Wind-evoked evasive responses in flying cockroaches

1. A standing cockroach (Periplaneta americana) responds to the air displacement made by an approaching predator, by turning away and running. The wind receptors on the cerci, two posterior sensory appendages, excite a group of ventral giant interneurons that mediate this response. While flying, these interneurons remain silent, owing to strong inhibition; however, the dorsal giant interneurons respond strongly to wind. Using behavioral and electromyographic analysis, we sought to determine whether flying cockroaches also turn away from air displacement like that produced by an approaching flying predator; and if so, whether the cerci and dorsal giant interneurons mediate this response.
2. When presented with a wind puff from the side, a flying cockroach carries out a variety of maneuvers that would cause a rapid turn away and perhaps a dive. These are not evoked if the cerci are ablated (Figs. 4, 5, 6).
3. This evasive response appears to be mediated by a circuit separate from that mediating escape when the cockroach is standing (Fig. 7).
4. The dorsal giant interneurons respond during flight in a directional manner that is suited to mediate this behavior (Fig. 8).
5. Recordings of the wind produced by a moving model predator (Fig. 9), together with measurements of the behavioral latency of tethered cockroaches, suggest that the evasive response would begin just milliseconds before a predator actually arrives. However, as explained in the Discussion section, under natural conditions, the evasive response may well begin earlier, and could indeed be useful in escaping from predators.
6. If cockroaches had a wind-mediated yaw-correcting behavior, as locusts have, this could conflict with the wind-evoked escape. In fact, cockroaches show the opposite, yaw-enhancing response, mediated by the cerci, that does not present a conflict with escape (Figs. 10–14).

     

Physiological and pharmacological properties of responses to GABA and ACh by abdominal motor neurons in Manduca sexta

1. GABA, ACh, and other agents were applied by pressure ejection to the neuropil of the third abdominal ganglion in the isolated nerve cord of Manduca sexta. Intersegmental muscle motor neurons with dendritic arborizations in the same hemiganglion were inhibited by GABA (Fig. 2) and excited by ACh (Fig. 5).
2. Picrotoxin was a potent antagonist of GABA (Fig. 4A). Bicuculline reduced GABA responses in some motor neurons (Fig. 4C), but had no effect on many other motor neurons. Curare reduced ACh responses (Fig. 6A). Bicuculline was an effective ACh antagonist in most motor neurons tested (Fig. 6B).
3. Motor neurons with dendrites across the ganglion from the ejection pipette exhibited different responses to GABA and ACh. Contralateral motor neurons often showed smaller, delayed hyperpolarizing GABA responses (Fig. 7). On two occasions, contralateral motor neurons had excitatory responses (Fig. 8). Contralateral motor neurons were hyperpolarized by ACh (Fig. 9). The inhibitory responses had only slightly longer latencies than ipsilateral excitatory ACh responses (Fig. 10A). The contralateral inhibitory ACh responses, but not the ipsilateral excitatory ACh responses, were eliminated by TTX (Fig. 10B).
4. A model, which includes inhibitory interneurons that cross the ganglionic midline to inhibit their contralateral homologs and motor neurons (Fig. 11), is proposed to account for contralateral responses to GABA and ACh and antagonistic patterns of activity of motor neurons during mechanosensory reflex responses.

     

Response characteristics of wide-field non-habituating non-directional motion detecting neurons in the optic lobe of the locust,Locusta migratoria

1. Extracellular recordings from wide-field nonhabituating non-directional (ND) motion detecting neurons in the second optic chiasma of the locust Locusta migratoria are presented. The responses to various types of stepwise moving spot and bar stimuli were monitored (Fig. 1)
2. Stepwise motion in all directions elicited bursts of spikes. The response is inhibited at stimulus velocities above 5°/s. At velocities above 10°/s the ND neurons are slightly more sensitive to motion in the horizontal direction than to motion in the vertical direction (Fig. 2). The ND cells have a preference for small moving stimuli (Fig. 3).
3. The motion response has two peaks. The latency of the second peak depends on stimulus size and stimulus velocity. Increasing the height from 0.1 to 23.5° of a 5°/s moving bar results in a lowering of this latency time from 176 to 130 ms (Fig. 4). When the velocity from a single 0.1° spot is increased from 1 to 16°/s, the latency decreases from 282 to 180 ms (Figs. 5–6).
4. A change-of-direction sensitivity is displayed. Stepwise motion in one particular direction produces a continuous burst of spike discharges. Reversal or change in direction leads to an inhibition of the response (Fig. 7).
5. It shows that non-directional motion perception of the wide-field ND cells can simply be explained by combining self-and lateral inhibition.

     

Time course of the Houseflies' landing response

The landing response of stationary flying houseflies Musca domestica has been recorded on video tape. The leg movements were quantitatively evaluated. It could be demonstrated that:

1. only the first two pairs of legs are involved in the reaction (Fig. 1). Prothoracic tarsi are lifted beyond the head, mesothoracic tarsi are lowered and moved sidewards (Fig. 2a and b).
2. the movement of the tarsal tips is mainly due to an opening of one single joint per leg, i.e. the femurtibia joint of the prothoracic leg (Fig. 2c), and the coxa-femur joint of the mesothoracic leg.
3. the landing reaction is a fixed action pattern which does not seem to require further sensory input once it is released (Fig. 4d).
4. the landing responses to a light-off stimulus and to expanding patterns with different angular velocities are indistinguishable (compare Fig. 3a-c with Fig. 2a-c). The only parameter that is obviously dependent on the stimulus conditions, is the latency of the reaction (Fig. 4a-c).

     

Discrimination of phase spectra in complex sounds by the bullfrog (Rana catesbeiana)

1. Male bullfrogs at two different natural calling sites were presented with playbacks of synthetic advertisement calls differing in phase spectra. Sounds were presented in a ABA design to analyze the ability of the animals to perceive changes in repeated series of stimuli.
2. The number of individual croaks in an answering call significantly increased over repeated presentations of two of the three stimulus phase types in condition A1. There were significantly fewer croaks to the third stimulus. These data suggest that two stimuli were perceived in a similar manner.
3. Latency of calling to stimuli presented in conditions A and B changed in response to shifts in phase spectrum at a low density calling site. These differences were significant when comparing latency to playbacks where shifts in the phase spectrum changed the temporal fine-structure and waveform periodicity of the stimulus.
4. The increase in number of croaks and decrease in response latency across condition A1 and the increase in latency in condition B suggest that discrimination may take the form of stimulus-specific sensitization. In this context, sensitization might reflect an increase in arousal due to repeated presentation of a salient stimulus.
5. The operation of a hypothetical ‘mating call detector,’ based on linear summation of temporal responses from the eighth nerve, provides output similar to the behavioral results.

     

How lesioning the nucleus praeeminentialis affects electrolocation behavior in the weakly electric fish, Apteronotus leptorhynchus

The ability to orient to and track moving electrolocation targets was assessed in normal Apteronotus leptorhynchus and in those with unilateral lesions of the nucleus praeeminentialis dorsalis.

1. Each fish was trained to hover between two vertical metal rods and track their movement. Two aspects of this behavior were measured: a) the hovering position of the fish relative to stationary rods; b) the latency between the onset of rod motion and the fish's tracking response. Control fish hovered midway between stationary rods, while lesioned fish hovered closer to the rod ipsilateral to the lesion. Response latency varied negatively with rod diameter in both sets of fish, and lesioned fish exhibited shorter latencies than control fish. While the response latencies of control fish were shortest when their starting position was midway between the rods, lesioned animals' latencies were shortest when they hovered closer to the rod ipsilateral to their lesion.
2. Control fish responded to the approach of a single metal ball to either side of the body with nearly equal latencies and fish-to-object distances. After lesioning, response latency increased and fish-to-object distance decreased for approaches to the side ipsilateral to the lesion; opposite changes occurred for contralateral approaches.

     

Polarization sensitivity in crayfish lamina monopolar neurons

1. Polarization sensitivity (PS) was examined in photoreceptors and lamina monopolar cells (LMCs) in two species of crayfish, Procambarus clarkii and Pacifasticus leniusculus. The measurements were made with intracellular recordings and broad field illumination.
2. PS is about 40% greater in Pacifasticus than in Procambarus (Table 1). In both species the LMC stationary PS profiles (estimated with flashes) are similar to those of receptors (Figs. 1 and 2). Both receptor and LMC sensitivity profiles are well described by cos² θ functions (Fig. 3). PS was observed in all receptors and 78% of LMCs.
3. When stimulated with a rotating polarizer, receptors and LMCs exhibit membrane potential modulation with phase predicted by the stationary PS profile (Fig. 5). In photoreceptors, the polarization-elicited percent modulation falls off steeply as intensity increases. The LMC modulation is stronger than that in receptors and relatively insensitive to the mean intensity (Figs. 6 to 8). For low intensities the LMC modulation is 100%. The LMC dynamic behavior is consistent with either an opponency mechanism or strong but polarization-insensitive lateral inhibition.
4. Receptors and LMCs exhibit steady-state differential sensitivity to stationary e-vector orientation (Fig. 9).
5. About 10% of the LMC neurons exhibit PS maxima separated by 90°. These results imply a nonlinear summation of signals from orthogonal receptor channels (Fig. 10).

     

Encoding of plant odours by receptor neurons in the pine weevil (Hylobius abietis) studied by linked gas chromatography-electrophysiology

Receptor neuron responses to plant volatiles, trapped by head-space procedures, were examined in the pine weevil Hylobius abietis, using gas chromatography linked with electrophysiological recordings from single neurons. Seventy-two receptor neurons were tested 173 times for various plant volatile mixtures, either via a polar or a non-polar column.

1. All responses appeared as increased firing rates which followed the concentration profiles of the GC-eluted compounds.
2. The neurons were classified separately for the two column types in 17 and 19 groups respectively, according to the compounds they responded to. It suggests that the plant odour information is encoded by a large, but limited number of receptor neuron types.
3. Most neurons responded to a limited number of compounds (1–5) and showed a marked best response to one of them, whereas additional responses to several other components which seems to be structurally similar, was recorded for some neurons. It suggests that the plant odour receptor neurons are rather narrowly than broadly tuned, and that each neuron is specialized for receiving information about one or a few related compounds.
4. Most neurons responded to monoterpenes, whereas the other neurons responded to compounds of other categories.
5. Both major and minor plant volatile components activated specifically receptor neurons.

     

Modulation of swimming in Tritonia: excitatory and inhibitory effects of serotonin

1. In the mollusc Tritonia escape swimming is produced by a network of central pattern generator (CPG) neurons. The purpose of this study was to determine which neurotransmitters might be involved in the swim system.
2. Injection of serotonin (5HT) into whole animals elicited swimming followed by a long-lasting inhibition of swimming. In isolated brain preparations, bath-applied 5HT elicited a swim pattern at short latency and also caused a long-lasting inhibition of the swim pattern. The activation of swimming by 5HT was associated with a tonic depolarization of cerebral cell 2 (C2) and the dorsal swim interneurons (DSI) which form part of the swim CPG network.
3. In isolated brain preparations, bath applied glycine, histamine, proctolin, and FMFRamide had no effect on the swim motor pattern elicited by electrical stimulation of a peripheral nerve. Aspartate, carbacol, dopamine, glutamate, octopamine, pilocarpine, and small cardioactive peptide-B (SCP_B) inhibited the activation of swimming by nerve stimulation.
4. The 5HT antagonists cyproheptidine, tryptamine, and 7-methyltryptamine had no effect on swimming, but methysergide and fenfluramine inhibited swimming to both normal sensory stimuli and exogenously applied 5HT.
5. Staining with a polyclonal antibody indicated that one class of CPG neurons, the dorsal swim interneurons (DSI), was immunoreactive for 5HT.
6. Taken together, the data suggest that pattern generator interneurons, particularly the DSIs, use 5HT as a neurotransmitter.

     

Morphological and physiological properties of pheromone-triggered flipflopping descending interneurons of the male silkworm moth,Bombyx mori

1. The morphology of descending interneurons (DNs) which have arborizations in the lateral accessory lobe (LAL) of the protocerebrum, the higher order olfactory center, and have an axon in the ventral nerve cord (VNC), were characterized in the male silkworm moth, Bombyx mori.
2. Two clusters (group I, group II) of DNs which have arborizations mainly in the LALs were morphologically characterized. The axons of these DNs are restricted to the dorsal part of the each connective (Figs. 1–5).
3. Pheromonal responses of the group I and group II DNs were characterized. Flipflopping activity patterns, which have two distinct firing frequencies (high and low) in response to sequential pheromonal stimulation, were usually recorded (Figs.6–10).
4. Two types of flipflopping activity patterns were classified into those that had an antiphasic relationship (called the ‘FF’ type) between the left and right connectives and those with a synchronized relationship (‘ff’ type) (Figs. 8–12). We propose that some group II DNs show ‘FF’ flipflopping activity patterns (Fig. 10).
5. A state transition was usually elicited by less than 10 ng bombykol, the principal pheromone component. Extra impulses were elicited during constant light stimulation (Fig. 9).
6. Our results suggest that the LAL olfactory pathways might be important for producing flipflopping activity patterns (Fig. 11).

     

Effect of external factors on phototaxis of Chlamydomonas reinhardtii

1. A fully automated phototaxis monitoring device is described for measuring photo-topatactic responses of flagellated organisms.
2. Photokinesis can be demonstrated in Chlamydomonas cells only after a dark period of about 72 hrs.
3. Pre-darkening of a few hours duration raises the phototactic disposition, whereas pre-illumination has no significant effect.
4. Circadian rhythms can be initiated by only one period of darkness or lower light intensity, whereas a period of higher intensity does not induce rhythms. The period length of the circadian rhythms is about 24 hrs.

     

Electric organ discharge frequency modulation evoked by water vibration in Gymnotus carapo

• 1.1. The electric organ discharge (EOD) frequency modulations evoked by brief water vibration were analysed in the pulse-type fish Gymnotus carapo.
• 2.2. The response consisted of a transient increase of the EOD frequency at short latency (30 msec). Response profiles were characteristic of the specimen and relatively independent on stimulus intensity.
• 3.3. Conversely, they were dependent on stimulation sequence, showing a rapid decrement along successive stimuli and high temporal discrimination.
• 4.4. The brief latencies indicate a relatively simple neural circuit.
• 5.5. The response may be an electrolocation enhancement strategy for the detection of moving objects based on “sampling” the periphery at a higher frequency.

     

The oxygen consumption of mayfly (Ephemeroptera) and stonefly (Plecoptera) larvae at different oxygen concentration

1. The aim of this investigation was to elucidate how four acquatic insect larvae, from different habitats and having different respiratory organs or types of respiratory regulation, react to a lowered oxygen concentration, and how their oxygen consumption is affected. The species investigated were the stoneflies Taeniopteryx nebulosa, Diura nanseni and Nemoura cincerea and the mayfly Cloëon dipterum.
2. The measurements were performed in a respiratory apparatus of open, flowing-water type. Its design is shown in Fig. 1. Water of known oxygen concentration was allowed to flow past the experimental larvae. The oxygen consumption of the larvae was calculated from the lowering of the oxygen concentration in which ensued.
3. The water used in the experiments was standardized, so that the electrode had the necessary stability (conductivity 470 micromhos/cm). The calcium ion was excluded in order to prevent the precipitation of CaCO₃ in the electrode capillary.
4. A large variation in the values of oxygen consumption was found as seen in Fig. 2–5. The reason for that is a corresponding variation in the motor activity of the experimental animals.
5. The physiological reasons for the general form of the curves A and C in Fig. 2–5 are discussed. The curves A and C represent oxygen consumption of the larvae at different degrees of stimulation, entailing different levels of motor activity. Curve A represents intentinally activated animals, curve C non-activated, motionless animals. The curves A and C are boundary curves corresponding to a sort of scope for activity of the animals. Over this scope area a series of curves of the same form could in principal be construed, representing different degrees of stimulation.
6. Within a certain oxygen concentration interval a motor activation was observed caused by a reduced oxygen concentration. The result of that activation can be seen in Fig. 2–5 as a zone with no or very few oxygen consumption values between curve C and D. The more easily activated the species is, the broader the zone will be. Cloën has the most narrow zone and was observed to be less activated than the other species.
7. Small larvae of Cloën (2–4 mm and 42–6 mm) and Nemoura (2–4 mm) showed clearly a greater ability to take up oxygen at low oxygen concentrations than full-grown larvae (see Fig. 8 and 9).
8. The critical point on the curve representing mean oxygen consumption as a function of oxygen concentrations was found to be at 2–5 mg O₂/1 for Taeniopteryx and Diura, at 2.2–2.5 mg O₂/1 for Cloëon, and at about 2–7 mg O₂/1 for Nemoura. The values refer to 8°. Cloëon is the species which is exposed to the greatest variations in oxygen concentration in its natural environment.
9. No influence on the oxygen consumption of starvation for 4 to 5 days was found. No difference between the oxygen consumption values obtained in the presence or in the absence of calcium ions could be observed during the experiments (Fig. 10, 11).
10. The basic picture obtained in this investigation is a set of oxygen consumption values scattered between a curve connecting highest values obtained and a curve of the standard metabolism together with a zone in which the larvae are activated by reduced oxygen concentrations. This picture is presumed to be general in aquatic animals with a well developed motor activity.

     

Basic processes of locomotor coordination in the rock lobster

Rock lobsters are able to perform long and stereotyped stepping sequences above a motor driven treadmill. Forward walking samples are estimated by mean of statistical methods to draw out the basic rules involved in the locomotor behaviour (Fig. 1).

- The spatial and temporal parameters defined in a single propulsive leg are either invariable in respect to the imposed speed, as the mean step length (L), the return stroke duration (Tr) and the pause times (T's, T'r), or speed dependent as the power stroke duration (Ts) and the whole period (Figs. 2 and 3).

- The interleg phase coupling is strong and stable in the ipsilateral rear pairs (4–5), these legs acting most of the time in absolute coordination (1:1) or in harmonic ratio (2:1). In the contralateral pairs (R4-L4, R5-L5) the legs roughly operate in antiphase, but the relationship appears much weaker and variable, with frequent episodes of relative coordination (Fig. 4).

- The time intervals between the ground contact of any leg and the swing initiation in the nearest ones appear somewhat constant and could be closely related to the mechanism of stepping synchronization. The “5 on - 4 off” delay, very stable and always positive, suggests that the rear legs could exert a predominant influence upon the rhythmical movements of the next anterior ipsilateral appendages (Fig. 5).

- To test the contralateral relationships, the treadmill belts can be decoupled in order to impose different walking speeds on each side. Such a conflicting stimulus reveals that:

1. The relative hierarchy always observed between the ipsilateral legs can be artificially created between the two sides (Fig. 6).
2. The driving influence of a given leg is closely linked to the intensity of EMG's discharges in its power stroke muscles.
3. The contralateral appendages are able to walk in absolute coordination despite a large speed difference between the two sides (up to 4 cm/s). Under such a constraint, the walking legs alter its invariable parameters (L and Tr) to reach a common step period and steadily maintain the alternating pattern (Figs. 6 and 7).

     

Patterns of monosynaptic input to the giant interneurons 1–3 in the cereal system of the adult cockroach

1. The cerci of the cockroach Periplaneta americana bear filiform hair mechanoreceptors that are arranged in segmentally repeated rows and longitudinal columns. The monosynaptic connections between receptors of the same column or row and the 3 largest giant interneurons (GIs) were compared using the oil-gap single fibre technique.
2. For many columns, the synaptic efficacy of the afferents decreased gradually from the base to the tip of the cercus, but columns with an inverted gradient or without any gradient were also observed. On the ipsilateral side (relative to the GI axon), the inverted gradients were exclusively found for columns with short proximal hairs. For one column (d) and GI3, the ipsilateral and contralateral gradients were opposite.
3. Monosynaptic EPSPs evoked by stimulating different receptors of the same segment (segment 3) were of very different amplitudes, which partially account for the directional sensitivity of the GIs. Differences in the location, shape and size of the afferent terminals were not sufficient to explain these differences in connection strength.
4. No correlation was found between the size of the EPSPs produced by a sensory neuron and the length of its associated hair.

     

‘Absolute steepness’ of ramps as an essential cue for auditory pattern recognition by a grasshopper (Orthoptera; Acrididae; Chorthippus biguttulus L.)

Male grasshoppers of the species Chorthippus biguttulus react to female songs with a characteristic turn towards the female. The probability of turning towards female song models was used to evaluate those parameters which are essential for a signal to be interpreted as female song.

1. The shape of sound pulses turned out to be most decisive; pulses with ramps rising gradually over 3 and more ms were efficient (Figs. 2, 3), whereas rectangularly modulated pulses evoked only weak responses and only when pulse intervals were between 3 and 5 ms (Fig. 2). The decline of a pulse did not influence its efficiency (Fig. 3). In particular, pulses with sudden onsets and gradual declines were as weakly effective as rectangularly modulated ones and thus remarkably less effective than pulses with ramp-like onsets (Fig. 4).
2. Intensity tuning curves suggest, that the absolute steepness of ramps (expressed as μbar/ms) is detected by the grasshopper nervous system (Figs. 6, 7), possibly by processing the delay in excitation onset of at least two receptor types differing in threshold sensitivity.
3. The sawtooth shape of pulses in female signals is suggested to be adaptive with respect to directional hearing.

     

In vivo buccal nerve activity that distinguishes ingestion from rejection can be used to predict behavioral transitions in Aplysia

1. We are studying the neural basis of consummatory feeding behavior in Aplysia using intact, freely moving animals.
2. Video records show that the timing of radula closure during the radula protraction-retraction cycle constitutes a major difference between ingestion (biting or swallowing) and rejection. During ingestion, the radula is closed as it retracts. During rejection, the radula is closed as it protracts.
3. We observed two patterns of activity in nerves which are likely to mediate these radula movements. Patterns I and II are associated with ingestion and rejection, respectively, and are distinguished by the timing of radula nerve activity with respect to the onset of buccal nerve 2 activity.
4. The association of ingestion with pattern I is maintained when the animal feeds on a polyethylene tube, the same food substrate used to elicit rejection responses. Under these conditions, pattern I is associated with either swallowing or no net tube movement.
5. Most transitions from swallowing to rejection were preceded by one or more occurrences of pattern I in which there was no net tube movement, suggesting that these transitions can be predicted.
6. Our data suggest that these two patterns can be used to distinguish ingestion from rejection.

     

Ontogeny of positive phonotaxis in female crickets,Gryllus bimaculatus De Geer: dynamics of sensitivity,frequency-intensity domain,and selectivity to temporal pattern of the male calling song

1. The ontogeny of positive phono taxis (PPT) in female crickets, Gryllus bimaculatus was followed in tethered flight. During the first day of adult life many females already demonstrated PPT to the calling song (CS) of conspecific males. The average threshold of PPT at 5 kHz, the dominant frequency of the CS, decreased by 30 dB by the time of sexual maturity (Fig. 1).
2. No correlates of this decrease were found in the activity of the most sensitive ascending prothoracic neuron tuned to 5 kHz recorded in the neck connective. This is presumably the AN1 neuron which is known to be involved in PPT realization. Its threshold at 5 kHz in young animals was the same as in adults. Therefore, ascending circuits of PPT seem to be mature by the first day of imago life and there should be some other mechanisms preventing performance of PPT by young walking females until maturation.
3. The PPT of females in flight is tuned to 5 kHz, much sharper than in walking (Fig. 2). In flight, the carrier frequency of a signal is probably an important parameter driving PPT, at least in a no-choice situation, whereas on the substrate, at close range, temporal parameters become decisive.
4. The ontogenetic development of the selectivity of a female's PPT to temporal parameters of a signal passes 3 successive steps: 1) response mainly to the trill with pulse repetition rate as in the CS; 2) response mainly to the actual CS with chirp structure; 3) destruction of selectivity (Figs. 3–6). The existence of steps 1 and 2 strengthens our hypothesis, that in phylogeny, the trill (pulse rate) detector of the CS “recognizer” in the CNS appeared earlier, and was later accompanied by the chirp detector.
5. Joint breeding of female larvae with males accelerates maturation of the CS recognizer.

     

Oxygen consumption and respiratory behaviour of three Nile fishes

1. The respiratory behaviour and patterns of oxygen consumption of three Nile species have been investigated.
2. Tilapia nilotica showed a typical pattern of oxygen consumption with an ambient region, adaptive plateau and lethal region (Fig. 2).
3. Specimens of Polypterus senegalus and Clarias lazera (body weights 20–30 and 30–45 g respectively) showed patterns of consumption comparable to that of Tilapia (Fig. 3a and 4a). In larger specimens of the two species the adaptive plateau was either insignificant or completely absent.
4. Specimens of Polypterus and Clarias (20–30 g and 30–45 g respectively) could survive in waters saturated with oxygen (7.4 mg/l) but their tolerance to lower oxygen concentrations was limited. Larger specimens of Polypterus and Clarias failed to survive in oxygen saturated waters.
5. The tolerance of Tilapia nilotica to extremely low oxygen concentration is an adaptation of a tropical and completely aquatic species. Polypterus and Clarias resort to their compensatory mechanisms only when the aquatic respiratory surface fails to satisfy their oxygen requirements.