25.05.2014 Views

Bat Echolocation Researc h - Bat Conservation International

Bat Echolocation Researc h - Bat Conservation International

Bat Echolocation Researc h - Bat Conservation International

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

and van Parijs 1993).<br />

Early studies using bat detectors indicated that different<br />

species of bats use different echolocation calls,<br />

thus suggesting that free-flying individuals could be<br />

identified on the basis of their calls (Ahlén 1981; Fenton<br />

and Bell 1981). However, one problem in using echolocation<br />

calls to identify species of bats is that the characteristics<br />

of the calls of each species (e.g., duration, bandwidth,<br />

frequency, harmonic structure) vary. To address<br />

questions requiring species identification, researchers<br />

rely on the fact that variation in call structure is generally<br />

greater among species than within species. Within<br />

many bat communities, the calls of some species are distinctive<br />

(Fig. 1A), and even though there is variation<br />

within each species, calls can be assigned unambiguously<br />

to particular species. In most cases, however, the call<br />

characteristics of various species overlap because on<br />

average the calls of different species are very similar to<br />

one another, or because there is greater variation in the<br />

calls of each species (Fig. 1B). Assigning calls to a particular<br />

species is less certain under these circumstances.<br />

Calls that fall within the zone of overlap (Fig. 1B) may<br />

be assigned to a species with a probability less than one,<br />

with the probability increasing the closer the combination<br />

of traits is to the mean value for that species. By definition,<br />

there is error involved in making such classifications.<br />

Often, multivariate statistical techniques such as<br />

discriminant function analysis (DFA) or neural networks<br />

are used to determine the probability of assignment<br />

(e.g., Jones et al. 2000; Parsons and Jones 2000; Vaughan<br />

et al. 1997).<br />

Variation in the echolocation-call characteristics of a<br />

particular species of bat clearly increases the uncertainty<br />

of identifying calls of unknown, free-flying individuals<br />

to that species. Such variation occurs for a number of<br />

reasons, including variation within a population related<br />

to characteristics of the individual, such as age, gender,<br />

and body mass (Betts 1998; Brigham et al. 1989; Jones et<br />

al. 1992; Jones and Ransome 1993; Obrist 1995). Some<br />

Section 4: Resources, <strong>Researc</strong>h and Study<br />

Figure 1: Variation in two<br />

characteristics of echolocation<br />

calls within and between<br />

two species of bats. A: Variation<br />

within each species,<br />

depicted as the shaded circles,<br />

is relatively small and<br />

calls can be unambiguously<br />

assigned to one species or<br />

the other. B: The mean values<br />

for the two call variables<br />

are the same as in A, for each<br />

species, but there is greater<br />

within-species variation. This<br />

means that there is overlap<br />

among the calls of the two<br />

species. Calls with characteristics<br />

in the zone of overlap<br />

cannot be assigned to a particular<br />

species with 100%<br />

certainty.<br />

of these factors may cause variation in calls from different<br />

locations (e.g., body mass), while others will not<br />

(e.g., age and gender), unless population structure varies<br />

(e.g., Barclay 1991). The purpose of this review is to<br />

consider geographic variation, especially among populations<br />

within a species, and to make recommendations as<br />

to how to account for it when conducting studies that<br />

depend on identifying bats by their echolocation calls.<br />

By geographic variation we mean variation in the<br />

structure of echolocation calls of individuals of the same<br />

species recorded in different locations. It is a potential<br />

problem if call characteristics from one population are<br />

used to identify the calls of unknown individuals from<br />

another population (Law et al. 2002). Most studies rely<br />

on a set of calls from known individuals (a reference<br />

library) to establish the characteristics used to identify<br />

unknown bats recorded by the monitoring system.<br />

Suppose there are two species in our study area and<br />

we base species identification on the call characteristics<br />

determined from another location (Fig. 2). Suppose further<br />

that the characteristics of the calls of the two<br />

species are distinct, but call characteristics from individuals<br />

in our study populations differ from those in the reference<br />

library. Such geographic variation could result in<br />

the inability to identify some calls and misidentification<br />

of others, depending on the amount of geographic variation<br />

(Fig. 2).<br />

While geographic variation is most commonly<br />

thought of on a large scale, an analogy being the dialects<br />

of humans and birds, there is relatively little information<br />

regarding this for the echolocation calls of bats. A large<br />

part of the problem is a lack of consistency in recording<br />

techniques and equipment among studies. There are,<br />

however, considerable data regarding smaller-scale geographic<br />

variation and variation caused by other factors<br />

that may result in large-scale geographic differences. We<br />

thus examine these sorts of variation, make predictions<br />

Figure 2: Potential problems associated with geographic variation in the<br />

characteristics of the echolocation calls of two species of bats. Characteristics<br />

from the reference library for each species (shaded circles) are<br />

distinctive, but differ from the characteristics of the calls from the study<br />

populations (open circles). Because of this, some calls will be unidentifiable<br />

(they do not fall within either reference set), and others will be<br />

misidentified (they fall within the reference set of the other species).<br />

145

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!