Pimachrysa (Neuroptera: Chrysopidae: Nothochrysinae): Larval Description and Support for Generic Relationships

We describe the previously unknown larval characteristics of Pimachrysa (second and third instars), and we provide new comparative data on the Nothochrysa californica Banks larvae that were described earlier. The Pimachrysa larvae (identified as Pimachrysa fuscaAdams) express the numerous features that characterize the chrysopid subfamily Nothochrysinae.They resemble Hypochrysa larvae in many respects, but several differences support retaining the two genera as separate. It now appears that the reputedly more archaic, small bodied, nothochrysine genera (e.g., Dictyochrysa, Hypochrysa, and Pimachrysa) have naked larvae that differ distinctly from the larvae of Nothochrysa species, all of which are debris-carriers. In addition, the pattern of variability in Nothochrysa larval features supports the proposal that this genus has a closer relationship with the other chrysopid subfamilies than do the small-bodied nothochrysine genera. The findings are consistent with the hypothesis that larval nakedness (as opposed to debris-carrying) was the ancestral life style for the extant chrysopid clade.


Introduction
Considerable effort has focused on elucidating the evolutionary history of the neuropteran family Chrysopidae, and yet significant challenges remain. The subfamily Nothochrysinae is of particular interest, largely because of its position at or very near the base of the chrysopid phylogenetic tree and also because of its substantial and ancient fossil record. Members of this subfamily have primitive patterns of wing venation and unique features in their larval morphology that have been used to explore evolutionary trends within the subfamily [1][2][3][4][5][6]. However, generic relationships among the extant taxa remain ambiguous. On-going molecular studies [4,7,8] thus far have yielded some interesting, but contradictory results concerning the origin of the subfamily, and they have not yet resolved the relationships among its constituent genera.
Ideally, the above issues would be met with systematic comparisons among all included taxa using a broad range of approaches (adult morphology, larval morphology, comparative biology, and molecular sequencing). In reality, a large barrier prevents such an analysis, that is, the rarity of specimens from relevant taxa. For example, except for a few species, specimens of Nothochrysinae and other basal chrysopid taxa are seldom encountered in the field; as a result the range of taxonomic coverage has been relatively narrow.
To continue efforts in overcoming this barrier, we describe the hitherto unknown larva (Semaphoront B) of the nothochrysine genus, Pimachrysa. Along with Nothochrysa, Dictyochrysa, and Hypochrysa, whose larvae were described earlier (Box 1), Pimachrysa is now the fourth genus of Nothochrysinae for which larvae are known. Also, we provide new data on the larva of Nothochrysa californica Banks that document a broader range of larval variation than previously reported for the genus. Finally, we discuss the pattern of variation among known nothochrysine larvae in relation to Psyche the current understanding of generic relationships and the phylogeny of the subfamily.

Larval Material and Procedures.
Our study is based on four specimens of Pimachrysa fusca Adams collected in late February and late April at Lower Covington Flat, Joshua Tree National Park (previously a national monument), Riverside Co., CA. Three of the specimens (two third instars and the single second instar) were used for morphological study, and one third instar was reared to confirm the specific identity of our larval specimens. The three specimens were treated in KAAD solution [9] for 24 hours and then transferred to 95% ethyl alcohol. In the laboratory the specimens were photographed, and the external gross features were described. Subsequently, one specimen from each instar was cleared in KOH and transferred to glycerin for examination of fine structures and setation. The second instar died soon after it was collected. We noted a globule of apparently sticky material on the labrum and base of the left mandible; presumably the material was exudate from aphids that we supplied the larva while we were in the field. We were able to remove the exudate from the preserved specimen without damage.
All voucher specimens are now in the Tauber Research Collection (Lot 2015: 01); a leg from one of the third instars was given to Shaun Winterton, California Department of Food and Agriculture, Sacramento, for DNA analysis.
To rear one of the third instars to the adult stage, we held it in a glass vial provided with frozen eggs of Ephestia kuehniella (Zeller), the Mediterranean flour moth. It was maintained at room temperature on a long-day light : dark cycle (∼LD 16 : 8).

Terminology and Comparisons.
The chrysopid larval stage typically includes three instars, the first of which (L1, Semaphoront A) differs markedly from the latter two (L2, L3), which, in turn, usually resemble each other (Semaphoront B). The main differences between second and third instars are in size, markings, and the number of thoracic and abdominal setae. In this paper, both instars of Semaphoront B are considered.
With minor exceptions, morphological terminology and chaetotaxy followed the usage of Rousset [10], Tsukaguchi [11], Tauber et al. [12], and Monserrat and Díaz-Aranda [13]. In one case there is a discrepancy. Some abdominal segments of Nothochrysa have a pair of laterodorsal tubercles (LDTs); in papers by Tauber and coauthors, the term tubercle refers to a protuberance bearing more than one seta. In Nothochrysa californica Banks, each LDT bears one long, filiform seta and one shorter seta. Thus, the tubercles of N. californica are consistent with those of other chrysopids. However, third instars of the two European species of Nothochrysa also were reported to have laterodorsal "tubercles" on abdominal segments A5 to A7; but in these species, the "tubercles" were reported to have only one, relatively long, seta [13]. If such a situation pertains (and no small seta is found on these "tubercles"), the homology of the structures should be examined and the usage of the term "tubercle" clarified. For convenience, in this paper, we refer to the structures reported for the European species as laterodorsal tubercles (LDTs).

Larval
Behavior. The specimens did not carry debris and they gave the overall impression of a Chrysoperla larva, but their bodies appeared thicker and more rounded (Figures 1(a)-1(c)). They remained motionless during much of the time we observed them; occasionally they moved quickly. The third instars, in the laboratory, fed readily on eggs of E. kuehniella; one specimen spun a delicate, oval shaped cocoon on the side of its vial 5-6 days after captivity; small pieces of debris were incorporated into the outer layers of the cocoon. The larva (within the cocoon) remained in diapause for 4 to 5 months, pupated ∼155 days after spinning, and emerged as an adult 15 days later. The pharate adult exited the cocoon via a circular opening that it cut at the head end of the cocoon.

Generic Diagnosis.
Our P. fusca larvae have the three diagnostic features that typify nothochrysine larvae. (i) The flagellum is very short relative to the pedicel (∼1/14th the length of the pedicel) (Figure 1(d)). (ii) The terminus of the flagellum has a group of short apical setae. (iii) More than three lateral sensilla are present on the terminal segment of the labial palps (L2 = 5, L3 = 4-5) (Figure 1(e)).
Among the Nothochrysinae with known larvae, Pimachrysa (P. fusca) closely resembles Hypochrysa (two species) (Box 2). That is, the larvae (second and third instars) from both genera have elongate, naked bodies; the thorax and abdomen lack lateral tubercles; the abdomen has no laterodorsal tubercles. In addition, none of the setae are hooked, and the primary cephalic setae of the Pimachrysa larvae are slightly blunt and thus similar to those of Hypochrysa. Finally, the Pimachrysa head and body markings resemble those of H. elegans; that is, larvae of both species have dark brown, elongate, dorsal head markings, and dark brown, sublateral stripes on the thorax, and abdomen (most noticeable on the L3; see Figures 1(a), 1(c), 2(a), 2(b), and 6). However, a number of distinguishing features differentiate the larvae of the two genera. First, other than the cephalic setae, all other setae on the Pimachrysa larvae, especially the abdominal submedian setae (SMS), are short and acutely tipped, whereas on Hypochrysa they are short and blunt or clavate. Second, the jaws of Hypochrysa are longer than those of the Pimachrysa (Hypochrysa: 1.7-1.8x head length; Pimachrysa 1.4-1.6x head length). Third, the white-to-tan body coloration of the Pimachrysa larvae differs markedly from the green-to-bright green coloration of both H. elegans and H. africana. In our opinion, the above differences offer significant supporting evidence for the separate identity of the two genera.

Body (Figures 2(a)-2(c)
). Length ∼5.5 mm (measured in lateral view through spiracles), height ∼0.94 mm (thickest section of abdomen). Cream-colored-to-white with dark brown dorsolateral and lateral markings on each segment; lateral surface with brown dorsal band above lateral band of white and lower narrow band of dark brown; pronotal sclerites brown. All setae short, smooth, pale, of two types: "simple" = with acute tip (most setae) or "slightly blunt" = with slightly rounded tip (some primary cranial setae). Thoracic, abdominal setae often with swollen, lightly sclerotized setal bases (Figures 3(a) and 3(c)), these setal bases becoming increasingly larger and more sclerotized posteriorly.

Head (Figures 2(a)-2(c), 4(a), and 4(b)
). Dorsum white, with dark brown markings as in figures; frontal marking elongate, without lateral arms; mesal epicranial and/or postfrontal markings reduced to pair of posterior spots below frontal marking; lateral epicranial marking elongate, extending to lateral side of antennal base; eyes with stemmata clear, integument between stemmata, surrounding all stemmata dark brown to black; cranium roughly triangular, with posterior margin (dorsal) rounded, ventral margin elongate; width (across eyes) ∼0.79 mm, length (dorsum) ∼0.64 mm, depth (midregion to top of eye) ∼0.20 mm; base mostly exposed. Clypeolabral area elongated distally, extending over basal segment of labial palps; with large brown mark mesally. All dorsal primary cephalic setae present except S10 and S12; S11 long, S3 short and shifted mesally, others of medium length, most slightly blunt apically; Vx setae obscured by cervix, probably with three small setae (as drawn on Figure 4(a)) or two setae and pore; labrum with three pairs of setae (one mesally, two laterally); dorsum without secondary setae. Cervix full laterally, ventrally, slightly withdrawn from cranium dorsally; with row of 6-8 pairs of setae along anterior margin, dorsum to venter.
Metathorax (T3) consisting of three well-delineated subsegments separated by transverse depressions, each with pair of small setae laterally (probably S1Sc1, S1Sc2), but sclerites Sc1, Sc2 not identified. Sc3 medium-sized, brown, circular, with small mesal seta (S1Sc3). (Figures 2(a), 2(b), 2(c), and 6). Lateral tubercles (LTs), laterodorsal tubercles (LDTs) absent; spiracles (A1-A8) circular, sessile, with simple atrium. Each segment (A1-A7) roughly quadrangular, divided into three subsegments (or raised folds) separated by transverse depressions, anterior one extending to margin of segment; spiracle located laterally on dorsum of second subsegment; dorsum of all subsegments with acute submedian setae (SMS) on chalazae of small to medium sizes, arranged in transverse rows on each subsegment; pleura with patches of small, mostly acute, sometimes blunt setae, spines. A1: anterior subsegment short, dorsum with four pairs of SMS; second, third subsegments longer, similar in size to each other, each with row of ∼ten to twelve pairs of SMS, most stemming from relatively small chalazae. A2-A7: dorsum of each subsegment with ∼nine to twelve pairs of SMS; number and length of SMS, size (plumpness) of chalazae becoming progressively larger posteriorly. A7: first subsegment with four pairs of short SMS stemming from relatively large, plump chalazae; second and third subsegments each with five to seven pairs of SMS, most with plump chalazae. A8: segment divided into three subsegments; anterior subsegment narrow, well-delineated from subsequent subsegment; second, third subsegments not well-delineated from each other; dorsum of anterior subsegment with one pair of SMS, middle subsegment with two pairs of SMS; posterior subsegment with ∼seven pairs of SMS; pleural region of segment with two, unsclerotized, lateral swellings, each bearing ∼five medium length setae. A9: cylindrical, with two visible subsegments; anterior subsegment without setae anteriorly, with ∼three pairs of robust dorsal setae posteriorly, lateral-most pair on swelling; posterior subsegment with row of ∼seven pairs of medium length setae on robust chalazae posteriorly. A10: dorsum with ∼six pairs of setae laterally, ∼eight pairs of setae extending from posterior membrane on terminus. Venter with subsegmentation poorly demarcated; each segment (A1-A8) with two transverse bands of acute setae and spines (microsetae), pair of large, sublateral patches of relatively robust setae and spines (microsetae); setae often longer than those on dorsum, but without enlarged chalazae.

Abdomen
A2-A7: dorsum of each subsegment with ∼nine to fifteen pairs of SMS-one to three more than L2.
A8: dorsum of anterior subsegment with two pairs of SMS; pleural region with single pair of broad, unsclerotized, lateral swellings, each bearing ∼ten setae.
A10: basal portion of segment withdrawn and shorter than on L2.
Venter similar to L2.

Size Differences between Second and Third Instars.
Among our specimens, the L2 ( = 1) and L3 ( = 2) express some unusual variation in the relative sizes of body parts. The L3s, as expected, have much larger bodies than the L2, and yet the head size is almost identical among our specimens of the two instars. Also, while the L3 mandibles appear somewhat larger than those of the L2, the antennae and labial palps do not. In addition, the L3 setal size is approximately the same as in the L2, but the degree of enlargement and sclerotization of the abdominal setal bases is less than on L2 (Figures 3(a) and 3(c) versus 3(b) and 3(d)). Thus, the L3 setae seem less prominent than those of the L2 (Figure 3). Given the small number of specimens, we are puzzled by the above variation in the relative sizes of larval body parts. First, we do not know whether our observations represent the normal situation for this species. Second, if it does, we suggest that the variation could be related to some unusual pattern of annual or seasonal variation in the availability of food or perhaps seasonal or generational differences in physical factors. (Figures 16(a) and 16(b)). White, oblong longitudinally, round transversely, with basal side spun directly against substrate; attached base with single layer of thin, transparent silk; upper surface with two layers of silk: inner layer: dense, relatively smooth and outer layer: sparse, mixed with debris.

Comparative Information on
Nothochrysa californica Banks Previously, the three instars and cocoon of N. californica were described in detail [16]. However, full comparisons with other described nothochrysine immatures are hampered because the descriptions lack adequate illustrations and several larval characters were not included. Here, we provide images and drawings of the N. californica second instar (Figures 9-13) and the following notes on all instars for use in comparisons. All observations and measurements below are based on one to two specimens per instar:

Unique Larval Characters for Nothochrysinae
Three larval features have been proposed as unique, and perhaps apomorphic, for the subfamily Nothochrysinae [5,13,18]. Given the recent data from Hypochrysa and now Pimachrysa, as well as our reexamination of the N. californica larvae, we were led to reevaluate the diagnostic and phylogenetic value of these characters. In addition, we propose an addition larval character that may be unique to the Nothochrysinae. For our larval comparisons, we used our own measured values and estimated others from published drawings (Dictyochrysa: Figure 3 by New [14]; N. capitata, N. fulviceps: Figures 46 and 47

Length of Flagellum.
The nothochrysine antenna is characterized as having a terminal segment (flagellum) that typically is short and stubby, that is, approximately 10 or more times shorter than the pedicel; in larvae of Apochrysinae and Chrysopinae, the flagellum is considerably longer-only about 3-6 times shorter than the pedicel. For Hypochrysa and Pimachrysa second and third instars, as well as N. fulviceps and N. californica third instars, this character is consistent (their flagella are ∼12-20 times shorter than the pedicel). Also, for the Dictyochrysa first instar, it is close at ∼9 times shorter than the pedicel. However, there are interesting examples of interspecific and developmental variation in Nothochrysa. For example, the third instar of N. fulviceps, with a flagellum that is 12.5 times shorter than the pedicel, is consistent with the character, and the third instars of both N. capitata and N. californica (with flagella 8.3-9.1 times shorter than the pedicel) are near the 10x designation (Box 2). However, the flagella of the first and second instars of N. californica are noticeably longer, with the first instar being the most deviant (L3: flagellum 9.4-9.5 times shorter, L2: 5-7 times shorter, L1: only 4 times shorter than the pedicel; see Figure 14). Whether this pattern of developmental variation applies to the first instars of the European Nothochrysa species is unknown. However, if the old maxim "ontogeny recapitulates phylogeny" holds for this genus, the pattern could have phylogenetic importance.

Terminal Flagellar Setae.
The nothochrysine flagellum is typified as bearing several, very small terminal setae; in contrast, the tip of the flagellum in Chrysopinae and Apochrysinae bears a single long seta with a robust base. Upon reevaluation, this trait holds for most of the known nothochrysine larvae-Dictyochrysa (L1, one species), Hypochrysa (L1, L2, and L3, one species each), Pimachrysa (L2, L3, one species, Figure 14(a)), and Nothochrysa (L3, two European species). However, the structure differs in N. californica; in this species the flagellum terminates with a single elongate seta having a broad base (Figure 14(b)). The length of the terminal seta varies from 0.18 to 0.25 times the length of the flagellum. Again, the deviation from the typical nothochrysine pattern is most pronounced in the N. californica first instar. Unfortunately, the situation with the first and second instars of the European species is unknown.

Number of Sensilla on Terminal Segment of Labial Palp.
Thus far, this character appears to be a consistent means of distinguishing nothochrysine larvae (second and third instars, Semaphoront B; there are no sensilla on the palps of the first instar). A number of four or more sensilla is typical of nothochrysine larvae. Three or fewer sensilla are present on larvae of Apochrysinae and Chrysopinae (note: this character is difficult to examine; it requires careful placement and often manipulation of a cleared specimen with labial palps in ventrolateral position and under sufficient lighting.) 6.4. Cocoon Structure. As described here, the Pimachrysa cocoon is sessile and strongly oblong; the basal membrane of the cocoon is single, thin, and highly transparent, whereas the upper layer is double, thick, and opaque. Among the Nothochrysinae, this cocoon structure is shared with Nothochrysa (Toschi 1965: N. californica); however, the situation is not clear for Hypochrysa. Hypochrysa africana shares the oblong ("barrel") shape of Pimachrysa and Nothochrysa (Duelli et al. 2010, as Kimochrysa), but it is unknown if its cocoon is sessile or entire. In any case, the Pimachrysa and Nothochrysa cocoons appear unlike the entire, more spherical cocoons spun by larvae of many Chrysopinae (Chrysopini, Leucochrysini) that we have observed.
6.5. Synapomorphies? Whether the above characters constitute apomorphies or symplesiomorphies is unconfirmed, and available evidence from the hemerobiids (the presumed sister group of Chrysopidae) is inconclusive. For example, among the hemerobiids, (1) the length of the flagellum exhibits intergeneric variability: in Boriomyia fidelis (Banks), the flagellum is short relative to the pedicel [19]; in Hemerobius species, it is elongate [20]; in Micromus species, it is indistinguishable from the pedicel [20]. (2) The length of the terminal flagellar seta also shows variability among hemerobiid genera: in B. fidelis and Drepanacra binocula (Newman) the flagellum is elongate [19,21]; in Micromus it is short and accompanied by a very small seta [20]; in Hemerobius it is also short, but without a small seta [20].  From the above, it is obvious that the choice of an out-group is of crucial importance for the interpretation of an apomorphic versus plesiomorphic state for the various larval characters. Unfortunately, there has been relatively little comparative systematic work that compares hemerobiid and chrysopid larvae (see [22]).

Nothochrysine Phylogeny
When Adams [1] revised the subfamily Nothochrysinae, it comprised seven extant genera and four fossil genera attributed to the Miocene. Later, two extant and eight fossil genera were added, and one extant genus was synonymized [3,6,23]. So, now the subfamily includes eight extant genera and twelve extinct fossil genera from the early Eocene (Ypresian) to the Miocene (Aquitanian, Burdigalian?).
Although Adams [1] included Nothochrysa within the subfamily because it retained many archaic and/or unusual features that are not expressed in the Chrysopinae, he pointed out that it has other advanced features that are more typical of Chrysopinae and Apochrysinae than Nothochrysinae (e.g., in its development of the pseudomedia and pseudocubitus, its wing folds). And recently, an extensive study of all known fossil chrysopids identified patterns of evolutionary progression in certain venational features among the nothochrysine fossils that support a distinct position for Nothochrysa [3]. But, the relationships among the extant genera remain unresolved.
Based on his morphological studies, Brooks [2] proposed that the extant genera of the subfamily fall into two groups. In his scheme Nothochrysa was included with Dictochrysa and Hypochrysa, separate from Pimachrysa and Kimochrysa.
These groupings subsequently were found to be weak; indeed Kimochrysa was synonymized with Hypochrysa [6]. Finally, in another study that used molecular and other data, Duelli et al. [4] proposed that the small bodied, pollen-feeding nothochrysine genera (e.g., Hypochrysa sen. lat. (including Kimochrysa), Pamochrysa, Pimachrysa, and Asthenochrysa) are very similar to each other. These authors concurred with the opinion that although Nothochrysa shows some affinities with these nothochrysine genera, they also resemble Apochrysinae and Chrysopinae in other respects. Given their findings, they conjectured that the small bodied, pollenfeeding nothochrysines represent the most ancient group of extant chrysopids. Moreover, they concluded that pollenfeeding is the likely ancestral condition for chrysopid adults. It appears from the descriptions of now four of the eight nothochrysine genera (Box 1) that larval characters support the conclusions of Adams, Archibald and Makarkin, and Duelli et al. [1,3,4]. The pattern of larval differences between