Page 728 - The Toxicology of Fishes

708 The Toxicology of Fishes

biomarker of respiratory function is selected, it is important to measure the variability of response in
addition to absolute values, so a normal range can be established (Handy and Depledge, 1999). Only
then will it be statistically possible to separate normal biological variation such as diurnal changes in
metabolism, food intake, water quality, and temperature (Gonzalez and McDonald, 1992; Lyndon et al.,
1992; Morgan and Kühn, 1984; Wilson et al., 1994) from the effects of pollution.

Behavior as a Biomarker
A variety of animal behaviors can be measured during pollutant exposure, including avoidance of the
pollution gradient, changes in feeding activity, predator avoidance, foraging behavior, reproductive
behavior, social behavior, and swimming behavior (Kasumyan, 2001; Little and Finger, 1990; Little et
al., 1985; Sandheinrich and Atchison, 1990; Schreck, 1990; Scott and Sloman, 2004). Behavioral
biomarkers offer three very important advantages over biochemical, morphological, or physiological
biomarkers: (1) The behavioral response is often an integrated effect of the underlying biochemical and
physiological disturbances and so may reflect a series of toxic effects and compensatory responses
(Campbell et al., 2005). (2) Behavioral responses are often more sensitive indicators of exposure than
other approaches (Little and Finger, 1990). (3) Some behavioral responses can be linked on an energetic
basis to population survival (e.g., locomotor activity) (Handy et al., 1999; Priede, 1977).
Ecotoxicological research on fish behavior in the laboratory has studied both metals and organic
pollutants (Little et al., 1985; Scott and Sloman, 2004) but has generally focused on relatively short-
term acute effects lasting a few hours or days (Little and Finger, 1990; Rice et al., 1997; Scarfe et al.,
1982). Most studies have focused on exposure via the water, where disturbances to respiration and
osmoregulation (Pilgaard et al., 1994) may limit locomotor capacity (Waiwood and Beamish, 1978) and
thus reduce the behavioral repertoire of the animal. In addition, alterations of olfaction by various organic
and inorganic contaminants may also significantly impact grouping and schooling behavior, thus allowing
animals to be more susceptible to predation (Sandahl et al., 2005; Scholz et al., 2000). Alternatively,
food route exposures may not damage the gills, so the behavioral activities of the fish are not limited
by respiratory distress (Handy et al., 1999). Furthermore, the etiology of brain injury may be different
in dietary compared to aqueous exposures (e.g., mercury) (Berntssen et al., 2003). We should therefore
consider that behaviors not only have the opportunity to increase in complexity over time in chronic
compared to acute exposures (Scott and Sloman, 2004) but may also vary with the route of exposure.
Some behaviors may also be attenuated, such as the loss of circadian rhythms and aggression during
dietary copper exposures (Campbell et al., 2002, 2005). This raises the possibility of biomarkers based
on the presence or absence of behavioral responses.
Sudden changes in fish behavior, such as altered swimming ability, have long been suggested as early
warning systems or as biological monitoring tools for acute pollution. Early biomonitoring devices
required animals to be housed in a flow-through system with the recording apparatus in close proximity.
Today, acoustic tags, radio tags and transponders, and cardiovascular monitoring devices have enabled
wildlife telemetry to be employed in more realistic field situations on unrestrained fish (for a discussion
of these technologies in biomonitoring, see Handy et al., 2002a, and references therein). Such biomon-
itoring technologies for measuring fish behavior could be applied to the biomarker concept. Similar to
biomonitoring, any device used to assess behavior as a biomarker must be robust, reliable, and validated
against a variety of manual or other techniques for observing behaviors (Craig and Laming, 2004).
Systems such as the Multispecies Freshwater Biomonitor have been used to assess locomotor behaviors
in the presence of environmental stressors such as ammonia (Craig and Laming, 2004) or municipal
wastewater (Gerhardt et al., 2002), and similar systems have been used for fish or invertebrates with a
variety of pollutants (Handy and Depledge, 1999; Handy et al., 2002a); however, it remains a challenge
to use these techniques as biomarkers of chronic pollution. Most of the existing data on behavioral
effects are over time scales of a few hours or days, although Scott and Sloman (2004) pointed out that
experiments on complex behaviors such as reproductive behaviors and some social behaviors involve
longer time scales. We are also now realizing that changes in behavior feed back at the physiological
level to produce adaptive changes in physiology (Scott and Sloman, 2004); for example, during metal
exposure such changes include changes in the cost of aerobic metabolism following the loss of circadian