Learning To Blog & Blogging To Learn

Blog With Me?

Month: November, 2012

The previous blog introduced us to the difficulties penguins face when trying to find their mates amongst their crowded colony.  As a reminder:

A bird returning from the ocean [. . .] goes back to the breeding area and then calls its mate using the mutual display call.  The partner, incubating the egg or rearing the chick, responds and thus gives its identity and its exact position in the colony.  After a few calls, the two individuals are able to find each other. [. . .] Colonial life requires that vocal recognition occurs in the continuous background noise of the colony (Lengagne et al., 1999).

The above research was conducted during dry, calm weather.  However, the natural conditions are hardly so.  Instead, penguins spend much of their lives trying to survive in windy, arctic conditions and are continuously subjected to the influence of strong polar wind streams.  Lengagne, Aubin, Lauga, and Jouventin have continued their research to look at if penguins have adopted some sort of adaptation to account for varying weather conditions, and if so, how they accomplish such a feat.

They propose that penguins take advantage of the mathematical theory of communication as follows.  The volume of information (V) contained in a produced signal is equal to the signals frequency (F), times the signal duration (T), times the log of one plus signal to noise ratio (g=log2(1+S/N)) such that V=FTg.  Because windy conditions decrease the amount of volume information transmitted by decreasing the signal to noise ratio, the producing signal must be modified to counteract such degradation.  It is assumed that penguins are already emitting signals at their maximum amplitude and therefore cannot directly increase their signal to noise ratio.  Instead, it is hypothesized that penguins increase their signal duration (T) in two ways, “birds can extend the duration of the mutual display call or enhance the emission rate of the call” (Lengagne et al., 1999).

To conduct such a study, Lengagne et al. observed 30 mating pairs amongst a colony consisting of 40,000 penguin pairs.

First, Lengagne et al. “analyzed the modification of the spectral composition of the ambient noise of the colony versus wind speed” through three frequency bands:

0-350 Hz corresponded to physical noise (such as wind),

350-2000 Hz corresponded to calls of birds,

2000-8000 Hz corresponded to the remaining noise of the colony (Lengagne et al., 1999).

They measured the percentage of the total energy in each category (taken from the middle of the colony) via Welch’s method of 80-second recordings.  The independent variable was presence of wind such that analyses were made in conditions with no wind and conditions with a wind speed of 11 meters per second.

Secondly, the entropy of broadcast calls was computed in the following three conditions.  Control conditions were taken as trails where wind speeds were 5 meters per second or less.  Accelerated wind conditions were taken as trials where wind speeds were 11 meters per second.  These high-speed conditions were studied in two ways: with the direction of the wind favorable to the transmittance of calls or with the direction of the wind opposing the transmittance of the calls.  In this part of the experiment, Lengagne et al. were interested in the “emergence of the penguin’s signal over the background noise (colony+wind) [. . .] measured by computing the entropy of the distribution of its energy” via a calculation proposed by Beecher in 1988 (Lengagne et al., 1999).

Thirdly, Lengagne et al. measured the number of syllables and call durations in six different categories: 5, 6, 7, 8, 9, and 11 meters per second, all at a direction opposing propagation.

Fourthly, the number of calls emitted by the returning penguins was observed in the same categories discussed in the second part of their experiment.

And lastly, Lengagne et al. used the time duration required for change-over as a means to assess the wind-cost for penguins.

Quantitatively, they found that for windy conditions, 23.8% additional energy corresponded to physical noises such as wind (which intuitively makes sense) and the distribution of energy corresponding to the calls of penguins and noises from the colony decreased.

From the second part of their experiment, Lengagne et al. found that the aforementioned observations had correlations with the amount of entropy of the broadcast calls.  Downwind conditions are correlated to wind direction favorable to the signal propagation while upwind conditions are correlated to wind direction against signal propagation.  As noted in the caption, “a value near 1 characterize[d] a signal almost lost in the background noise” (Lengagne et al., 1999).  It can be seen that in windy situations with the wind going against signal propagation, most of the signal is lost due to the background noise (again, agreeing with intuition).

When looking at the number of syllables and call duration, the “two separate models are in surprisingly good agreement with the values of their respective wind speed threshold, W,” duration of call being 7.70 meters per second and 7.52 meters per second for number of syllables (Lengagne et al., 1999).  Above those two thresholds, the data fits a linear graph with positive slopes as follows:

In the fourth part of their research, they found that in the upwind condition, the number of calls emitted by both the returning and the brooding penguins were 11.6 calls, which exceeded that of the downwind condition (7.7 calls) and that of lack of wind (5.3 calls) with statistical significance.

To quantitatively show the cost of windy environments, Lengagne et al. showed that when graphing the duration of change-over versus wind speed, the difference between the slopes of the regression lines were statistically significant “showing that the time necessary for change-over was more important in windy situations than without wind” (Lengagne et al., 1999).

The data found by Lengagne et al. shows that the modality of emission of penguin calls changes as wind speed increases.  Wind increases the environmental ambient noise which affects both the amount of information transmitted and the distance over which the information is transmitted such that an increase in background noise leads to a diminution of the signal to noise ratio.  Sometimes, the “attenuation becomes so strong, the signal disappears” (Lengagne et al., 1999).  In order to account for such attenuation and increase the success of their signals, penguins increase their call duration and enhance the number of syllables within each call.  Overall, the total number of calls between the mates increases as well.  These results are linked to an increase in the “redundancy process” where, “by repeating the same information many times, the birds may increase the probability of communicating during a short-window during which the wind speed suddenly drops” (Lengagne et al., 1999).  In other words, an increased redundancy in penguin signaling increases the probability of receiving the intended message.  Such an acoustic adaption supports the finding that the animal adjusts its behavior in response to wind noise, at least in the temporary sense.  The penguin “adapts in some manner to wind speed and wind direction or noise generated by the wind or indirectly some other parameters linked to the wind modifying the number of syllables that must be emitted” (Lengagne et al., 1999).  It is fascinating to see how communication is not static, but a dynamic process that take into consideration many inputs at the current situation.  Communication, therefore, is not just an output system that is straight forward.  It’s sensitive to many variables such that the efficiency of communication can be enhanced.

 

Lengagne, T., Aubin, T., Lauga, J., & Jouventin, P. (1999). How do king penguins (Aptenodytes patagonicus apply the mathematical theory of information to communicate in windy conditions?. Proceedings of the Royal Society of London. Series B: Biological Sciences266(1429), 1623-1628.

As the holiday season rolls in full swing and the festivities begin, Christmas decorations become delightfully impossible to avoid.  Snowmen, gingerbread families, doves, and penguins seem to pop-up everywhere amongst decorated Christmas trees, wreaths, and lights.  To keep with the spirit of the season, my blog this week will revolve around those mini-tuxedo dawning creatures, the penguins.

During three months out of the year, king penguins alternate care duties on land with foraging trips on the sea between mates.  However, king penguins live in extremely large colonies consisting of anywhere from a few hundred to 500,000 mating pairs, which poses a major problem.  Upon returning from sea, finding a single penguin in such a massively dense crowd, about 2.2 breeders per square meter, seems nearly impossible.  To make matters more difficult, king penguins lack a nest.  Instead, the mate on land incubates the egg on his feet, allowing him to move within the colony during storms or disputes with neighbors.  The resulting “short distance movement [. . .] creates an important problem in relocating one’s mate” (Lengagne et al., 1999).  The work of Lengagne, Jouventin, and Aubin aimed to study how penguins are able to handle such a problem and have found that the “difficulty in relocating mates posed by wandering incubation has been partially solved by the use of acoustic signals” (Lengagne et al., 1999).  Acoustic calls are first produced by the returning penguin as he aims to identify his mate.  His mate calls in reply, providing more information on her location within the colony.  They repeat this process until they are reunited.  In addition to the aforementioned, each pair faces difficulties due to their environment.  Not only do windy and snowy conditions drown out and degrade the sound of their calls, but the background noises due to conspecific calls with similar temporal and spectral characteristics from other calling penguins generate an extreme jamming effect.

Lengagne et al. observed 28 pairs marked for individual identification from the laying of their egg to the end of the brooding stage (when both parents leave the chick to forage for food).  The calls of each individual of interest were recorded.  While one of the penguins were out foraging for food, the calls of that absent penguin were played to its mate at a distance of 20, 15, 14, 13, 12, 11, 10, 9, 8, and 7 meters away in a randomized fashion.  It is important to note that the calls were played back in the morning during calm and dry weather.  They then noted the response of the penguin and categorized them either as positive (where the penguin calls in reply to the broadcast signal of its absent mate, interpreted as recognition of its mate’s signal) or negative (where there is no vocal response and the penguin fails to reply, interpreted as failure to discriminate the broadcast signal of its absent mate).  They also observed what they termed the Distance of First Emission, or the distance between the two penguins at the beginning of the acoustic search when the first call by the returning penguin was made.  They kept track of the Number of Display Calls emitted by the Arriving and Incubating penguins, termed NCA and NCI, in addition to the Time Delay, or the time taken between the first call of the arriving penguin and the moment of unification.

Lengagne et al. found that 71% of the incubating or brooding penguins moved, on average, a distance of 4.4 meters while their mate was away foraging (total, combining the results from both the incubating and brooding stages).  Transforming this linear distance into the maximum radius of circular areas produced an area of presence of 65.7 meters squared.  Within this area, the number of mating pairs was estimated to about 145.

Image

They also found the average discrimination distance to be 8.8 meters such that at 8.8 meters, the penguins called back in response to their returning mate indicating that there was recognition of the initial call.

Image

Image

The average distance between the two penguins when the returning penguin made his first call was 8.3 meters.

Lengagne et al. found that there were strong correlations between Time Delay and the DFE, between the NCA and DFE, and the NCA and Time Delay.

ImageImageImage

The data showed that the returning penguins called an average of 5 times over an average time of 114 seconds to reunite with its partner, and 70.17% of the time, its partner was able to discriminate the very first call of the returning penguin.

Lengagne et al. discuss how the efficiency of these uniting calls is accomplished.  The returning penguin “progresses silently towards a preferred area of the colony, named the attachment zone, probably using topographic cues” then starts its acoustic search (Lengagne et al., 1999).  However, because the individually specific calls propagates among the bodies and background noise of thousands of penguins, the broadcast distance is reduced, the signal degrades, and masking effects alter the frequency and temporal domains of the calls, which all ultimately impairs the communication process.  However, the fact that the majority of the incubating king penguins discriminate their incoming mates at their first emission implies that acoustic communication is a particularly efficient strategy in this species” (Lengagne et al., 1999).

One thing to bring up, though, is the distinction limitation of observing performance.  Just because the penguin does not respond to the returning bird (negative response) does not mean that it does not recognize its partners call.  Maybe the penguins hear their mates’ calls, however it takes energy to call back out to them (they don’t have much energy taken that they haven’t eaten for so long while their mates are out obtaining food).  Therefore, even if they hear that their mate is far away, they could wait a little longer until they know their mate is close enough so it is more likely they will be reunited.  It could be argued that this increases efficiency and energy utilization.  In addition, there is a difference between hearing their mates call as opposed to recognizing it.  The distance of 8.8 meters relates to a clear, calm, and dry environment where the main variable would be recognition.  However, if the environment were snowy or windy, it becomes an issue of whether or not the penguins can even hear each other in the first place.

Another thing of discussion revolves around their methods.  At what loudness did they play back the calls?  Although they played the calls at set distances, maybe the loudness they played it out actually corresponded to a different distance.  For example, if they played the recorded call back too loudly, having the speaker 11 meters away could actually be interpreted as the penguin being only 9 meters away.

Lengagne et al. looked at how far away the returning bird is from the brooding penguin once the brooding penguin calls back, indicating recognition by the brooding penguin.  However, I am curious to know if this is the same distance required for the returning bird to recognize the calls of the brooding bird.  Maybe there is something about moving around (by the returning penguin) that makes it different when compared to staying in one general location (the brooding penguin) in regards to hearing and recognizing calls.  When all is said and done, however, this is still an amazing feat penguins accomplish each year.

To obtain a better understanding of the difficult situation penguins are faced with, please see the following video at 1:00 (1:00 on will show you how noisy and crowded the colonies are):

Attached are some pictures illustrating just how densely populated the penguin colonies are.  The difficulties of finding ones mate can further be appreciated.

Lengagne, T., Jouventin, P., & Aubin, T. (1999). Finding one’s mate in a king penguin colony: efficiency of acoustic communication. Behaviour, 833-846.

Like that of monkeys, the brain of another animal is thought to be comparable to the brain of humans.  Dolphins are commonly labeled as “smart,” and their levels of intelligence have made them a popular animal of study.  V. Janik, L. Sayigh, and R. Wells followed this trend and conducted their research on the identity calls of bottlenose dolphins.   Identity information, for almost all animal calls, give specific details of species, population, group, family line, and even in some cases individual identity (Janik et al. 8293).  However, dolphins use distinctive calls unique on the level of the individual.  Individual identity information requires a tremendous amount of interindividual variability and is hypothesized to be encoded by the “pattern of frequency modulation over time that gives a spectrographic contour its distinctive shape” (Janik et al. 8293).  Janik et al. however addressed the problem of whether this type of individual discrimination occurs dependently of voice features.  For example, human naming is accomplished dependently of voice features.  You could recognize your name if you heard it in your own voice, in your friend’s voice, or in a stranger’s voice.  We recognize that identity call based on the name itself and not on the voice.  Dolphins can use distinctive call types as descriptive labels in referential communication, however, the research of Janik et al. aimed to see how dolphins do so.

Janik et al. produced “synthetic whistles that had the same frequency modulation but none of the voice features of known signature whistles” in the following manner (Janik et al. 8293).  They recorded adult dolphins and transformed their calls into a synthetic form via SIGNAL 3.1, a program that essentially stripped down the whistles to its basic frequency modulation pattern.  Then, over four seasons between June 2002 and February 2005, they played back these calls and measured the target dolphins’ responses.  Each animal was held in a net by several people such that the dolphin’s head was free to move from side to side.  Fifteen minutes after the dolphin became acclimated to the aforementioned position, the speaker was introduced 2 meters to the side and the calls were played.  Each subject was exposed to a “sequence of synthetic whistles resembling the signature whistle of a related individual (as determined through long-term observations and confirmed through genetic testing) and a sequence of synthetic whistles resembling the signature whistle of an unrelated but known individual” (Janik et al. 3296).  To avoid confounds, the calls were matched between the related and non-related individuals on the levels of approximate age, sex, and time spent together.  Any head turn that was greater than 20 degrees was counted as a turn related to the call.  (Head turns less than 20 degrees were not counted since dolphins naturally move their heads back and forth within this range).

The results of Janik et al. demonstrated that “bottle nose dolphins from Sarasota can recognize structure whistles of individuals [. . .] [through] distinctive frequency modulation patterns [dependent of] voice features” (Janik et al. 8295).  Individual dolphins turned their head more towards the speaker if the playback was the synthetic signature whistle of a close relative rather than that of an unrelated individual.  The whistle modulation contour themselves therefore carry the identity information required.  They addressed the possibility that related dolphins had more similar whistle calls than unrelated dolphins as follows.  Janik et al. tested to see if the response strength of targeted dolphins responded related to stimuli significantly more or less similar to their own whistle, independent of kin relationship, and found no significant differences (Janik et al. 8295).  Therefore, the orienting responses of the dolphins were not due to similarities to their whistle.

As stated before, individual identity information is encoded independent of the signaler’s voice or location such that the frequency modulation pattern of signature whistles is sufficient for individual discrimination (Janik et al. 8295).  It does not mean, however, that dolphins do not have or benefit from voice features specific to each individual.  In their discussion, Janik et al. suggest that their findings support the notion that dolphins are recognizing whistles as opposed to merely discriminating between them.  They define recognition as “perceiving something to be identical with something previously known,” and discrimination as “the comparison of distinctive features that can use, but does not require, such previous knowledge” (Janik et al. 8295).  Therefore, this reveals the complicated abilities dolphins encompass.

I would like to just address one concern.  When they tested the dolphins, they were constraining the subject in an unnatural manner.  Dolphins are free-swimming animals, and I wonder if such methods caused any distress which could therefore alter any results.

In closing, the following is a pictorial representation of the dolphin whistles.  The left side shows the natural original whistles while the right side shows the corresponding synthetic versions.

 

Janik, V., Sayigh, L., & Wells, R. (2006) Signature whistle shape conveys identity information to bottlenose dolphins. Proceedings of the National Academy of Sciences, 103(21), 8293-8297.

According to Robert Seyfarth, Dorothy Cheney, and Peter Marler, the semantics of animal communication is a “central but neglected issue” (Seyfarth et al. 801). Therefore, they conducted research on predator classification to provide the necessary attention to such matters. Their focus was on the systematic use of distinguishing signals which represented a distinction of objects and a sorting of objects into groups. Essentially, they wanted to see how animas “categorize” objects in their external world, specifically speaking, how free-ranging vervet monkeys categorize predators. In order to do so, they studied three groups of vervet monkeys comprised of adult males, adult females, juveniles, and infants. Previous studies showed that vervets give “acoustically different alarm calls to at least three different predators: leopards, martial eagles, and pythons” in the following manner (Seyfarth et al. 802).

Leopard: short, tonal calls produced in a series on both exhalation and inhalation

Eagle: low-pitched, staccato grunts

Snake: high-pitched, stutters

These alarm calls were distinct from nonalarm vocalizations and associated with different types of vervet monkey responses. Alarm calls produced by separate vervet monkeys were recorded previously. Through a hidden speaker, Seyfarth et al. presented each type of alarm call to the groups of monkeys and observed their reactions. (By using a speaker, they eliminated the confound of monkeys reacting due to visual cues or sightings of the predator.) Equal presentations of each alarm were played in a systematically varied fashion. Fifty trials were conducted when monkeys were on the ground, and thirty-eight trials were conducted when the monkeys were in trees.

From the table, we see that these responses were specific to each alarm presented and “seemed to represent adaptive strategies for coping with the hunting behavior of the predators involved” (Seyfarth et al. 802). When leopard alarms were played, the monkeys most often ran up into trees. In a natural setting, this would locate them in an area safe from the ambush style of attack characteristic of leopards. When eagle alarms were played, the monkeys most often looked upward, ran into dense bushes for cover, or executed both behaviors. In a natural setting, this would allow the monkeys to avoid attacks from the air. When snake alarms were played, the monkeys most often looked down towards the ground, which in a natural setting, would allow them to locate and avoid their predator (Seyfarth et al. 802). For all calls, all monkeys looked toward the speaker, the source of the alarm, and scanned their surroundings as if searching for additional cues; they responded “as though each type of alarm call designated different external objects or events” (Seyfarth et al. 802).

Therefore, Seyfarth’s et al. findings (that acoustically distinct alarms were assigned to different predators) support the notion that vervet monkeys can effectively categorize other species, particularly their predators. What’s interesting is that there is an age discrimination for this behavior. Adults were the most selective such that leopard alarms were primarily performed to leopards, eagle alarms to eagles, and snake alarms to pythons. Younger monkeys, however, did not distinguish between the different predators as well as adult monkeys did. According to Seyfarth et al., infants gave leopard alarms to a wide range of terrestrial mammals, eagle alarms to many different kinds of birds, and snake alarms to snakes or long objects on the ground, therefore failing to distinguish between particular predator species within such classes (Seyfarth et al. 802). In addition, young monkeys were more likely to give alarms to things that posed no danger. However as they matured and gained more experience they “sharpened the association between predator species and the type of alarm call” (Seyfarth et al. 803).

The findings of Seyfarth et al. are interesting because this systematic use of signals means that the animals must understand contextual communication cues. An alarm signal can elicit general behavior to remove oneself from danger. However, vervet monkeys show much more complexed behavior; they have to know how exactly to remove themselves from danger. It brings into discussion discriminative stimulus, where the specific type of alarm call is the discriminative stimulus, and the response outcome relationship varies. If the wrong response is performed based on the context, they will not effectively avoid the predator and therefore will not experience a desired outcome. However, if the alarm call is different, that same response can allow avoidance of the predator and produce a desire outcome. It all depends on the discriminative stimulus, or specific alarm call, that is produced by the vervet monkeys.

Seyfarth, R., Cheney, D., & Marler, P. (1980) Monkey responses to three different alarm calls: evidence of predator classification and semantic communication. Science, 210(14), 801-803.