Mate preferences in zebra finches: color bands and the role of sexual imprinting

Report by: Anja Jansen

Supervision: Tim Fawcett

Nikolaus von Engelhardt

CONTENTS

Summary3

General introduction4

Experiment one: Observations vs PIR’s

Introduction7

Materials & methods8

Results11

Discussion13

Experiment two: Color-band preferences in zebra finches from Bielefeld

Introduction15

Materials & methods15

Results16

Discussion17

Experiment three: Effect of paternal phenotype on color-band preferences

Introduction18

Materials & methods19

Results20

Discussion22

General discussion24

Acknowledgements25

References26

SUMMARY

The zebra finch (Taeniopygia guttata) is a commonly used subject in studies of mate choice. One interesting finding is that females appear to pay attention to artificial traits, preferring males wearing red plastic leg-bands over those wearing green ones. However, while some researchers have managed to repeat this famous result, others have found no preference for red leg-bands, and some even report a preference for green.

In this study, I investigated whether this is likely to be due to differences in set-up and methodology, or differences between laboratory stocks of birds. I then investigated one possible mechanism by which such differences in preference might arise, namely sexual imprinting on different leg-band colors.

In a preliminary experiment, I compared two different methods of measuring preference, to determine whether an automatic recording system (passive infrared detectors) would give similar results to direct observations by an experimenter.

For these preference tests, I let a zebra-finch female choose between a zebra-finch male and a Bengalese-finch male. The automatic recordings correlated very strongly with the manual observations.

In a second experiment, I tested the color-band preferences of zebra finches from the University of Bielefeld, Germany, in our preference-test set-up at the University of Groningen, Netherlands.

A preference for red-banded over green-banded males had previously been found in Bielefeld birds tested in Bielefeld, but not in Groningen birds tested in Groningen. Groningen females previously tested in Bielefeld had also shown no preference. Here, I report no color-band preference in the Bielefeld females tested in Groningen.

In a final experiment, I assessed the color-band preferences of female zebra finches raised by fathers wearing red, orange or green color bands, to test the hypothesis that these females had sexually imprinted on their father's band color.

I found no significant differences between the groups in their mate-choice behavior.

GENERAL INTRODUCTION

Scientists have long been interested in the mechanisms of mate choice in animals, since sexual preference is a driving force for evolution. Novel traits can be selected for by a sexual preference, and these preferences are one important reason why species look like they do today.

One particular focus of research is how and why these preferences develop. Sexual imprinting, the process by which young acquire mate preferences based on the characteristics of the parent of the opposite sex, is an important mechanism in this respect, and seems to be widespread in birds. In fact, in every bird species in which the phenomenon has been investigated, it has been found to occur (ten Cate & Vos, 1999).

Zebra finches (Taeniopygia guttata) are an ideal species for studying this phenomenon, since they are highly suitable for studies of mate choice. They are easy to keep in captivity and are ready to breed all year round, show clear evidence of sexual imprinting (Vos et al.,1993; Vos,1994) and have preferences for artificial traits (Burley & Symanski,1998; Witte & Sawka,2003) which can easily be manipulated. For these reasons, I chose the zebra finch as the main subject species for my research.

The zebra finch is a small songbird native to the semi-arid regions of Australia and Timor (see figure 1). They are highly social, nesting colonially in groups of 20 to 1000 individuals.

Figure 1: The distribution of the zebra finch.

Zebra finches are sexually dimorphic in color: the male has orange cheek patches, a black-and-white-striped throat and breast, a black breast bar and brown flanking with white spots, all of which the female lacks (see figure 2). The male has usually a redder beak than the female, whose beak is mostly orange.

Figure 2: The characteristics of a male and a female zebra finch

The zebra finch is the 'white lab rat' of the bird world. No other finch has been studied as intensively or had as much written about it as the zebra finch. There are numerous studies of wild zebra finches in their natural environment as well as laboratory studies of song learning, sexual imprinting and mate selection to name just a few (Beckham, 2004).

In my research I used the zebra-finch subspecies from Australia, T. g. castanotis. The stock I used has been bred in captivity for several generations.

In one experiment I also used Bengalese finches (Lonchura domestica; see figure 3). Another common name for the Bengalese finch is the society finch.

The Bengalese finch is a small songbird which has been raised by the Chinese and Japanese for hundreds of years. It was probably first cage-bred by the Chinese who later brought it to Japan. By this time, thousands of generations had been raised and selected to breed in even the smallest of cages with the simplest of supplies.

The Bengalese finch was originally believed to be a fertile hybrid developed from unknown members of the Lonchura genus. Recent tests have shown, however, that the Bengalese finch is in fact a domesticated form of the white-rump mannikin (Lonchura striata).

Males and females are identical in appearance and therefore cannot be sexed on the basis of their morphological characteristics. The easiest way to sex them is by behavior, since males sing more than females. I sexed them by putting two birds in separate cages next to each other, and waited until one started singing. If neither bird did so, I switched one of them with another bird.

The Bengalese finch is such an excellent parent that it is often used to foster other species of estrildids, even ones that are quite different from the young of its own species (Beckham, 2004).

Zebra finch Society/Bengalese finch

Taeniopygia guttata Lonchura domestica

Figure 3 (a) Figure 3 (b)

My main goal in this research was to investigate how sexual preferences develop, and how important certain morphological traits are for female preference. I tested this by measuring the preference of female zebra finches raised by fathers with a manipulated morphological trait, namely the leg-band color.

Prior to this, I compared different methods of measuring preference, and investigated whether the preference for leg-band color is consistent across different test situations.

Experiment One: Observations vs PIRs

INTRODUCTION

There are different methods of assessing preference in choice tests.

In this experiment I compared two of them, one involving direct observation by an experimenter (manual observations) and the other involving automatic, computer-processed recording, in which the location of the female was monitored by infrared detectors (PIR’s).

The purpose of this experiment was to test the validity of the PIR data, in the hope that I could use this system in the later experiments instead of the more time-consuming manual observations.

Manual observations

To observe the birds' behavior manually I used a method of scan sampling, in which the current position and activity of the birds was noted at fixed time intervals. I recorded the position and behavior of the test female and both of the stimulus males every five seconds, ignoring any movement and other behavior that occurred in between the scans.

This method has the advantage that it can distinguish between different types of behavior, rather than simply recording the spatial position of the test female. Therefore it is possible to distinguish between cases when, for example, the female is asleep, and when she is actively interested in the male. Interest in the male can also be judged by the female's distance from him.

With manual observation, one can simultaneously record male behavior. This is also possible with other methods of assessing preferences, but extra equipment would be needed.

The major disadvantages of manual observation are that all events in between the scans are missed, there is greater risk of human error influencing the results, and the method is highly time-intensive.

Passive infrared (PIR) detectors

The infrared detection system consisted of two infrared detectors per female cage (one on the left side and one on the right, to count the preference for the male on each side), an event recording system (ERS) box which counted the detections, and a computer with an ERS program running on it, which showed and saved the tally of detections in each one-minute period.

This system records automatically, and several tests can be run concurrently provided there is sufficient equipment. It records continuously, so there are no time gaps in the data. The PIRs are activated by movement, so when the test bird is sleeping or is highly inactive, and presumably is not expressing a preference, nothing is registered. However, it is possible that sometimes a female is expressing an interest in a particular male even when she is sitting still. Using the PIRs, if the test female is very active, this is always recorded as a stronger preference than if she is less active. As a technical problem, the PIRs can be difficult to align in exactly the same orientation, so it is possible that the PIR at one side of the test cage records better than that at the other. This can be accounted for by switching the stimulus males between the cages half-way through a test.

In this preliminary experiment, I compared data from manual observations with those from PIR recordings. For this I let a female (wild-type) zebra finch (Figure. 3a) choose between a male Bengalese finch (Figure. 3b) and a male (wild-type) zebra finch.

To select the Bengalese-finch males for this from a mixed-sex group, I first placed pairs together and noted which one sang, replacing one of the pair if neither bird sang. In the subsequent two-way choice tests between a Bengalese-finch male and a zebra-finch male, I scored the female's behavior by direct observation while, at the same time, the PIRs detected movement automatically.

Materials & methods

In this experiment and the subsequent two experiments, I used testing cages for the females (60 × 30 cm and 40 cm high), stimulus cages for the males (30 × 30 cm and 40 cm high), passive infrared (PIR) detectors, an event recording system (ERS) box for registration, and a computer running an ERS program to show and record the number of detections during a one minute period.

I used two different set-ups for the testing cages (see pictures). In set-up one (stimulus cages opposite; Fig. 1a) the males were able to see each other, whereas in set-up two (stimulus cages adjacent; Fig. 1b) they could not. It is possible that competition between the males, and thus the effect on female preferences, differs according to whether they can see each other.

(a) “Opposite” (set-up one):

Figure 1 (a)

(b) “Adjacent” (set-up two):

Housing and general care

The subjects (eight female zebra finches, four male zebra finches and four Bengalese finches) were fed with tropical bird seed (Teurlings, The Netherlands), which was replenished every Monday, Wednesday and Friday. Drinking water and this seed were available ad libitum. Lighting in the room included UV wavelengths and operated on a 14:10 light:dark cycle, with lights on at 0700 hours. All the birds in this test were ringed for identification with orange plastic split-ring leg bands (obtained from A.C. Hughes, Middlesex, U.K.).

Throughout this experiment the males and females were housed in the testing cages. Both sexes were housed in the same room but were unable to see each other.

Test procedure

All tests were conducted between 1000 and 1700 hours.

I started with a series of pilot tests, to see if the birds responded well in the set-up under my protocol. I observed four birds (numbers 1, 2, 4 and 5) for 20 minutes and one (number 6) for 30 minutes, to see if a longer observation period produced better results. All these tests used the “adjacent” set-up.

I gave the birds 1-2 minutes to acclimatize to their new set-up. Prior to the start of a test, the cages of the males and the female were separated by wooden screens, preventing visual contact. At the start of the test I removed these screens. For my observations I used a method of scan sampling, in which I noted the position and behavior of the female and each of the two stimulus males at 5-s intervals.

Half-way through the test, the male positions were switched (left to right and right to left) to control for any preference the female might have for one side over the other, or for differences in the sensitivity or orientation of the two PIRs. I replaced the screens before switching the males, removed them after switching and replaced them at the end of the test.

After this preliminary trial, I repeated the tests with all the birds (numbers 1 - 8) for 30 minutes, so that the test duration for each female was constant. Next, I tested all these birds for 30 minutes in the “opposite” set-up.

The reason I used both cage set-ups in this experiment was to compare behavior between the two conditions. For every test she took part in, I ensured that each female saw a new combination of males.

Statistical analysis

Prior to analysis, I converted the counts registered by the PIR detectors to per-1-minute values. This involved dividing the test duration into 1-minute intervals and counting up the number of such intervals in which the male of interest received the majority of the PIR counts for that minute. This number was then expressed as a proportion of the total number of 1-minute intervals in the test, to indicate the overall preference for that male. Converting the PIR counts in this way ensured that all 1-minute intervals contributed equally to the final preference measure, and therefore that short periods in which the female was unusually active did not have an inflated influence on the result.

First, I tested the correlation between manual observations and PIR recordings. For this, I used two different kinds of PIR measurements, namely the proportion of raw counts (the counts a PIR detector registers when a bird moves in front of it) and the proportion of 1-min values. With the raw counts, if the bird sat in the middle, 'neutral' zone, neither of the PIRs recorded anything, so it wasn't counted as a preference. For the manual observations, I also excluded scans where the female was in the middle. With the proportion of 1-min values, however, if the bird sat in the middle, this was counted as equal preference for both sides. I calculated the Pearson product-moment correlation coefficients and their associated p values for the correlations between these three forms of preference data. In each case, the normality of both data sets was confirmed using a Kolmogorov-Smirnov (K-S) test.

To test the overall preference for a conspecific I used the PIR data, using a one-sample t test to investigate whether the proportion of 1-min values to the zebra-finch male differed significantly from 0.5. The test was two-tailed and I confirmed that the data were normally distributed using a K-S test.

I also tested whether there were side preferences, again using the PIR data. This time I used a one-sample t test (two-tailed) to investigate whether the proportion of 1-min values to the left stimulus cage differed significantly from 0.5. Normality of the data was tested with a K-S test.

RESULTS

All data and test residuals satisfied assumptions of normality and homoscedasticity, so parametric statistics could be applied throughout.

Correlation between manual observations and PIR recordings

There was a highly significant correlation between the preference measure derived from the PIRs and that derived from my own observations, both when the raw PIR counts were used (Pearson correlation: r = 0.752, N = 21, p < 0.001; Figure 1a) and when the PIR counts were converted to 1-min values (r = 0.872, N = 21, p < 0.001; Figure 1b). Since the latter correlation was stronger, I based all subsequent analyses on the 1-min values.

Overall preference for a conspecific

On average, females directed the majority of their behavior to the zebra-finch male in 62.7% of the 1-min intervals, indicating a highly significant preference for the conspecific (one-sample t test: t = −4.165, d.f. = 19, p = 0.001). When the data from my own observations were used, this preference was even stronger, with females directing 74.5% of the recorded behavior to the zebra-finch male (t = −5.475, d.f. = 19, p < 0.001).