Geoscience Reference
In-Depth Information
lack of control animals was common during the initial period of marker eval-
uation studies (i.e., 1940-1960), but more recent evaluation studies also pos-
sess this flaw (Davis et al. 1984; Eagle et al. 1984; Reid et al. 1986; Koehler et
al. 1987; Mullican 1988). In other cases, authors include in the design a
cohort of previously marked (Fairley 1982) or alternatively marked (Garrott et
al. 1985; Wood and Slade 1990) animals as controls. However, this approach
assumes that “control” animals are representative of the unmarked population
even though alternative markers might cause important effects on their own.
In this case, any comparison of marked versus “control” animals could result
in an underestimation of effects of the targeted marker.
In some studies, control animals are not subjected to the same handling
procedure as marked animals, thereby making marking effects indistinguish-
able from those of handling (Mears and Hatch 1976; Scheirer and Coble
1991). This can be particularly problematic in situations where handling
causes significant stress or long-term effects, and as a result researchers may
find it difficult to identify which procedure (marking or handling) requires
modification. However, some studies (Lucas 1989) have correctly subjected
controls to all the same handling procedures as the marked sample, thus allow-
ing a more rigorous evaluation of the effects of the marker itself.
Marker evaluation studies often have sample sizes that are simply too small
to detect a reasonable difference between marked and unmarked samples
(White and Garrott 1990; Daly et al. 1992). Inadequate statistical power
increases the likelihood of committing a type II error (Sokal and Rohlf 1981),
thereby increasing the chance of failing to reject a null hypothesis of no signif-
icant marking effects when effects actually occur. Marker effects tend to be
more readily detected in the laboratory because field studies often have smaller
sample sizes and larger within-sample variance. Many field studies for which
marker evaluation is apparently an offshoot (Guynn et al. 1987; Douglass
1992), or those that evaluate marker effects on large mammals (Hamlin et al.
1982), lack statistical power. Thus determining the detectable effect size and
statistical power associated with a given marker evaluation study should be a
necessary precursor to implementation of that study. Also, whenever possible,
studies probably should be initiated under controlled laboratory conditions to
reduce confounding effects of the environment. However, laboratory studies
should be followed by evaluations in the field.
A common characteristic of marker evaluation studies is the use of subjec-
tive or qualitative measures of marking effects (Seale and Boraas 1974; Gold-
berg and Haas 1978; Andelt and Gipson 1980, 1981; Garshelis and Siniff
1983; Griben et al. 1984; Reid et al. 1986; Van Vuren 1989). Without rigor-
