tion to produce vocalizations that differ systematically among populations.
This can be independent of any possible advantages to vocal variation and
need not involve vocal learning. In this case, morphological and behavioral
traits will also reflect population substructure. This expectation provides
one way to test the influence of population substructure on vocal differences. Finding vocal variation without concomitant variation in other traits
would refute the population substructure hypothesis.
The same kinds of dialect patterns can result from local adaptation, which
can also result in genetic variation among populations. Where population
substructure can result from genetic drift generating random differences
between isolated populations, the local adaptation hypothesis argues that
the action of divergent selection is important. Populations that inhabit areas
differing in physical features may produce vocalizations whose characteristics allow for efficient transmission in their respective habitat (Slobidchikoff
and Coast 1980; Rydell 1993; Barclay et al. 1999; Bass and Clark, Chapter
2; Ryan and Kime, Chapter 5). Characteristics can include frequency shifts,
changes in the proportion of tonal, trilled, and noisy elements, and temporal pattern (Wiley and Richards 1982). In captive studies, the physical
characteristics of housing facilities can influence frequency and temporal
characteristics of vocalizations. Individuals or groups housed in different
facilities can produce vocalizations that differ systematically, whereas
those housed in similar facilities can sound similar, independent of actual
vocal learning. Changes in housing can induce changes in vocalizations,
again independent of social influences and vocal learning. Predation can
also exert selection on signal structure by selecting for vocalizations that
are hard for predators to detect, localize, or recognize. Consequently,
vocalizations can vary among areas that vary in predation risk or predator
identity.
To unequivocally demonstrate vocal learning, the alternative mechanisms
described above must be ruled out. When a species’ biology implicates one
or more as possible explanations for observed patterns, observational
studies are not conclusive. Descriptions of dialects are especially vulnerable
to this. Demonstrating that vocal learning produces the patterns observed
is especially challenging in animal systems where experimental manipulations are difficult and direct observations of behavioral interactions and
context are not possible. However, we suggest possible approaches and
point out those that have been successfully used. Experiments are necessary, and these should be carefully designed to control or directly test the
likely alternatives. Determining which of these alternatives is involved is
difficult because they can predict the same kinds of patterns—in many cases
only the process differs.
In the next sections we begin by describing vocal learning in the context
of individual signatures, group signatures, and dialects. We specifically
focus on expected patterns of vocal variation at each of these levels and
with respect to call function to highlight potential sources of selection
144
J.W. Boughman and C.F. Moss
This can be independent of any possible advantages to vocal variation and
need not involve vocal learning. In this case, morphological and behavioral
traits will also reflect population substructure. This expectation provides
one way to test the influence of population substructure on vocal differences. Finding vocal variation without concomitant variation in other traits
would refute the population substructure hypothesis.
The same kinds of dialect patterns can result from local adaptation, which
can also result in genetic variation among populations. Where population
substructure can result from genetic drift generating random differences
between isolated populations, the local adaptation hypothesis argues that
the action of divergent selection is important. Populations that inhabit areas
differing in physical features may produce vocalizations whose characteristics allow for efficient transmission in their respective habitat (Slobidchikoff
and Coast 1980; Rydell 1993; Barclay et al. 1999; Bass and Clark, Chapter
2; Ryan and Kime, Chapter 5). Characteristics can include frequency shifts,
changes in the proportion of tonal, trilled, and noisy elements, and temporal pattern (Wiley and Richards 1982). In captive studies, the physical
characteristics of housing facilities can influence frequency and temporal
characteristics of vocalizations. Individuals or groups housed in different
facilities can produce vocalizations that differ systematically, whereas
those housed in similar facilities can sound similar, independent of actual
vocal learning. Changes in housing can induce changes in vocalizations,
again independent of social influences and vocal learning. Predation can
also exert selection on signal structure by selecting for vocalizations that
are hard for predators to detect, localize, or recognize. Consequently,
vocalizations can vary among areas that vary in predation risk or predator
identity.
To unequivocally demonstrate vocal learning, the alternative mechanisms
described above must be ruled out. When a species’ biology implicates one
or more as possible explanations for observed patterns, observational
studies are not conclusive. Descriptions of dialects are especially vulnerable
to this. Demonstrating that vocal learning produces the patterns observed
is especially challenging in animal systems where experimental manipulations are difficult and direct observations of behavioral interactions and
context are not possible. However, we suggest possible approaches and
point out those that have been successfully used. Experiments are necessary, and these should be carefully designed to control or directly test the
likely alternatives. Determining which of these alternatives is involved is
difficult because they can predict the same kinds of patterns—in many cases
only the process differs.
In the next sections we begin by describing vocal learning in the context
of individual signatures, group signatures, and dialects. We specifically
focus on expected patterns of vocal variation at each of these levels and
with respect to call function to highlight potential sources of selection
144
J.W. Boughman and C.F. Moss