Species identification by calling song

Introduction.—The ease of identifying Anaxipha species by their songs contrasts with the difficulty of identifying many of the species by morphological means. In fact, when geographic locality and habitat are taken into account, the local species are usually easy to distinguish by ear. Careful morphological studies of adult males have made identifying dead specimens of all species possible but often only after manipulating the specimen and examining the results with a quality microscope. When identification by song requires the analysis of a recording, advances in electronics have led to digital cameras with video modes that will do the recording. When electronic analysis of the recorded song is needed, an online PC will do the job if combined with free software and instructions on how to use it (Capinera et al. 2004).

Necessary terminology.—The basic unit of Anaxipha calling songs is a pulse, the sound made when the scraper on the edge of one forewing engages the file on the underside of the other forewing during a portion of a wing movement cycle. These cycles occur at uniform rates for brief to prolonged periods, producing pulse trains. The pulse rate (PR) during a pulse train (PT) is an important character, because it is species specific (though temperature sensitive). At 25°C, the mean PRs of North American Anaxipha species range from 5 to 79 p/s (Fig. 4). Within a pulse train, each pulse has a pulse duration (PD) and is followed by a pulse interval (PI). The PD plus the PI is a pulse period (PP). The proportion or percent of the PP that is occupied by the PD is the pulse duty cycle (Pdc). See Fig. 5 for 3 s waveforms of calling songs that illustrate the full range of pulse rates and duty cycles.

The calling songs of Anaxipha are musical in quality because their pulses and pulse trains are nearly pure in frequency, with the carrier frequency (CF) of the song being characteristic of the species but, as with PR, temperature sensitive. At 25°C, the mean carrier frequencies of the 13 species range from 5.5 to 7.3 kHz (inset of Fig. 4).

The calling songs of species that produce pulse trains that usually last 4 s or longer are said to be continuous and if the PR is high enough to make the individual pulses hard to detect, the effect is termed a trill (Fig. 5B). If the PR is so slow that each pulse is easily distinguished, the song can be rendered as a continuing tink, tink, tink, tink and described as a slow tinkle (Fig. 5E). In between are fast tinkles (Fig. 5D) and slow-pulsed trills (Fig. 5C).

Fig. 7.

Trendlines for carrier frequency plotted as a function of pulse rate in North American Anaxipha. The bottom end of each line is the expected kHz of the lowest recorded PR for the species and the top end is the expected kHz for the highest recorded PR for the species. The y-axis span of each line (i.e., max kHz-min kHz) thus depends not only on the range of temperatures at which recordings were made but also on the rate of change of kHz with PR. The lines for calusa and imitator have the least spans (1.5 and 1.6 kHz) but are based on recordings made over a range of only 9.5 and 6.4° C. The continuities of the trendlines reflect important differences in the continuities of the calling songs (see key in Fig. 4). Lines for species of the exigua group end in triangles; for the delicatula group, in circles; for the litarena group, in tipped squares. Lines for species without sister species in the region end in a square (fultoni) or open symbols (calusa and imitator). In the key, species are listed in the left-to-right order of their trendlines. Formulas for trendlines are in Table 6. Only for delicatula do the data appear nonlinear (see Fig. 8).

Pulse rate.—At every temperature at which calling occurs, pulse rate (PR) is a valuable means of identifying Anaxipha species. Because the downward extrapolated PR vs°C trendlines tend to converge at 0 p/s (Fig. 4, Table 2), the differences in PR between species with similar rates will be greater at higher temperatures. Knowing the PR without knowing the temperature of the caller is practically useless, because, except for a large gap between trillers and tinklers, the active space in Fig. 4 is pretty well filled with lines. For example, a calling song with a PR of 45 p/s and no temperature data would only eliminate five of the 13 North American Anaxipha species. The pulse rates of the four species with tinkling songs that occur in the same geographical area can usually be distinguished by ear (with subjective adjustments for temperature), but accurate PR determinations at carefully measured temperatures are required in many cases for the species that trill and cannot be distinguished by whether the trill is continuous or intermittent.

Using pulse rates to identify calling songs that were recorded at temperatures other than the standard 25°C can be done by plotting them on a copy of Fig. 4 and connecting each point to 3°C, but it can be done quicker and more accurately by using a spreadsheet-based application ( SMTbl_PRcalculator (SMTbl_PRcalculator.xlsx)).

Those wishing to record and analyze Anaxipha songs and lack access to specialized audio equipment and software may find help in the section on Sound Production in the field guide by Capinera et al. (2004).

Pulse-train phrasing.—All species that interrupt their pulse trains too frequently to be considered to have continuous calling songs have rapid pulse rates (42 to 79 p/s at 25°C). For these six species, the phrasing of the pulse trains is an important aid to identification (Fig. 6A–F) but the six species fall into three pairs with the members of each pair sometimes difficult or impossible to identify by PT phrasing alone.

The A, B-1, and B-2 waveforms in Fig. 6 illustrate that delicatula and litarena have songs that are repetitions, at a nearly uniform rate, of pulse trains of nearly uniform lengths. In delicatula the PT rate is nearly always faster, the PT durations are usually shorter, and the PT intervals are relatively longer. This causes the pulse train duty cycle (PTdc) to be less than 50%. In litarena the PT rate is usually slower, the PT duration usually greater, and PTdc usually greater than 50%. In Florida the PT phrasing seems always to be useful in separating the two species by ear but this may not be the case in coastal North Carolina.

The 10 s song samples in Fig. 6E, F illustrate how rosamacula and fultoni may be distinguished by ear by their calling songs. In rosamacula nearly all intervals between pulse trains are momentary, whereas in fultoni few if any PT intervals would be so interpreted.

Even when samples are taken from the same locality and within a narrow range of temperatures, intraspecific and intra-individual variation in PT phrasing in these six species is so great that an attempt to quantify the phrasing and to statistically analyze the interspecific differences proved too complex to continue ( SM_PTpilotStudy (SM_PTPilotStudy.pdf)).

Carrier frequency.—In Anaxipha, as in most crickets, the CF of the calling song increases markedly with ambient temperature. Fig. 7 shows this effect but does so by using PR as a proxy for temperature. This is valid because PR is linearly related to temperature (Table 2), and useful because it allows the CF of a recorded song to be used for identification even when temperature was poorly measured or not measured at all. The six species in Fig. 8 are five members of the exigua group plus vernalis. All six have continuous calling songs that are cleanly separated based on kHz/PR data pairs alone. The four species with the slowest pulse rates are very similar in their dominant frequencies but are well separated by their pulse rates. The two species with the highest pulse rates have almost identical PR vs temperature relationships but at any PR have discrete ranges of carrier frequencies.

Fig. 8.

Carrier frequency vs pulse rate pairs as an aid to identifying six Anaxipha species that produce continuous calling songs — with no need that the temperature of the calling male be known! The data points are from DF recordings of the six species (n= 792). Nearly all are from the mid-Atlantic states but a few are from Georgia (details are in SM).

Air temperature measurements in the field may be a poor indicator of what an individual cricket, whether a signaler or the receiver of a signal, is experiencing. For example, species of the exigua group often bask in sunlight, especially late in the season when air temperatures may be low, with the result that an individual's body temperature may be significantly higher than ambient air. The fact that all six northeastern species can be distinguished by the relationship between PR and CF, as illustrated in Fig. 8, suggests the crickets themselves may use that relationship to distinguish each other. A possible case in point is that exigua and vernalis have overlapping seasons of calling and call at essentially the same PR (Fig. 4). They also have the greatest differences in their CF: 1.7 kHz at 25 °C (inset of Fig. 4). Of relevance here is that recent work by Mhatre & Robert (2013) has demonstrated that the tympanal ears of a tree cricket have previously unsuspected abilities to actively amplify and adaptively tune to conspecific carrier frequencies.

Fig. 9.

A–L. Representative femoral stripes of 12 North American Anaxipha species. (Images not at a common scale.) Similar view of calusa femur is in Fig. 3C. Photos by David Funk.

Key to species based primarily on calling song (for recorded songs and live males)

The following key will enable the identification of captive Anaxipha species by their song. Under field conditions, when the singer is not definitely known to be an Anaxipha, several other, unrelated, cricket species could mistakenly ‘key’ to an Anaxipha species. For example, in the mid-Atlantic region, the songs of several common Allonemobius species can be confused with some of the continuous-calling Anaxipha. The pulse rate and carrier frequency of Allonemobius allardi is very similar to that of Anaxipha tinnulacita. Similarly, Al. tinnulus could be confused with An. tinnula or tinnulenta, or Al. walkeri with An. thomasi. However, Allonemobius species nearly always call from ground level, while most Anaxipha call from just above to about 2m above ground level. Also, there are distinct differences in habitat preferences: most Allonemobius prefer low open grassland such as pasture or lawn while Anaxipha tend toward taller, mixed vegetation. Confusion is most likely in transitional areas such as field to woods.

Fig. 10.

Range and mean of two ovipositor-related characters of North American Anaxipha. Number of individuals measured is in parentheses.

Fig. 11.

Box plots of the number of teeth in the stridulatory files of North American Anaxipha. For each species, the vertical white line is the median value, the black box shows the 2^nd and 3^rd quartiles, and the horizontal black line shows the range of values. The number after each species name is the sample size. (Detailed data are in  SMTbl_StridFiles (SMTbl_StridFiles.xls).)

Species identification based primarily on morphology

Easy to observe characters.—The presence of a conspicuous dark stripe on the lateral face of the hind femur (Fig. 9A–F) identifies an Anaxipha individual as being a member of the exigua group. This stripe is easily seen in live or freshly killed individuals, but may be obscure in older, pinned material.

Most Anaxipha have rather conspicuous teeth on the tarsal claws (although a high quality stereomicroscope at 50× or higher may be required to observe them). In one species (tinnula) these teeth are absent and the surface of the claw is smooth or only slightly roughened Fig. 2C). In another species (scia) the teeth are present but small (Fig. 2B). With the exception of some tinnulacita individuals from mangrove habitat in southern Florida, all other North American Anaxipha have well developed teeth on the claws as in Fig. 2A.

On the six large spines on the distal half of the hind tibia there is, in most North American Anaxipha species, a fringe of long hairs arranged in a single line on the side nearest the tibia (Fig. 2D). This fringe is absent in fultoni, imitator and calusa. In some instances, especially older museum specimens of species that have the fringe, many of these hairs may have been broken off. However, there is almost always some remnant of them.

The length of the ovipositor varies among species and can be useful in distinguishing females (Fig. 2E shows method of measurement). Measuring the length of the ovipositor relative to the length of the hind femur is a good way to normalize for overall body size (Fig. 10).

The importance of macropters to identification is that in imitator and calusa only macropters are known and that in delicatula and fultoni they are sometimes numerous at light but rarely collected otherwise.

Fig. 12.

Male genitalia (dorsal view) of Anaxipha vernalis with pertinent structures labeled. Terminology follows that of Desutter (1987).

Fig. 13

A–L. Representative male genitalia of North American Anaxipha species (for calusa; see Fig 3A). All images at scale at lower left. A–F. exigua group. G–H. delicatula group. I–J. litarena group. K. fultoni. L. imitator. Fig. 3A shows genitalia of calusa. (Additional images are in  SMFig_Genitalia (SMFig_Genitalia.pdf).)(Preparations and photos by David Funk.)

Stridulatory file characters.—The number of teeth on the stridulatory file is an important character for distinguishing many Anaxipha species — perhaps the only way to reliably distinguish most members of the exigua group in the absence of calling song data. Enumeration requires removal of the right forewing, making at least a temporary slide mount and observation under a compound microscope at ≥100×. Fig. 11 illustrates the distribution of tooth counts among the North American species.

Male genitalia.—Characters of the male genitalia are useful foridentification to at least the level of species group (Figs 12, 13). They are best observed following dissection and clearing with KOH. For the purposes of illustration, we made temporary slide mounts and photographed under a compound microscope in dorsal aspect. However, in practice it is best to view specimens from several angles in liquid under a good stereomicroscope because some structures (especially the parameres) sometimes distort or rotate. It is a good practice to evert the genitalia of freshly killed males which often obviates the need for dissection. When dissection and clearing are necessary, removal is much easier (and less destructive) if the genitalia are already everted.

Key to species based primarily on morphology

Relationships among species

Mating and mating tests. —When DF first became interested in crickets and their calling songs he soon saw the remarkable courtship and mating that generally follows when a female cricket is attracted to a calling male (Alexander & Otte 1967). This led him to observe carefully what he saw and to start making notable photographs of the details. His initial studies were of the larger species in his area — the oecanthines Oecanthus spp. and Neoxabea bipunctata; the eneopterines Orocharis saltator and Hapithus agitator; and the nemobnnes Allonemobius fasciatus and allardi (results for oecanthines and Orocharis reported in Funk 1989). Later he did an exhaustive study, in microscopic detail, of the mating behavior and reproductive structures of the trigonidiine Phyllopalpus pulchellus (SM_PhyllopalpusMating). Thus when DF became aware that he had in abundance, in the area that he lived, the four inland Anaxipha species of the exigua group that Fulton (1956) could not distinguish morphologically, he thought he might learn something of their reproductive isolation by studying the details of their mating behavior and comparing the results of intra and interspecific pairings.

The pattern of Anaxipha mating as observed in the laboratory for six of the 13 North American species is that courtship is initiated following antennal contact between a male and a female. (It is presumed that such a female will usually have been attracted to the male's vicinity by his calling song.) Courting males sing and this song does not differ in pulse rate or dominant frequency from the male's calling song (although the phrasing may be different). If a female remains attentive, the male soon extrudes a small spermatophore. After a maturation period (generally 10 to 20 minutes) he encourages the female to mount him whereupon he inserts the sperm tube into her genital opening, leaving the ampulla hanging free. She then dismounts taking the spermatophore with her and the male then extrudes a second, much larger spermatophore (Fig. 2F). This pattern of producing an initial spermatophore with a very small ampulla, followed by a second spermatophore with a much larger ampulla, seems to be the norm for trigs generally. For the Hawaiian trig, Laupala pacifica, deCarvalho and Shaw (2005) have termed these two types micro- and macro-spermatophores. Anaxipha males have not been observed to produce a macrospermatophore without having first transferred a microspermatophore. In Phyllopalpus pulchellus the ampullae of microspermatophores contain no sperm, while those of macrospermatophores contain many (up to 35,000; SM_PhyllopalpusMating). Asimilar situation probably exists in Anaxipha. If a female remains attentive after receiving a macrospermatophore, the male will usually produce a second microspermatophore, followed by another macrospermatophore. DF has not observed more than two of each type transferred in a single bout in Anaxipha, but up to three rounds were observed in Phyllopalpus.

It is common in the mid-Atlantic region for as many as three exigua-group species to be found in very close proximity (i.e., within centimeters of each other). Although the rather large differences in calling song are presumed to minimize long-distance attraction between heterospecifics (first barrier to hybridization), individuals in these habitats are still likely to encounter one another if only by chance. Since antennal contact alone can result in the initiation of courtship of conspecifics, DF decided to test whether encounters by heterospecifics in the lab might result in courtship, spermatophore production and/or mating.

In mating tests among the four inland species of the exigua group, DF tried all possible pairings one or more times. In 8 of 16 attempted heterospecific pairings males sang following antennal contact, but in only two cases was a spermatophore produced. These two pairings went to the stage of microspermatophore transfer and involved males of thomasi and females of tinnulenta and tinnulacita. Unlike the other three species, thomasi is strongly restricted to pines. In our experience, the only other exigua-group species found in this habitat along with thomasi is exigua. Of the eight conspecific pairings attempted, six resulted in spermatophore production but only two were allowed to continue through spermatophore transfer ( SMTbl_MatingTests (SMTbl_MatingTests.xlsx)).

Relationships among species as revealed by morphology.—The 13 North American Anaxipha species recognized herein can be divided into six groups based on morphology, three of which are monotypic.

Anaxipha imitator and calusa are the only North American species with the basal segment of the hind tarsus distinctly longer than segment 2+3 combined, a condition shared with some (most?) other Neotropical species. The male genitalia are quite different from other North American species. Both lack the stout terminal setae present in other North American species (Fig. 13L); in other respects imitator and calusa show little resemblance to each other.

Like the remaining North American species, fultoni has a shortened first segment of the hind tarsus, but the spines of the hind tibiae lack the distinctive row of long, fine setae that characterize the remaining species (Fig. 2D). The shape and pattern of sclerotization of the epiphallus is also distinct (Fig. 13K).

The remaining three groups are distinguished primarily by differences in the male genitalia and coloration of the hind femur. Members of the exigua group, which includes exigua, thomasi, tinnulacita, tinnula, tinnulenta and scia, all have very similar male genitalia with distinct median lobes on the parameres that protrude caudally as far or further than the remainder of the paramere (Fig. 13A–F). Exigua-group species generally have a distinct dark stripe on the hind femur (Fig. 9A–F). This stripe usually has distinct margins (obscure in scia) and contrasts strongly with the background coloration (but may be faint in tinnula and tends to fade in older pinned specimens of all species). Males of exigua-group species can be distinguished by calling song and the number of teeth on the stridulatory file. Females of some species can be distinguished by ovipositor length and tarsal claw dentition, but no reliable characters have been found to distinguish those of exigua, tinnulacita and tinnulenta.

Fig. 14.

Maximum likelihood tree based on COI sequences. Members of the exigua group are color-coded to facilitate viewing. Monophyletic species clades as well as clades within the exigua group with branch lengths less than ∼0.25% have been collapsed. The vertical extent of each triangle is proportional to the number of individuals contained and the horizontal extent represents the greatest distance between any two members of that clade (see scale at bottom). Numbers on branches are bootstrap values. [See  SMFig_MaxLikliTree (SMFig_MaxLikeliTree.pdf) for complete tree and all data.]

The litarena group includes litarena and rosamacula. Male genitalia have a distinctive shape, with recessed median lobes on the parameres (Fig. 13I–J). Males can be distinguished by the number of teeth on the stridulatory file and minor differences in the genitalia. Females may be separated by differences in body coloration.

The delicatula group includes delicatula and vernalis. The male genitalia lack distinctive median lobes on the parameres but otherwise resemble those of the exigua group. Males of the two can be distinguished by the number of teeth on the stridulatory file (Fig. 11) and females, in most cases, by the relative length of the ovipositor (Fig. 10).

Identification and relationships among species based on DNA barcoding.—DNA barcoding involves the sequencing of short, standardized gene regions to aid in species identification and discovery. The term barcoding gene loosely describes a fragment of DNA that has low sequence divergence within species but high divergence among species, from which unknown samples can be placed accurately into species groups simply by calculating their pairwise genetic distances (Hebert et al. 2003).

A 658-bp region of the mitochondrial cytochrome c oxidase I (COI) is currently the most commonly used gene for barcoding animals and major efforts are underway to construct reference libraries of sequences for known species (Ratnasingham & Hebert 2007). Many studies have shown that intraspecific differences in COI sequences typically fall in the range of 0–2%, whereas interspecific comparisons are usually characterized by >5% divergence. The use of COI for accurate species identification depends on the existence of this “barcoding gap”. Some studies have shown more than 95% of species in test assemblages possess distinctive COI sequences (e.g., Hebert et al. 2003).

COI was sequenced for 132 of our specimens including representatives of all 13 of the North American Anaxipha plus two other North American trigs, Phyllopalpus pulchellus and Cyrtoxipha columbiana. Table 3 shows minimum and maximum intra- and minimum inter-specific Kimura-2-Parameter (K2P) distances. For 10 of the 15 species maximum intraspecific distances were less than 0.7% while minimum interspecific values were >4%, thus barcoding might enable accurate identification of these species. However, most members of the exigua group were not resolved by COI. For tinnula, only two haplotypes were found (differing by 0.15%), but minimum interspecific distance was only slightly higher (0.77%). For exigua, thomasi, tinnulacita and tinnulenta minimum interspecific distances were considerably lower than maximum intraspecific comparisons. In the case of tinnulacita and tinnulenta, individuals were found with identical haplotypes.

Table 3.

Summary of intra- and inter-specific K-2P distances calculated from COI sequences. (Complete data are in SMTbls_COIspecimens.)

Three types of trees were constructed using MEGA5 software: Neighbor-joining, Maximum Likelihood and Maximum Parsimony. In all trees, Phyllopalpus pulchellus, Cyrtoxipha Columbiana and Anaxipha imitator and A. calusa appeared strongly differentiated from each other and from the remaining Anaxipha species. Branchingtopology differed among methods but in each case confidence was low, as indicated by low bootstrap values. However, these trees reinforce the idea that Anaxipha as currently recognized is likely polyphyletic. For the remaining North American Anaxipha species the three methods rendered similar results. The Maximum Likelihood tree (based on the method of Tamura & Nei 1993) is illustrated in Fig. 14. Anaxipha fultoni is well differentiated from the remaining Anaxipha. The delicatula and litarena groups are well supported, as is the exigua group. Within the exigua group, scia appears to be well differentiated from the other exigua group species, and the 10 tinnula specimens form a coherent group, but COI relationships among the remaining exigua group specimens are not consistent with the species concepts presented in this paper. These include exigua, thomasi, tinnulacita and tinnulenta (all previously included under exigua). K2P distances among species in this group are quite small — in a few cases individuals we consider different species have identical COI haplotypes. Conversely, there is considerable diversity within our species. Thus, all four appear polyphyletic in this tree.

The inability to resolve closely related species using COI is not unique to the Anaxipha exigua group. A similar situation has been documented in other insect taxa (e.g., blow flies; Whitworth et al. 2007). One of us (DF) has discovered identical COI haplotypes in specimens belonging to different but closely related species of the nemobiine cricket genera Allonemobius (allardi and walkeri) and Eunemobius (carolinus and melodius) (unpublished; sequences public and available at BOLD). Shaw (2002) reported that mitochondrial markers gave a misleading picture of evolution in Hawaiian Laupala (Trigonidiinae) when compared to nuclear DNA and morphological evidence, and that some patterns of mtDNA variation were likely the result of persistent interspecific hybridization during the history of this group. Hurst and Jiggins (2005) suggested that cases such as these might be the result of endosymbiont-driven introgression. Endosymbionts such as Wolbachia are common in crickets generally (although they have yet to be documented in Anaxipha). As a rule, these microorganisms are transmitted vertically and thus are in linkage disequilibrium with maternally-inherited mtDNA in their hosts. New infections among populations (by occasional movement of hosts) or even between sibling species (via hybridization events) can result in indirect selection on mtDNA, leading to major changes in the patterns of mitochondrial diversity. The direction of these changes depends on the direct effects of endosymbionts (e.g., cytoplasmic incompatibility, male-killing).

Indirect selection on mtDNA resulting from endosymbionts is predicted to result in patterns similar to what we observe in Anaxipha, that is, COI barcoding effectively places individuals to genus and separates more distantly related congeners, but may fail to resolve very closely related sympatric species. Preliminary AFLP (amplified fragment length polymorphism) data taken by Tamra Mendelson (unpublished) using nuclear DNA from a subset of our specimens suggests that each of our exigua group species really is monophyletic. Any future attempts to assess phylogenetic relationships between closely related Anaxipha species and populations from molecular data should include nuclear markers.

Mechanics of sound production in Anaxipha

Introduction.—Terminology and information about cricket sound production that was essential to identifying Anaxipha species by their calling songs was introduced in an earlier section. These eight features of song production in Anaxipha were not needed there but are basic to the subject of this section.

(1) Unlike those cricket taxa whose sound mechanics have been most thoroughly studied, which usually call from the ground and hold their wings at an angle of about 45° to the body axis when calling, Anaxipha species usually call from foliage and hold their forewings approximately perpendicular to the body axis while calling (Fig. 2F and).

Fig. 15.

Waveforms in calling songs of the Anaxipha species with the highest and lowest pulse rates (both at about 25°C). Duration of each waveform is to the right of the species name. Horizontal bars with arrows beneath waveforms show the portion of each waveform that is expanded by the waveform below. A–C, delicatula (MLNS 121494), 76 p/s, 5.64 kHz. A, Three pulse periods B, One pulse. C, Twenty fundamental vibrations. D–E, tinnulenta (R09_0756DF), 5.4 p/s, 6.94 kHz. D.Three pulse periods. E, One pulse. The bar beneath this wave form is 2.9 ms, the duration of the waveform that would produce 20 sine waves, the equivalent of C. (Note that waveforms A and E are the same duration.)

(2) In most crickets, including Anaxipha, but with exceptions in Mogoplistinae and Gryllotalpidae (Masaki et al. 1987, Forrest 1987), the right forewing is nearly always above the left when the wings are at rest.

(3) Both male forewings generally have files, and when the files of the two wings differ, the file of the right wing (the wing on top at rest) is the better developed one and is the one that functions in sound production (Masaki et al. 1987).

(4) Each pulse in a cricket calling song corresponds to a wing closure — i.e., when the distal portions of the raised wings move mesad; wing opening is silent or nearly so (Pierce 1948 and all later authors).

(5) The teeth of cricket stridulatory files project toward the anal margin of the wing. Thus when the scraper is engaged with the file during wing closure, the scraper is moving “against the grain” of the teeth and the resistance to moving is greater than it would be if the file were engaged during wing opening (Walker & Carlysle 1975).

(6) Pulses in cricket calling songs are nearly pure tones — i.e., their waveforms are nearly sinusoidal, showing no evidence of strong overtones (Fig. 15) (Pasquinelly & Busnel 1954, Dumortier 1963, and all later authors).

(7) The instantaneous frequency during a calling song pulse may remain almost constant or change slowly (usually downward) during the main portion of the pulse (audiospectrograms in Leroy 1966 and in SINA 2014).

(8) During the main portion of the pulse, impacts of the scraper on the file teeth are synchronous with the fundamental oscillations of the song and hence with the principal vibrating areas of the wings (Kreidl & Regen 1905 according to Dumortier 1963, Pierce 1948, Walker 1962b, Dumortier 1963, Nocke 1971, Elliott & Koch 1985, Montealegre-Z et al. 2011).

Synchronization of forewing oscillations and file tooth impacts.—How insects as small as crickets can produce sounds that are nearly pure in frequency has long been of interest and talented researchers have studied the problem using increasingly sophisticated physical and electronic methods (e.g., Nocke 1971, Sismondo 1979, Koch et al. 1988, Bennet-Clark 1999, Bennet-Clark & Bailey 2002, Montealegre-Z et al. 2009, Mhatre et al. 2012, Robillard et al. 2013). Most of these authors relied on crickets that are easily reared or otherwise conveniently available for laboratory studies and few were concerned with the effects of temperatures on the frequencies produced. In the following account we briefly review the history of these studies and the background for our conclusion that CF control in Anaxipha is different from that of most crickets previously studied yet probably similar to CF control in most crickets.

Currently the most widely accepted means of CF control in crickets, and the best substantiated one, is that the resonances of certain membranes in the forewings control the toothstrike rate (which corresponds to CF) via some sort of escapement mechanism that releases one file tooth at a time. Early adapters of this idea were Pasquinelly & Busnel (1954), who proposed that during calling, the raised forewings of Oecanthus pellucens acted like the prongs of a tuning fork. Specifically, they conjectured that the resonant frequency of the forewings in that species controls the movement of the scraper during its engagement with the file and, that during each fundamental vibration of this tuning fork, the file and scraper move slightly apart, allowing a tooth to be released, and return, catching the next tooth. Decades later, using Gryllus campestris, Koch et al. (1988) explored this proposed mechanism in convincing detail, showing that as the forewings close during the production of a pulse, the motion momentarily stops and reverses as each successive file tooth is struck. In exploring the effect of temperature on stridulation in that species, they showed that between 20 and 30° C the wing closing speed, tooth impact rate, and CF were nearly constant, whereas the rates at which chirps and pulses were produced were strongly temperature dependent. Koch et al. (1988) concluded that their results fit a “clock escapement model” but not a “cog-rattle” model, which they described as a system in which the scraper impacts a resonant structure but is driven along the file unrestrained by an escapement mechanism. The tooth strike rate is free to vary with the temperature because “no mechanical feedback occurs between oscillator and wing movement; i.e., the wing motion is not controlled by the harp motion.” This would allow CF to vary substantially with temperature.

Walker (1962b) studied the effects of temperature on the songs of 19 species of crickets — including 10 species of Oecanthus but representing five subfamilies. In attempting to explain substantial changes in CF as a function of pulse rate in 18 of these 19 species ( SMTbl_CFTempCoefs (SMTbl_CFTempCoefs.xlsx)), he rejected the escapement model (as formulated by Pasquinelly & Busnel 1954) and proposed that the forewings act like the sounding board of a musical instrument in that they can be made to vibrate at a range of tooth impact frequencies (i.e., a cog-rattle model)

Sismondo (1979) was first to measure the resonances of the forewings of a cricket with a wide range of temperature-dependant CFs in its calling song. The species was Oecanthus nigricornis, and he used live males and ingenious techniques to make repeatable measurements of a remarkably diverse set of phenomena associated with CF determination. For example, he demonstrated the modes of vibration produced in each forewing by stroking its file with an artificial scraper, and he compared the effects on CF of gradual changes in ambient temperature with the effects of a sudden decrease of 12°C or more. Significantly, he showed that the forewings had free resonances that ranged between 1.0 and 6.9 kHz, that the usual CF of the O. nigricornis calling song changed from 3.0 to 4.6 kHz over a range of 15 to 32°C, and that between 3.8 and 4.6 kHz the resonances formed “a virtually continuous series.” His summary stated that “the concept of a resonator with continuously variable tuning is supported.”

More than 30 years later, using state-of-the-art technology, Mhatre et al. (2012) extensively studied the resonances of the wings of Oecanthus henryi. With laser Doppler vibrometry they measured the resonances of the five major vibrating areas of the left and right forewings of five males at 18, 22, and 27 °C and reported that they “vibrated maximally and most coherently in response to sound between 2.5 and 4.5 kHz at all temperatures.” When they studied the effect of aspect ratio on the resonances of plates that resembled the shape and structure of the wings of O. henryi, they found that changing the aspect ratio beyond the range of natural wings of O. henryi changed the deflection modes from their more efficient states in the real wings. Unlike Sismondo (1979), who concluded that different cells of the forewings became the dominant resonators at different temperatures, Mhatre et al. (2012) showed that wing cells did not resonate independently from one another but formed a single wing-shaped plate with different modes of vibration at different temperatures.

To summarize, the forewings of two species of Oecanthus have been studied and found to have resonances that could be driven by continuously varying tooth-strike frequency. Sismondo's techniques for determining resonances in O. nigricornis were so imprecise compared to those used to study O. henryi that it seems highly likely that the differences in results are entirely technique related, However, it should be noted that the wings of O. henryi are unusually elongate for the genus, a conclusion supported by the aspect ratios of the forewings (length/dorsal width) of six eastern U.S. species having a range of 2.07 (O. latipennis) to 2.80 (O. angustipennis) (Fulton 1915, Plates IV & V) and four Indian species having a range of 2.78 (O. bilineatus) to 3.20 (O. henryi) (Metrani & Balakrishnan 2005, Table 1).

Table 4.

Temperature coefficients for pulse rates and carrier frequencies between 20 and 30°C (based on detailed data in  SMTbl_CFTempCoefs (SMTbl_CFTempCoefs.xlsx)).

Currently then, there is convincing evidence that in certain grylline crickets, CF is determined by a clockwork-like escapement mechanism in which the resonant frequency of wing membranes (especially the harp) control the tooth-strike rate. This results in the CF being little affected by changes in temperature. However, all other crickets that have been carefully studied for temperature effects on CF have proved to be tree-cricket-like rather than field-cricket-like in this respect.

Grylline species, such as those of Giyllus and Teleogryllus, are among the most robust and sclerotized of crickets and are adapted to singing from the ground. On the other hand, Oecanthus species sing from vegetation and are among the most delicate of crickets. None of these three taxa seems especially likely to be representative of the prototypical system for producing the nearly pure frequencies characteristic of cricket calling songs. Cricket taxa that seem more likely to have species that use prototypical means to produce pure frequencies include Nemobiinae, Trigonidiinae and a group of closely related subfamilies currently lumped with the Eneopterinae (OSFO 2014).

Table 5.

Carrier frequencies in the calling songs of North American Anaxipha. (Plots of trendlines and all source data are in  SMTbls_Songs (SMTbl_Songs.xlsx).) (I.D.=insufficient data.)

Fig. 16.

SEM images of selected segments of the stridulatory files of four crickets from three subfamilies (from Walker & Carlysle 1975) 3. Anaxipha latipennis (Trigonidiinae). 11. Gryllus ovisopis (Gryllinae). 12. Anurogryllus arboreus (Gryllinae). 14. Amphiacusta sp. (Phalangopsinae). The (a) image for each species is from the central portion of the file; whereas (b) images are enlargements of a small number of the teeth. The A. latipennis file had 485 teeth; its (a) image shows about a quarter of the teeth. For A. arboreus (Gryllinae), the source paper has both an (a) image and a second close-up image.

These subfamilies make up the majority of the cricket fauna of the moist tropics and most sing with their wings in a near-perpendicular position. Song production in these neglected groups deserves careful study by modern methods. In our opinion, published data (top section of Table 4) suggest that all these groups have CFs that vary enough with temperature and pulse rate to question the hypothesis that the resonances of one or a very few wing cells control toothstrike rate. The data for Anaxipha (Table 5 and bottom section of Table 4) are more extensive than for any other “neglected group” and are consistent with earlier data from the literature (Table 4). In all cases the coefficients for CF seem large enough to reject the clock escapement theory of CF control. If the control of CF in Anaxipha spp. is by the rate of file tooth impacts (“cog-rattle” control of Koch et al. 1988), their files (and those of Oecanthus) might be expected to contrast with the files of Gryllus spp. This is because files used in escapement control must bring the scraper to a stop before it is allowed to move forward again. In their detailed study of wing movements during calling in G. campestris, Koch et al. (1988, Fig.3B) showed that at each file tooth impact the scraper comes to a complete stop and must “move a bit backwards to escape from locking position.” In contrast, files that function to control CF by tooth impact rate need to have teeth that allow the scraper to move continuously forward as it impacts successive teeth. Nearly 40 years ago, Walker & Carlysle (1975) sought correlations between the structure of cricket file teeth and the parameters of the songs produced. They studied the files of 24 species of crickets selected for their taxonomic and calling song diversity. They found no correlations, but now, looking at the 28 scanning electron micrographs of file teeth they published, we find that the files of only three species seem especially adapted to stopping the scraper at each tooth. The files of these three have deep pockets between sturdy teeth into which the scraper might fit. Should this occur, the forewings would need to temporarily reverse directions in order to continue closing. The file teeth of these three species share an additional noteworthy feature not evident among Walker & Carlysle's other SEM images — i.e., they have very thin, lateral extensions that, hypothetically, might function to dampen the shocks of the repeated sudden stops during wing closures (Fig. 16). The three species are two unidentified Amphiacusta spp (Phalangopsinae) and Gryllus ovisopis (Gryllinae). (Although G. ovisopis has no calling song, Gryllus that call have similar file teeth.)

Fig. 17.

These graphs show three morphological characters and three calling song characters plotted as functions of calling song pulse rate for 13 Anaxipha species. A. Number of teeth in the stridulatory file. B. File tooth density. C. File length. D. Pulse duration. E. Pulse duty cycle. F. Carrier frequency. Because the x-axis is the same for all graphs, the 13 species are always in the same left to right order and can be identified by the sequence of pulse rates and species as listed in the text boxes of B and C. The range of values for each of the six characters graphed is expressed by dividing the highest value by the lowest value. The results varied from a high of ×4.8 in D to a low of ×l.3 in F; each such multiplier is displayed at the intersection of the axes. (The multiplier for PR values is ×15.5!) Sources of data: Tables 2, 6, Fig. 7,  SMTbl_StridFiles (SMTbl_StridFiles.xls).

Table 6.

Pulse durations and pulse duty cycles for Fig. 17.

The 10 genera, from five subfamilies, with files that do not seem as well adapted to stopping wing closure at every tooth impact are these: (After each genus is the number of SEMs that Walker & Carlysle used to illustrate the files of that genus.) Anurogryllus (Gryllinae) (3), Allonemobius (3), Eunemobius (2), Hapithus (3), Anaxipha (2) (see also Fig. 16), Cyrtoxipha (2), Phyllopalpus (2), Orocharis (3), Oecanthus (4), and Neoxabea (1).

Wing movement cycles during calling.—In most crickets the wing movement cycle (WMC) for each pulse is a wing closure (C), with the scraper and file engaged, followed by a wing opening (O), with the scraper and file making scant if any contact. The forewings would be expected to move more slowly during C than during O because the resistance to movement is greater. This is probably always the case, but instances in which the PI is longer than the PD make it necessary to either suppose that the wings move more slowly during a near frictionless O than in a file-engaged C or that at least one hold (H), with no movement along the file, has been added. Among our 13 species, these five seem likely to have an H in their WMC: thomasi (approximate Pdc at 25°C: 35%), tinnulacita (28%), tinnula (22%), calusa (21%), and tinnulenta (20%). All but calusa are in the exigua group, which is distant from that species group in both morphology and genetics (Figs 3, 14).

If only one H is added per WMC the sequence becomes either C-H-O-or C-O-H-. Determining which is the case requires monitoring wing position at intervals sufficiently brief to detect the H. Pierce (1948, Fig. 93) used a 50 frames per second film camera to reveal that the WMC in Allonemobius allardi, has the sequence C-O-H-with the H being about as long as the C and O combined. Walker et al. (1970) used an ultra-high-speed camera to make films of calling crickets and katydids at a thousand or more frames per second (e.g., Walker & Dew 1972, Morris & Walker 1976). Allonemobius sparsalus, one of several crickets filmed, was selected as a subject because its song consists of triplets of pulses, with the pulses within the triplets and the triplets themselves produced at a regular rate ( http://entnemdept.ufl.edu/walker/buzz/536a.htm). When the film was analyzed using the same techniques as in the examples cited above, an H was revealed between the triplets while the wings were in the O position making the sequence for three pulses and a pause C-O-C-O-C-O-H-rather than O-C-O-C-O-C-H- ( SM_WMCsparsalsus (SM_WMCsparsalsus.pdf)). These two examples from nemobiines suggest that any Anaxipha with a Pdc low enough to have an H in its WMC would likely have a C-O-H- sequence. However, when DF made and slowed a video of calusa, we agreed that the sequence was O-C-H-

Fig. 18.

Log/log plots of pulse period, pulse duration and pulse interval against pulse rate at 25°C, for 13 species of Anaxipha. Trendlines: pulse duration = -0.541(PR) + 4.586 (r², 0.96); pulse interval = -0.136(PR) + 7.368 (r² = 0.99).

Fig. 19.

Sexual differences in length of hind femur and in dorsally measured body length in 13 Anaxipha spp. Species are arranged, left to right, in approximate^* descending order of body mass based on femoral lengths. Tabular data for this figure are in SMTbls_HFandBLmeasurements.xlsx.

^*The summed average femoral lengths of male and female thomasi and vernalis were 8.81 and 8.85mm, which seemed close enough to a tie to excuse moving thomasi next to its four closest relatives.

A recent study of eneopterine crickets of the tribe Lebinthini (Robillard et al. 2013) documented WMCs in Lehinthus n. sp. that were markedly more complex than any previously reported for crickets and were suggestive of some of the ones known in phaneropterine katydids (Walker & Dew 1972, Heller 1990). The song consisted of short trains of pulses produced within a WMC, with successive pulses within a WMC initially increasing in intensity. Each song consisted of 15 or more WMC of this pattern: O-(H-) C-H-C-H-…C-H- with each C-H- unit producing a pulse and the number of C-H- units gradually increasing from a few to >10. The (H-) occurred only in the first few WMCs of a song.

Evolution of pulse rates among North American species of Anaxipha.— The 13 species of North American Anaxipha provide an unusual opportunity to examine the relationship of calling song pulse rates to changes in the stridulatory files and in other features of the calling song. This is because the range of pulse rates is extraordinarily wide (5.1 to 79 p/s at 25°C) and the genetic relationships among the species are well established — through concordant molecular and morphological studies. The genetic relationships of certain species are extremely close, as in the case of the five species of the exigua group that cannot be distinguished by COI analysis or male genitalia; and in others extremely distant, as in the cases of calusa and imitator, which are likely to be placed in other genera in the future.

In Fig. 17, three features of the stridulatory file and three features of the calling song are plotted as functions of pulse rate at 25°C. Pulse rate is in the position of an independent variable for these graphs because of its demonstrated importance to cricket females in distinguishing the songs of conspecific males from those of other species. In Fig. 17A–C, number of file teeth, file length, and file tooth density have general trends that apply to all species groups over the full range of pulse rates. This is not unexpected because, all other things being equal, producing faster and slower pulse rates should be facilitated by changes in the file in the directions that the trends indicate. As faster pulse rates evolve, pulse duration may be constrained by the shorter pulse period, and shorter pulses are facilitated by shorter files. Conversely, when slower pulse rates evolve, longer pulses seem to be favored, as evidenced by Fig. 17D, which shows a strong relationship for all species between PR and PD. Longer pulses are facilitated by longer files and/or higher tooth densities, and they could be constrained by physical limitations to the length of the file and the density of teeth thereon. An advantage of longer pulses over shorter ones in the attraction of conspecific females might result from more easy detection by listening females, or from increased likelihood that they would interfere with females detecting pulses produced by rival males, or their being an honest signal of male vigor. Whatever the selective forces, the wide range of PD values (×4.8) and the tightness of the relation of PD to PR suggest that the optimal PD changes in step with the evolution of higher and lower pulse rates.

The measurements of PD made for Fig. 17D were used to calculate the pulse duty cycles for Fig. 17E: (Pdc = PD/PP) or (Pdc = PD/ (1/PR)). PD measurements from the beginning to end of a pulse's wave form include a short time at the end during which the wings continue to vibrate at their own rather than a driven frequency. Because duty cycle is defined as the active or powered portion of a cyclic event, this non-powered portion should be excluded in calculating Pdc. Zero-crossing analysis (Bennet-Clark & Bailey 2002) is a method of detecting the switch from powered to non-powered vibration of the forewings, but we opted not to use it in determining Pdc. Fig. 17E shows a steady increase in duty cycle from 0.20 to ca 0.70 where it plateaus. Because deducting the non-powered portion of the raw PD would have slightly reduced the Pdc values that this figure displays, we assume that the actual plateau values are slightly less than shown.

When CF@25°C is displayed as a function of PR@25°C (Fig. 17F) no pattern is evident other than that members of the same species group tend to have similar carrier frequencies. This is compatible with the low value of CF as a character by which to distinguish closely related Anaxipha species. The one case where CF does usefully distinguish species is between exigua and vernalis, which have nearly identical pulse rates, are sympatric, and overlap seasonally. A. exigua has an unusually high CF for its group and pulse rate, and vernalis is in a tie for the lowest CF of the 13 species. These observations suggest a fresh look at Fig. 17B, C. It is vernalis (the triangle at 45 p/s) that is well below the general trend in 17B and above it in 17C. And it is true that evolution of a lower CF would be aided by a longer file (Fig. 17C) with lower tooth density (Fig. 17B).

Fig. 18 shows the trends in PP, PD and PI in relation to PR at 25°C on a log/log plot. The trends in PD and PI are linear and tight (with r² of 0.95 and 0.99, respectively) but the slope for PI is much steeper, indicating that as species evolve higher or lower pulse rates, PI changes much faster than PD. It is noteworthy that the data in Fig. 18 are for 13 species at their 25°C PR but that similar changes occur within the songs of each of these species as their pulse rates change with temperature.

Early in the development of this paper, in hope of documenting such changes in PD and PI in several species, TW searched his Anaxipha song data for individual males that had quality song recordings made at a wide range of tightly controlled temperatures. He soon found a number of individuals that qualified in all respects except sound quality. This was because the temperature control room that provided ambient temperatures well above and below 25° C had been quiet enough to make recordings that permitted precise measurements of PR and PP, but not of PD or PI. Much later, DF, while using his Anaxipha song data to complete Fig. 17E, realized that he had data that would provide useful estimates of PD and PI over wide temperature ranges for six mid-Atlantic species. These estimates and what might be concluded from them are in  SMFig_PD&PIbySixSpp (SMFig_PD&PIbySixSpp.pdf).

SMFig_OecanthusFiles compares features of file and song of 13 species of North American Oecanthus spp plotted against PR in a manner paralleling that of Fig. 17 for the 13 Anaxipha species recognized here. In number of file teeth and length of file the relationship to PR proved similar to that in Anaxipha spp. However, for file tooth density and carrier frequency, no trends similar to those in Anaxipha are evident.

Are there crickets smaller than North American Anaxipha that are as loud?—The males of certain species of Anaxipha are among the smallest crickets known by us to make calling songs easily heard at several meters by persons of normal hearing. Their only rivals in this respect are certain similarly tiny and easily heard mogoplistines in which the males have an elongate pronotum that covers all or most of the short forewings (Love & Walker 1979). During calling these crickets drop their abdomen to create a space that may serve as an amplifying chamber, whereas calling Anaxipha produce their calls with the sound producing system that we propose is prototypical for crickets.

In measuring Anaxipha for this study, we noticed that the males of the smaller species did not decrease in body length as much as did the females. Our convention for measuring body length is explained in Methods, but it may help to repeat here that the forewings were included when they extended at rest beyond the tip of the abdomen, and, in females, the ovipositor was excluded from the measurement. This method insured that for nearly all specimens our measurement was independent of the unpredictable changes in body length that can be caused by varying degrees of abdominal extension or contraction after death and preservation. In practice, for all Anaxipha we measured, except for a few females of litarena and rosamacula, our measure of body length ended with the tips of the forewings. Knowing that this body length might be a poor predictor of body mass and lacking weights of specimens of Anaxipha, we settled on hind femur length as the most practical attribute for estimating body mass.

Fig. 19 summarizes the sexual differences in body length and femur length of 13 Anaxipha species arranged from left-to-right largely by descending estimated body mass (all data are in SMTbls_HFandBLmeasurements). The two bars for each species show the amount by which the male mean values for femoral and body lengths are exceeded by the female values. Because cricket females generally average larger than their males, the expectation would be that all differences would be positive—i. e., that females would exceed males in both measures. This is not true for calusa (the heaviest species) but is true for the next five in Fig. 19, which are the five most closely related species of the exigua group. In the remaining seven species, males exceed females in body length, and in three of these, including the two smallest, males exceed females in femoral length as well. These results suggest that our smallest Anaxipha are near the size limit for producing an effective air-borne calling song. (It also reinforces our conclusion that calusa is a distant relative of our other Anaxipha.)

Supporting the notion that trigs smaller than fultoni are unlikely to be able to produce long-range calling songs is Falcicula hebardi, a North American trig that is smaller than any Anaxipha (Rehn 1905, Blatchley 1920). F. hebardi males stridulate weakly during courtship but make no calling song (Spooner 1972). TW, in the 1960's, examined series of about 40 of each sex of F. hebardi at UMMZ and noted that males averaged substantially smaller than females. When he measured the body lengths of the smallest and largest of each sex, he recorded that body lengths of males were 3.2 to 3.7mm, and of females 3.3 to 4.2mm. As reported in  SM_Falcicula (SM_Falcicula.pdf), DF's examination of the male genitalia, hind tarsi, and setation of the hind tibial spines of F. hebardi suggested that this species is more closely related to the clade that includes the fultoni, delicatula, litarena and exigua groups than either calusa or imitator — which brings into question the current generic status of F. hebardi.