English & Montgomerie ESM Page 1
Electronic Supplementary Material
Robin’s egg blue: does egg color influence male parental care?
JOURNAL: Behavioral Ecology and Sociobiology
Philina English† and Robert Montgomerie*
Department of Biology, Queen’s University, Kingston, ON K7L 3N6, Canada
emails: †, *
In this online supplement (ESM) we present some additional details of our Methods and Results, especially to provide full details of the statistical methods used and the models evaluated using the IT approach.
American robins are socially monogamous passerine birds that usually lay 3-4 immaculate blue-green eggs in a bulky nest of grass and mud, most often placed on a firm branch or ledge that is sheltered from above. Throughout most of their range, robins can rear as many as three broods in a single breeding season despite suffering frequent nest depredation from squirrels, snakes, a variety of corvids and blackbirds (Sallabanks and James 1999). During the incubation period, males often visit the nest (personal observations) where they would have ample opportunity to see the eggs and evaluate egg color.
Though egg-laying began in mid April, breeding densities and synchrony were not sufficient for experimental manipulation until the beginning of May. Incubation typically begins when the penultimate egg is laid and lasts 12-13 d. Therefore, in a typical four-egg clutch, the first chicks hatch 12-15 d after the first egg is laid. Immediately after the eggs hatch, males provide the greater proportion of the food to the nestling, while females do all of the brooding. As the nestlings age and require less thermoregulation by the mother, females increase their feeding rates (Collar 2005).
Measuring egg color
To measure egg color,we mounted the spectrometer’s reflectance probe in a holder designed to minimize the effects of ambient light during measurements, and placed it touching and perpendicular to the egg surface. Reflectance was measured relative to a standard white reference, and was calibrated against both this and a dark standard prior to the measurement of each clutch. As a white reference standard, we used a fresh piece of Teflon tape wrapped 3 times around a small piece of white plastic and made a new standard for each clutch. This standard was tested against an optical grade Spectralon® (Labsphere, USA) standard and was equivalent in spectral quality, with the advantage of always being clean. We used the software RCLR (version 1.0; Montgomerie 2010) for all calculations of color variables.
Luminance (total area under the reflectance curve) is negatively related to biliverdin concentration in eggshells (López-Rull et al. 2008), but is also sensitive to calibration and measurement error and can have low repeatability. Various indices of the chroma, or spectral purity, of eggshells are also correlated with their biliverdin concentration (Moreno et al. 2006a; López-Rull et al. 2008), and, in turn, with female, egg and nestling quality in other passerine species (Morales et al. 2006; Moreno et al. 2006a; Siefferman et al. 2006; Hargitai et al. 2008; Lopez-Rull et al. 2008; Morales et al. 2008; Hanley et al. 2008; but see Cassey et al. 2008). The majority of these studies of eggshell coloration have used an index of blue-green chroma, but there is considerable variety in the way this index has been calculated (ESM Table 1).
ESM Table 1 Different indices of blue-green chroma that have been used to quantify the intensity of color of eggs pigmented with biliverdin from different bird species. In this table, Ri-j is total reflectance from wavelengths i to j.Blue-green chroma index / Reference
R400-576/R380-700 / Hanley et al. 2008
R400-570/R360-700 / Moreno et al. 2006a, b; Morales et al. 2008
R400-580/R360-700 / Morales et al. 2006
R400-580/R320-700 / Hargitai et al. 2008
R400-570/R300-700 / Moreno et al. 2006a; López-Rull et al. 2008
R400-575/R300-700 / Siefferman et al. 2006; Soler et al. 2008
R401-600/R301-700 / Krist and Grim 2007
Band colors were chosen so that the sex of the parent could be easily determined on video recordings during feeding visits, and to avoid red and orange colors similar to those in that adult plumage that might confound investment decisions (Burley et al. 1982). Thus we attached pale blue, yellow or white bands to females, and dark blue or purple bands to males. To reduce nest abandonment, the female at each nest was caught only after incubation began; males were captured >48 h prior to monitoring provisioning rates.
No video recording was done during extreme heat or heavy rain. Regional air temperature at the start of each recording was estimated from Environment Canada data for Kingston, ON. Videos were reviewed using Sony Vegas Platinum 8 software. Recording was done continuously for about 2 h, but sample duration [experimental, 2.01 h (1.92-2.11), N = 31 recordings; unmanipulated, 1.95 h (1.91-1.98), N = 98] was measured from the beginning of the first parental visit after start of video, to control for potential differences in individual responses to human activity in the vicinity of the nest. Delay to first visit ranged from 0 to 57 min after video recording began, but was usually short [experimental: 6.8 (2.6-11.1) min, N = 31; unmanipulated: 5.4 (3.8-7.1) min, N=98].
The feeding rate of the other parent was included in each model to control for unmeasured variation in habitat and environmental conditions that might influence both parents, and to account for the possibility that each parent might modify its effort to compensate for the investment of the other parent. Female time at the nest was included as a measure of the total time that the female spent brooding and feeding nestlings. We included this measure as a predictor both because it limits the time available to the female for foraging, and because the male might assess female effort by her total time at the nest rather than the frequency of her feeding visits. Time of day might also influence feeding visit rates due to hunger after a night of fasting and because potential food sources might change through the course of the morning. Extremes in ambient air temperature would be expected to increase the time a female spends brooding/shading, and to affect the availability of food. Likewise, variation in the number of nestlings and their age could influence both the amount of food and thermoregulatory care required. Finally, the average egg-laying date for each clutch and the number of previous nesting attempts by the parents were included to control for the possibility that parental behavior changes either through exhaustion of intrinsic or extrinsic resources or with reduced chances of producing subsequent broods. Of all these variables, those found to be good predictors of parental feeding visit rates based on hierarchical partitioning (Mac Nally 2002) were explored further using correlation analysis, and included in subsequent multiple regression analyses.
Permits and ethical considerations
This research was conducted under the authority of Queen’s University Animal Care Protocol 2005-044-R2, and Bird Banding (10448) and Scientific (CA 0221) Permits from the Canadian Wildlife Service, as well as permission from local landowners, Queen’s University Biological Station, and Prince Edward Point Bird Observatory. Every effort was made to minimize sample sizes by conducting a carefully controlled experiment and thus maximizing statistical power.
Models to predict male provisioning rates
To identify which factors best predicted male provisioning rate [log(x+1)-transformed], we used the Information-Theoretic approach to model selection (Anderson 2008; Symonds and Moussalli 2010) for Tables 1 and 3 in the main paper, and associated material in the Results. In the following tables the models are ranked from lowest AICc (best model) to highest, and we list all of the models that have any support (ΔAICc <10) compared to the best model. Generally models with ΔAICc <2 are considered to have equivalent support to that of the best model in each set (Anderson 2008). We also list the weight (ω) for each model, which is the probability that each model is the correct model, given the data, and the evidence ratios (ER) comparing the support for each model relative to the best model in each set.
In the following tables, the predictors are: Nage = nestling age (days since hatch), SexParent = male or female parent; Nchicks = number of nestlings; Ftime = amount of time female spent at the nest (min/h) either brooding or feeding nestlings; CHR = chroma of eggshell; BGC = blue-green chroma of eggshell; Treat = pale or vivid egg colour treatment.
ESM Table 2 Models to predict nestling provisioning rates by each parent at 40 unmanipulated nests. Note that color variables are not evaluated in these models.Rank / Model / AICc / ΔAICc / ω / ER
1 / Nage + SexParent + Nchicks + Ftime – Nchicks*Ftime / –161.2 / 0 / 0.89
2 / Nage + SexParent + Nchicks – Ftime / –155.6 / 5.64 / 0.05 / 16.8
3 / Nage + SexParent + Nchicks / –155.5 / 5.72 / 0.05 / 17.5
4 / SexParent + Nchicks + Ftime – Nchicks*Ftime / –152.5 / 8.74 / 0.01 / 79.0
ESM Table 3 Models to predict male provisioning rates to 3-day-old nestlings at experimental nests where natural eggs were replaced with artificial ones (Treatment nests = 5 pale, 8 vivid).Rank / Model / AICc / Δ AICc / ω / ER
1 / Treat + Ftime / –17.1 / 0 / 0.62
2 / Treat / –15.4 / 1.72 / 0.26 / 2.4
3 / Treat + Ftime + Nage / –11.6 / 5.49 / 0.04 / 15.6
4 / Nage + Treat / –11.3 / 5.86 / 0.03 / 18.7
5 / null / –11.1 / 6.01 / 0.03 / 20.2
6 / Ftime / –8.6 / 8.54 / 0.009 / 71.5
7 / Nage / -7.9 / 9.19 / 0.006 / 99.0
We used the same procedure as above to test for an effect of egg colour on parental feeding visit rates at natural, unmanipulated broods (ESM Table 4). In this dataset, one brood—at which the male did not feed during our sampling period for 3-d-old nestlings, even though he was known to be alive and in the area—was a large outlier in many analyses and had high leverage in multiple regression analyses (see ESM Fig. 1). To meet model assumptions with respect to residuals, we omitted this sample for 3-d-old nestlings, and log-transformed male feeding visit rates.
ESM Table 4 Models to predict male provisioning rates at different nestling ages. Each model includes either (A-C) CHR or (D-F) BGC as a predictor.
A Models that include CHR as a predictor for 3-d-old nestlings.
A1 All nests (n = 28), including a statistical outlier where the male did not feed nestlings during the sampling period (see also ESM Fig 1).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks / –6.7 / 0 / 0.21
2 / null / –6.1 / 0.54 / 0.16 / 1.3
3 / CHR + Nchicks / –6.0 / 0.72 / 0.15 / 1.4
4 / CHR + Nchicks + Ftime – Nchicks*Ftime / –5.9 / 0.80 / 0.14 / 1.5
5 / CHR / –5.1 / 1.58 / 0.10 / 2.2
6 / Nchicks – Ftime / –4.3 / 2.39 / 0.07 / 3.3
7 / Nchicks + Ftime – Nchicks*Ftime / –3.9 / 2.81 / 0.05 / 4.1
8 / – Ftime / –3.8 / 2.87 / 0.05 / 4.2
9 / CHR + Nchicks – Ftime / –3.3 / 3.37 / 0.04 / 5.4
10 / CHR – Ftime / –2.5 / 4.15 / 0.03 / 8.0
A2 As above, but with the outlier removed (n = 27 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks + Ftime – Nchicks*Ftime / –24.1 / 0 / 0.45
2 / CHR + Nchicks + Ftime – Nchicks*Ftime / –23.4 / 0.66 / 0.33 / 1.4
3 / Nchicks / –21.1 / 2.91 / 0.11 / 4.3
4 / Nchicks – Ftime / –20.1 / 3.99 / 0.06 / 7.4
5 / CHR + Nchicks / –18.6 / 5.45 / 0.03 / 15.3
6 / CHR + Nchicks – Ftime / –17.2 / 6.85 / 0.02 / 30.7
7 / null / –14.7 / 9.33 / 0.004 / 106
B Models that include CHR as a predictor for 6-d-old nestlings (n = 31 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks + Ftime – Nchicks*Ftime / –17.2 / 0 / 0.57
2 / Nchicks + Ftime / –15.4 / 1.86 / 0.22 / 2.5
3 / CHR + Nchicks + Ftime – Nchicks*Ftime / –14.2 / 2.97 / 0.13 / 4.4
4 / CHR + Nchicks + Ftime / –12.8 / 4.44 / 0.06 / 9.2
5 / Nchicks / –9.8 / 7.41 / 0.01 / 40.7
6 / CHR + Nchicks / –7.6 / 9.64 / 0.005 / 124
C Models that include CHR as a predictor for 9-d-old nestlings (n = 25 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks / –26.1 / 0 / 0.55
2 / Nchicks – Ftime / –24.8 / 2.17 / 0.18 / 3.0
3 / CHR + Nchicks / –22.1 / 2.86 / 0.13 / 4.2
4 / Nchicks + Ftime – Nchicks*Ftime / –21.2 / 3.76 / 0.08 / 6.6
5 / Nchicks – CHR – Ftime / –19.6 / 5.31 / 0.04 / 14.2
6 / Nchicks – CHR + Ftime – Nchicks*Ftime / –17.7 / 7.27 / 0.14 / 37.9
D Models that include BGC as a predictor for 3-d-old nestlings (n = 27 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks + Ftime – Nchicks*Ftime / –24.1 / 0 / 0.60
2 / Nchicks / –21.1 / 2.91 / 0.14 / 4.3
3 / BGC + Nchicks + Ftime – Nchicks*Ftime / –20.7 / 3.31 / 0.11 / 5.2
4 / Nchicks – Ftime / –20.1 / 3.99 / 0.08 / 7.4
5 / Nchicks – BGC / –18.6 / 5.43 / 0.04 / 15.1
6 / BGC + Nchicks – Ftime / –17.1 / 7.00 / 0.02 / 33.1
7 / null / –14.7 / 9.33 / 0.006 / 106
E Models that include BGC as a predictor for 6-d-old nestlings (n = 31 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks + Ftime – Nchicks*Ftime / –17.2 / 0 / 0.52
2 / Nchicks + Ftime / –15.4 / 1.86 / 0.21 / 2.5
3 / Nchicks – BGC + Ftime – Nchicks*Ftime / –15.1 / 2.10 / 0.18 / 2.9
4 / Nchicks – BGC + Ftime / –13.1 / 4.07 / 0.07 / 7.7
5 / Nchicks / –9.80 / 7.41 / 0.01 / 40.7
F Models that include BGC as a predictor for 9-d-old nestlings (n = 25 nests).Rank / Model / AICc / Δ AICc / ω / ER
1 / Nchicks / –24.9 / 0 / 0.50
2 / Nchicks – BGC / –23.0 / 1.94 / 0.19 / 2.6
3 / Nchicks – Ftime / –22.8 / 2.17 / 0.17 / 3.0
4 / Nchicks + Ftime – Nchicks*Ftime / –21.2 / 3.76 / 0.08 / 6.6
5 / Nchicks – BGC – Ftime / –20.1 / 4.82 / 0.05 / 11.1
6 / Nchicks – BGC + Ftime – Nchicks*Ftime / –17.9 / 7.03 / 0.02 / 33.6
ESM Fig 1Partial residual plot of relation between log male feeding visit rate and eggshell chroma, both controlled for the amount of time the female spent at the nest, the number of chicks, and the interaction between those two predictors. Statistical outlier shown as solid dot. The full model here is statistically significant (F = 2.9, P = 0.047, df = 4,23, R2 = 0.33)
Anderson DR (2008) Model based inference in the life sciences: a primer on evidence. Springer Science+Business Media,New York
Burley N, Krantzberg G, Radman P (1982) Influence of colour-banding on the conspecific preferences of zebra finches. Anim Behav 30:444-455
Cassey P, Ewen JG, Blackburn TM, Hauber ME, Vorobyev M, Marshall NJ (2008) Eggshell colour does not predict measures of maternal investment in eggs of Turdus thrushes. Naturwissenschaften 95:713-721
Collar N (2005) Family Turdidae (thrushes). In: del Hoyo J, Elliot A, Christie D (eds) Handbook of birds of the world, volume 10: Cuckoo-shrikes to thrushes. Lynx Editions,Barcelona
Hanley D, Heiber G, Dearborn DC (2008) Testing an assumption of the sexual-signaling hypothesis: does blue-green egg color reflect maternal antioxidant capacity? Condor 110:767-771
Hargitai R, Herényi M, Török J (2008) Eggshell coloration in relation to male ornamentation, female condition and egg quality in the collard flycatcher Ficedula albicollis. J Avian Biol 39:413-422
Krist M. and T. Grim (2007) Are blue eggs a sexually selected signal of female collared flycatchers? A cross-fostering experiment. Behav Ecol Sociobiol 61:863-876
López-Rull I, Miksik I, Gil D (2008) Egg pigmentation reflects female and egg quality in the spotless starling Sturnus unicolor. Behav Ecol Sociobiol 62:1877-1884
Mac Nally R (2002) Multiple regression and inference in conservation biology and ecology: further comments on identifying important predictor variables. Biodiversity and Conservation 11:1397-1401
Montgomerie R (2010) RCLR, version 1.0. Kingston, Canada: Queen’s University. Available at
Morales J, Sanz JJ, Moreno J (2006) Egg colour reflects the amount of yolk maternal antibodies and fledging success in a songbird. Biol Letters 2:334-336
Morales J, Velando A, Moreno J (2008) Pigment allocation to eggs decreases plasma antioxidants in a songbird. Behav Ecol Sociobiol 63:227-233
Moreno J, Lobato E, Morales J, Merino S, Tomas G, Martinez-de la Puente J, Sanz JJ, Mateo R, Soler JJ (2006a) Experimental evidence that egg colour indicates female condition at laying in a songbird. Behav Ecol 17:651-655
Moreno J, Morales J, Lobato E, Merino S, Tomas G, Martınez-de la Puente J (2006b) More colourful eggs induce a higher relative paternal investment in the pied flycatcher Ficedula hypoleuca: a cross-fostering experiment. J Avian Biol 37:555-560
Sallabanks R, James FC (1999) American robin (Turdus migratorius). InPoole A (ed) The birds of North America online. Cornell Lab of Ornithology, Ithaca, NY. Retrieved in Nov 2009 from The Birds of North America Online database at http://bna.birds.cornell.edu/bna/species/462
Siefferman L, Navara KJ, Hill GE (2006) Egg colouration is correlated with female condition in eastern bluebirds (Sialia sialis). Behav Ecol Sociobiol 59:651-656
Soler JJ, Navarro C, Contreras TP, Aviles JM, Cuervo JJ (2008) Sexually selected egg coloration in Spotless Starlings. Am Nat 171:183-194
Symonds MRE, Moussalli A (2010) A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behav Ecol Sociobiol in press, doi:10.1007/s00265-010-1037-6