- Top of page
- Conflict of Interest
Due to the unforgiving nature of predation, most animals have invested heavily in antipredator defense mechanisms (Crowl and Covich 1990; Lima and Dill 1990; Brönmark and Miner 1992). Morphological defenses, such as protective spines and armor, deter attacks and reduce the probability of capture in a variety of animals (Arnqvist and Johansson 1998; Hoverman et al. 2005). Many prey species have cryptic coloration to avoid detection or alternatively are brightly colored advertising noxious or toxic properties to would-be predators (Cuthill et al. 2005; Stankowich et al. 2011). Behavioral defenses also limit the success of predators, with prey avoiding specific locations and/or limiting their activity during times of day that predators are hunting (Lima 1998; Ferrari et al. 2009). When prey do encounter predators, they can also flee or hide to escape an attack (Sih 2005).
One of the prerequisites for successful predator avoidance is that prey animals recognize predators or high-risk situations as dangerous (Brown and Chivers 2006). Information that allows prey to recognize risk can come from a variety of sources, including nearby conspecifics (Griffin 2004; Crane and Ferrari 2013). Prey may respond to the visible flight responses or alarm calls of conspecifics (Blumstein and Armitage 1997). They may also respond to chemical cues released by nearby individuals that have detected a predator (often referred to as disturbance cues) or to chemical cues released by prey that have been attacked by predators (often referred to as damage-released alarm cues) (Chivers and Smith 1998; Ferrari et al. 2010). Both of these chemical sources of information could provide prey with an early warning of a potential attack. Indeed, several studies have shown that this early warning increases the probability of survival during a staged encounter with a predator (Hews 1988; Mathis and Smith 1993a; Mirza and Chivers 2001).
Damage-released alarm cues are common in freshwater and marine organisms and known to induce adaptive changes in morphology in prey animals (Stabell and Lwin 1997; Chivers et al. 2008) and are crucial in facilitating learned recognition of predators (Mathis and Smith 1993c; Ferrari et al. 2005) and dangerous habitats (Chivers and Smith 1995). There are many hundreds of studies that have investigated the importance of these cues in mediating predator–prey interactions (Chivers and Smith 1998; Ferrari et al. 2010). It is therefore surprising that we know almost nothing about the temporal aspects of risk assessment using alarm cues (Ferrari et al. 2010). When a prey individual is captured and alarm chemicals are released, the prey detecting the chemical cues are aware that a nearby individual was recently captured. However, to understand the value of this information, we need to begin to consider what recently actually means. In systems where predators have the ability to eat multiple prey in a short time, an alarm cue released 1 min ago would probably be important as a risk assessment cue. What about an alarm cue released 10 min ago or one released an hour or even a day ago? How long do the chemicals actually persist in the environment? Short-lived chemicals could provide very accurate temporal information about risk. In contrast, those cues that last for many hours would be much less temporally reliable, but nonetheless would provide at least some information.
Three studies have investigated these sorts of temporal risk assessment questions using chemical alarm cues. In a laboratory experiment, Hazlett (1999) showed that alarm cues of crayfish (Orconectes virilis) can persist (i.e., be detectable by conspecifics) for more than 6 h. Likewise, Wisenden et al. (2009) demonstrated that alarm cues of freshwater fish (fathead minnows, Pimephales promelas) and amphipods (Gammaruslacustris) may last at least 3 h, but not more than 6 h. In these experiments, the alarm cues were prepared in the laboratory with clean dechlorinated tap water. We caution that such designs need to be interpreted carefully because the absence of sunlight could influence photodegradation of the alarm chemicals and the absence of appropriate biofauna could influence the rate of biodegradation. Wisenden et al. (2009) attempted to use a natural trapping experiment to confirm their laboratory findings, but the results of their experiment were somewhat ambiguous. Ferrari et al. (2007b) tested the rate of breakdown of wood frog tadpole alarm cues in a natural pond and found that tadpoles responded to cues released 5 min following injection into the pond, but did not respond to the same cue after 2 h. Clearly, there is a dire need for experiments designed to determine the rate of breakdown of alarm cues to understand the role of the cues in mediating predator–prey interactions. The goal of our current work was to determine the duration that alarm cues of juvenile damselfishes (Pomacentrus amboinensis) persist under natural conditions around Lizard Island in the Great Barrier Reef. Several studies have recently shown the importance of alarm cues in risk assessment in coral reef damselfishes (McCormick and Larson 2008; Ferrari et al. 2011; Lonnstedt et al. 2013; Chivers et al. in press).
Here, we used a two-channel choice flume to assess avoidance behavior of damselfish to the alarm cues of conspecifics. First, we identified a concentration of fresh alarm cues that consistently lead to a strong avoidance response. We then prepared additional batches of fresh alarm cues and introduced them into ambient seawater held in floating containers in the ocean where they were exposed to a natural temperature and light regime. The containers also held a natural sand substrate to ensure that the water was in contact with natural substrate-born biofauna that could breakdown the alarm cues. For each container, we sampled the water immediately after placing the alarm cues into the container and at 10, 20, and 30 min post-injection. We used the choice flume to determine whether the cues remained active after various amounts of time had elapsed. The breakdown of alarm cues could be influenced by both photodegradation and biodegradation, the rates of which could vary based on abiotic conditions including temperature, pH, dissolved oxygen, and solar radiation. Consequently, we replicated the start time of the experiment to test whether the rate of degradation varied throughout the day and could be correlated with any abiotic factors.
- Top of page
- Conflict of Interest
The results of our study highlight that chemical alarm cues of coral reef damselfish degrade rather quickly under natural conditions. In our bioassay, we found nearly 100% avoidance of freshly prepared alarm cues. However, after the cues aged for only 30 min, we observed little avoidance. The responses for cues aged 10 and 20 min were intermediate. What was most striking was that we observed very different patterns of responses to aged cues at different times of the day. Given that we failed to find a temporal effect on the response of fish to fresh cues, we concluded that the differential responses to aged cues were driven by the aging of the cues, rather than a diel change in the fish's response. Aged cues failed to evoke antipredator responses at midday, but they did both early and late in the day. We speculate that this corresponds with the highest rate of breakdown of alarm cues occurring in the afternoon.
The relatively short active time for alarm cues that we documented contrasts an extensive literature showing that prey animals frequently respond to odors of predators fed conspecifics of the prey, but not to odors of predators fed a different diet (Mathis and Smith 1993c; Chivers and Mirza 2001). Indeed, one study showed that minnows (Pimephales promelas) responded to odors of predators fed minnows that have alarm cues, but did not respond to predators fed minnows that lacked alarm cues (Mathis and Smith 1993b). In predator-diet studies, predators are fed specific diets for days and then not fed for a day or two before odor cues are collected. The fact that odors that are days old can evoke antipredator responses implies that alarm cues survive digestion for days in the predators' gut or that the breakdown products of the alarm cues can last for days.
Our work provides some evidence that the rate of breakdown of alarm cues is dependent on solar radiation. Indeed, early in the day and late in the evening when the sun was not shining directly into the pails, the rate of alarm cue breakdown was significantly lower. This raises the interesting question of whether fishes can gain different temporal information from chemical cues depending on time of day and ambient weather conditions. Do chemical cues last longer on cloudy days? Do alarm cues released near the surface have a shorter half-life than those released at a depth where solar penetration is reduced? The specific wavelengths of light that could be responsible for the photodegradation are unknown to us, but UV radiation is known to cause the breakdown of many organic molecules (Hays et al. 1996). If this is the case, then any factor that influences the level of UV radiation will alter the rate of degradation. Stratospheric ozone depletion, a major environmental concern, particularly in the Southern Hemisphere (Smith et al. 1992), could lead to increased rates of alarm cue degradation, while the addition of turbidity and dissolved organic carbon associated with anthropogenic change (Wenger and McCormick, 2013) could lead to decreased rates of degradation.
We must be cautious in our conclusion that solar radiation is primarily responsible for differences in the rate of degradation of alarm cues. Clearly, additional manipulative experiments are in order. Temperature, pH, and dissolved oxygen followed the same general temporal pattern as solar radiation with peaks in the mid-afternoon, and hence, these factors could be responsible for the effects we observed. We had a relatively narrow window of temperatures, pH, and dissolved oxygen in our study; hence, these factors seem much less plausible than solar radiation. However, other prey species in other systems do experience extreme changes in physicochemical conditions. For example, temperate fishes frequently experience greater than a 20°C change in temperature throughout the year. If temperature is a factor mediating the rate of alarm cue breakdown, then we could easily imagine seasonal differences in temporal information use. The IPCC predicts a 3°C increase in ocean temperatures by the end of the century (IPCC 2007). It seems plausible that such warming could increase the rate of alarm cue degradation and alter chemosensory risk assessment, even for fishes in warm tropical waters. The acidity of our oceans is also predicted to change considerably over the next century (Kleypas et al. 2006). If alarm cue breakdown is linked to ocean pH, then we should expect to see the opposite effect as with an increase in temperature; ocean acidification should reduce degradation of alarm cues. This is based on the observation in our study that there is a higher rate of breakdown in mid-afternoon when pH is the highest.
We are in dire need of comparative field experiments designed to test the rate of breakdown of chemical cues that indicate risk. Besides the work of Ferrari et al. (2007b) and Wisenden et al. (2009) examining the rate of breakdown of alarm cues in freshwater ponds and lakes, we have little information in other systems. However, there are a few other studies that have attempted to determine temporal aspects of risk assessment using predator odors. Bytheway et al. (2013) recently showed that rats (Rattus fuscipes) avoid fresh dog (Canis lupus familiaris) scent, but fail to respond to dog scent that was aged for two days. Peacor (2006) found that the time period that bullfrogs (Ranacates beiana) responded to odors of larval dragonflies (Anax junius) was in the order of 2–4 days. Interestingly, the length of time was shorter when the dragonfly cue was aged in pond water compared with when it was aged in well water. In a similar experiment, Fraker (2009) concluded that greenfrog (Rana clamitans) tadpoles lacked the perceptive ability to reliably assess the age of predator odors. Tadpoles responded to dragonfly cues aged up to 48 h but not 72 h. He argues that tadpoles overemphasize risk resulting in a disproportionately strong antipredator response. Unfortunately, Fraker (2009) used well water not pond water, and like Peacor (2006), he aged the cues in the laboratory not under natural solar conditions. Our work here suggests that we should use caution in interpreting results of laboratory-based experiments. In the absence of solar radiation and natural biofauna that could breakdown the cues, the rate of degradation could be misleading. In fact, when we conducted preliminary trials to establish the timeframe and concentrations to use in these experiments, we found that the cues prepared in fresh ocean water and held indoors, in pails containing sand substrates, could last for over 3 h in the laboratory. This is vastly different from what we observed under natural conditions.
When a prey animal is captured by a predator and alarm cues are released, the cues not only breakdown, but also become dispersed due to water movements, etc. Our first experiment provided a clear indication of a threat-sensitive response to varying concentrations of alarm cues. As the concentration of cues increased, we saw greater avoidance of the cue. This work is in accordance with several studies that have demonstrated similar concentration effects in other systems (Ferrari et al. 2005). The challenge for researchers that want to understand the importance of chemical cues as information sources will be to understand how prey integrate information about degradation and cue dispersal to make informed decisions. Indeed, there is another intriguing possibility that may also come into play. Are prey able to determine the age of chemical cues irrespective of dilution and degradation? If this were the case, we should expect that the shape of the degradation curve would be different than the shape of the dilution curve. We did not have a consistent degradation curve in our experiment; the shape of the curve (i.e., the rate at which the fish quit responding to the cues) was dependent on time of day. This result is consistent with the possibility that fish can determine the age of chemical information irrespective of degradation and dilution. This hypothesis would be much easier to address if we knew the chemical identity of the alarm cues, but to date, we do not have this information (Ferrari et al. 2010). Bytheway et al. (2013) recently used GC-MS to show that aged predator odors are indeed different than new predator odors and that the difference in the cues may allow rats the ability to age predator cues. Ecologists often think of chemical cues released during predator attacks as long-lasting chemicals that linger in the area. As such, they provide some information about risk but the spatial and temporal aspects of the information are somewhat unreliable. We need to think about the natural rate of breakdown of the cues and the environment in which the cues are dispersing to gain a full appreciation of the spatial and temporal limitations of chemical information sources.