There are four pairs of neurosecretory cells producing bombyxin, an insulin-related peptide, in the brain of Bombyx mori. In the present study, electrical signals of the bombyxin-producing cells were identified and short- and long-term activity patterns of the cells throughout the pupal period were analyzed. Activity pattern of the cells was characterized by periodic discharges of a volley of action potentials. Periodicity of the electrical activity depended on the temperature and mean±SD of periods was 34.7±4.8 min at 26°C. Four cells in a brain hemisphere were functionally heterogeneous: individual cells had specific activity amplitudes and rhythms, thereby suggesting that an ultradian activity rhythm of a cell originates from its intrinsic pacemaker activity. There was no significant diel (circadian) oscillation in the electrical activity. Daily activity patterns had a maximum at an early stage (day-2 or 3) of pupal period. Periodic secretion of bombyxin, like insulin, seems to be an efficient temporal pattern for hormonal actions.
Insulin is a peptide hormone regulating cellular metabolism and growth of vertebrates. Secretion of insulin, like many other hormones, is characterized by a pulsatile ultradian rhythm and fluctuation of plasma insulin concentration due to the pulsatile secretion is important for maintaining hormonal actions (e.g., Matthews, et al., 1983; Matthews, 1991).
Insects have an insulin-related peptide hormone, bombyxin that was first identified in the head of silkmoth, Bombyx mori (Nagasawa et al., 1984, 1986). Bombyxin comprises highly heterogeneous molecular forms that are generated from complex multi-family genes (Iwami, 1995; Tsuzuki et al., 1997; Yoshida et al., 1997, 1998). Bombyxin secretion is assigned to four pairs of neurosecretory cells with a particularly large soma in Bombyx (Mizoguchi et al., 1987) or to a large number of cells in another silkmoth Samia cynthia cicini (Yagi et al., 1995). These characteristic molecular and cellular organizations for production and secretion of bombyxin indicate that bombyxin is one of the most important peptides in insects.
Physiological functions of bombyxin are not fully understood. Bombyxin obtained from Bombyx elicited metamorphosis in brain-removed dormant pupae of the moth Samia, by acting on the prothoracic gland to stimulate the synthesis and release of ecdysteroid (Nagasawa et al., 1984). However, the peptide failed to activate the prothoracic gland of Bombyx (Ishizaki et al., 1983). Bombyxin injected into a feeding larva induced decreases in haemolymph trehalose and fat body glycogen, major storage carbohydrates in insects (Satake et al., 1997), but the peptide was ineffective for carbohydrate metabolism in the adult moth (Satake et al., 1999). The titer of bombyxin in the haemolymph becomes maximal during adult development (Saegusa et al., 1992), thereby suggesting that bombyxin is a factor regulating growth and development. It has been reported that bombyxin binds to specific receptors on ovarian cells (Fullbright et al., 1997) and that it induces meiosis in ovarian cells (Orikasa et al., 1993) or morphological changes in a cell line derived from the ovarian tissue (Tanaka et al., 1995).
To elucidate physiological functions of a neurosecretory cell and its products, it is vital to determine temporal patterns of it activity and mechanisms that control secretion from the cell. As noted with neurosecretory cells releasing pheromonotropic peptides (Ichikawa, 1998), monitoring electrical activities of a cell (or a set of cells) is suitable for the purpose.
Four pairs of bombyxin producing cells of Bombyx have a readily identifiable soma in the dorsomedial region of the brain and thick axon terminals in the corpora allata (CA) to release the hormone into the haemolymph (Ichikawa, 1991). In the present study, electrical signals of the bombyxin producing (BP) cells were detected extracellularly from the CA in pupae and pulsatile activity patterns were revealed.
MATERIALS AND METHODS
Fifth instar larvae of commercially available F1 hybrid of Bombyx mori (Kinshu × Showa) were purchased from a supplier and reared on an artificial diet. The larvae were placed at 26±1°C under a 16-hr light/8-hr dark photoperiod.
The ventral part of the thorax of a pupa at 6–10 hr after ecdysis was fixed with paraffin to a platform at an angle of 45°. The whole body except for the head of the pupa was placed in a plastic chamber and the chamber was filled with CO2 gas for purposes of immobilization. Small pieces of head cuticle and part of the fat body in the head capsule were removed to expose the brain and the CA.
For long-term monitoring of BP cells' activities, CA was introduced into a suction electrode filled with a physiological saline containing 1% streptomycin sulfate (Sigma). A piece of silver wire serving as an indifferent electrode was made to contact the haemolymph, and the cuticular window was sealed with melted paraffin. The pupae was usually placed in an incubator kept at 26±1°C and illuminated at the same light/dark cycles, using a fluorescent lamp. Light intensity was about 100 lx. Under these experimental conditions, pupae usually shed abdominal cuticle 10 or 11 days after pupation.
For intracellular stimulation and recording, the sheath of the brain was partially removed, using a fine tungsten needle. Under visual guidance, a soma of a BP cell was penetrated with a glass microelectrode filled with 3M KCl. The microelectrode was connected to a preamplifier, which allowed simultaneous recording and current injection via a bridge circuit. An action potential was induced from the soma by passing a depolarizing current pulse (1–5 nA, 3 ms in duration). After all four somata of BP cells were impaled, they were removed, using a fine tungsten needle.
All electrical signals were amplified, digitized, and stored on a computer equipped with an analog/digital converter (1401plus, Cambridge Electronic Design, Cambridge). Spikes from BP cells were discriminated using spike-sorting software (Spike 2, Cambridge Electronic Design). Periods of firing activity rhythm were estimated by chi square periodogram (Sokolove and Bushell, 1978).
Identification of spikes of Bombyxin-producing cells
Electrical signals were recorded from the CA where BP cells have axon terminals. Because other types of neurosecre-tory cells such as PTTH producing cells also terminate there (Ichikawa, 1991), one should extract electrical signals (action potentials or spikes) of BP cells from multiunit neural signals. Several classes of spikes, usually discriminated based on their specific amplitudes and/or waveforms, were divided into two groups with significantly different amplitudes, large and small. To identify spikes of BP cells, each soma of a BP cell located in the dorsomedial region of a contralateral brain hemisphere was impaled and injected with current pulses. The electrical stimulation always induced large spikes regardless of the BP cell impaled (Fig. 1). When four large somata of BP cells were surgically removed, large spikes disappeared and only the small group of spikes remained (data not shown). Thus, it became evident that the large spikes originated from the BP cells.
Ultradian rhythmic activity
Spontaneous firing activities of BP cells were recorded from 40 male and 10 female pupae for 7–11 consecutive days. There were no significant differences in activity patterns between sexes. Although amplitudes and waveforms of BP cells' spikes gradually changed with adult development, significantly large spikes of the cells were maintained throughout development. This feature greatly facilitated analyses of the firing activity of BP cells.
A train of spikes produced by a single BP cell could be often discriminated from those of other BP cells because of a specific amplitude and different timing of firing. A BP cell usu-cell 1, 33.7 min, 3070 (37%); cell 2, 31.5 min, 2120 (26%); cell 3, 31.7min, 1492 (18%); cell 4, 39.4 min, 1565 (19%). ally produced single spikes at a relatively constant interval of about 2 s, except at the start and end of a spike train (Fig. 2A). Some cells showed a burst of a few or several spikes generated at a short interval of about 0.5 s (Fig. 2B).
Activity pattern of a population of BP cells was characterized by rhythmic discharges of trains of spikes, though rhythmicity did vary from animal to animal. In about half the number of the pupae examined, the rhythmic pattern was relatively regular and simple (Fig. 3A), while activity patterns of others were complex and appeared to be a mixture of different rhythms (Fig. 3C). The distinct rhythmic pattern in the former type was due to a dominance of a particular single unit that accounted for about 60% of the entire population of spikes (Fig. 3B). Pulsatile activity of the dominant unit was quite regular and a maximum firing rate was usually about 20 spikes/min. There was no apparent diel change in the activity. Rhythm periods of dominant units measured on day-3 in 16 pupae varied from 26 to 48 min and mean±SD of such ultradian rhythm periods was 34.7±4.8 min at 26±1°C. Dominant single units usually accounted for 50-65% of the total activity of BP cells. Minor components of activity patterns consisted of spikes originating from one or two units with a periodicity that usually differed from that of the dominant unit.
When there were significant differences in spike amplitudes among the four units, four classes of spikes could be clearly discriminated in the record showing a complex pattern of multiunit activity (Fig. 4). Production of a train of spikes in each cell lasted for 7–20 min and such bursting activities were repeated at an interval of 25–40 min. Each cell revealed a specific periodicity and mean periods of the four cells were no longer identical. The shortest period was 31.5 min and the longest 39.4 min; the periods differing by 7.9 min. There was no stable phase-relationship between activity rhythms of any pair of cells and this situation made the apparent pattern of multiunit spikes complex.
The number of spikes produced by a single cell during a long period of time (12 hr) differed significantly: spikes of the most active cell accounted for 38% of the total number of spikes while those of the least active cell made up for only 17%. Similar cell-specific rhythms and heterogeneous activity profiles in a population of BP cells were observed in two other pupae.
Daily activity patterns during the pupal period
Total numbers of spikes of BP cells counted daily usually became maximal at an early pupal stage, day-2 or day-3, and gradually decreased, though there was considerable variability in the spike numbers among individuals (Fig. 5).
Mean periods of the ultradian rhythms changed considerably within a range between 25 min and 55 min as pupal stages proceeded, but no common trend of changes was evident. The mean periods gradually shortened (n=2) or lengthened (n=2), while they became shortest at a middle pupal stage (n=3). A pupa maintained a relatively constant period around 30 min throughout the pupal period.
Temperature dependence of ultradian rhythm
To know a nature of the pulsatile activity of BP cells, effects of temperature on the rhythm were examined.
When the temperature was elevated from 25°C to 30°C for 8 hr during the middle photophase, the period of ultradian rhythm shortened and a maximum rate of firing increased (Fig. 6). In contrast, the period lengthened and the rate of firing decreased, when the temperature was shifted down from 25°C to 20°C (Fig. 6). Periodograms show that warming by 5°C shortened the mean period by 6 min and cooling by 5°C lengthened it by 8 min (Fig. 7). Thus, the activity rhythm of BP cells is highly sensitive to the temperature. The relationship between the temperature and the period of the ultradian rhythm is shown in Fig. 8. The regression line indicates that the rhythm becomes fast 1.83 times as the temperature is increased by 10°C. The value is the temperature coefficient (Q10) and mean±SD of Q10 was 2.21±0.41 (n=5).
Long-term, chronic recordings of insect neurosecretory cells have been done on PBAN-producing cells in the suboesophageal ganglion of Bombyx mori (Ichikawa, 1998). Because the cells have an unique axonal pathway, there is no problem in identifying action potentials recorded extracellularly. In contrast, it is important to determine if a recorded action potential originated from a BP cell, because the cell sends an axon to the CA via the NCC 1+2, the common axonal pathway of cerebral neurosecretory cells. There are at least three types of neurosecretory cells terminating in the CA: BP cells, PTTH producing cells and unknown ones (Ichikawa, 1991). Because terminal branches of a BP cell are thicker and more abundant than those of cells of other types, a spike potential originating from the cell is expected to be large due to a large extracellular current flowing out from thick branches. The spike of BP cells was apparently large (Fig. 1) and this facilitated identification of spikes. Origins of small spikes were not determined. However, analysis and identification of small spikes may be important because a fraction of small spikes possibly originates from a putative clock neuron that expresses period gene and is involved in circadian rhythm in the silkmoth, Antheraea pernyi (Sauman and Reppert, 1996).
Each cell appears to have specific rhythmicity (Fig. 4). If electrical activities of four cells are regulated by a putative, common pacemaker (ultradian oscillator), their activity rhythms should either be synchronized or significantly phase-locked. The free running rhythm of a cell may represent its inherent oscillatory property. The period of the rhythm strongly depends on the temperature (Fig. 5) and Q10 is about 2.2 (Fig. 7), thereby suggesting that biochemical oscillatory mechanisms may drive the rhythmic activity of the cell. However, the result can not exclude a possible mechanism that coordinates activity rhythms of whole cells. In a preliminary experiment, simultaneous recordings from right and left corpora allata revealed a significant correlation between electrical activities of right and left groups of BP cells.
Many hormones, neuropeptides and biogenic amines in vertebrates are secreted in an ultradian pulsatile fashion (Matthews, 1991). For example, insulin secretion shows a rapid (10–14 min) and a slow oscillation (50–100 min), superimposed on circadian fluctuation. The fast oscillation appears to be an inherent feature of cellular mechanisms of secretion and the slow, circhoral oscillation is in part generated by a negative feedback loop linking glucose and insulin secretion (Polonsky et al., 1998). To examine feedback mechanisms involved in bombyxin secretion, a large amount of glucose (or bombyxin II) was injected into a pupa in preliminary experiments, but it induced no significant change in the activity pattern of BP cells. Glucose may be no longer effective in a pupa, as demonstrated in an adult moth, though a glucose injection into a feeding larva does induce bombyxin secretion (Satake et al., 1999; Masumura et al., 2000).
It is evident that many neuroendocrine systems rely on pulsatile patterns of secretory activity to maintain biological functions (Belchetz et al., 1978; Norstedt and Palmiter, 1984; Clark et al., 1985; Matthews, 1991; Windle et al., 1998). Hormone-specific periodicity of secretion seems to be important for efficient actions of the hormone. For example, circhoral (one pulse per h) stimulation of the pituitary gland with gonadotropin-releasing hormone (GnRH) is effective for a maintained gonadotropin secretion while high-frequency or continuous exposure to GnRH results in desensitization of the pituitary response (Belchetz et al., 1978; Wildt et al., 1981). Strong desensitization or down-regulation due to prolonged exposure to an agonist is a common property of many membrane receptors, including insulin receptors (Gustafson et al., 1999). Ultradian activity rhythm of the bombyxin-releasing cells may also be important for actions of bombyxin on target tissues. Because bombyxin II injected into a neck-ligated larva seems to be rapidly removed from the haemolymph by degradation, binding to target cells' surface receptors and subsequent incorporation into the cells (Satake et al., 1997), a similar rapid removal of bombyxin secreted into the pupal haemolymph may also occur. Since a half-life period of bombyxin in the haemolymph is less than 1 hr during an initial phase of removal, pulsatile secretion of bombyxin at an interval of 30–40 min may be significant to fluctuate the hormone titer in the haemolymph. A putative bombyxin receptor has been found on ovarian cells of Bombyx mori and two other species of moth and it has structural similarities to the mammalian insulin receptor (Fullbright et al., 1997).
Secretion of an insect peptide hormone is usually assigned to multiple, clustered neurosecretory cells. For example, five pairs of cells produce a neuropeptide for activating sex pheromone production in Bombyx mori (Ichikawa et al., 1995, 1996), and 16 pairs of cells are involved in the production of a bombyxin-immunoreactive material in Samia cynthia ricini (Yagi, et al., 1995). One important question is whether all those cells contribute equally to secretion of the hormone. The present study revealed that four pairs of cells are electrophysiologically heterogeneous. The amount of hormone released from a cell may be practically in proportion to the electrical activity of the cell because there is a close similarity between changes in electrical activity (Fig. 5) and in the titer of bombyxin immunoreactive material in the haemolymph during adult development (Saegusa et al., 1992). Thus, they do not equally contribute to the secretion of bombyxin at least during the pupal stage. One or two dominant cells are significantly more active than others: the former produces >50% of the entire number of action potentials (Figs. 3 and g4). Mizoguchi et al. (1990) reported stage-specific, unequal distribution of bombyxin-immunoreactive materials among the four cells in a hemisphere of the larval Bombyx brain: two are stained more intensely than were the other two cells, thereby suggesting different secretory activities of those cells. Similar heterogeneous distributions of firing activities were observed among five neurosecretory cells releasing PBAN in a female moth of Bombyx (T. Ichikawa, unpublished observation). If all bombyxin-producing cells become equally active as the dominant one, they can secret at least a 2- or 3-fold larger amount of hormone. Such an active state may be evident in an adult male 3–6 hr after eclosion, at which the bombyxin titer transiently increases to 5-fold larger than the titer at the pupal stage (Satake et al., 1999).
I thank M. Ohara for language assistance. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan.
- P. E. Belchetz, T. M. Plant, Y. Nakai, E. J. Keogh, and E. Knobil . 1978. Hypophysial responses to continuous and intermittent delivery of hypothalamic ganadotropin-releasing hormone. Science 202:631–633. Google Scholar
- R. G. Clark, J. O. Jansson, O. Isaksson, and I. C. A. F. Robinson . 1985. Intravenous growth hormone: growth responses to patterned infusions in hypophysectomized rats. J. Endocrinol 104:53–61. Google Scholar
- G. Fullbright, E. R. Lacy, and E. E. Bülesbach . 1997. The prothrocicotropic hormone bombyxin has specific receptors on insect ovarian cells. Eur J Biochem 245:774–780. Google Scholar
- T. A. Gustafson, S. A. Moodie, and B. E. Lavan . 1999. The insulin receptor and metabolic signaling. Rev. Physiol. Biochem Pharmacol 137:71–190. Google Scholar
- T. Ichikawa 1991. Architecture of cerebral neurosecretory cell systems in the silkworm Bombyx mori. J exp Biol 161:217–237. Google Scholar
- T. Ichikawa 1998. Activity patterns of neurosecretory cells releasing pheromonotropic neuropeptides in the moth Bombyx mori. Proc Natl Acad Sci USA 95:4055–4060. Google Scholar
- T. Ichikawa, K. Hasegawa, I. Shimizu, K. Katsuno, H. Kataoka, and A. Suzuki . 1995. Structure of neurosecretory cells with immunoreactive diapause hormone and pheromone biosynthesis activation neuropeptide in the silkworm, Bombyx mori. Zool Sci 12:703–712. Google Scholar
- T. Ichikawa, T. Shiota, I. Shimizu, and H. Kataoka . 1996. Functional differentiation of neurosecretory cells with immunoreactive diapause hormone and pheromone biosynthesis activating neuropeptide of the silkworm, Bombyx mori. Zool Sci 13:21–25. Google Scholar
- H. Ishizaki, A. Suzuki, I. Moriya, A. Mizoguchi, M. Fujishita, H. O'oka, H. Kataoka, A. Isogai, H. Nagasawa, and A. Suzuki . 1983. Prothoracicotropic hormone bioassay: Pupal-adult Bombyx assay. Develop Growth Differ 25:585–592. Google Scholar
- M. Iwami 1995. Multi-family genes encoding bombyxin and bombyxin-related peptides. In A. Suzuki, H. Kataoka, and S. Matsumoto . Ed by. “Molecular Mechanisms of Insect Metamorphosis and Diapause”. Tokyo Industrial Publishing & Consulting. 65–74. Google Scholar
- M. Masumura, S. Satake, H. Saegusa, and A. Mizoguchi . 2000. Glucose stimulates the release of Bombyxin, an insulin-related peptide of the silkworm Bombyx mori. Gen Com Endocrinol 118:393–399. Google Scholar
- D. R. Matthews 1991. Physiological implications of pulsatile hormone secretion. Ann New York Acad Sci 618:28–37. Google Scholar
- D. R. Matthews, B. A. Naylor, R. G. Johns, G. M. Ward, and R. C. Turner . 1983. Pulsatile insulin has greater hypoglycemic effect than continuous delivery. Diabetes 32:617–621. Google Scholar
- A. Mizoguchi, H. Ishizaki, H. Nagasawa, H. Kataoka, A. Isogai, S. Tamura, A. Suzuki, M. Fujino, and C. Kitada . 1987. A monoclonal antibody against a synthetic fragment of bombyxin (4K-prothoracicotropic hormone) from the silkmoth, Bombyx mori: characterization and immunohistochemistry. Molec Cell Endocrinol 51:227–235. Google Scholar
- A. Mizoguchi, M. Hatta, S. Sato, H. Nagasawa, A. Suzuki, and H. Ishizaki . 1990. Developmental change of bombyxin content in the brain of the silkmoth Bombyx mori. J Insect Physiol 36:655–664. Google Scholar
- H. Nagasawa, H. Kataoka, A. Isogai, S. Tamura, A. Suzuki, H. Ishizaki, A. Mizoguchi, Y. Fujiwara, and A. Suzuki . 1984. Amino-terminal amino acid sequence of the silkworm prothoracicotropic hormone: homology with insulin. Science 226:1344–1345. Google Scholar
- H. Nagasawa, H. Kataoka, A. Isogai, S. Tamura, A. Suzuki, A. Mizoguchi, Y. Fujiwara, A. Suzuki, S. Y. Takahashi, and H. Ishizaki . 1986. Amino acid sequence of a prothoracicotropic hormone of the silkworm Bombyx mori. Proc Natl Acad Sci USA 83:5840–5843. Google Scholar
- G. Norstedt and R. Palmiter . 1984. Secretory rhythm of growth hormone regulates sexual differentiation of mouse liver. Cell 36:805–812. Google Scholar
- C. Orikasa, H. Yamauchi, H. Nagasawa, A. Suzuki, and M. Nagata . 1993. Induction of oocyte nurse cell differentiation in the ovary by the brain during the initial stage of oogenesis in the siikworm, Bombyx mori (Lepidoptera: Bombycidae). Appl Entomol Zool 28:303–311. Google Scholar
- K. S. Polonsky, J. Sturis, and E. Vancauter . 1998. Temporal profiles and clinical significance of pulsatile insulin secretion. Hormone Res 49:178–184. Google Scholar
- H. Saegusa, A. Mizoguchi, H. Kitahara, H. Nagasawa, A. Suzuki, and H. Ishizaki . 1992. Changes in the titer of bombyxin-immunoreactive material in hemolymph during the postembryonic development of the silkmoth Bombyx mori. Develop Growth Differ 34:595–605. Google Scholar
- S. Satake, M. Masumura, H. Ishizaki, H. Kataoka, A. Suzuki, and A. Mizoguchi . 1997. Bombyxin, an insulin-related peptide of insects, reduces the major storage carbohydrates in the silkworm Bombyx mori. Comp Biochem Physiol 118/B:349–357. Google Scholar
- S. Satake, K. Nagata, H. Kataoka, and A. Mizoguchi . 1999. Bombyxin secretion in the adult silkmoth Bombyx mori: sex-specificity and its correlation with metabolism. J Insect Physiol 45:939–945. Google Scholar
- I. Sauman and S. M. Reppert . 1996. Circadian clock neurons in the silkmoth Antheraea pernyi: novel mechanisms of period protein regulation Neuron 17:889–900. Google Scholar
- P. G. Sokolove and W. N. Bushell . 1978. The chi square periodogram: its utility for analysis of circadian rhythms. J Theor Biol 72:131–160. Google Scholar
- M. Tanaka, H. Kataoka, K. Nagata, H. Nagasawa, and A. Suzuki . 1995. Morphological changes of BM-N4 cells induced by bombyxin, an insulin-related peptide of Bombyx mori. Regul Peptides 57:311–318. Google Scholar
- S. Tsuzuki, T. Masuta, M. Furuno, S. Sakurai, and M. Iwami . 1997. Structure and expression of bombyxin E1 gene, a novel family gene that encodes bombyxin-IV, an insect insulin-related neurosecretory peptide. Comp Biochem Physiol 117/B:409–416. Google Scholar
- L. Wildt, A. Häusler, G. Marshall, J. S. Hutchison, T. M. Plant, P. E. Belchetz, and E. Knobil . 1981. Frequency and amplitude of gonadotropin-releasing hormone stimulation and gonadotropin secretion in the rhesus monkey. Endocrinol 109:376–385. Google Scholar
- R. J. Windle, S. A. Wood, N. Shanks, S. L. Lightman, and C. D. Ingram . 1998. Ultradian rhythm of basal corticosteron release in the female rat: dynamic interaction with the response to acute stress. Endocrinol 139:443–450. Google Scholar
- Y. Yagi, J. Ishibashi, K. Nagata, H. Kataoka, A. Suzuki, A. Mizoguchi, and H. Ishizaki . 1995. The brain neurosecretory cells of the moth Samia cynthia ricini: immunohistochemical localization and developmental changes of the Samia homologues of the Bombyx prothoracicotropic hormone and bombyxin. Develop Growth Differ 37:505–516. Google Scholar
- I. Yoshida, S. Tsuzuki, S. E. Abdel Salam, M. Ino, A. M. Karayem, S. Sakurai, and M. Iwami . 1997. Bombyxin F1 gene: structure and expression of a new bombyxin family gene that forms a pair with bombyxin B10 gene. Zool Sci 14:615–622. Google Scholar
- I. Yoshida, K. Moto, S. Sakurai, and M. Iwami . 1998. A novel member of the bombyxin gene family: structure and expression of bombyxin G1 gene, an insulin-related peptide gene of the silkmoth Bombyx mori. Dev Genes Evol 208:407–410. Google Scholar