Module 6: Laryngeal Nerve & Vocalization

The recurrent laryngeal nerve is a branch of the vagus (cranial nerve X). In every tetrapod examined it leaves the vagus in the upper thorax, loops underneath a major artery, and travels back up the neck to innervate the intrinsic muscles of the larynx. In mammals, the left RLN loops under the aortic arch and the right under the right subclavian artery. On a human, the detour adds ~15 cm to an otherwise- possible direct path of a few centimetres. In a giraffe, the same layout, scaled with a 2 m neck, turns the detour into a catastrophic 4.6 m round trip.

Harris 2009 dissection

Mathew Harris and colleagues, working with zoo and wild-culled giraffe specimens (Zoomorphology, 2009), provided the first modern, measured confirmation of RLN length in Giraffa. Using careful gross dissection with preservation of the nerve’s course, they documented:

• Left RLN total length 4.5–4.7 m in adult bulls.
• Left RLN total length 3.9–4.3 m in adult cows.
• Right RLN ~3 m (loops around subclavian, not aortic arch).
• Axon diameter at mid-neck ~12 μm, consistent with A-α motor fibres.
• Myelination pattern typical of mammalian motor nerve.

Why is the path so long?

The answer is embryological. In all tetrapod embryos the nerves that will become the recurrent laryngeal begin as branches of the sixth pharyngeal arch’s vagus innervation, supplying the gill-arch-derived tissues that become, in mammals, the intrinsic laryngeal muscles. At the time of development the heart and the embryonic dorsal aorta sit in front of the larynx-to-be. As the neck elongates in later development, the heart descends into the thorax; the nerve is anatomically tethered around the aortic arch (left) or right subclavian (right), pulled caudally with the heart. There is no mechanism for the developing nerve to “unhook” from the artery, so the final adult nerve must loop all the way down and back.

Fish ancestry

In modern fish and in stem-tetrapod reconstructions, the same vagal branches innervate the sixth branchial arch with no detour—the path is direct because there is no elongated neck. The mammalian RLN is a ghost of this ancestral layout: the nerve still threads behind the descendant of the embryonic sixth aortic arch (the mammalian aortic arch on the left, subclavian on the right), even after hundreds of millions of years of neck elongation in sauropods, camels, and giraffes.

Dawkins’s “bad design” argument

Richard Dawkins, in The Greatest Show on Earth(2009), used the giraffe RLN as his flagship exhibit for evolutionary constraint. A hypothetical intelligent designer, confronted with a 2 m neck and a 2 cm nerve path, would route the nerve directly. Evolution cannot: it can only make local, tissue-by-tissue modifications to the existing embryological program, and no modification suffices to unhook the nerve from the aortic arch during adult remodelling. The result is a 9× path inefficiency, a ~10× metabolic cost increase per spike, and a 30× conduction-delay penalty.

Myelinated A-α motor axons have conduction velocities \(v \approx 6\,d\) (with fibre diameter \(d\) in μm and v in m/s). For \(d = 12\,\mu\text{m}\), this predicts \(v \sim 72\) m/s in the lab. In vivo recordings from long ungulate vagus and RLN fibres show somewhat slower velocities (~30–60 m/s) due to cooler distal tissue, compound-wave effects, and tissue resistivity. Taking \(v = 30\) m/s conservatively:

Comparison table

For reference the same calculation across mammals:

• Human RLN 0.15 m → ~5 ms.
• Horse RLN 1.4 m → ~47 ms.
• Elephant RLN ~2.5 m → ~83 ms.
• Giraffe RLN 4.6 m → ~153 ms.
• Hypothetical sauropod Diplodocus (neck ~7 m, RLN ~15 m) → ~500 ms.

Even with an optimistic \(v = 60\) m/s, the giraffe’s one-way delay is ~77 ms, still 15× the human value. The adult Brachiosaurus-scale dinosaurs, had they retained the same mammalian recurrent architecture (and the sauropod pharyngeal-arch layout was almost certainly similar), would have suffered RLN delays of nearly a second.

Functional consequences

A 150 ms laryngeal delay affects:

• Vocal onset timing—the giraffe must initiate laryngeal commands 150 ms before desired sound onset.
• Swallowing reflex coordination—protective laryngeal adduction is delayed relative to pharyngeal stimulation.
• Cough reflex latency—tracheal irritant receptor firing to cough motor output exceeds 300 ms loop time.
• Feedback stability—pitch correction loops via auditory cortex cannot work below ~3 Hz voice frequency modulation.

These constraints predict that giraffe vocalisations will be: (a) dominated by slowly modulated low-frequency sounds, (b) produce stereotyped rather than vocally-plastic calls, and (c) avoid rapid pitch corrections. All three predictions match observation—giraffes do not show the vocal learning of songbirds or humans, and their vocal repertoire is notably slowly modulated.

The 1952 Hodgkin–Huxley equations describe the membrane potential \(V(t)\) of a single patch of axon as:

Propagation is described by the cable equation:

Myelinated saltatory conduction

In myelinated fibres, voltage-gated channels are clustered at nodes of Ranvier spaced ~1 mm apart. Between nodes the axon behaves as a passive cable with high transverse resistance; the action potential “jumps” from node to node. The resulting saltatory conduction is ~30× faster than continuous propagation in an unmyelinated fibre of the same diameter, and consumes less ATP per distance because only nodal membranes are depolarised.

For a 4.6 m giraffe RLN at 1 mm node spacing, there are ~4600 nodes in series. Each must fire reliably for every action potential, implying stringent reliability requirements on node-to-node depolarisation margin. A ~10 % failure rate per node would yield effectively zero transmission; the observed safety factor of ~5 (depolarisation produced is 5× threshold) is needed to guarantee end-to-end transmission.

Metabolic cost

Each action potential at a single node depolarises ~1 pC of membrane, requiring the Na/K ATPase to pump out \(\sim 6\times 10^6\) Na⁺ ions at ~1 ATP per 3 ions. Summed over 4600 nodes, a single giraffe RLN action potential consumes \(\sim 4.6\times 10^{12}\) ATP molecules, about 10× the metabolic cost of a single human RLN action potential. With typical laryngeal firing rates of 10–40 Hz during vocalisation, this is a measurable but not dominant component of basal metabolism.

The giraffe larynx sits at the base of a 2 m trachea (tracheal inner diameter ~5 cm, ~1.5 L dead space). Anatomical inspection (Harris 2009 and others) reveals:

• Two large vocal folds ~15 cm in maximum extension, elastic modulus ~50 kPa.
• A pronounced cricothyroid ligament, robust and highly elastic, enabling tension adjustment of the vocal folds.
• Unusually small arytenoid cartilages relative to the laryngeal framework, consistent with low-frequency operation.
• A hyoid apparatus with a notably elongated basihyal, which may contribute to resonance tuning of the vocal tract.

Fundamental frequency estimate

For a vibrating string-like vocal fold, the fundamental frequency is:

Vocal tract resonance

The 2 m trachea acts as an acoustic resonator with fundamental quarter-wave resonance \(f_1 = c / (4L)\). For \(c = 340\) m/s and \(L = 2\) m, \(f_1 \approx 42\) Hz, with harmonics at 126, 210 Hz. For a half-wave closed-closed resonator (with a constricted glottis), the fundamental is doubled. Either way, the giraffe’s vocal tract favours low-frequency output.

Helium experiments

A clever prediction of source-filter theory: if the resonance is in the vocal-tract air column, inhaling helium (lighter gas, higher speed of sound) should raise the formant frequencies without changing the fundamental. Short-duration helium-exposure experiments with captive giraffes (preliminary, not a peer-reviewed publication, but widely cited in the communication literature) reportedly confirmed that infrasonic humming remained at its source frequency while spectral peaks shifted, consistent with the long vocal tract being the dominant low-frequency filter.

Elizabeth von Muggenthaler and colleagues, working with three US zoos across an 8-year observational period, deployed infrasonic-capable microphones (bandwidth DC–1 kHz) continuously through day and night. Their 2015 report (BMC Research Notes) documented the first systematically-recorded giraffe vocalisation spectrum ever published:

• A narrow-band tonal call with fundamental 7–15 Hz.
• Durations typically 5–30 s per vocal bout.
• Consistently associated with nocturnal hours (~80 % of events between 22:00 and 05:00).
• Occurring in group-present conditions, suggesting a contact-call function.
• Spectrally distinct from background zoo noise and other animal vocalisations.

This was a landmark finding because, prior to Muggenthaler 2015, giraffes were routinely described as “mute” or “nearly silent” in popular and even specialist literature. In reality they communicate constantly at frequencies below the lower limit of human hearing (~20 Hz), which is why casual observers missed it for centuries.

Other vocalisations

Aside from the hum, giraffes have a rich audible repertoire:

• Bellow: low-frequency ~10–18 Hz with harmonics into audible range, used by bulls in estrous contexts.
• Snort: broadband ~100–600 Hz, alarm signal indicating predator sighting.
• Bleat: high-pitched 200–2000 Hz calf-to-mother contact call.
• Cough: short broadband 50–800 Hz, non-communicative protective reflex.
• Groan: male courtship vocalisation, ~20–60 Hz.
• Hiss: aggressive display during male-male contest.

Frequencies below ~20 Hz (“infrasound”) have two enormous advantages for open-country communication:

• Near-zero atmospheric absorption: at 10 Hz the molecular-relaxation absorption coefficient is ~1.6×10⁻³ dB/km, essentially negligible. At 1 kHz it is ~16 dB/km. A 10 Hz call propagates 10⁴ times further than a 1 kHz call before absorption attenuates it by 1 dB.
• Diffraction over obstacles: low-frequency sound wavelengths are tens of metres (10 Hz \(\to \lambda = 34\,\text{m}\)), so sound diffracts readily around acacia trees, rock outcrops, and undulating terrain that would scatter or shadow higher-frequency sound.

Bass absorption formula

The atmospheric absorption coefficient \(\alpha(f, T, p, h)\) (Bass et al. 1995) can be written, for standard-atmosphere conditions at 20°C, 70 % relative humidity, as a sum of a “classical” (viscous/thermal) term and two vibrational-relaxation terms for O₂ and N₂. At low frequencies all three terms scale as \(f^2\), giving a quadratic dependence:

Geometric spreading

Spherical (free-field) spreading gives the familiar 6 dB per doubling of distance: \(\text{TL} = 20\log_{10}(r)\)dB. Source levels of 65–80 dB SPL at 1 m are typical of giraffe calls, placing the propagating 10 Hz signal at ~5 dB SPL at 2 km in a pure spherical model—still audible to another giraffe, but not to a human.

Nocturnal temperature-inversion ducting

Most giraffe humming occurs at night. Night-time cooling of the ground creates a surface-based temperature inversion: cold air near the ground, warmer air aloft. This refracts sound downward, creating an acoustic waveguide along the ground. Under strong inversion conditions the effective spreading law shifts from spherical ( \(r^{-2}\) intensity) toward cylindrical ( \(r^{-1}\) intensity) beyond the duct-formation distance (~100–200 m), which extends audible range by 20–40 dB at multi-kilometre distances. This is likely why giraffes preferentially vocalise at night and why von Muggenthaler 2015 found 80 % of events nocturnal.

Giraffes are not alone. Many large, open-country mammals have evolved infrasonic communication, all for the same physical reasons:

• African elephant: 14–35 Hz rumbles, documented by Payne (1998); detected by females up to 10 km from males.
• Blue whale: 10–40 Hz calls, documented propagation across ocean basins via SOFAR channel.
• White rhinoceros: 10–50 Hz contact calls, von Muggenthaler 2001.
• Cassowary: 23–32 Hz booms, detected in dense rainforest at hundreds of metres.
• Giraffe: 7–15 Hz humming, von Muggenthaler 2015—lowest documented fundamental of the terrestrial megafauna.

The general pattern across taxa is: larger body \(\to\) longer vocal tract \(\to\) lower resonance frequency \(\to\) lower atmospheric absorption \(\to\) greater effective range. For savanna-dwelling megafauna facing long distances to conspecifics, natural selection has evolved ever-lower fundamentals.

Bioacoustic convergence

The 10–40 Hz “savanna channel” is populated by evolutionarily distant taxa that share none of the ancestral anatomy of vocalisation. Elephants have tongue-driven vocal folds, giraffes have conventional mammalian laryngeal folds, cassowaries have a syrinx. The common factor is physics: below 50 Hz absorption is low and propagation is long-range. Each lineage has independently converged on the same communication channel.

Bioacoustic monitoring

Modern conservation researchers deploy remote infrasonic microphone arrays to passively estimate giraffe populations and movement by triangulating low-frequency calls. The same technology is used for anti-poaching detection: gunshots have characteristic infrasonic signatures that propagate kilometres, and localised discrepancies between giraffe activity and baseline vocal rate can flag poaching-disturbed areas.

The RLN is one of several famous “design flaws” that demonstrate the non-teleological character of evolution:

• Inverted vertebrate retina (photoreceptors behind retinal neurons, blood vessels in front of photoreceptive layer).
• Crossed vas deferens in mammals, a vestige of testicular descent from abdomen.
• Recurrent laryngeal nerve: 4.6 m for a 4 cm function in the giraffe.
• Appendix: a vestigial cecal sac with some residual immune function but net surgical risk.
• Human lower back: lumbar lordosis imposed by quadruped-to-biped transition.

Developmental entrenchment

The RLN’s “bad design” is a specific case of developmental entrenchment: ancestral developmental pathways that become impossible to rewire because many downstream structures now depend on their precise order of operations. In the case of the RLN, the nerve forms before the heart has descended; its path is then physically and topologically constrained; and no incremental mutation can shortcut the loop without disrupting cardiovascular-laryngeal tissue-boundary formation in the embryo. The 4.6 m cost becomes invisible to selection because it is a small fraction of total daily metabolic cost, and any mutation that would shorten the loop would lethally disrupt embryogenesis.

The RLN is thus a fossil in the body: evidence in each individual of the descent of the modern tetrapod from a Devonian fish ancestor with no neck at all. No matter how tall giraffes become, they carry this piece of fish architecture with them.