.22 8.64**69.87 6.89**86.38 5.2090.75 2.8589.68 5.11**90.24 4.04**58.53 3.58109.51 2.28107.36 5.61**108.48 4.27**31.09 3.08355.96 53.25401.53 62.02**377.83 61.00**629.91 70.001128.72 107.361073.93 191.47*1102.42 152.80*1169.26 126.771111Each value is the mean of the means of n individuals in the population subsample, SD (one standard deviation). Although the songs of the twospecies were measurably different (**P 0.0001, two-tailed t-tests), they were nearly the same in volley period, differing by only 5% (*p 0.021,two-tailed t-test). Sex differences within C. heidarii were not significant. SRU, shortest repeated unit exchanged between individuals while duetting.determined by playing back 5–10 songs of each species to fiveindividuals of C. heidarii .Adult morphologyThirty-seven specimens (21 males, 16 females) identifiedacoustically as C. heidarii were examined for external morphological features that might distinguish this species fromother members of the carnea-group. Localities were Herher(three ), Sulema Valley (four , one ), and Kotayk (16 ,ten ) in Armenia; the single locality in Greece was Pandrosoon the island of Samos (one , two ). GPS coordinates, elevations, vegetation and collecting dates are in Table 1. Adultswere examined for the states of 19 characters. These includedground colour of body; presence, extent and colour of markingson stipes, palps, gena, clypeus, frons, antenna and postoccipitalregion; relative abundance and distribution of black and blondsetae on pronotum; relative size of basal dilation of metathoracic tarsal claw expressed as a ratio (AB/BD, fig. 4 in Henryet al., 2002a); extent to which forewing is rounded or taperedat apex; forewing length and relative width expressed as theratio of length to breadth at widest point; presence or absenceof black markings on wing veins; length of costal setae; relative abundance and distribution of black and blond setae onthe abdominal sternites; shape and relative proportions of malegenital ‘lip’ and ‘chin’ at apex of sternite 8 9 (fig. 5 in Henryet al., 2002a); shape of tignum of male genitalia (figs 17–22 in 2014 The Royal Entomological Society, Systematic Entomology, 39, 380–393

Chrysoperla heidarii, a new Eurasian species with an American song385900Volley duration (ms)800700600 C. plorabundaMandible500Antenna400 C. heidariiLabial ial spot2200Volley period (ms)Dorso-lateral spotFig. 3. Plot of volley duration versus volley period of the songsof Chrysoperla plorabunda and C. heidarii sp.n., illustrating theirconvergent phenotypes. Each datum represents a single individual,coded by taxon. Oscillograms, drawn to the same time scale, comparethe typical duetting songs of the two species.Brooks, 1994); and presence or absence of dark brown stripe onpleural membrane of second abdominal segment. Because thepresence or absence of this stripe could not be discerned easily in dried material due to discolouration of the abdomen, weassessed this last feature in 92 additional freshly killed deepfrozen specimens, enumerated by collecting site in Table 1.Larval morphologyWe examined 35 larvae reared from adults taken at threesites in Armenia (Yeghegnadzor, Herher and Sulema Valley;see Table 1). These included 9 first-instar, 5 second-instarand 21 third-instar larvae. They were boiled for 1 min indistilled water with a drop of liquid detergent and transferredafter cooling to 70% ethanol with 5% glycerine. Specimenswere viewed at 50 . Head markings were compared tothose characterising specimens of other, previously describedPalearctic species. A selected, representative third-instarindividual is illustrated in Fig. 4.Life history and ecologyHabitat data and plant associations were recorded for eachspecimen of C. heidarii at the time of collection, as hadbeen done previously for C. plorabunda and other lacewingsin the carnea-group (Wells & Henry, 1998; Duelli et al.,2002). Using song analysis of progeny grown from fieldcollected gravid females, we also determined which otherspecies of the carnea-group were associated with C. heidariiat each collecting site. A small number of the offspringof field-collected C. heidarii were reared from third-instarlarvae until adult eclosion under short-day, cool conditions( 8:16 h L:D; 22:16 C warm:cool) to induce winter diapauseand concomitant colour changes (Duelli, 1992).Dorso-lateral stripedarkenedposterior areaFig. 4. Dorsal view of third-instar larval head capsule (left half) ofChrysoperla heidarii sp.n. from western Asia. The drawing representsthe typical condition seen in 21 individuals from several populationswithin the range of the species.Evolutionary relationshipsUsing techniques described previously (Takano et al., 2011),DNA was extracted and 4630 base-pairs (bp) of the mitochondrial genome were amplified, sequenced and aligned for 17species and two subspecies (85 OTUs) of the carnea-group,including three individuals of C. heidarii from the Lesser Caucasus of Asia and four of C. plorabunda from western NorthAmerica. Outgroup taxa included C. pudica (Navás) fromWolfdrift, Republic of South Africa; C. comanche (Banks)from southern California, USA; and C. rufilabris (Burmeister) from southeastern Arizona, USA. The dataset comprisedfour protein-coding mitochondrial genes: 1000 bp of NADHdehydrogenase subunit-2 (ND2 ), 1226 bp of cytochrome c oxidase subunit I (COI ), 685 bp of cytochrome c oxidase subunitII (COII ), and 1719 bp of NADH dehydrogenase subunit5 (ND5 ). Primers and PCR conditions were the same asspecified in table 4 in Henry et al. (2012). INDELs werenot present and alignment of all sequences for all OTUswas unambiguous. Mitochondrial gene sequences for thethree specimens of C. heidarii have been deposited in theDNA Data Bank of Japan (DDBJ) under accession numbersAB836669–AB836680. Accession numbers for the other 82OTUs were published previously (Henry et al., 2012).Phylogenetic analyses were performed in MEGA v5.2(Tamura et al., 2011), garli v2.0 (Zwickl, 2006) andMrBayes v3.1.2 (Ronquist & Huelsenbeck, 2003). Optimalitycriteria of maximum parsimony (MP, using MEGA), maximum likelihood (ML, using GARLI), and Bayesian inference (BI, using MrBayes) were applied to the mitochondrialsequences. Prior to ML and BI analyses, we partitioned by 2014 The Royal Entomological Society, Systematic Entomology, 39, 380–393

386C. S. Henry et al.Table 3. Optimal data partitions and best nucleotide substitutionmodels selected by PartitionFinder v1.1 (Lanfear et al., 2012) forthe mitochondrial DNA sequences used to analyse the evolutionaryrelationships of C. heidarii sp.n. within the carnea-group.SubsetBest model123GTR I GTR I HKY I 45GTR I GTR Subset partitions (mtDNA genesand codon positions)COII pos.3, COIND2 pos.2, ND5COII pos.2, COIND5 pos.2COII pos.1, COIND5 pos. 3pos.3, ND2 pos.1pos.1pos.2, ND2 pos.3,pos.1gene and by codon position, using the ‘Greedy algorithm’ ofPartitionFinder v1.1 (Lanfear et al., 2012) to detect optimalpartitions in the dataset and to determine the best nucleotidesubstitution model for each partition when all partitions werelinked (Table 3). Five partitions were identified and modelled(number of parameters 216; global Bayesian InformationCriterion 28 774.44). ML and BI analyses were run usingthe model appropriate for each partition, with all parametersunlinked across partitions. For BI analyses, we performed twoseparate MCMC runs with four chains each for 10 million generations, sampling every 1000 generations. Data from the tworuns were then combined. We confirmed with Tracer v1.5(Rambaut & Drummond, 2007) that all likelihoods converged.Phylogenetic trees from MP, ML and BI analyses were constructed and viewed but are not shown, because they did notdiffer significantly from a recently published Bayesian phylogram (fig. 8 in Henry et al., 2012).Evolutionary divergences among selected taxa and groups oftaxa were estimated as the number of base substitutions per siteaveraged over all sequence pairs between the taxa or groups.Methods of analysis included uncorrected p-distances anddistances calculated using the Jukes–Cantor, Tajima–Nei andTamura–Nei models as implemented in MEGA v5.2 (Tamuraet al., 2011).Heterosexual duetting in C. heidarii consisted of eachindividual repeatedly answering single volleys of its partnerwith the same kind of volley. The result of that interactionwas a long series of repeated volleys, wherein the responses(SRUs) of the partners alternated with one another (Fig. 1A).During these duets, songs of males and females were notstatistically distinguishable (Table 2), and did not overlapor interfere with each other’s volleys (Fig. 1A). As in otherlacewings of the carnea-group (Henry, 1979), the duet endedwith the male repeatedly tapping the genital area of the femalewith the tip of his abdomen just prior to copulation.The song of Asian C. heidarii closely resembled that ofNorth American C. plorabunda in its volley period (Fig. 3)and pattern of volley alternation during duets (Fig. 1A, B).For example, the mean volley period of C. heidarii (1.10 s)was only about 5% shorter than that of C. plorabunda (1.17 s;see Table 2), and duets in the two species looked and soundedremarkably alike (Fig. 1). However, their songs differedfrom one another in every other measurement; for example,whereas the dominant frequency within each C. heidariivolley was upwardly modulated, it was downwardly modulated in C. plorabunda (Table 2; Fig. 1). Also, an averagevolley in C. heidarii was only two-thirds as long as that inC. plorabunda (Table 2; Fig. 3).Chrysoperla heidarii was collected with as many assix other co-occurring species of the carnea-group, whichincluded C. carnea, C. mediterranea (Hölzel), C. pallida,C. lucasina (Lacroix), C. agilis and C. zastrowi sillemi (EsbenPetersen). Any similarity in song phenotype among thesesympatric species could make species recognition difficult forinsects and entomologists alike. However, the short, singlevolley SRU that comprised the duetting song of C. heidariiwas totally unlike the long multi-volley SRUs of C. carnea,C. mediterranea, C. pallida and C. lucasina, and it was onlya small fraction of the duration of the single-volley SRUscharacteristic of C. agilis and C. z. sillemi (Fig. 2).Behavioural testsResultsSong phenotypeAdult males and females of C. heidarii produced a singletype of simple song, which was used both in solitarycalling and duetting. The shortest repeated unit (SRU) wasmonosyllabic but iterative, consisting of a single volley about378 ms long repeated on average every 1.10 s (Fig. 1A;Table 2). During the course of the first half of the volley,the dominant frequency increased from an average minimumof 70 Hz to a maximum of about 108 Hz (25 C); the lasthalf of the volley was a toneless low-frequency rumble.Coefficients of variation among individuals within populationsor between collecting localities in western Asia and Greecewere 14–16% for temporal features and 4–10% for frequencyfeatures (Table 2).Living individuals of C. heidarii and C. plorabundaresponded to the other specie

A related process – parallel speciation – lends powerful support to any hypothesis of ecological speciation, because it demonstrates that a particular selective environment can repeat-edly generate the same phenotype during speciation. In parallel speciation, a single species invading several new places gives