protecting egg prey from carrion crows: the potential of aversive conditioning

Applied Animal Behaviour Science 87 (2004) 325–342

Protecting egg prey from Carrion Crows:the potential of aversive conditioning

Ruth Coxa,∗, Sandra E. Bakera, David W. Macdonalda,Manuel Berdoya,b

a Department of Zoology, Wildlife Conservation Research Unit, South Parks Road, Oxford OX1 3PS, UKb Oxford University Veterinary Sciences, Parks Road, Oxford OX1 3PT, UK

Received 12 May 2003; received in revised form 23 December 2003; accepted 25 January 2004

Abstract

Carrion Crows,Corvus corone, are held responsible for taking the eggs and chicks of many birdspecies. In areas of conservation significance, intervention may be required. Traditionally, managershave attempted to control predation by killing predators, but this may not be the most effective ordesirable approach. A non-lethal alternative, which might protect vulnerable eggs from damage bycorvids, is conditioned taste aversion (CTA), a process in which animals learn to avoid certain foodsfollowing consumption of a toxin. CTA evolved to reduce the likelihood of poisoning but may beinduced deliberately using an emetic.

In this study, we attempted to train wild-caught Carrion Crows to avoid eggs of a preferred colour,using three different doses of the emetic Carbachol. During pre-conditioning, we established eachcrow’s preferred egg colour (subsequently known as ‘toxic’). During conditioning, each bird wasallowed access to a single Carbachol-treated ‘toxic’ egg for 4 h per day. We compared pre- andpost-conditioning responses to untreated eggs of ‘toxic’ and ‘non-toxic’ colour.

Crows took longer to attack untreated eggs after conditioning with Carbachol (all doses combined).This was the result of a reduction in attacks on ‘toxic’ eggs after 2 h, while there was no change inattacks on ‘non-toxic’ eggs.

The total number of eggs attacked after 4 h did not change following conditioning, however theamount of egg consumed increased. This resulted from an increase in the number of ‘non-toxic’eggs attacked and consumed. There was no change in the attack or consumption of ‘toxic’ eggs. Thehighest Carbachol dose tested (381 mg kg−1 body weight) created aversions as described above forthe combined results.

We have shown that Carbachol can be used to manipulate crow predation on eggs in captivity.Future work should focus on captive and field trials based on a dose of 381 mg kg−1 body weight.

∗ Corresponding author. Present address: Department of Biological and Biomedical Sciences, University ofDurham, South Road, Durham DH1 3LE, UK. Tel.:+44-191-334-1297; fax:+44-191-334-1201.E-mail address:[email protected] (R. Cox).

0168-1591/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.applanim.2004.01.008

326 R. Cox et al. / Applied Animal Behaviour Science 87 (2004) 325–342

CTA could provide a more effective and more desirable alternative to culling for controlling predationby corvids on eggs of conservation significance. Such techniques might have widespread applicationto predation situations across a range of animal species worldwide.© 2004 Elsevier B.V. All rights reserved.

Keywords:Carbachol; Carrion Crow; Conditioned taste aversion; Conservation management; Corvid; Emetic

1. Introduction

Corvids are often held responsible for taking the eggs and chicks of many bird speciesand are therefore of conservation concern. Traditionally, predators have been controlled bylethal means however these methods are becoming unpopular for ethical reasons, and may beinappropriate if the predator is of conservation significance (Baker and Macdonald, 1999).Moreover, culling may be ineffective as a management tool (Draulans, 1987), not leastbecause predators that have been removed may be replaced quickly through immigrationby conspecifics, or through density-dependent compensations in the fertility or survival ofremaining members of the population (Reynolds et al., 1993). This is an instance of thewider perturbation hypothesis, which raises the concern that by disturbing the behaviourof survivors, lethal control may prove counter-productive (Tuyttens and Macdonald, 2000).Modifying the behaviour of a resident predator population through a non-lethal processcalled conditioned taste aversion (CTA) might avoid some of these problems by allowingthe predator to remain in its ecological niche (Cowan et al., 2000). It may also be more costeffective than other non-lethal methods such as repellents, which can allow predators todiscriminate between baits and untreated food (Conover, 1984) or startle devices, to whichpredators may become habituated (Conover, 1979).

CTA is a process that has evolved in natural populations to reduce the risk of poisoningthrough the ingestion of toxins (Garcia and Hankins, 1977). This involves an animal learningto avoid a specific food as a result of illness. To form a CTA an animal must ingest a toxicfood item and associate the resulting illness with the taste of that food after a single or smallnumber of exposures to it (Nicolaus and Nellis, 1987). CTA is widespread across the taxafrom molluscs to humans (Gustavson, 1977), and was first recognised in rats byGarciaet al. (1955). The phenomenon is one of the most intensively studied learning processes(Cowan et al., 2000), with the laboratory rat as the predominant subject (see review papersbyGamzu (1977),Garcia and Koelling (1972)andKalat (1977), bibliographies byRiley andBaril (1976), Riley and Clarke (1977)andRiley and Tuck (1985), and screening studiesby Gill et al. (2000)and Massei and Cowan (2002)). CTA can be caused deliberatelyby administering an emetic (e.g.,Reynolds, 1999; Gustavson et al., 1974; Nicolaus andNellis, 1987; Conover, 1990) and can potentially be exploited for purposes of non-lethalpredator management (Baker and Macdonald, 1999). The aim is to elicit prey avoidanceduring future encounters (Horn, 1983). Predators may learn to avoid prey at a distancebased on distal identity cues such as colour, size, shape or smell, through one of twoprocesses. The first is ‘second-order conditioning’, a two-stage process in which, initially,taste becomes associated with illness via CTA (but non-taste cues may still elicit attack), andthen the aversive taste conditions other senses, so that, for example, the food odour becomes

R. Cox et al. / Applied Animal Behaviour Science 87 (2004) 325–342 327

punishing, and subsequently inhibits approach or attack (Garcia et al., 1973; Gustavson et al.,1974, 1976). The second is ‘potentiation’, a single-stage process in which exposure to tastedirectly facilitates aversion to weaker food cues, e.g. odour or colour, so that they becomestrong cues in their own right (Braun and Ryugo, 1974; Clarke et al., 1979; Rusiniak et al.,1979; Westbrook et al., 1980).

Several studies have assessed the protection afforded by CTA to commercially impor-tant species such as sheep (Gustavson, 1977; Burns, 1983; Horn, 1983). A larger bodyof work has been concerned primarily with using CTA to reduce egg predation by bothbirds (e.g. American Crows:Corvus branchrynchus(Nicolaus et al., 1983, 1989; Dimmickand Nicolaus, 1990); Ravens:Corvus corax(Avery et al., 1995); Hooded Crows:Corvuscorone cornix(Bogliani and Bellinato, 1998)) and mammals (Small Indian Mongooses:Herpestes auropunctatus(Nicolaus and Nellis, 1987; Conover, 1990); Raccoons:Procyonlotor (Semel and Nicolaus, 1992; Ratnaswamy et al., 1997); laboratory rats:Rattus norvegi-cus(Gill et al., 2000)). Studies such as these have recognised that social and environmentalfactors can influence the development of CTA and that there may be variation between thesexes and between individuals. Important factors include the predator’s experience of prey,the importance of target prey in the predator’s diet, species differences in the cues usedto identify noxious prey, the influence of conspecifics in developing food preferences oraversion and the conditioning agent used. To constitute an effective wildlife managementtool, CTA would need to produce a robust, long-lasting avoidance of the referent food (Gillet al., 2000). It is crucial that safe, effective aversive agents are identified for the species andsituation in question before CTA can be applied as a non-lethal control technique (Masseiand Cowan, 2002).

There is potential to manage corvids using CTA as a humane alternative to culling.Nicolaus et al. (1983)first tested the concept and reported that free-ranging AmericanCrows stopped eating chicken eggs after they had been exposed to eggs treated with theillness inducing chemical trimethacarb. In our study, we investigated the use of CTA withwild-caught Carrion Crows,Corvus coroneL., using the chemical Carbachol (carbamylcholine chloride). The biology of Carrion Crows meets the minimum requirements forCTA, as summarised byNicolaus and Nellis (1987). Adult crows vary little in size (Perrins,1994) and so their meal size and dose of aversive agent received (per kg body weight) willbe relatively uniform. Carrion Crows feed on a variety of food items (Perrins, 1994), soremoval of eggs from their diet should not result in food deprivation. Finally, because theydefend territories, areas where aversions are established are unlikely to receive a flow ofnew immigrants (which would also require conditioning).

Carbachol is a widely available emetic that is water soluble, odourless and tasteless atdoses known to be capable of producing CTA in other species (L.K. Nicolaus, unpublisheddata, 1986) including mongooses (Nicolaus and Nellis, 1987), American Crows (Nicolauset al., 1989) and Hooded Crows (Bogliani and Bellinato, 1998). However there is one po-tential disadvantage of cholinergic agonist emetics (which increase neurotransmission andresult in vomiting), such as Carbachol. They are toxic and repeated doses could poisonboth target and non-target species (Conover, 1990). Although some work has been donein closely related species, the response of Carrion Crows to repellents has not been testedand a safe dose that may be used successfully as an aversive agent in the field has notbeen established. It is important to identify an optimal effective dose because crows that


learn to avoid treated eggs entirely tend to extend their avoidance response to untreatedeggs for longer periods. In contrast, smaller doses of emetic, which only cause slight ill-ness, result in crows repeatedly sampling treated and untreated eggs (Avery and Decker,1994).

We used captive wild-caught Carrion Crows to examine aversive conditioning to twoegg types: green egg on green background (‘cryptic’) and yellow egg on green background(‘conspicuous’). Our aims were as follows:

(1) To generate a colour-specific aversion to preferred egg types using Carbachol.(2) To establish whether crows developed a generalised aversion to chicken eggs.(3) To assess the duration of any aversion produced, in terms of the number and type

(colour) of eggs attacked and the amount of egg consumed.(4) To examine the safety and effectiveness of each dosage rate of Carbachol.(5) To make recommendations regarding how CTA could be developed for future use in

wildlife management.

2. Materials and methods

2.1. Subjects

Eleven wild-caught crows were housed individually in outdoor aviaries between late Apriland July 1999. Each aviary was divided into two units, with individuals separated from oneanother by a see-through net partition (Fig. 1). Each unit consisted of a sheltered and anoutdoor section and contained similar apparatus including one horizontal beam across therear of the aviary (this supported an egg-eating platform (30 cm× 30 cm) with a 1 cm rimto prevent eggs from rolling off), plus several other perches in standard positions. Some ofthe aviaries were equipped with a video recorder so that we could film a total of six birdsduring the trial.

2m

6m

Egg box

Sheltered sectionSheltered section Net partition

Outdoorsection

Outdoorsection

Fig. 1. A two-cage unit. One crow was kept on either side of the central net partition; both had access to shelteredand outdoor sections.


All birds were weighed immediately after capture (mean weight 426.8 ± S.E. 9.00 g)and marked with two coloured leg-rings for identification purposes. They were weighedagain a maximum of 2 days before conditioning and at the end of the trial. Birds had freeaccess to water throughout and (except during egg presentation), were fed an ad libitumdiet of tinned pet food, and soaked cat biscuits (Go Cat Friskies Complete, duck, rabbit andchicken), as well as occasional mealworms. The birds were fed each day and the aviariescleaned every 2 days. For the first few days the birds were otherwise left undisturbed. Priorto the main experiment (with eleven birds), we conducted a pilot study with four birds 6months previously.

2.2. Experimental design

In addition to their staple diet, birds were presented with raw medium-size chickeneggs on a daily basis. Eggs were painted yellow or green with domestic food dye (see baitpreparation). The experiment was divided into three phases: pre-conditioning (25–39 days),conditioning (9–14 days) and post-conditioning (4–12 days).Fig. 2shows the duration ofeach experimental phase. During pre- and post-conditioning, each crow was offered fouruntreated eggs per day, while during conditioning, each bird received a single egg treatedwith Carbachol, of the colour preferred during pre-conditioning. This treatment regimeallowed comparison of the birds’ responses before and after conditioning.

2.3. Bait selection

Carrion Crows naturally predate the eggs of lapwings, curlews and other wading birds,which approximately match the size of the chicken eggs that we used (average weight of48 g). Results of the pilot study indicated that four chicken eggs were sufficient to assessthe amount of egg eaten per day and therefore to detect changes in the birds’ consumptionfollowing conditioning.

123456789

101112131415

0 10 20 30 40 50 60 70 80

Days post capture

Bir

d ID

Captivity

Habituation

Pre-conditioning

Conditioning

Post-conditioning

Fig. 2. Calendar illustrating the duration of each experimental phase. Bird 11 and 15 died on the first day ofconditioning (see ‘safe dosing’). Birds 12–15 inclusive were those used for the pilot study.


2.4. Establishing safe and effective doses of Carbachol

When administering drugs to wild animals it is common to extrapolate doses from otherspecies, where the dose rate is known, by allometric scaling (Wolfensohn and Lloyd, 1998).We referred to previous work as a guide to determine a dose of Carbachol that mightsafely create a lasting aversion after one or two exposures.Prescott et al. (1997)used threedifferent doses of Carbachol in an attempt to deter Magpies (Pica picaL.) (average bodyweight 216 g) from eating quail eggs. The highest dose of 244 mg kg−1 body weight (BW)of Carbachol (50 mg in one quail egg per day) was the most effective at producing anaversion, with no observable long-term ill effects. We began by testing an equivalent dose(dose 1) with Carrion Crows (calculated by allometric scaling) of 190.5 mg kg−1 BW ofCarbachol. The appropriate dose was calculated for each individual bird according to itsown body weight measured a maximum of 2 days prior to conditioning.

2.5. Bait preparation

Eggs were soaked in vinegar for 1 min, rinsed thoroughly in water and left to dry. Thevinegar treatment made the shell surfaces more accepting of dye. Eggs were painted withtwo layers of food dye. Treated eggs for the conditioning phase were tailor-made for specificbirds according to their pre-conditioning egg colour preference and their body weight. Eggswere treated as follows. Opposite ends of the shell were pierced with a pin and air wasblown through one hole forcing the contents out through the opposite hole. One hole wasthen sealed with glue using a hot-glue gun. Carbachol was weighed on electric scales (tofive decimal places), and mixed thoroughly with 0.5 g of water. The emetic solution wasmade up to 48 g (the average contents of the eggs used) with whisked egg, and this mixturethen injected into the empty eggshell. The remaining hole was sealed with glue and theegg painted the appropriate colour according to the preference of the bird for which it wasprepared. Eggs were refrigerated and used within 3 days of preparation.

2.6. Bait presentation

Presentation of eggs began on average 7 days after captivity. Initially some birds hadto be trained to eat the eggs. This involved offering opened eggs in their food bowls andegg boxes (see below) and removing other food for a few hours per day (no longer than16 h).

A trial began once a bird had eaten eggs consistently for at least three consecutive days.Eggs were presented to the crows for a total of 4 h at approximately the same time eachday (late morning). We found that 4 h was a suitable amount of time to allow all birdsto eat (taking into account that birds feed at different rates), whilst encouraging them tofeed immediately and rapidly. Previous conditioning trials have suggested that captive RedFoxes (Vulpes vulpes) (Macdonald and Baker, 2004) and magpies (Prescott et al., 1997)may have been tempted to sample untreated food during post-conditioning as a result ofbeing kept in close proximity with it for 24 h a day. Under natural conditions, wild predatorsare unlikely to be faced with potential prey for very long, but rather eat the prey or leave thearea.


100cm

50cm

50cm

Green presentation box Central partition

Eggs placed on topof plastic tubes

Fig. 3. Egg presentation box. Eggs were placed on top of plastic tube stands, situated on each side of the centralpartition.

Eggs were presented in green painted chipboard ‘egg boxes’ (50 cm× 50 cm× 100 cm).They were placed on top of short lengths of plastic tube, situated on either side of the centralpartition (Fig. 3) (yellow eggs on one side, green on the other). The green backgroundcontrasted with the yellow (conspicuous) eggs and matched the green (cryptic) eggs. Eggcolours were alternated daily between the left- and right-hand sides of the box. During theconditioning phase, the position of the single treated egg was switched daily between frontand back, and left or right compartment of the egg box. In the pilot study, birds were ableto see their neighbour’s eggs, however during the main experiment the boxes were turnedaround to prevent birds from observing and being influenced by each other’s choices.

2.7. Establishing preference

Direct observations were made 1, 2 and 4 h after the eggs had been placed in the enclosure.These observation intervals were selected to reduce the chance that a bird would eat all foureggs before their colour preference could be established, while restricting human contactto minimise disturbance to their feeding behaviour.

At each observation, we recorded an ‘attack score’ for each egg—whether it was‘untouched’, ‘moved’, ‘attacked’ and/or ‘consumed’. After 4 h, the eggs (and any remains)were collected and, as well as the ‘attack score’, each egg was assigned a ‘consumptionscore’–a number between 0 and 10 according to the amount consumed: 0 indicating nothingconsumed, 0.5 for a small crack in the shell (too small for fluid loss), 1–9 for relative amountconsumed and 10 if the whole egg, including the shell, was eaten. This scoring system wasbased on the design used byPrescott et al. (1997). Egg scoring was practiced for severaldays under experimental conditions before pre-conditioning began.

Each bird’s preferred egg type was determined from pre-conditioning data (seeSection 3).The preferred egg colour was treated during conditioning and is referred to as ‘toxic’, theother colour is referred to as ‘non-toxic’ during all phases.


2.8. Dosing

One of the main aims of the experiment was to establish a safe and effective dose ofCarbachol to produce an aversion to eggs in captive crows. We tested the effect of threedifferent doses. Dose 1 (190.5 mg kg−1 BW) was tested on six birds (four in the pilot study,two in the main study). Having assessed the results of dose 1 (seeSection 3), we decided toincrease the concentration of Carbachol by 50% (dose 2—285.75 mg kg−1 BW) and testedthe effect on five birds. Doses 1 and 2 had very little effect and so we tested three birds withdose 3—a 100% increase of dose 1 (381 mg kg−1 BW). Although this approach meant thatthe duration of captivity before conditioning varied for each bird, it allowed us to assess theeffectiveness of a dose before attempting to condition the remaining birds.

During the experiment, two of the birds died; one during the pilot study and the otherduring the main experiment. The deaths both occurred on the first day of conditioning. Onebird received a dose of 190.5 mg kg−1 BW (dose 1), and the other 381 mg kg−1 BW (dose3). All of the birds in the pilot study lost between 10 and 21% of their original body weight(average 580 g at the start of the experiment). As a result the birds in the main experimentwere weighed on the day of capture, 1 or 2 days before conditioning began, and at the end ofpost-conditioning. None of these, including the second bird that died lost any weight duringtheir time in captivity. These birds had a lower average body weight (427 g) than those ofthe pilot study (that were caught in summer). Mortality did not appear to be associated withbody weight as both birds showed a similar pattern to others caught at the same time. Ratherthe deaths are more likely to be a result a combination of factors including variations inbody condition at the time of capture and individual variation in the birds’ physiologicalresponse to Carbachol.

2.9. Data analysis

Analysis was carried out on the results from the main study only. Consumption andattack data for each bird were averaged over 10 days before conditioning began and overall post-conditioning days (maximum of 14 days). We tested the development of aversionby comparing birds’ preference for attacking or consuming toxic and non-toxic eggs duringthe post-conditioning period with that recorded before conditioning. Analyses (two-tailed)involved using the GLM procedure, and a repeated measures approach to control for a birdeffect (SAS, 1996). Normality of residuals was checked, and the data transformed wherenecessary.

3. Results

3.1. Video observation

We used video surveillance in six of the enclosures to gather information on the birds’behavioural responses. Three (of the six) birds showed evidence of Carbachol inducedsickness. One bird (treated with dose 1) was seen vomiting shortly after consuming a treatedegg and only attacked eggs on the first day of post-conditioning, but none thereafter. This


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

PRE POST

Conditioning phase(a)

(b)

PRE POST

Conditioning phase

Atta

ck s

core

Atta

ck s

core

Non toxic egg Toxic egg (P=0.0007) Total (P=0.041)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Non toxic egg Toxic egg (P=0.001) Total

Fig. 4. (a) Mean (±S.E.) egg attack scores after 1 h of presentation during pre- and post-conditioning (all doses com-bined) (n = 10 birds). (b) Mean (±S.E.) egg attack scores after 2 h of presentation during pre- and post-conditioning(all doses combined) (n = 10 birds). (c) Mean (±S.E.) egg attack scores after 4 h of presentation during pre- andpost-conditioning (all doses combined) (n = 10 birds).

individual was not observed vomiting during either pre- or post-conditioning. The other twobirds (treated with doses 1 and 3) were not seen vomiting, although their activity appearedsubdued and they performed large amounts of beak wiping for some time after eating treatedegg. Both birds continued to attack and consume egg during post-conditioning.


0

0.5

1

1.5

2

2.5

3

PRE POST

Conditioning phase

Atta

ck s

core

Non toxic egg (P=0.036) Toxic egg Total

(c)

Fig. 4. (Continued).

3.2. Egg attack and consumption

Eggs that were classed as ‘moved’, ‘attacked’ and ‘consumed’ were pooled during anal-ysis of data collected in the first 1 and 2 h to represent eggs that had not ‘survived’. We com-pared the number of toxic and non-toxic eggs surviving between pre- and post-conditioning.(In all figures, significant changes between pre- and post-conditioning are indicated in thelegend withP values.)

Overall (all doses combined) birds began attacking eggs later during post-conditioningthan during pre-conditioning. One hour after egg presentation the total number of eggsattacked was significantly lower than during pre-conditioning (F1,9 = 5.72, P = 0.041)(Fig. 4a), but this difference, while approaching significance, was less marked 2 h after pre-sentation (F1,9 = 4.85,P = 0.055) (Fig. 4b). Attacks on toxic eggs decreased significantlyfor the first 2 h of presentation (after 1 hF1,9 = 25.44,P = 0.0007, after 2 hF1,9 = 21.96,P = 0.001) whilst the number of non-toxic eggs attacked did not change during this time(after 1 hF1,9 = 0.17,P = 0.691, after 2 hF1,9 = 1.34,P = 0.277).

After 4 h of egg presentation the total number of eggs attacked did not change (F1,9 =3.79, P = 0.083) (Fig. 4c), however the total amount consumed increased significantly(F1,9 = 5.57, P = 0.043) (Fig. 5). This resulted from an increase in the total number ofnon-toxic eggs attacked (F1,9 = 6.05,P = 0.036) and consumed (F1,9 = 10.59,P = 0.01).

Analysis of individual doses showed that doses 1 and 2 had little influence on egg preda-tion however dose 3 did have significant effects. We did not find a significant effect of anydose after 1 h, although doses 1 and 3 caused the number of attacks on toxic eggs (Fig. 6a


0

1

2

3

4

5

6

7

8

9

10

PRE POST

Con

sum

ptio

n sc

ore

Non toxic egg (P=0.01) Toxic egg Total (P=0.043)

Conditioning phase

Fig. 5. Mean (±S.E.) egg consumption after 4 h of presentation during pre- and post-conditioning (all dosescombined) (n = 10 birds).

for dose 3) to decrease to zero for the first hour of post-conditioning. No significant effectsof dose 1 or 2 were found after 2 or 4 h. However dose 3 resulted in significant changes.After 2 h this dose caused a decrease in the number of toxic eggs attacked post-conditioning(F1,2 = 29.58, P = 0.032) (Fig. 6b) while the number of non-toxic eggs (F1,2 = 1.47,P = 0.349) and the total number of eggs (F1,2 = 0.65, P = 0.504) attacked did notchange. After 4 h dose 3 caused a significant increase in the number of non-toxic eggsattacked (F1,2 = 22.18,P = 0.042), whilst there was no change in number of toxic eggs(F1,2 = 1.00, P = 0.423) and total number of eggs attacked (F1,2 = 3.73, P = 0.193)(Fig. 6c) or the amount of egg consumed (F1,2 = 1.74,P = 0.318).

4. Discussion

In this study, we attempted to generate CTA in Carrion Crows to raw chicken eggs usingthe conditioning agent Carbachol. We aimed to establish whether crows would develop acolour-specific or a generalised aversion to chicken eggs. We assess the effectiveness of theemetic and make recommendations for future use.

4.1. Strength of aversion

Birds began attacking toxic eggs later during post-conditioning than pre-conditioning.Doses 1 and 2 had no significant effect at these sample sizes, however dose 3 significantlyreduced attacks on toxic eggs for at least 2 h after egg presentation. Overall, after 4 h


0

0.1

0.2

0.3

0.4

0.5

PRE POST

Conditioning phase(a)

(b)PRE POST

Conditioning phase

Atta

ck s

core

Atta

ck s

core

Non toxic egg Toxic egg Total

0

0.2

0.4

0.6

0.8

1

1.2

Non toxic egg Toxic egg (P=0.032) Total

Fig. 6. (a) Mean (±S.E.) egg attack scores after 1 h of presentation during pre- and post-conditioning (birds exposedto dose 3: 381 mg kg−1 body weight of Carbachol) (n = 3 birds). (b) Mean (±S.E.) egg attack scores after 2 h ofpresentation during pre- and post-conditioning (birds exposed to dose 3: 381 mg kg−1 body weight of Carbachol)(n = 3 birds). (c) Mean (±S.E.) egg attack scores after 4 h of presentation during pre- and post-conditioning (birdsexposed to dose 3: 381 mg kg−1 body weight of Carbachol) (n = 3 birds).

there was an increase in the number of non-toxic eggs attacked and amount consumedpost-conditioning (this change was significant for dose 3 only). Why an aversion should beexpressed in this way (as an increase in attack and consumption of non-toxic eggs) is unclear.We suggest that birds increased their (non-toxic) egg intake during post-conditioning to


0

0.5

1

1.5

2

2.5

3

PRE POST

Conditioning phase(c)

Atta

ck s

core

Non toxic egg (P=0.042) Toxic egg Total

Fig. 6. (Continued).

compensate for the period of conditioning (during which the only eggs available weretoxic-coloured and treated with Carbachol). Although attack and consumption of non-toxiceggs increased, the total number of eggs attacked and consumed did not change. The dosemay have been strong enough to produce a change in colour preference, but no overallreduction in the total number of eggs attacked or the amount eaten.

Conditioning needs to provide total protection to the target egg as a small amount ofdamage may render an egg unviable. Although we did not prevent eggs from being attackedcompletely, we did delay sampling for at least 2 h. This can be seen clearly by comparingFig. 4a–c. It is also apparent inFig. 6a–c which show the effect of dose 3 over 1, 2 and 4 h.It is probable that after 2 h birds were tempted to try the eggs due to the proximity enforcedby captive conditions (seeMacdonald and Baker, 2004; Prescott et al., 1997). Higher dosesthan those tested here may produce a longer lasting conditioned aversion than we haveobtained.

One can only speculate as to how such an aversion would be manifest in a field situation,however, a Carbachol dose that increases protection for a period of time in captivity mayprovide much greater protection for eggs in the wild, where crows have alternative foodsources and freedom of movement.

Proximity has played a part in other conditioning studies.Brett et al. (1976)noted such aneffect while studying prey-lithium aversions in raptors. They argued that a raptor in the wildmight be more selective than one in captivity (and therefore more likely to be conditionedafter consuming treated prey), since the costs involved in sampling a ‘suspicious-looking’prey could be greater.Nicolaus and Nellis (1987)showed that captive mongooses formedaversions to egg flavour after consuming a dose of Carbachol in egg. However the mongoosesdid not avoid toxic eggs at a distance based upon an olfactory cue, instead they attacked,


tasted and then avoided eggs, whether they contained Carbachol or not. When tested inthe field, free ranging mongooses avoided eggs at a distance after consuming Carbachollaced eggs, in contrast to their behaviour at control sites. This avoidance resulted in intactsurvival of 72% of available treated eggs (compared to 10% during the baseline tests), andalso extended to eggs that did not contain Carbachol. Avoidance at a distance enhances thelongevity of CTA because extinction of the aversion often results from repeated samplingof prey without illness consequences (Testa and Ternes, 1977).

Restriction of dietary diversity can play a part in the development of aversion.Nicolauset al. (1989)found that food deprived ravens given a 10 mg dose of Carbachol in a mealacquired a lasting aversion, but that the effect was probably exaggerated compared to thelikely outcome if this dose had been tested on wild, ad libitum feeding ravens. In thesame study 18 mg of Carbachol in sweet green eggs produced long-term changes in foodpreferences among free-ranging American Crows (which were not food deprived), but moreeggs were consumed before the aversion became evident.

Interestingly, other studies have recorded similar results to ours in terms of an increasein post-conditioning consumption.Nicolaus et al. (1989)found that captive food deprivedravens treated with a dose of 10 mg of Carbachol initially increased their post-test con-sumption, but that subsequent consumption levelled off below baseline rates.Prescott et al.(1997)noted that magpies exposed to the lower Carbachol doses used in their experiments(16 and 32 mg per egg, equivalent to 70.5 and 141.0 mg kg−1 BW, respectively) attackedmore non-toxic eggs post-conditioning, while there was no change in the number of at-tacks on toxic eggs. They suggest that the dose produced a change in colour preferenceonly. Magpies presented with eggs containing a higher dose of 50 mg Carbachol per egg(243.9 mg kg−1 BW) developed a general aversion to the taste of raw egg, reducing theamount of both toxic and non-toxic colour eggs that they sampled.

4.2. Efficacy of aversion

An important factor in the success of aversion learning is individual variation (e.g.Semeland Nicolaus (1992)on free ranging raccoons;Prescott et al. (1997)on captive magpies).Egg predation, before and after exposure to Carbachol, varied between individuals in ourstudy. Possible reasons for a difference before conditioning include prior experience (whichmay be influenced by parental guidance), while differences after conditioning could alsobe attributed to the intensity and rapidity of illness experienced by different birds.

Inevitably birds will have differed in their experiences of consuming eggs prior to capture,which may have influenced their food preferences and their responses to conditioning duringthe trial. During captivity all birds were familiarised with ‘safe’ eggs during habituationand pre-conditioning. The step-by-step method that we used for testing three differentconcentrations of Carbachol (allowing us to assess the effectiveness of a range of doses)meant that birds were exposed to ‘safe’ eggs for different periods of time. Stimulus of a CTAdepends, among other factors, on whether the conditioned stimulus (here the taste of egg)is novel or familiar to the target species (Massei and Cowan, 2002; Nachman et al., 1977).Ours was therefore a conservative test of repellency, because repeated sampling of ‘safe’untreated baits before conditioning began may have endowed treated baits with ‘learnedsafety’. This is a phenomenon whereby familiarity with a particular food attenuates the


acquisition of aversions to that food (Kalat and Rozin, 1973). ‘Learned safety’ may havereduced the likelihood of producing an effective CTA in this study, thereby adding weight toour significant results. Our approach (of exposing birds to ‘safe’ foods in a pre-conditioningphase) is justified because wild animals are likely to have had at least some access to safetarget food before conditioning commences for management purposes. Many other studieshave used a similar design for this reason (Dimmick and Nicolaus, 1990; Gustavson et al.,1974, 1976; Nicolaus et al., 1982; Nicolaus and Nellis, 1987; Semel and Nicolaus, 1992).Clearly, prior natural experience of a ‘safe’ target food could influence the stimulus of anaversion towards that food. However, familiarity with the target food among free-rangingtarget animals could also work to the manager’s advantage. For example, animals that arenaturally more active egg predators may be more likely to eat treated eggs, and, in sodoing, dose themselves during a conditioning exercise (Semel and Nicolaus, 1992). Thefood preferences of young animals may be influenced by those of their parents. For similarreasons, parents may also (indirectly) pass learned aversions on to their offspring, by denyingthem the opportunity to experience foods that they themselves have been conditioned toavoid (Semel and Nicolaus, 1992).

Nicolaus and Nellis (1987)cite intensity of illness as a variable in the acquisition ofan aversion. During our experiments, Carbachol was evenly distributed within the treatedwhisked egg, although it is likely that individual birds experienced different degrees of ill-ness based on variable intake. It is also possible that different birds’ physiological responsesto the same emetic dose will depend on other factors such as their state of health.

The length of conditioning phase for our birds varied between eight and 14 days, howeverthis did not seem to determine whether an aversion was induced.Prescott et al. (1997)suggested that there was little advantage to extending the conditioning phase beyond aweek and concluded that aiming for a ‘short, sharp shock’ was the most effective strategyfor establishing an aversion.

4.3. The potential of Carbachol as an aversive agent

Carbachol has been safe and very effective at inducing CTA to raw egg in other birds,e.g. American Crows (Nicolaus et al., 1989) and magpies (Prescott et al., 1997). Theirresults suggest that by treating eggs, which mimic those of prey species (with doses upto 244 mg kg−1 BW), it may be possible to confer strong CTA protection on a particularspecies of egg, as well as eggs in general.

In ways similar to those suggested bySemel and Nicolaus (1992)(who studied CTAin raccoons at an artificial feeding site), andPrescott et al. (1997)(on captive magpies),this study probably represents a conservative test of the power of Carbachol based CTA toalter the food preferences of wild crows. Captive circumstances such as forced proximityto eggs, access to ‘safe’ eggs and restriction of dietary diversity may have increased the rateof attack in our experiment.

Under the experimental conditions of this study, Carbachol proved to be a potentiallyuseful aversive agent for deterring crows from eating eggs. Although we did not producedramatic aversions, we have shown that Carbachol can be used to manipulate crow predationon eggs in captivity, and that illness-based avoidance of eggs could be achieved. Furtherwork should attempt to establish a safe dose that could be effective after one or two exposures


and have a lasting effect. Aversive conditioning should ideally give total protection to thetarget egg. Alteration of the egg colour preference or the length of time taken to attack anegg, may, however, offer some degree of protection in the wild. With a change in the numberof chicken eggs attacked as the key criteria, the highest Carbachol dose (381 mg kg−1 BW)would appear to be a good starting point on which to base further captive and field trials.

CTA and other non-lethal approaches to predator management merit further investigationand could provide an effective alternative to culling. Potential problems with using CTAfor wildlife management purposes include extinction of aversions (Testa and Ternes, 1977),and incursions by new individuals (Gustavson et al., 1976). Conditioned taste aversions aretherefore most likely to be effective where the target animal is territorial (Cowan et al., 2000;Nicolaus, 1987; Nicolaus et al., 1992; Reynolds, 1999), as are Carrion Crows. And, sinceeggs are vulnerable to predation for just a few weeks each year, protection would be requiredfor a limited period only. Rather than exposing crows to ‘safe’ untreated eggs (and therebyrisking providing Carbachol-treated eggs with ‘learned safety’ (Kalat and Rozin, 1973),presentation of treated eggs should commence prior to the predicted time of damage (Averyand Decker, 1994). Because the onset of the vulnerable period is predictable, distributionof treated eggs should occur prior to the onset of egg laying to maximise effectiveness, andtreatment should continue throughout the incubation period (Bogliani and Bellinato, 1998;Conover, 1990; Semel and Nicolaus, 1992). The broadly synchronous production of eggsby different species improves the feasibility of this prospect.

Acknowledgements

This study was funded by the Royal Society for the Protection of Birds. We are gratefulfor the help of Dr. Rhys Green, James Davys, Amy Dickman, Stephen Ellwood, BeckyLander, Judith Lloyd, Matt Prescott, Barry Sparrowhawk and Dave Wilson.

References

Avery, M.L., Decker, D.G., 1994. Responses of captive fish crows to eggs treated with chemical repellents. J.Wildlife Manage. 58, 261–266.

Avery, M.L., Pavelka, M.A., Bergman, D.L., Decker, D.G., Knittle, C.E., Linz, G.M., 1995. Aversive conditioningto reduce raven predation on California least tern eggs. Colon. Waterbird 18, 131–245.

Baker, S.E., Macdonald, D.W., 1999. Non-lethal predator control-exploring the options. In: Proceedings of theFirst Vertebrate Control Conference, 1997, York, UK.

Bogliani, G., Bellinato, F., 1998. Conditioned aversion as a tool to protect eggs from avian predators in heroncolonies. Colon. Waterbird 21, 69–72.

Braun, J.J., Ryugo, R.A., 1974. Taste facilitation of a learned odor aversion. B. Psychonomic Soc. 4, 253.Brett, L.P., Hankins, W.G., Garcia, J., 1976. Prey-lithium aversions. III. Buteo Hawks. Behav. Biol. 17, 87–98.Burns, R.J., 1983. Microencapsulated lithium chloride bait aversion did not stop coyote predation on sheep. J.

Wildlife Manage. 47, 1010–1017.Clarke, J.C., Westbrook, R.F., Irwin, J., 1979. Potentiation instead of overshadowing in the pigeon. Behav. Neural

Biol. 25, 18–29.Conover, M.R., 1979. Response of birds to raptor models. Proc. Bird Contr. Semin. 8, 16–24.Conover, M.R., 1984. Response of birds to different types of food repellents. J. Appl. Ecol. 21, 437–443.


Conover, M.R., 1990. Reducing mammalian predation on eggs by using a conditioned taste aversion to deceivepredators. J. Wildlife Manage. 54, 360–365.

Cowan, D.P., Reynolds, J.C., Gill, E.L., 2000. Reducing predation through conditioned aversion. In: Gosling, L.M.,Sutherland, W.J. (Eds.), Behaviour and Conservation, vol. 2. Cambridge University Press, UK, pp. 281–299.

Dimmick, C.R., Nicolaus, C.K., 1990. Efficiency of a conditioned aversion in reducing depredation by crows. J.Appl. Ecol. 27, 200–209.

Draulans, D., 1987. The effectiveness of attempts to reduce predation by fish-eating birds. A review. Biol. Conserv.41, 219–232.

Gamzu, E., 1977. The multi-faceted nature of taste-aversion-inducing agents: is there a common factor? In: Barker,L.M., Best, M.R., Domjan, M. (Eds.), Learning Mechanisms in Food Selection. Baylor University Press, Waco,TX, USA, pp. 477–510.

Garcia, J., Clarke, J.C., Hankins, W.G., 1973. Natural responses to scheduled rewards. In: Bateson, P.P.G., Klopfer,P.H. (Eds.), Perspectives in Ethology. Plenum Press, New York, pp. 1–41.

Garcia, J., Hankins, W.G., 1977. On the origin of taste aversion paradigms. In: Barker, L.M., Best, M.R., Domjan,M. (Eds.), Learning Mechanisms in Food Selection. Baylor University Press, Waco, TX, USA, pp. 3–22.

Garcia, J., Kimeldorf, D.J., Koelling, R.A., 1955. Conditioned aversion to saccharin resulting from exposure togamma radiation. Science 112, 157–158.

Garcia, J., Koelling, R.A., 1972. Relation of cue to consequence in avoidance learning. In: Seligman, M., Hager,J. (Eds.), Biological Boundaries of Learning. Appleton Century Crofts, New York, pp. 10–15.

Gill, E.L., Whiterow, A., Cowan, D.P., 2000. A comparative assessment of potential conditioned taste aversionagents for vertebrate management. Appl. Anim. Behav. Sci. 67, 229–240.

Gustavson, C.R., 1977. Comparative and field aspects of learned food aversions. In: Barker, L.M., Best, M.R.,Domjan, M. (Eds.), Learning Mechanisms in Food Selection. Baylor University Press, Waco, TX, USA,pp. 23–43.

Gustavson, C.R., Garcia, J., Hankins, W.G., Rusiniak, K.W., 1974. Coyote predation control by aversiveconditioning. Science 184, 581–583.

Gustavson, C.R., Kelly, D.J., Sweeney, M., Garcia, J., 1976. Prey-lithium aversions. I. Coyotes and wolves. Behav.Biol. 17, 61–72.

Horn, S.W., 1983. An evaluation of predatory suppression in coyotes using lithium chloride-induced illness. J.Wildlife Manage. 47, 999–1009.

Kalat, J.W., 1977. Biological significance of food aversion learning. In: Milgram, N.W., Krames, L., Alloway,T.M. (Eds.), Food Aversion Learning. Plenum Press, New York, pp. 73–103.

Kalat, J.W., Rozin, P., 1973. ‘Learned safety’ as a mechanism in long delay taste-aversion learning in rats. J.Comp. Physiol. Psychol. 83, 198–207.

Macdonald, D.W., Baker, S.E., 2004. Non-lethal control of fox predation: the potential of generalised aversion.Anim. Welfare 13, 77–85.

Massei, G., Cowan, D.P., 2002. Strength and persistence of conditioned taste aversion in rats: evaluation of 11compounds. Appl. Anim. Behav. Sci. 75, 249–260.

Nachman, M., Rauschenberger, J., Ashe, J.H., 1977. Stimulus characteristics in food aversion learning. In: Milgram,N.W., Krames, L., Alloway, T.M. (Eds.), Food Aversion Learning. Plenum Press, New York, pp. 105–131.

Nicolaus, L.K., 1987. Conditioned aversions in a guild of egg predators: implications for aposematism and preydefense mimicry. Am. Midl. Nat. 117, 405–419.

Nicolaus, L.K., Cassel, J.F., Carlson, R.B., Gustavson, C.R., 1983. Taste-aversion conditioning of crows to controlpredation on eggs. Science 220, 212–214.

Nicolaus, L.K., Crowe, M., Lundquist, R., 1992. Oral estrogen retains potency as an aversion agent in eggs:implications to studies of community ecology and wildlife management. Physiol. Behav. 51, 1281–1284.

Nicolaus, L.K., Herrera, J., Nicolaus, J.C., Dimmick, C.R., 1989. Carbachol as a conditioned taste aversion agentto control avian predation. Agric. Ecosyst. Environ. 26, 13–21.

Nicolaus, L.K., Hoffman, T.E., Gustavson, C.R., 1982. Taste aversion conditioning in free-ranging raccoons,Procyon lotor. Northwest Sci. 56, 165–169.

Nicolaus, L.K., Nellis, D.W., 1987. The first evaluation of the use of a conditioned taste aversion to controlpredation by mongooses on eggs. Appl. Anim. Behav. Sci. 17, 329–346.

Perrins, C.M. (Ed.), 1994. Handbook of the Birds of Europe, The Middle East and North Africa: The Birds of theWestern Palearctic, Crows to Finches (Chief Ed. Cramp, S.), vol. 8. Oxford University Press, Oxford, UK.


Prescott, M.N., Berdoy, M., Macdonald, D.W., 1997. Protecting egg prey from magpies (Pica pica): the potentialof aversive conditioning. A Report for the Royal Society for the Protection of Birds, Bedfordshire, UK.

Ratnaswamy, M.J., Warren, R.J., Kramer, M.T., Adam, M.D., 1997. Comparisons of lethal and nonlethal techniquesto reduce raccoon depredation of sea turtle nests. J. Wildlife Manage. 61, 368–376.

Reynolds, J.C., 1999. The potential for exploiting conditioned taste aversion (CTA) in wildlife management. In:Cowan, D.P., Feare, C.J. (Eds.), Advances in Vertebrate Pest Management. Filander Verlag, Furth, pp. 267–282.

Reynolds, J.C., Goddard, H.N., Brockless, M.H., 1993. The impact of local foxVulpes vulpesremoval on foxpopulations at two sites in southern England. Gibier Faune Sauvage 10, 319–334.

Riley, A.L., Baril, L.L., 1976. Conditioned taste aversions: a bibliography. Anim. Learn. Behav. 4, 1S–13S.Riley, A.L., Clarke, C.M., 1977. Conditioned taste aversions: a bibliography. In: Barker, L.M., Best, M.R., Domjan,

M. (Eds.), Learning Mechanisms in Food Selection. Baylor University Press, Waco, TX, USA, pp. 593–610.Riley, A.L., Tuck, D.L., 1985. Conditioned food aversions: a bibliography. In: Braveman, N.S., Bronstein, P.

(Eds.), Experimental Assessments and Clinical Applications of Conditioned Food Aversions. Ann. N.Y. Acad.Sci. 443, 381–437.

Rusiniak, K.W., Hankins, W.G., Garcia, J., Brett, L.P., 1979. Flavor-illness aversions: potentiation of odor by tastein rats. Behav. Neural Biol. 25, 1–17.

SAS, 1996. SAS/STATTM Users Guide, Release 7.01 ed. SAS Institute, Cary, NC, USA.Semel, B., Nicolaus, L.K., 1992. Estrogen based aversion to eggs in raccoons. Ecol. Appl. 2, 439–449.Testa, T.J., Ternes, J.W., 1977. Specificity of conditioning mechanisms in the modification of food preferences.

In: Barker, L.M., Best, M.R., Domjan, M. (Eds.), Learning Mechanisms in Food Selection. Baylor UniversityPress, Waco, TX, USA, pp. 229–253.

Tuyttens, F.A.M., Macdonald, D.W., 2000. Consequences of social perturbation for wildlife management andconservation. Conserv. Biol. Ser. 2, 315–329.

Westbrook, R.F., Clarke, J.C., Provost, S., 1980. Long-delay learning in the pigeon: flavor, color, andflavor-mediated color aversions. Behav. Neural Biol. 28, 398–407.

Wolfensohn, S., Lloyd, M., 1998. Handbook of Laboratory Animal Management and Welfare, 2nd ed. BlackwellScience, Oxford, UK.

protecting egg prey from carrion crows: the potential of aversive conditioning

Documents