Activity, Behaviour and Interactions ofParrot Species at a Peruvian Clay Lick

Thenumber of individuals of each species on the lick itself was recorded at five-minute intervalsfrom the time the first bird landed on the clay, and the maximum number on the lick throu

Unit 61BL4161 (60 credit route)

Activity, Behaviour and Interactions of Parrot Species at a Peruvian Clay Lick

Elisabeth Mary ShawStudent Number: 05977369

Submitted by Elisabeth M Shaw, in part fulfilment of the Master of Science degree in Animal Behaviour, awarded by Manchester Metropolitan University, through the Division

of Biology in the School of Biology, Chemistry and Health Sciences

Submitted February 2008

With the exception of any statements to the contrary, all the data presented in thisreport are the results of my own efforts and have not previously been submitted incandidature for any other degree or diploma In addition, no parts of this report havebeen copied from other sources I understand that any evidence of plagiarism and/orthe use of unacknowledged third party data will be dealt with as a very serious matter.

Signed ………

There has been much investigation in recent years into the reasons for and

factors affecting geophagy behaviour by parrots and macaws at clay licks There

has been much less focus on the behaviour of individual birds at these sites,

despite the unique opportunity they present for behavioural investigation, and the

potential such research may have to contribute to our understanding of lick use

and of parrot behaviour in general This study presents some of the first detailed

empirical data on the behaviours of eight parrot species at a Peruvian lick Focal

animal samples performed in the early dry season of 2007 allowed the

construction of an ethogram and of time budgets for birds both on the lick and in

the surrounding trees and vegetation

Birds in the trees around the lick spent most of their time vigilant, with the rest

engaged mainly in a variety of preening and social behaviours Vigilance did not

depend on flock size or body size Behaviour patterns changed little with time of

day, but did change significantly during periods of lick activity There were also

differences between species, particularly in Chestnut-fronted Macaws, which may

have had fledglings present Birds often ate clay away from the lick face, a

previously unreported behaviour that may relate to predation, competition, or the

mechanics of clay processing

Individual birds spent a mean of 2.35 minutes on the lick in any one feeding

bout, taking a mean of 13 bites from the clay, and in a third of cases taking clay

away from the lick at the end of the bout Some preference was seen for feeding

from a vine rather than perched on the clay Most time on the lick was spent

chewing clay, and associated behaviours such as gagging and head shaking

suggested some difficulties with this Agonistic interaction rates were five times

higher on the lick than in the surrounding trees, with a dominance hierarchy based

on body size Most aggression occurred between conspecifics, as did most

displacement in the trees, whereas most displacement on the lick itself was

between heterospecifics and probably due to competition for space

2004a,b, Brightsmith & Aramburú 2004, Burger & Gochfeld 2003, Diamond et al 1999, Duffie

2003, Emmons & Stark 1979, Gilardi et al 1999, Mee et al 2005, Symes et al 2006, Symes &

Marsden 2003) The best known are in south-east Peru, where there may be up to 100 or so licks

of various sizes (C Munn unpubl data), visited by at least 17 species of psittacids, as well as

various pigeons and caracids, and mammals including monkeys and squirrels (Brightsmith2004a,b, Brightsmith & Aramburú 2004, Brightsmith & Figari 2003, Brightsmith & Houtan

2000, Hammer 2001, Hammer & Tatum-Hume 2003, Munn 1988, Tatum-Hume et al 2003,

2005) The licks are important tourist attractions, and bring in much foreign revenue every year(Brightsmith 2002a, Munn 1992, 1998)

There are various theories to explain geophagy, though it may perform different functions indifferent species, or even multiple functions in some species (Abrahams & Parsons 1996,

Diamond et al 1999, Gilardi et al 1999, Wilson 2003) Some reasons cited for soil

consumption - obtaining grit for the mechanical breakdown of food, buffering of gastric pH, andtreatment for diarrhoea - are largely discounted for parrots (Brightsmith & Aramburú 2004,

Diamond et al 1999, Gilardi et al 1999) Instead, parrots may consume soil to obtain sodium

and / or for the adsorption of toxins from the seeds and unripe fruits common in their diets

(Brightsmith 2002a, 2003, 2004a,b, Brightsmith & Aramburú 2004, Diamond 1999, Diamond et

al 1999, Emmons & Stark 1979, Gilardi et al 1999, Munn 1988, 1998, Symes et al 2006).

Clay may also stimulate mucus production in the gut, forming a protective layer that further

defends against chemical insult (Brightsmith 2002a, Gilardi et al 1999) Such protection against

toxins would allow parrots to exploit a wider range of food resources, many of which are

unavailable to other animals (Diamond 1999, Gilardi et al 1999).

Despite advances in our understanding of parrot geophagy, there are still many unansweredquestions, particularly in relation to the importance of clay licks to parrot populations and

distribution (Diamond 1999, Diamond et al 1999, Mee et al 2005, Symes et al 2006) Around 28% of parrot species are threatened in the wild (Snyder et al 2000) Parrots and macaws are

ideal flagship species, and a better understanding of all aspects of their ecology and behaviour –including a better understanding of geophagy – may help conserve not only them but also the

many other species which share their forest habitats (Munn 1998, Snyder et al 2000).

Most research at parrot clay licks has focused on the reasons for and factors affecting lick use,such as weather, season, and human disturbance (Brightsmith 2002a, 2004a,b, Brightsmith &

Figari 2003, Brightsmith & Houtan 2000, Hammer & Tatum-Hume 2003, Tatum-Hume et al.

2003, 2005, 2006) There has so far been little focus on the behaviour of individual birds MostNeotropical parrots live in dense, closed-canopy forest, making them difficult to observe andstudy (Beissinger & Snyder, 1992, Gilardi & Munn 1998), and few studies have looked at theirwild behaviour (Seibert 2006, Pitter & Christiansen 1997); open licks, where birds are highlyvisible, therefore provide unique opportunities to study them in their natural habitats A fewinvestigations have looked at certain aspects, such as aggression and competition, responses toboats and other disturbances, and behaviours of birds waiting to descend to the lick (Burger &

Gochfeld 2003, H Clegg unpubl data, Hammer & Tatum-Hume 2003, Tatum-Hume et al 2003,

2005, 2006) However, there has been a lack of detailed empirical studies

Understanding behaviour at the lick may help us further understand lick use It has beensuggested that one reason parrots visit is to socialise and find a mate (e.g., Brightsmith 2002a),but there are few data on social or courtship behaviours More detailed information on how birdsspend their time at the lick, particularly when this can be related to lick activity, may also help

us better understand anti-predator behaviour, responses to disturbance, and factors influencingfeeding, as well as provide additional information on behaviours such as feeding of clay toyoung There has to date been no detailed study of feeding behaviours of individual birds on thelick itself, which would seem important for determining, for example, how much clay is eaten,how birds select which soil to eat, how they process it, and factors affecting these More detailed

study of interactions between birds on and around the lick would also be useful in determiningthe extent to which competition affects lick use.

The aims of this study were to categorise and describe behaviours shown by parrot speciesboth on and in the trees and vegetation immediately surrounding a clay lick, to quantify patterns

of behaviour and examine factors influencing it, and to assess levels of intra- and interspecificcompetition Some baseline lick monitoring (recording parrot numbers, species and timing oflick use) was also performed, to provide a context within which to interpret behavioural data

METHODS

Study area

The study was performed at a clay lick on the south bank of the Tambopata River, eastern Peru (S 12°49’23”, W 69°17’25”), at the edge of the Tambopata National Reserve Thesite is 30km from the town of Puerto Maldonado and 1.6km from Explorer’s Inn tourist lodge,from which it was reached either by boat or on foot along trails from the lodge The lick itself is

south-a northwest-fsouth-acing cliff south-approximsouth-ately 8m long, 3m high, south-and comprising three msouth-ain sections ofroughly equal width (Figure 1) It is set back 30m from the river, which is approximately 200mwide at this point Behind the site is primary lowland floodplain forest, and dense understoreyvegetation overhangs the top of the lick, though the lick face itself is almost totally bare Thearea between the lick and the river is less densely vegetated, and includes several cecropia trees

(Cecropia sp.) of 15-20m high in which birds often perch before descending to feed.

The lick is regularly visited by up to nine species of parrots, parakeets and macaws, andsquirrels and monkeys also occasionally feed and are regularly seen in the area Most speciesvisit in the early morning and spend some time in the surrounding trees and vegetation beforedescending to feed, although feeding does not always occur for every species every day Touristgroups from Explorer’s Inn and Inotawa lodge visit most days during the high season (April -October), the birds being viewed from two small hides 20m downstream, on the same bank as

the lick itself Groups arrive in the early morning, before the birds, and usually leave an hour ortwo later when feeding activity is over These groups are clearly visible to the birds whenarriving and leaving by boat Previous study here has found no evidence that tourist presence

affects bird numbers or feeding, though it may make the birds more wary (H Clegg unpubl.

data) Both local and tourist boats regularly pass on the river.

Figure 1: Photograph of the clay lick, as viewed from the Explorer’s Inn hide,

showing birds using the first lick section (nearest the hide) Second section is set back

from the others (centre); third section is on the right, and includes a hollow out of

sight on the far right Vegetation immediately in front of the hide had been cut back

half-way through the study by a member of the lodge staff (Photo: EM Shaw 2007.)

Monitoring of daily lick activity

Behavioural data were collected alongside monitoring of lick activity as part of ongoing work

by the Tambopata Macaw Project Monitoring data were collected by between one and threeobservers over 30 days between 21 April and 31 May 2007 (early dry season), with most datacollection taking place in the early morning between 05:45 and 08:35 Observers usually stayeduntil all birds had left the area and showed no signs of returning One full day (to 15:50), onelate morning and one afternoon monitoring session were also performed to check whether any

birds fed later in the day Heavy rain and flooding prevented access to the site on several daysduring the study period.

All monitoring was performed from the Explorer’s Inn hide, with observations made with thenaked eye or using 10x25 binoculars Observers recorded the start and end time of monitoring,species present, the time each arrived and the time each first landed on the lick, and themaximum number on and in the area around the lick, as visible from the hide; this was defined

as the area between the lick and the river, as well as the trees and vegetation immediately aboveand to each side of the lick, and the trees immediately above the hide Birds out of sight behindthe lick were excluded, as were those flying over and any on the opposite side of the river Thenumber of individuals of each species on the lick itself was recorded at five-minute intervalsfrom the time the first bird landed on the clay, and the maximum number on the lick throughoutthe observation period noted

Video recordings were made of lick activity on most days, using a JVC GR-D370EK MiniDVDigital Video Camera at up to 32x optical zoom; on days when the amount of activity andnumber of behavioural samples being performed made lick counts difficult, these were takenfrom the recordings Recordings were made from the same position in the hide each time andusually continued for as long as birds remained on the lick

Behavioural sampling

Behaviour samples were recorded alongside daily monitoring and performed from theExplorer’s Inn hide; all were performed by a single observer (the author), thus avoiding inter-observer bias After a few days of preliminary observations, behaviours of all species weredivided into categories and an ethogram produced Any additional behaviours observed duringthe study, or on subsequent playback of video recordings, were added to this ethogram as andwhen they were identified Behaviours were sampled using focal animal sampling andcontinuous recording; two sets of samples were performed, the first for birds in the trees andvegetation in the area around the lick (‘behaviours around the lick’), and the second for birds onthe lick itself (‘behaviours on the lick face’) All behaviours were treated as states (having

duration) and given a minimum duration of one second, except Displacement, which wasconsidered an event, and was recorded as a frequency.

Behaviours around the lick

Focal samples for birds around the lick were performed using 10x25 binoculars, andbehaviours recorded into an Olympus DS-50 Digital Voice Recorder The same recorder wasused to play back samples, with behaviours transcribed by hand Where possible, focal birdswere chosen at random, though samples were limited to species present Samples were

performed on an ad hoc basis throughout the monitoring period, with up to 35 samples in any

one morning For each sample, the observer recorded the date, time of day, species, and whetherany birds used the lick during the sample Samples often had to be fitted between five-minutelick counts; sample length was therefore limited to one minute If the focal bird was lost fromsight or left the area during a sample, the sample was ended, unless the bird could be re-sightedwithin ten seconds of being lost In this case, the time spent out of sight was excluded from theoverall sample time Incomplete samples were included in the analysis to avoid biasing againstbehaviours shown just before birds left the area; however, the sampling time was ended from themoment the bird took off, and time spent flying as it left was not included If a bird flew to thelick during a sample, it was followed on the lick, and the sample resumed if it returned to thetrees before the end of the sixty seconds; flying to and from the lick was included, but behaviourwhile on the lick itself was not

To quantify competition, the number of agonistic interactions (Displacement and Aggression)

in each sample was noted, along with the species with which the focal bird was interacting, andwhich bird initiated the interaction In cases where the focal bird interacted with more than oneother (for example, displacing two others), the number of interactions was recorded as thenumber of birds with which it interacted

Behaviours on the lick face

It proved impossible to sample birds on the lick at the same time as other behaviour samplesand general monitoring; these samples were therefore taken from video footage recorded duringmonitoring sessions Video recordings were replayed on Windows Media® Player 11, with thetiming of each behaviour taken directly from the time displayed on the screen and behaviourstranscribed by hand Focal birds were chosen at random, though again sampling was limited tospecies present and to individuals clearly visible on the video Sampling more than one birdduring the same time period was avoided, though in a few cases some overlap of samples didoccur.

Samples were started from the moment the focal bird landed on the lick face (or on a vine orbranch crossing the lick face), and behaviours recorded until the moment it left the lick (flewaway and out of view of the camera) If the bird was lost from sight (for example, behindvegetation) and could not be re-sighted before it left the lick, the sample was ended from themoment it was lost If the bird could be re-sighted the sample was resumed, and any time out ofsight was recorded as such

The reason the bird left the lick at the end of the sample was noted, and divided into thefollowing categories: unknown (no clear reason for leaving), lost (sample ended before bird hadleft, either because it was lost, or, on at least one occasion, because the video cut off), flush(some or all of the birds, including the focal individual, suddenly fly from the lick),displacement (focal bird is displaced by another), or leaving with clay (no other clear reason forleaving, but focal bird can be seen holding clay in its beak as it flies away, or is assumed withsome certainty to be holding clay, for example because it flies immediately after biting the lick).The observer also recorded the species, time and date, and which lick section was used Inaddition, the position of the bird (perched on the clay itself, on a vine or branch, or with one foot

on each) was recorded throughout the sample

Agonistic interactions were recorded as before, and included Displacement occurring as thebird landed at the start of the sample and Displacement of the focal bird when this caused thebird to leave at the end of the sample

Data analysis

Non-parametric statistical tests were used as most of the data were not normally distributed.Statistical analyses were performed using Minitab® 14.

Monitoring data were used to calculate mean flock sizes, mean maximum number of birds onthe lick on days species fed, and the mean time each species spent on the lick, as well as lagtimes to feed (the time between a species first arriving at the lick and the first individuals of thatspecies feeding) and the number of feeding bouts each day (periods of continuous lick use,defined as the presence of birds on the lick at successive five-minute counts) Daily presence,numbers and feeding activity were compared between species Mean lick activity (mean of dailyfive-minute lick counts) was also calculated for fifteen-minute periods throughout the morning,for comparison with behavioural data at these times

For focal samples, the total percentage of time spent on each behaviour in each sample wascalculated, and a mean percentage calculated over all samples and used to construct timebudgets, which were compared between species, different times of the morning, and differentconditions of lick use In some cases, less common behaviours were grouped together(‘OTHERS’) to simplify analyses The short length of focal samples for birds in the trees meantmean durations of individual behaviours could not be calculated (since it was unlikely allbehaviours went to completion in this time), but the mean percentage of time on individualbehaviours was used to compare, for example, Vigilant behaviour to the number of birds presentthat day and to species body size, and Eat Clay to bird numbers Mean durations could becalculated for individual behaviours on the lick itself, and were used to compare biting andchewing of clay between species Time chewing and head shaking (which may be associatedwith difficulty swallowing) were also compared to recent rainfall; rainfall totals were taken from

a weather log recorded daily at 18:30 at Explorer’s Inn Bite rates were also calculated, as wasthe total time each bird spent on the lick and the mean proportion of time spent feeding fromdifferent positions

Rates of agonistic interactions were calculated as the mean number of incidences ofAggression and Displacement per bird per hour of sample time, and compared between speciesand between birds on the lick and in the trees The percentages of intra- versus interspecificinteractions were also compared, and a dominance index calculated for each species; this wascompared to species body size and body mass (taken from the literature), and to mean flock size

Daily activity patterns

Eight species of psittacid were recorded at the lick: Blue-headed Parrot (Pionus menstruus), Chestnut-fronted Macaw (Ara severa), Dusky-headed Parakeet (Aratinga weddellii), Mealy Parrot (Amazona farinosa), Orange-cheeked Parrot (Pionopsitta barrabandi), Red-and-green Macaw (Ara chloroptera), White-eyed Parakeet (Aratinga leucophthalmus), and Yellow- crowned Parrot (Amazona ochrocephala) All except Red-and-green Macaws fed on the clay A

mean of five species visited each day, with a mean of three species per day feeding on the clay.Table 1 compares presence, numbers and feeding activity between species Most feedingoccurred on the first lick section (Figure 1), but the other two sections were also used by allspecies, though Mealy Parrots tended to dominate the third (furthest) section There was nosignificant difference in numbers at the lick between days with or without lick use (W=68.0,p=0.20)

On days birds fed, the mean number of feeding bouts (continuous periods of lick use) was 1.67(SD 0.87), with a minimum of 1 and maximum of 4 Birds fed between 06:10 and 08:35, thoughDusky-headed Parakeets were also seen feeding during the afternoon monitoring session Lagtime to feed ranged from 4 to 111 minutes, with a mean of 30.8 mins (SD 20.5) There was nosignificant difference in lag time between species (H=8.13, df=6, p=0.23)

Table 1: Presence, numbers and feeding activity of parrot species during early morning monitoring at the clay lick Days present gives number of days, out of 30, on which species was seen at the lick; days fed gives number of days, out of 30, species was seen feeding on the clay Flock size gives mean and maximum daily species counts for days species was present; maximum on lick gives mean and maximum of maximum daily lick counts, for days species fed Mean values are given ±SD Excludes feeding seen by Dusky-headed Parakeets during an afternoon monitoring session Species abbreviations are as follows: CFMA – Chestnut-fronted Macaw; DHPA – Dusky-headed Parakeet; BHPA – Blue-headed Parrot; MEPA – Mealy Parrot; YCPA – Yellow-crowned Parrot; OCPA – Orange-cheeked Parrot; WEPA – White-eyed Parakeet; RGMA – Red-and-green Macaw ‘All species’ gives values for all species combined.

SPECIES Days

present Days fed

FLOCK SIZE MAXIMUM ON LICK Mean

bout length (mins) a

Mean total duration on lick (mins) b

Mean number of birds

Maximum count

Mean (maximum) number

Maximum count

a 5x(mean number of successive five-minute counts at which species was recorded on lick in single feeding bouts).

b For days species fed, 5x(mean of total five-minute counts per day at which species was recorded on lick).

c Minimum number of birds on any day was 21.

d Mean of daily maximum number of birds (of all species) on lick at any one time, for days birds fed; maximum is

maximum number of birds on lick at any one time on any of these days (not sum of individual species maximum

counts).

Behaviours around the lick

Focal samples (n=524) were performed for birds around the lick on 26 monitoring days Intotal, behaviours were recorded for 8.04 hours, and all eight species were sampled Mostsamples were collected between 06:00 and 08:35 Fifteen behaviour categories were identified,and the ethogram is given in the Appendix

Table 2 gives the overall time budget of birds around the lick and compares this betweenspecies (excluding less common species for which sample sizes were small (<10)) Three

behaviours (Allofeed, Wing Flip, Hang-Play) were only observed in Chestnut-fronted Macaws.There was no significant correlation between the proportion of time a bird spent Vigilant and themaximum number of birds present (rs=0.01, p=0.90) or body size (rs=0.31, p=0.61; body lengthsfrom Clements & Shany 2001) The proportion of time spent in Eat Clay was not significantlycorrelated to the maximum number of birds present (rs=0.07, p=0.11), however there was asignificant positive correlation with the maximum lick count for that day (rs=0.13, p<0.01).

Overall, birds spent a mean of 1.0% of their time hanging (SD 8.0) Chestnut-fronted Macawsspent the largest proportion of time hanging (3.2%), and none was recorded at all in Blue-headed Parrots, Orange-cheeked Parrots, White-eyed Parakeets or Red-and-green Macaws(though sample sizes were small for the latter three) Of the time hanging, most of that not spent

in Hang-Play was spent Vigilant, with other behaviours recorded only twice (ManipulateVegetation and Preen)

Table 2: Time budgets of five parrot species in the trees / vegetation around the lick, and of all eight species combined (‘all species’) Values are mean percentage of time a bird spent on each behaviour (±SD) Behaviours are listed in order

of overall mean percentage time, and are as follows: VG – Vigilant; PR – Preen; ALLPR – Allopreen; CALL – Call; CLAY – Eat Clay; VEG – Manipulate Vegetation; FL – Fly; MV – Move; WF – Wing Flip; AGGR – Aggression; SCRAPE – Scrape Beak; INTER – Other Interaction; PLAY – Hang-Play; UN – Unidentified Behaviour; ALLFE – Allofeed Species abbreviations are given in Table 1 Sample sizes (n) are given in parentheses Kruskal-Wallis tests were used to compare median percentage times between the five species for which data are given; values give test statistic and level of significance NS indicates test was not significant at p<0.05.

BEHAVIOUR ALL SPECIES

(06:00-behaviours (Vigilant, Preen, Allopreen, Call, Eat Clay, Manipulate Vegetation, Fly, Move)

compared between different periods using Kruskal-Wallis tests There was no significant

difference (at p<0.05) in the median percentage of time spent on any behaviours between

different fifteen-minute periods, other than for Move (H=22.0, df=9, p<0.01) Changes in

median percentage of time on Eat Clay were significant at p<0.1 (H=14.9, df=9, p=0.09) Figure

2 compares changes in the mean percentage of time on Move and Eat Clay with mean lick use

during the morning There was no significant correlation between mean number of birds on the

lick and mean percentage time on Eat Clay (rs=0.33, p=0.35) or Move (rs=-0.30, p=0.41)

Time budgets were also calculated for various conditions of lick use (Table 3) and behaviours

compared using Mann-Whitney-U tests

0 5 10 15 20 25

06:15 (n=27)

06:00- 06:30 (n=70)

06:15- 06:45 (n=115)

06:30- 07:00 (n=99)

06:45- 07:15 (n=80)

07:00- 07:30 (n=55)

07:15- 07:45 (n=28)

07:30- 08:00 (n=24)

07:45- 08:15 (n=15)

08:00- 08:30 (n=8)

08:15-Time period (am)

Figure 2: Changes in mean lick use, and in percentages of Eat Clay and Move in time budgets

of birds around the lick, with time of morning Mean lick use gives mean number of birds (of all species) on lick during given time period, taken from five-minute lick counts for the 26 days behaviours were recorded Mean percentage of time budget gives mean percentage of time spent on given behaviour in all samples (for all species) during given time period;

sample sizes for each period are given in parentheses.

Table 3: Time budgets of birds in the trees / vegetation around the lick (all species combined) under various conditions of lick use Values are mean percentage of time spent on each behaviour (±SD) ‘Current lick use’ indicates whether any birds were on the lick when the sample was taken; ‘focal bird used lick’ indicates whether sample was interrupted by the focal bird flying to the lick; ‘day of lick use’ indicates whether there was lick use on the morning the sample was taken (regardless of whether or not this occurred at the time of sampling) Behaviours are as given in Table 2; OTHERS - all other behaviour categories Sample sizes (n) are given in parentheses Mann-Whitney-U tests were used to compare median percentage times between conditions; values give test statistic and

level of significance NS indicates test was not significant at p<0.05; dashes indicate test could not be performed as thegiven behaviour did not occur under one of the conditions.

a Of these, PLAY only occurred when no birds were on the lick, when focal bird did not use the lick, and on days without lick use

Behaviours on the lick face

Focal samples (n=60) were performed for birds on the lick itself on 17 monitoring days, with

behaviours recorded for a total of 2.35 hours Levels of lick activity only allowed three species

to be sampled at reasonable sample sizes All samples were collected between 06:15 and 08:35

Most behaviours were recorded for birds on the first lick section, though four birds were

sampled on the third section Thirteen behaviour categories were identified, and are given in the

ethogram in the Appendix Eat Clay was divided into Clay From Foot and Vigilant-Chew for

lick samples, to allow more detailed analysis of clay processing during feeding Table 4 gives

the time budgets for the three species sampled Again, Wing Flip was shown only by

Chestnut-fronted Macaws

Table 4: Time budgets of three species on the lick face, and for all three species combined (‘all species’) Values

are mean percentage of time a bird spent on each behaviour (±SD) Behaviours are listed in order of overall mean

percentage time, and are as follows: VGCH - Vigilant-Chew; BT - Bite; VG - Vigilant; MV - Move; GRIP - Grip;

UN - Unidentified Behaviour; HSH - Head Shake; FL - Fly; CFF - Clay From Foot; INTER - Other Interaction;

AGGR - Aggression; CALL - Call; WF - Wing-Flip Species abbreviations are as given in Table 1 Sample sizes

(n) are given in parentheses Kruskal-Wallis tests were used to compare median percentage times between the three species; values give test statistic and level of significance NS indicates test was not significant at p<0.05.

BEHAVIOUR ALL SPECIES

of birds on the lick itself during a sample (estimated from five-minute lick counts) and timespent on the lick (rs=0.003, p=0.98)

The total number of bites per sample ranged from 1 to 69, with a mean of 12.6 (SD 13.9) Bitelength ranged from 1 to 34 seconds, with a mean of 2.98 seconds (SD 2.93) Mean Bite rate(total number of bites divided by total sample time) per sample was 5.9 bites per minute (SD2.7); median bite rate did not differ significantly between species (H=1.89, df=2, p=0.39) Time

on the lick was not significantly correlated to Bite rate (rs=-0.19, p=0.15) or to mean percentagetime spent biting (rs=-0.20, p=0.13)

Bout length of Vigilant-Chew ranged from 1 to 42 seconds, with a mean of 5.17 seconds (SD4.61) Median bout length did not differ significantly between species (H=9.78, df=2, p=0.08).There was no significant difference between the percentage of time spent chewing clay(Vigilant-Chew) or performing Head Shake between samples for which there had been rainfall

or no rainfall the previous day (Vigilant-Chew: W=790, p=0.16; Head Shake: W=871, p=0.80;

31 samples with rainfall the day before), or between samples for which there had been rainfall

or no rainfall in the previous two days (Vigilant-Chew: W=517, p=0.61; Head Shake: W=535,p=0.77; 42 samples with rainfall in the previous two days).

Birds spent a mean of 73.2% of time on the lick perched on the clay itself, 25.1% perched on(and feeding from) a nearby vine or branch, and the remaining 1.7% perched on a vine or branchbut with one foot on the clay to stabilise themselves Percentage time in each position did notvary significantly between the three species (clay: H=1.85, p=0.40; vine: H=1.63, p=0.44; onefoot on both: H=1.10, p=0.58)

For 13% of samples, the reason for the focal bird leaving the lick at the end of the samplecould not be recorded because the sample was ended when the bird was lost from sight (but still

on the lick), or because the video clip cut off before the end Of the remaining samples, in 35%there was no obvious reason for the bird to leave the lick, 23% of departures were due to flushes(full or partial) and in 8% the bird was displaced by another (either deliberately or due toanother bird slipping and knocking the focal bird from the clay) In 17% the bird left with apiece of clay in its beak, and in another 17% it was thought to almost certainly be leaving withclay in its beak, though the clay itself could not be clearly seen Dusky-headed Parakeets werelost more than other species; excluding this, there was a significant difference between species

in the number of birds that left the lick for each reason (χ2=12.9, df=6, p<0.05), in particularwith Chestnut-fronted Macaws leaving with clay and Mealy Parrots being flushed more thanothers, though counts were small (<5) in some cases Flushes were sometimes in response tolarge birds flying over or tourists walking onto the riverbank, but in most cases the cause couldnot be determined

Macaws, Dusky-headed Parakeets and Mealy Parrots, both on and around the lick: H=5.06,df=2, p=0.08) No agonistic interactions were recorded in White-eyed Parakeets or Red-and-green Macaws, probably because of their small sample sizes There was a significant difference

in the frequencies of intra- versus interspecific Displacements between the tree and lick(χ2=28.5, df=1, p<0.001), but no significant difference in the frequencies of intra- versusinterspecific Aggression between the two locations (χ2=0.36, df=1, p=0.55)

The relative dominance of each species is given by the interaction matrix and dominanceindices in Table 6 Dominance index was significantly positively correlated to body size(rs=0.96, p<0.01) and to mass (rs=0.81, p=0.05), but not to mean flock size (rs=0.00, p=1.00;flock sizes in Table 1) Mean interaction rate of birds in the trees around the lick wassignificantly negatively correlated to body size (rs=-0.93, p<0.01); smaller size was alsosignificantly correlated to the percentage of Displacements (rs=0.84, p=0.04), and there was anon-significant trend towards a greater percentage of intraspecific interactions in smaller species(rs=-0.78, p=0.07)

Table 5: Frequency of agonistic interactions of birds around and on the lick Interaction rates give the mean number of agonistic interactions (Displacement or Aggression) per focal bird per hour of sampling time, for given species and location (±SD) ‘Within’ – intraspecific interactions, ‘Between’ – interspecific interactions; values are percentage of agonistic interactions in given location within each category of interaction, to the nearest whole number ‘All species’ gives values for all eight species combined Dashes indicate species not sampled on the lick Species abbreviations are

as given in Table 1; species are listed in order of interaction rate.