All field work was conducted during May and June of 2012. Cicadas identified as Tibicen chiricahua were observed and audio recorded in the Magdalena Mountains west of Socorro, New Mexico, and at the type locality for Tibicen chiricahua, Pinery Canyon in the Chiricahua Mountains of southeastern Arizona (Davis 1923). Specimens of the new species were observed and recorded at its type locality.

To estimate the geographic ranges of the two species and better understand morphological variation across these ranges, I examined a total of 202 specimens previously identified as Tibicen chiricahua from the collections of the Arthropod Museum at New Mexico State University (NMSU), the C. P. Gillette Museum of Arthropod Diversity at Colorado State University (CSUC), the Frank M. Hasbrouck Insect Collection at Arizona State University (ASUT), the Snow Entomological Museum at the University of Kansas (SEMC), the Texas A&M University Insect Collection (TAMU), the University of Arizona Insect Collection (UAIC), and the University of Colorado Museum of Natural History (UCMC). The SEMC specimens included a male paratype of Tibicen chiricahua from Davis’s original type series. I also examined high-resolution digital photographs of the holotype male and allotype female of Tibicen chiricahua, which are currently housed in the collection of the Academy of Natural Sciences of Drexel University (ANSP).

Morphological terminology follows Moulds (2005, 2012). Morphometric measurements were made with a digital caliper. Fore wing width was measured from the node to the posterior edge, head width was measured between the eyes, and pronotum width was taken at the widest point between the lateral angles.

Cicada calling songs were recorded in the field using a Sennheiser ME 66 shotgun microphone with an MZW 66 PRO windscreen connected to a Sony PCMM10 digital audio recorder. All recordings were made as uncompressed, 16-bit PCM audio at a sampling rate of 44.1 kHz. For each recording, the microphone was held between 0.5 and 2 meters away from the calling cicada. This was close enough to minimize background noise, but far enough away to avoid any near-field acoustic effects in the frequency range of the calling songs (Michelsen and Nocke 1974, Peterson 1980).

Cicada calls were analyzed to determine peak frequencies, amplitude burst rates, and the number of sound pulses per amplitude burst. In this paper, I use the term “pulse” in the sense of Broughton (1963) and “amplitude burst” to mean a single group of high-amplitude pulses in an amplitude-modulated pulse train (see Figures 4 and 5). I elected to use “amplitude burst” rather than “syllable, ” which has been used inconsistently in cicada bioacoustics and usually with disregard to the precise definitions of Broughton (1963) and Ragge and Reynolds (1998).

Analyses were conducted using Audacity® (Audacity Team 2012) and custom-written software. Peak frequency was estimated by identifying the highest peak in a power spectral density plot generated by a 512-sample Fast Fourier Transform with the Hamming window function. If there were two or more peak frequencies that differed by less than 0.5 dB, their average was taken as the overall peak frequency. The amplitude burst rate (i.e., pulse amplitude modulation rate) was calculated by first estimating the call’s amplitude envelope, then using a gate function to identify the amplitude peaks in the signal (Beeman 1998). To estimate the number of sound pulses per amplitude burst, a sequence of 12 bursts was selected from the middle of each call, the audio data were normalized so that the maximum signal amplitude was at 0 dBFS, and the beginning and ending pulses of each amplitude burst were determined by identifying the first and last pulses with absolute sample values that exceeded 50% of the maximum sample value (that is, -6.02 dBFS).

The calls of both Tibicen chiricahua and Tibicen neomexicensis can be divided into three phrases (see results below), but the boundaries between phrases are often indistinct. To avoid the non-repeatability and potential bias of estimating the phrase durations by simple visual or aural inspection of the call oscillograms, I used objective criteria based directly on the audio data. All audio data were first normalized so that the peak amplitude was at 0 dBFS. For both Tibicen chiricahua and Tibicen neomexicensis, the first phrase began at the start of the call, and the end of the first phrase and beginning of the second phrase was defined by the first amplitude burst that reached -3 dBFS. For Tibicen chiricahua, the end of the second phrase was defined by the last amplitude burst to reach -3 dBFS, while for Tibicen neomexicensis, the end of the second phrase was defined as the end of the modulated portion of the call. For both species, the third phrase consisted of all audio from the end of the second phrase to call termination.

I did not include ambient air temperatures in the acoustic analyses. North American cicadas utilize a variety of behavioral and physiological thermoregulation tactics, so ambient temperature is often a poor indicator of a calling cicada’s body temperature (Toolson 1987, Hastings 1989, Sanborn et al. 1992, Sanborn 2000, 2004).

The locations of field sites that I personally visited were determined using a Garmin nüvi 260 GPS receiver. Specimen label data lacking latitude and longitude information were georeferenced primarily using data from the Geographic Names Information System of the United States Geological Survey (http://geonames.usgs.gov/), and in some cases using Google Earth (http://earth.google.com/). Landcover data were from the Southwest Regional Gap Analysis Project (USGS National Gap Analysis Program 2004). The distribution of pinyon-juniper woodlands was estimated by mapping all land cover types that included both pinyon pines (Pinus edulis, Pinus monophylla) and junipers (Juniperus spp.) as dominant tree or shrub species (codes S038, S039, S040, S052, and S112). Landcover types with junipers but not pinyon pines and sparsely vegetated types (< 10% plant cover) were excluded. QuantumGIS (Quantum GIS Development Team 2012) was used to produce the distribution map.

Acoustic and morphometric data were analyzed in R (R Core Team 2013) using univariate multiple linear regression with categorical predictor variables (i.e., ANOVA models). Morphometric data were modeled with species and sex as predictors, while acoustic data were modeled with species as the sole predictor. For all analyses, preliminary F-tests were used to compare models with locality as a predictor (two localities for Tibicen chiricahua and the type locality for Tibicen neomexicensis) to models that grouped all Tibicen chiricahua data together (i.e., used species as a predictor). In all cases, there was not a significant difference between the models (all p-values > 0.0788), so the data from the two locations for Tibicen chiricahua were grouped together for both the morphometric and acoustic analyses. Plots of the standardized residuals were examined to verify that the data met the model assumptions. Because Tibicen neomexicensis is most easily separated from Tibicen chiricahua by its distinctive calling song, the statistical analyses only included specimens from localities that had been acoustically surveyed.

Tibicen neomexicensis can be separated from all other North American species of Tibicen except for Tibicen chiricahua by the combination of its size, almost entirely black dorsal color pattern (Figure 1), and the male’s genitalia, particularly the shape of the uncus (Figure 2). Within Tibicen, this uncal structure is unique to Tibicen neomexicensis and Tibicen chiricahua (see Davis 1923 for a figure of the uncus of Tibicen chiricahua).

In the field, Tibicen neomexicensis and Tibicen chiricahua are most easily distinguished by the unique calling songs of the males. Audio recordings of both species are available as online supplementary data for this paper. To human ears, the first and second phrases of the call of Tibicen neomexicensis sound like a high-pitched, whiny buzz with easily discernible pulsations that correspond to the amplitude and frequency modulations. At the beginning of the third phrase, the abrupt transition to an unmodulated, uniform whine is perhaps the most aurally distinctive feature of the calling song.

In contrast, the call of Tibicen chiricahua sounds like a monotonous, coarse buzz that rapidly increases in amplitude during the first phrase and then slowly fades away during the final phrase. Apart from the amplitude changes in the first and third phrases, there are no obvious changes in sound quality during the course of the call.

Both male and female specimens of Tibicen neomexicensis and Tibicen chiricahua can usually also be separated by the coloration of the wing venation. In Tibicen neomexicensis, the anterior margin of the subcostal vein (Sc) of the fore wing is usually yellowish or at least noticeably lighter in color than the main part of the vein, which is dark black (Figure 3). In Tibicen chiricahua, both the vein and its anterior margin are black (Figure 3). In addition, the cubitus anterior vein (CuA) in the hind wing of Tibicen neomexicensis is yellow from its base to the junction with its second branch (CuA₂), and the basal two-thirds or more of CuA₂ is usually also yellow (Figure 3). In Tibicen chiricahua, these two veins tend to be mostly or entirely black (Figure 3).

Although none of these morphological characters are 100% reliable, when used in combination, they identify nearly all specimens. The color of the margin of the Sc vein is the most reliable single morphological diagnostic character. Out of nearly 200 specimens examined, only 7 might have been misidentified by the color of the Sc vein alone. The color of CuA and CuA₂ is more variable, with some overlap between the two species, and the utility of this character seems to vary among populations of Tibicen chiricahua. Unfortunately, in very old specimens, the colors of the wing veins sometimes fade, making identification difficult. Fore wing length can also be used to help confirm an identification, especially when the wing vein colors are ambiguous.

Both Tibicen neomexicensis and Tibicen chiricahua are associated with pinyon-juniper woodlands, and neither species seems to occur in habitats where both pinyon pines (Pinus edulis, primarily) and junipers (Juniperus sp.) are absent (B. Stucky, pers. observation; Hastings and Toolson 1991). Although Davis (1925), citing a correspondence from Douglas K. Duncan, reported specimens of Tibicen chiricahua collected “on a high mountain plateau … devoid of any vegetation except many clumps of a large heavy grass, ” he also noted that, “There is much timber around the edges of this plateau, pine, cedar, and juniper.” Overall, records of these cicadas from Arizona and New Mexico closely overlap with the distribution of pinyon-juniper forests in those states (Figure 6).

Specimen label data and field observations indicate that adults of Tibicen neomexicensis and Tibicen chiricahua emerge in early summer and are mostly gone by the end of July. The earliest record for Tibicen chiricahua is May 25 (in 1997) (specimen, CSUC). Although the UAIC has a specimen of Tibicen chiricahua from Arizona with the date recorded as “September, ” the next latest collecting date is July 22 (in 1975) (UAIC), so the September date is either very unusual or in error. The majority of collecting events for Tibicen chiricahua were in June, and Hastings and Toolson (1991) reported that June 1–8 was approximately the middle of the adult active season of Tibicen chiricahua in the San Mateo Mountains of New Mexico in 1989.

Phenological data for Tibicen neomexicensis are much more limited, but consistent with an annual pattern similar to Tibicen chiricahua. The earliest record for Tibicen neomexicensis is for May 30 (in 2012), at the type locality (B. Stucky, pers. observation), but at this time, there were already large numbers of females ovipositing, so the cicadas must have emerged some number of days earlier. The latest record is from June 7 (in 2005) (specimen, NMSU).

The daily activity patterns of these two cicada species are also similar. Once the sun warms them sufficiently, males of both species will sing throughout much of the day, with peak calling activity occurring from about mid-day through early afternoon (B. Stucky, pers. observation; Hastings and Toolson 1991). Calling activity greatly diminishes during the late afternoon and evening.

Although the nymphal host plants of Tibicen chiricahua and Tibicen neomexicensis are not known with certainty, there is anecdotal evidence that females of these species have different oviposition preferences. Hastings and Toolson (1991) reported Tibicen chiricahua females ovipositing in dead pinyon pine and juniper branches. In contrast, numerous ovipositing females of Tibicen neomexicensis were observed at the type locality, most of which were placing their eggs in the dead, dried stems of grasses and forbs, often quite near to the ground.

Both Tibicen chiricahua and Tibicen neomexicensis are commonly found with Tibicen duryi Davis, another species that is specialized on pinyon-juniper habitats (Hastings et al. 1991, Kondratieff et al. 2002). At the type locality, Tibicen neomexicensis was also syntopic with Okanagana bella Davis.

Tibicen neomexicensis and Tibicen chiricahua are not only extremely similar morphologically, but the shared structure of the male genitalia separates them from all other species of Tibicen. It therefore seems probable that Tibicen neomexicensis and Tibicen chiricahua are sister species, although a broader phylogenetic analysis of Tibicen is needed to confirm this.

Today, these species are apparently entirely allopatric, separated from one another by the uninhabitable Albuquerque Basin and Chihuahuan Desert. This might be a relatively recent phenomenon, though. At the time of the last glacial maximum, pinyons and junipers were widespread across much of what is today the Chihuahuan Desert (Betancourt et al. 1993, Lanner and Van Devender 1998, Thompson and Anderson 2001). As the climate changed at the end of the Pleistocene and beginning of the Holocene, pinyons and junipers retreated to higher elevations and became extirpated from modern desert regions (Betancourt et al. 1993). Moreover, climatological data and models suggest that this pattern probably repeated multiple times during the Pleistocene (Paillard 1997, Petit et al. 1999, Smith and Farrell 2005).

These habitat changes must have certainly affected the distributions of and interactions among the ancestors of modern Tibicen neomexicensis and Tibicen chiricahua. What impact, if any, this had on population divergence and speciation is unknown. However, theory predicts that secondary sexual traits can diverge rapidly in allopatry (Pomiankowski and Iwasa 1998), and if Tibicen neomexicensis and Tibicen chiricahua are sister species, it seems likely that geographic isolation caused by habitat shifts played at least some role in their evolution.