Back to: Anabat Technical Notes, Anabat Contents, Home
Active monitoring involves humans. There are many ways this can be done, but the common element is that a human is present and influencing the process of monitoring bats. At one extreme, Active Monitoring could be just like bird watching, where an observer goes out watching bats and using a combination of acoustic and visual cues to help identify them. But Active Monitoring doesn't necessarily involve collection of visual cues. Just holding the bat detector in your hand will have a significant impact on how you record bats, because you will tend to orient the detector towards the bats, thereby improving the quality and quantity of calls you record. I would say that Active Monitoring must involve some direct human control over the recording process. But like everything in biology, the boundaries can be fuzzy! I would argue that someone sitting on the roof of a moving vehicle watching bats is undeniably Actively Monitoring, but someone driving a car with a detector mounted on the roof is another proposition!
Active Monitoring, compared to Passive Monitoring, results in more and better quality calls being recorded, it often leads to much higher identification rates because of the better quality recordings and the presence of helpful visual cues, and it allows the observer to actively explore for bats instead of just waiting for the bats to appear.
Passive Monitoring takes place when the bat detector records bats in the absence of direct, human control over the recording process. In effect, this is data logging of bat calls. Passive Monitoring has the advantage that bats can be monitored for very long periods. It is not uncommon for Passive Monitoring stations to record all night, every night, for weeks, months or years. Thus the sampling effort achievable using Passive Monitoring is vastly greater than for Active Monitoring. This means there is a greatly enhanced chance of detecting rare or hard-to-detect bats, because the effort being put into looking for them is so great. Passive Monitoring is ideally suited to looking at temporal patterns of activity, but because several detectors can be deployed by a single human with little need for servicing, it is also an excellent tool for looking at spatial heterogeneities. Another aspect of Passive Monitoring is that it collects data without any human influence on the recording process, so it has the potential to produce highly objective measures uninfluenced by observer biases. This assumes, of course, that identifications can be made without human subjectivity, but that is the role of automated identification systems, which can identify and collate all the recordings by machine. Such systems are getting better and will continue to do so, but for many purposes, they are already useful and for handling large data sets from multiple Passive Monitoring stations, they are virtually essential.
This is a sequence of bat PULSES given as a bat approaches a potential prey item. It is typically characterised by a progressive reduction in PULSE DURATION and increase in FREQUENCY SWEEP as the bat transitions from SEARCH PHASE towards a FEEDING BUZZ. However, PULSE DURATION may actually increase initially in the APPROACH PHASE, and not all species increase the FREQUENCY SWEEP during approach. APPROACH PHASE can usually be recognised as a period of transition, where the PULSES progressively change their nature towards a FEEDING BUZZ. But on their own, APPROACH PHASE calls typically look very like SEARCH PHASE calls given in CLUTTER.
A single, complete burst of sound emitted by a bat, separated by silence from other calls. Most bat calls are given in a single breath, and are supposedly synchronised to wingbeats, but this obviously isn't always the case, when you consider rapidly repeated vocalisations such as those in a feeding buzz. If you compare the use of the term CALL with that applied to birds or frogs, it isn't analogous. In the bat world, a CALL could be just a single element of a FEEDING BUZZ, whereas in a bird or a frog, such an element would usually be called a PULSE, and the FEEDING BUZZ as a whole would be called a CALL. Yet in a bat, there isn't always a clear distinction between a FEEDING BUZZ and the CALLS leading up to it, so perhaps the term PULSE might be better for bat vocalisations, as this can always be equivalent to a PULSE in a frog or bird call.
This is a call parameter that is easily visualised if you look at an Anabat display. It can be defined as the frequency at the right hand end of the flattest portion of a call. It is by far the most important single parameter for distinguishing species, though it will rarely be diagnostic in itself.
Fc is often close to the minimum frequency of a call, but is less variable because the minimum frequency will often occur in a downsweep at the end of a call, where the amplitude is decreasing rapidly. In other words, minimum frequency will typically occur when the call is dying out at the end, and just how much of the end of the call will be detected will depend on distance to the bat amongst other things. In a call such as produced by a Horseshoe Bat, Fc will be equal to the maximum frequency.
Some calls will contain more than one flat portion. In that case, it might not be evident which flat portion should be used to define Fc if you rigidly follow the above definition. But usually such calls will be given in a social context, and you can see that the second flat portion is a variation on a normal call. Bear in mind what the normal call would have looked like if the variation hadn't occurred, but better still, exclude such variations from your analysis.
Fc will most often not be diagnostic of a particular species in itself, but in any given locality, it will always limit the range of possibilities to a very small number. For identification purposes, you don't have to worry about measuring Fc - just look at the display and see what ballpark it lies in. Remember that it is subject to variation due to DOPPLER SHIFTS, like any parameter, and that it will usually vary between individuals, and according to what the bat is doing. Many species vary Fc depending on how many individuals are flying in the area. Again, it isn't the exact value which matters, but the general area where it lies.
Sc is another very important parameter for species identification, closely tied to the call SHAPE. It can be defined as the slope of the flattest part of the call. Sometimes, though, the flattest part of the call may occur in a short added on section at the end, often called a TOE. In such a case, it might be better to take the Sc as the slope of the section preceding the TOE, because that section is always present, whereas the TOE can be just an occasional add-on. For example, many Myotis calls have an essentially linear shape, in which case the Sc will just be the slope of the call as a whole. But in some cases, a short flat section might be added at the end of the call. In that case, it would be more meaningful to take Sc as the slope of the bulk of the call.
If ever you feel the need to make measurements of call parameters, you should always tailor your measurements to what makes most sense for the species you are working with. Don't feel constrained by traditional views of what should be measured, or by trying to force calls into a stereotyped model.
CLUTTER can be thought of as just the distance to the nearest object from which a bat can pick up an echo. So a bat flying in the open, well above the ground and well away from any objects, may be in zero CLUTTER, while a bat flying amongst treetrunks or along water close to a river bank would be in high CLUTTER. A bat emitting high frequency, low intensity CALLS may still be in low CLUTTER even when flying just a few metres from objects, while a bat emitting loud, low frequency calls might be in high CLUTTER even 30 m above the ground. So the degree of CLUTTER depends on the bat, as well as on the distance bewteen the bat and something else.
CLUTTER is the major factor determining the type of calls a bat produces when it is in SEARCH PHASE. Typically, a bat in high CLUTTER will produce calls of shorter DURATION, longer FREQUENCY SWEEPS, more rapid repetition rate and lower intensity, than the same bat would in low CLUTTER. So SEARCH PHASE calls tend to fall along a continuum, depending on the CLUTTER. At one extreme, a bat in zero CLUTTER will produce the flattest, longest DURATION, most widely spaced and loudest calls you will ever see from that bat. Such calls are often referred to as COMMUTING CALLS, as they are the sort of calls the bat will produce when flying directly from one place to another through open space. At the other extreme, calls produced in very high clutter may be very quiet, so hard to detect, and often just consist of very brief, steep DOWNSWEEPS. High CLUTTER calls are usually much more difficult to identify than low CLUTTER calls, because they tend to be similar even between very different species. But in some cases, higher clutter calls may be more distinctive. Every case should be treated on its own merits.
A useful term to describe the sort of SEARCH PHASE calls a bat produces in zero CLUTTER. These calls typically are the loudest and have the longest DURATION, shortest FREQUENCY SWEEP and are produced at the lowest rate of any that particular bat will emit.
Many species produce a distinctive series of CALLS when they come down to water to drink. These DRINKING BUZZES usually resemble a very long, drawn out FEEDING BUZZ.
The total time that a single CALL, or PULSE, lasts. It is important when measuring DURATION not to include any echoes. Usually echoes are distinctly separated from the main part of a call, but in some cases the distinction can be hard to make.
The Duty Cycle of a call sequence is just the proportion of time in which a call is actually being produced. It could be expressed as: DC = Dur / TBC * 100 where DC is Duty Cycle, Dur is call Duration and TBC is the Time Between Calls. As long as Dur and TBC have the same units, the DC is a simple percentage. Typical bat call sequences have a Duty Cycle of about 10%, but some species, such as the Horseshoe bats, have much larger Duty Cycles.
A FEEDING BUZZ is a rapid series of PULSES given as a bat approaches potential prey. In much bat literature, three echolocation phases are recognised, SEARCH PHASE, APPROACH PHASE and TERMINAL PHASE (which is the same as a FEEDING BUZZ). A FEEDING BUZZ is usually, but not always, a very distinct entity, clearly recognisable from other vocalisations by its rapid repetition rate.
Most bat CALLS consist at least partially of a FREQUENCY SWEEP, where the frequency of the CALL changes in time. The only real alternative is a CALL which stays on the same frequency for its entire DURATION, which hardly ever happens. The total FREQUENCY SWEEP of a CALL can be thought of as the total range of frequencies between the maximum and minimum frequencies. This is a better term than BANDWIDTH, often used in traditional bat literature, but quite inappropriate as the term has other meanings in engineering, and in any case, should apply across all the harmonics the bat emits.
Slope is an important feature of a bat call, both because variation in slope defines the shape of a call, and also because the minimum slope of a call is one of the most useful features for distinguishing between species, or between different types of calls from the same species as given in different degrees of clutter. Technically, the slope can be thought of as the rate of change of frequency with respect to time. Slope represents the steepness or flatness of a call (or portion of a call), but independently of how the call is viewed. This last point is very important. If you magnify a call in the horizontal (time) dimension, it inevitably reduces the apparent slope of the call. You can easily see this by viewing a call in Analook at a range of magnifications from F1 to F10. At F1, a call is likely to look like a vertical straight line, while at F10 the same call is more likely to look like a horizontal straight line. So the apparent slope of a call as you look at it on the screen is a function of the relative magnifications of the vertical and horizontal axes. Note that it is easiest to see changes in slope, and therefore the shape of a call, if the overall slope of the call appears on the screen to be about 45 degrees. However, this isn't necessarily the best viewing choice, as it has to be weighed against the fact that you can more quickly appreciate the nature of a call if you become familiar with using just one or two standard magnifications.
There are several ways that the slope of a call, or portion of a call, could be measured. A traditional way is to use kHz/ms which is simply the change in frequency over a given time period. It could also be measured as the angle of the call as it appears on a screen at a predefined magnification. However, in Anabat, it has been traditional to measure slope in Octaves Per Second (OPS) and to define it as the rate of decrease of frequency with respect to time. This means that Anabat treats slopes as negative when the frequency increases with time. The reason for this is very simple - most bat calls consist of downsweeps, in which the frequency drops as the call progresses. If we defined bat call slope in the usual way, most calls would have negative slopes, and it gets tedious putting minus signs in front of everything! So it is better to think of a downsweep as having a positive slope. This way, the slope decreases as the call becomes less steep.
OPS is a fundamentally logarithmic scale. Many people have aversions to logarithms, which is a shame because the way we think about most things is fundamentally logarithmic. This is because we tend to think in proportions, rather than in absolute terms. To illustrate this point, consider the distance scale, which is entirely linear. This means that a metre has the same meaning whether we are talking about the width of a hair or the distance to the moon. But in terms of human perception, the significance of a metre depends on the context, and particularly, on how a metre compares to the distance of interest at the time. If we are thinking of the height of a person, then a metre is a huge distance, about 50% of the typical human height, and two people differing in height by one metre would look immensely different because of that height difference. On the other hand, if we are thinking about walking 2 kilometres, a metre is completely insignificant - it makes no difference whether we walk 2 km or 2.001 km. The significance of something depends on its proportion to something else. If two items differ by 50%, that is a big difference which we can easily perceive. If they differ by one tenth of a percent, that is a tiny difference which we will generally need instruments to detect.
The musical scale is a good example of a logarithmic scale in common human use, and bat call frequency is just an extension of the musical scale. It makes sense to think of bat call frequency in logarithmic terms, because a difference of 1 kHz means much more when looking at a call at 10 kHz than it does when looking at a call at 40 kHz. It is the proportion which matters. In the same way, it makes sense to think of slope in logarithmic terms, because OPS has the same significance irrespective of the frequency of concern. A call of low slope (what people have often called QCF) will have a slope of say 10 OPS whether the call is at 5 or 50 kHz. Another consequence of using OPS is that different harmonics will have the same slope.
Mathematically, OPS is defined thus:
Slope = log2( F1 / F2 ) / ( T2 - T1 )
where Slope is the value of the slope in OPS between two points, the first point at time T1 with frequency F1 and the second point at T2 with F2. The frequency units are irrelevant but the times are in seconds. log2 means the logarithm to the base 2.
One octave is a change by a factor of 2, so if F2 is one quarter of F1, then F1/F2 = 4, which is two octaves, and log2( 4 ) = 2. So the slope is 2 divided by the time it takes for the frequency to change that much.
A series of CALLS given by one bat, typically the CALLS detected by a bat detector during a single pass of one bat past the detector. Since a bat almost has no alternative than to approach the bat detector and then move away, it can do little else! Obviously, the concept will get a bit hairy if a bat flies continuous circles around the detector. Often, bats will make repeated PASSES past a single point, and the definition of when PASSES start and end can be completely arbitrary. For this reason, people often define a PASS as having to meet certain criteria. For example, a PASS might only end if the bat has gone away for more than one second.
A single vocalization produced by a bat, separated from other vocalizations by silence. Synonomous with CALL in typical usage applied to bats. Note that a PULSE given by a bat can always be thought of as analogous to a PULSE in a frog call. In a frog call, there are usually many PULSES, and the PULSES may blend together to a variable degree. With bats, the term CALL applies to just one PULSE. The problem arises when you think of a FEEDING BUZZ, which contains a number of PULSES. Is a feeding buzz a single, multi-PULSEd call, or is it a series of single-PULSEd calls? For this reason, PULSE is probably a better term, as it can always be thought of as analogous to a PULSE in a frog or bird call.
In the traditional jargon applied to bat echolocation, three phases of echolocation calls are recognised, SEARCH PHASE, APPROACH PHASE and TERMINAL PHASE, more descriptively called a FEEDING BUZZ. In the SEARCH PHASE, the idea is that the bat is just motoring along looking for something to eat. SEARCH PHASE calls, then tend to be characterised by their relative uniformity. As long as the bat stays in the same degree of CLUTTER, flies straight and doesn't detect anything it can eat or it should be frightened of, it will produce similar-looking calls. Of course, on an Anabat screen, the calls won't all look the same, because when the bat is further away than some critical distance, you won't be able to see the quieter parts of the calls (typically the start, and sometimes also the end of each call tend to go missing).
This traditional jargon isn't much help to those interested in acoustic identification of bats, because it completely fails to take into account the fact that SEARCH PHASE covers a very wide range of different call types. Most of the time you detect a bat it will be giving SEARCH PHASE calls, yet these will probably vary immensely (depending on species). It will be more helpful to think of the variation in call types to be related to the degree of CLUTTER in which the bat is flying. In fact this approach could even be said to include all the traditional calling phases, since in a sense, a FEEDING BUZZ is just a specialised call type given in extremely high CLUTTER.
A SEQUENCE just refers to the series of CALLS which you happen to be dealing with at the time. So a SEQUENCE could be the same as a PASS, or it might contain several PASSES. It might refer to the contents of a single Anabat file, or it might be spread over several files. It could be that all the calls in a SEQUENCE were produced by a single bat, or by several bats. Think of a SEQUENCE as just a bunch of calls.
The traditional way of revealing the frequency component of a signal is SPECTRAL ANALYSIS. This is a very complex process in which the original signal is scanned by a large number of filters, each filter tuned to a narrow part of the frequency spectrum being investigated. The output of each filter represents the content of the signal within the bandwidth of that filter, so if you use lots of filters, you can build up a three dimensional image of a signal, showing the amplitude of each frequency component and how that varies in time ( a sonogram). There are other displays you could use also.
In the past, this was carried out by a sonograph, a large machine which recorded the signal on a magnetic medium and then repeatedly scanned it with a tuneable filter. The frequency to which the filter was tuned was controlled by a physical arm which was moved gradually upwards by a rotating thread. As the arm moved upwards, it not only tuned the filter, but it also carried a high voltage which burned a trace onto a sheet of paper wrapped around a rotating metal drum. The louder the signal at the frequency the filter was tuned to, the blacker the trace burned into the paper. It was a brilliant piece of equipment but exasperatingly slow. I don't want to think how many weeks of my life I spent burning traces of frog calls into paper on a sonograph.
Nowadays, spectral analysis of audio sounds is easy to carry out on your PC, thanks to the universal occurrence of soundcards in PCs and to the ready availability of software, some expensive, some free, which performs a process called FFT (Fast Fourier Transform). FFT simulates the bank of filters approach using a very cunning, very efficient software algorithm. On a reasonably fast PC, you can even analyse audio signals in real time, though at a reduced resolution. It probably won't be long till a laptop will have the power to do realtime FFT analysis of bat calls, though using special hardware, not a PC soundcard.
The final phase of vocalisations as a bat zeroes in on a potential prey item. FEEDING BUZZ is usually a more decriptive term, since the TERMINAL PHASE is usually very distinct and easily recognised from other aspects of a bat's vocal repertoire.
This is simply the time (typically measured in milliseconds) between the start of one call and the start of the next call. A curious fact is that so many species, when they are flying in the open, produce calls at a rate of about 10 per second, giving a TBC of 100 milliseconds. There is evidence that in at least some situations, bat calls are synchronised to their wingbeats, and the evidence is such that it seems likely this will usually be the case in SEARCH PHASE. If you look at a frequency distribution of the TBC values in a SEARCH PHASE call sequence, you will typically see a major peak near 100 milliseconds, and a second, smaller peak at about 200. The latter comes about because bats often reduce their calling rate by missing some calls, and it supports the view that calls are sychronised to wingbeats, since this would require that a missed call will double the TBC and that intermediate values won't occur. In practice, of course, calls can be decoupled from wingbeats and the extreme example of this would be the DRINKING BUZZ.
ZCA is the approach used by Anabat to analyse the frequencies in bat calls. It works as follows:
Every sound consists of waves of pressure where pressure rapidly alternates between being higher and lower than the average value. In air, this means that at any one point, as the sound wave passes, the air pressure fluctuates above and below the average air pressure (atmospheric pressure) These fluctuations are tiny, but they can be converted to electrical signals with a suitable microphone. These electrical signals alternate between positive and negative with respect to the average voltage at that point in the circuit. The average voltage isn't of interest here, the fluctuations are what matter. The ZeroCrossings are just the points in time when the electrical signal crosses over the average value from negative to positive (a positive-going transition) or from positive to negative (a negative-going transition). Consecutive transitions are always in opposite directions, so every second transition will be going in the same direction and transitions going in the same direction are called LIKE TRANSITIONS.
If you measure the time between LIKE TRANSITIONS, you get a value which is equal to the PERIOD of the waveform being examined. The PERIOD is just the reciprocal of the FREQUENCY, so the FREQUENCY can be calculated easily from the times of these transitions. All you need is a list of the transition times, and you can recover both frequency and time data. These data can be plotted on a graph to give a visual representation of the frequency characteristics of a bat call. Each DOT on the display represents the average frequency over the period since the last dot.
Anabat uses a more complex scheme. Firstly, bat echolocation calls can use frequencies over the range of 4,000 to 210,000 Herz. Most of this is outside the range of human hearing. In order to hear the bat calls, you need to generate a representation of them at audible frequencies. Anabat does this by counting transitions, and outputting one transition for every 8 LIKE TRANSITIONS of the original signal. This is the so-called "frequency-division" or "countdown" scheme, and results in an output frequency one sixteenth that of the original signal. Anabat can also use DIVISION RATIOS of 4, 8 or 32. The signal entering the ZCAIM is an audible signal at some fixed fraction of the frequency of the bat call.
The ZCAIM again detects the ZERO CROSSINGS of the audible signal output by the bat detector. The Anabat software produces a DOT on the screen for every CONSECUTIVE TRANSITION at the input to the ZCAIM, but it calculates the frequency for each dot by taking the reciprocal of the time between LIKE TRANSITIONS. This scheme results in twice as many DOTS for the same input frequency as you would get if you used only LIKE TRANSITIONS.
Back to: Anabat Technical Notes, Anabat Contents, Home
I hope to build on this page, including information on more of the jargon of bat-detecting. I would be the first to admit that not everyone will agree with everything I have written here. If you don't like what I've said about anything, please Email me and if I think your comments have merit, I'll incorporate them as presenting an alternative view (if you want me to!). I'd also very much like feedback about any errors, or suggestions for improvements, especially about other words you would like included.