Doi:10.1111/j.1600-048x.2008.04632.x

# 2009 The Authors. J. Compilation # 2009 J. Avian Biol.
Received 16 June 2008, accepted 16 November 2008 Test of an adaptive hypothesis for egg speckling along an elevationalgradient in a population of Mexican jays Aphelocoma ultramarina Elena C. Berg, John E. McCormack and Thomas B. Smith E. C. Berg (correspondence), J. E. McCormack and T. B. Smith, Center for Tropical Res., Inst. of the Environment, Univ. of California,Los Angeles, La Kretz Hall Suite 300, Box 951496, Los Angeles, USA 90095-1496. E-mail: [email protected] The adaptive significance of avian egg speckling patterns has been a subject of ongoing debate. We examined speckling ina population of Mexican jays Aphelocoma ultramarina exhibiting extreme eggshell variability. We sampled 167 eggs at 55nests from sites ranging across a steep elevation gradient within the Sierra del Carmen mountain range in Coahuila,Mexico, in order to test the recent hypothesis that egg speckling lends structural support to eggs and should therefore bemore prevalent in females subject to reduced environmental calcium. Although we documented high variation in theamount and distribution of eggshell speckling within the Sierra del Carmen jays, we found no relationship between localsoil calcium levels and the pattern of speckling. Our results indicate that explanations in addition to soil calcium levels arenecessary to explain extreme variation in eggshell speckling in birds.
The wide variety of colours and patterning of the avian One prediction following from Gosler et al.’s (2005) eggshell both among and within species has invited con- hypothesis is that when calcium is limited, speckling should siderable speculation as to its potential adaptive signifi- be densest around the crown of the egg, where the shells are cance. While many hypotheses exist to explain differences in thinnest (‘‘calcium deficiency hypothesis’’). Gosler et al.
egg appearance, speckling, in particular, is puzzling from an (2005) found a strong relationship between local soil adaptive perspective because it occurs in a wide variety of calcium availability and both eggshell mass and the dis- species in many ecological contexts Á open-cup nesters, tribution of speckling on the egg of the great tit Parus hole-nesters, and species with and without brood parasites major. Eggshells from low calcium areas were lighter and (Kilner 2006). In some ground-nesting birds, eggshell thinner, and speckles were more densely distributed at the speckling may camouflage the egg and decrease predation crown of the egg. Thus, while speckling appears to act as risk (e.g. blackheaded gulls Larus ridibundus, Tinbergen structural support for eggs of great tits, we do not know if et al. 1962). However, this hypothesis is less compelling for the calcium deficiency hypothesis for speckling holds cup-nesting and hole-nesting species, which are unlikely to generally for other bird species, and in particular for receive the same benefits from egg camouflage (Underwood medium and large birds, where calcium reserves in larger bones could mitigate environmental calcium deficiencies Gosler et al. (2005) proposed a novel hypothesis for egg speckling, suggesting that brown spots, which are produced Here, we test the hypothesis that egg speckling patterns by the pigment protoporphyrin and have qualities similar to are related to environmental calcium available to breeding solid-state lubricants used in engineering (Solomon 1997), females of the Mexican jay Aphelocoma ultramarina, a might provide structural support in areas of the shell that are medium-sized, cooperatively-breeding passerine bird that thin or weak. Egg breakage caused by thin or weak eggshells shows extreme intrapopulation variation in eggshell speckl- is exacerbated by a deficiency of calcium, which is a limiting ing. In the Sierra del Carmen of Coahuila, Mexico, where resource for breeding female birds (Dhondt and Hochachka we conducted our study, Mexican jays inhabit a steep 2001). The calcium content of soil is highly correlated with elevation gradient. They are found in highly-divergent the abundance of calcium-rich invertebrates such as snails, habitats ranging from arid woodland in the low canyons and calcium-specific foraging is a common way for many (1,400 m) to mixed conifer forest at the highest peaks bird species to supplement calcium levels during the egg- (2,750 m) (Miller 1955), making small-scale variation in laying period (Graveland and van der Wal 1996).
of eggshell quality from a limiting-resource perspective, wehave focused exclusively on calcium. For the two nests Our study was conducted on a population of the couchii where we collected four samples instead of just one, we also subspecies of Mexican jays, which ranges from southwestern conducted an analysis of alkaline earth carbonates (percent Texas to northern Nuevo Leo´n, Mexico, and is known to limestone). The first test provides a measure of the amount have speckled eggs (Ligon and Husar 1974). We collected of calcium found in organic matter (in parts per million, data on egg speckling and soil calcium from sites compris- ppm), while the second test accounts for inorganic sources ing the full elevational range of Mexican jays in the Sierra of calcium, such as calcium carbonate, that females might del Carmen of Coahuila, Mexico, mainly focusing on one low-elevation oak woodland site (San Isidro Canyon, All statistical tests were carried out using the statistical 1,458Á1,614 m), one mid-elevation pine-oak-juniper wood- software package Stata Intercooled (version 10.0, StataCorp land site (Campo Uno, 1,801Á1,865 m), and two high- 2003). We tested continuous data for normality and elevation conifer-dominated sites (Campo Dos and Tres, normalized data when necessary using standard transforma- 2,221Á2,585 m). GPS data including location and elevation tions. We assessed relationships among variables with were taken for each nest. Distances between nests ranged simple regression and multiple regression, using speckling or speckle clumping as the dependent variable and soil Using a scalpel, we first confirmed that the speckles on calcium, elevation, year, clutch size, and interaction terms as Mexican jay eggs are not superficial markings (as in many independent variables, removing non-significant terms from raptor and Charadriiforme eggs; Mikhailov 1997), but that multiple regression analyses with backward selection. We they penetrate into the shell. Thus, speckles form part of the tested for relationships among variables in two ways. First ground colour and might conceivably lend structural we assessed the intraclass correlation coefficient for the support to the shells. We assume that speckles occur at dependent variables to determine if these variables were thin parts of the eggshell, as has been shown in other species highly correlated among eggs within nests. If so, we used (Gosler et al. 2005, Jagannath et al. 2008). Speckling nest averages in regressions. Otherwise, we controlled for percentage and degree of clumping on the crown of the egg correlations among eggs within clutches with the cluster were quantified from digital photographs taken using a (variable) command in Stata, which calculates p-values Canon PowerShot A40 fitted with a macro lens to prevent using robust standard errors (Williams 2000). We also parallax. We used the Java image processing program analyzed the data with mixed-model ANOVA. Results for ImageJ (version 1.36b, Abramoff et al. 2004) to quantify both methods were similar, but we preferred the regression the area and percentage coverage of speckling on each egg model because mixed-model ANOVAs can be sensitive and on each third of each egg: crown, middle, and foot. To to small sample sizes at the level of the clutch (where 1 B avoid problems associated with inter-observer variability, n B5). Means are given9one standard deviation.
only one of us (ECB) conducted all speckling analysis.
We quantified speckling in two ways. First, we calculated the proportion of the entire (two-dimensional) egg area that was covered in speckles by dividing the total speckle area bythe total surface area of the egg. Second, we measured the We collected representative soil samples from 51 of the 55 distribution, or degree of clumping, of speckles across the nest sites spanning a broad elevation gradient (1,421Á egg’s surface, per Gosler et al. (2005), using three methods.
2,585 m). Within-nest calcium values were highly corre- First, we calculated the proportion of the total speckle area lated (intraclass correlation coefficient 00.97), indicating that was contained within the crown. Second, to control for that a single sample is probably sufficient to represent soil the relative differences in total speckle area across eggs, we calcium levels in the broader nest area. The two calcium ran a regression of the crown value against total speckle area measures Á extractable calcium and calcium carbonates Á and calculated the residuals. Third, we assigned each egg to were also highly correlated (Pearson R2 00.72, P 00.044, one of five categories, ranging from an even speckle n 08); thus, our analyses focus only on extractable calcium.
distribution (score of 0) to a highly clumped distribution Across the study site, soil calcium ranged from 783 to (score of 4). For all analyses of speckling distribution we 6,138 ppm. There was a highly significant negative relation- excluded eggs with no speckles (i.e. speckle area of zero).
ship between soil calcium and elevation (regression: F 0 Soil samples of 1 L were collected to a depth of 68.80, df 01,49, P B0.001, R200.58, n 051; Fig. 1a).
approximately six inches from the base of each nest tree.
For example, the lowest-elevation site, San Isidro Canyon To determine whether soil calcium values varied signifi- (1,458Á1,614 m), had relatively high soil calcium levels cantly within the nest area (where the nesting female was (4,51991,024 ppm, n 020), whereas soil calcium at the most likely to forage), we collected three additional soil highest-elevation site, Campo Tres (2,476Á2,585 m), was samples from randomly-selected locations within a 50- relatively low (1,1989410 ppm, n 05). Campo Uno meter radius of the nest tree at one low-elevation and one (1,801Á1,865 m) and Campo Dos (2,221Á2,400 m) were high-elevation nest site. Samples were analyzed for extrac- similar to each other with intermediate calcium levels table cations, including calcium, at A and L Western (2,6109712 ppm, n 021 and 2,4889944 ppm, n 05, Agricultural Laboratories in Modesto, USA. Other minerals are known to be important to eggshell formation, especially From 2003Á2007, we measured 167 eggs from 55 nests in their ratios with calcium (e.g. phosphorus and magne- along a steep elevation gradient in the Sierra del Carmen.
sium; Fox 1976). However, as these other minerals occur in From knowledge of flock location, we were reasonably trace amounts in the eggshell and we approach the question certain that 42 of these nests were from different flocks and was no relationship between average speckling per nest(arcsine-square-root transformed to normalize) and eleva-tion (regression: F 00.50, df 054, R2 B0.01, P 00.48,n 055; Fig. 1b), or soil calcium (regression: F 00.49, df 050, R2 00.01, P00.49, n 051).
There was relatively low correlation for speckle clumping within nests (intraclass correlation coefficient 00.33), so wetreated eggs as individual data points while controlling forcorrelations among eggs within nests (see Methods). Resultsfrom a multiple regression analysis indicated no significantrelationship between speckle clumping and a modelincluding calcium, elevation, year, and calcium)elevation,year)elevation, and year)calcium interactions (regres-sion: F 00.96, df 08,37, R200.05, P 00.46, n 0123).
Non-significant terms were removed in stepwise fashionstarting with non-significant interaction terms: calcium )elevation (marginal t 0(0.07; P 00.94), elevation (mar-ginal t 0(1.72; P 00.09) and year )elevation (marginalt 01.93; P00.06), year )calcium (marginal t 01.42;P 00.16), and year (marginal t 0(0.91; P 00.37). Theresulting simple regression between speckle clumping andcalcium was not significant (F 00.08, df01,42, R2 B0.001, P00.78, n 0123; Fig. 2). Even when these datawere analyzed without adjusting degrees of freedom toaccount for non-independence of eggs within nests, therewas no significant relationship between speckle clumpingand calcium (F 00.07, df 01,121, R2 B0.001, P 00.79, Figure 1. a: Negative relationship between soil calcium and elevation at nest sites in the Sierra del Carmen (regression: F 068.80, df 01,49, R2 00.58, P B0.001, n 051). b: No relation-ship between average speckling of clutches and elevation (regres-sion: F 00.50, df054, R2 B0.01, P 00.48, n 055).
The population of Mexican jays we studied was particularlywell-suited to a test of the calcium-deficiency hypothesis therefore highly likely to be from different females. The because we found high variation in egg speckling (0Á41% remaining 13 nests were from 10 flocks that were already surface area speckled) and soil calcium, the latter varying represented by one nest. Because dominant females often along an elevation gradient. However, we found that eggs remain as breeders within flocks over multiple years from nests at low-calcium sites were not more speckled and (McCormack and Brown 2008), we were therefore not speckles were not more clumped toward the crown of the certain if these 13 nests represented different females.
However, because our dependent variables were not highlycorrelated among the nests within a flock (intraflockcorrelation coefficient for speckle clumping 00.36 and forpercent speckling 00.45), we included these nests in ouranalyses. The results did not change when these data wereexcluded.
Speckles ranged from being completely absent to cover- ing nearly half of the shell’s surface (mean 00.0690.09,range 00 to 0.41). Speckle distribution ranged from beingevenly distributed across the egg to highly clumped towardthe crown, with the proportion of the egg’s speckles locatedon the crown ranging from 0 to 1 (mean 00.4390.29).
Our two continuous metrics of speckle clumping werehighly correlated (Pearson r2 00.94, P B0.001, n 0167),and these were correlated with our categorical metric(Pearson r2 00.58, P B0.001, n 073). Thus, we reportonly the results of analyses using the residuals of crown Figure 2. No relationship between soil calcium and clumping ofspeckles on the crown of the egg (crown speckling) (regression: speckle area against total speckle area.
F00.08, df01,42, R2 B0.001, P 00.78, n 0123). Values for Because percentage of the egg’s surface covered with crown speckling were normalized with arcsine-square-root trans- speckles (hereafter ‘speckling’) was highly correlated within formation. Eggs are represented here as independent data points, nests (intraclass correlation coefficient 00.73), we calcu- but correlations within nests were taken into account when lated nest averages for this variable (see Methods). There calculating the significance of the relationship (see Methods).
egg than at high-calcium sites. Likewise, elevation itself was tackled by returning to and rigorously testing hypotheses not correlated with either speckling or speckle clumping.
previously considered unlikely in this species, such as egg Thus, our results do not support the calcium-deficiency There are two possible explanations for why we might have failed to detect a relationship between soil calcium and Acknowledgements Á We thank CEMEX, the McKinneys, D Roe, speckling. One is that soil calcium levels in the Sierra del the Del Carmen Project staff, J Brotman, A Byrd, G Castan˜eda, Carmen might not be low enough to limit eggshell E Landay, G Levandoski, E Miller, E Pen˜aloza, and M Starling, formation in Mexican jays. In our study, extractable D Parker, UCLA Statistical Consulting, and C Bonneaud.
calcium ranged from 783 to 6,138 ppm. In Oxford’s A Gosler, T Langen, and A Maklakov commented on drafts.
Financial support was provided by a UC Mexus Doctoral Wytham Woods where Gosler et al.’s (2005) study took Dissertation Grant and a Sigma Xi Grant-in-Aid-of-Research (to place, extractable calcium ranged from 410 to 230,000 ppm JEM) and an American Philosophical Society Franklin Research (Farmer 1995). While the range of values was more extreme Grant (to ECB). All field procedures were approved by a UCLA in Gosler et al.’s (2005) study, the low-calcium values Animal Use and Care Protocol #2005-126-02.
were similar. Thus, all else being equal, female Mexican jaysfrom low-calcium nest sites in the Sierra del Carmen wereexpected to experience similar levels of calcium deficiency asthe great tits in Gosler et al.’s (2005) study. It is important to note that the range of calcium values that we recordedis similar to the variation found in other studies that Abramoff, M. D., Magelhaes, P. J. and Ram, S. J. 2004. Image demonstrated a strong relationship between soil calcium processing with Image J. Á Biophotonics Intl. 11: 36Á42.
Dacke, C. G., Arkle, S., Cook, D. J., Wormstone, I. M., Jones, S., and eggshell characteristics in birds (e.g. Graveland et al.
Zaidi, M. and Bascal, Z. A. 1993. Medullary bone and avian 1994), suggesting that if soil calcium does indeed influence calcium regulation. Á J. Exp. Biol. 184: 63Á88.
speckling in this species we would have been able to detect Dhondt, A. A. and Hochachka, W. M. 2001. Variations in calcium use by birds during the breeding season. Á Condor Alternatively, Gosler et al.’s (2005) hypothesis might not be generally applicable to all bird species. Kilner (2006) Farmer, A. M. 1995. Soil chemistry change in a lowland English reasons that a ‘‘hierarchy of selective forces’’ has probably deciduous woodland 1974Á1991. Á Water Air Soil Poll. 85: influenced the evolution of eggshell polymorphism, with crypsis providing the most general explanation for variation Fox, G. A. 1976. Eggshell quality: Its ecological and physiological significance in a DDE-contaminated common tern popula- in eggshell colour and patterning, brood parasitism playing a secondary role, and finally, the need for structural support Gosler, A. G., Higham, J. P. and Reynolds, S. J. 2005. Why are driving variation in just a few species. Medium to large- birds’ eggs speckled? Á Ecol. Lett. 8: 1105Á1113.
bodied birds might not be as dependent on environmental Graveland, J., van der Wal, R., van Balen, J. H. and van calcium during the egg formation stage. For instance, Pahl Noordwijk, A. J. 1994. Poor reproduction in forest passerines et al. (1997) showed that great tits cannot store calcium in from decline of snail abundance on acidified soils. Á Nature their bones, whereas a study of Japanese quail Coturnix japonica suggested that larger species do draw on medullary Graveland, J. and van der Wal, R. 1996. Decline in snail abundance due to soil acidification causes eggshell defects in bone as a source of calcium during egg-laying (Dacke et al.
forest passerines. Á Oecologia 105: 351Á360.
1993). Mexican jays are much larger than great tits (mean Jagannath, A., Shore, R. F., Walker, L. A., Ferns, P. N. and of 120 g versus 19 g), and female jays may thus be able to Gosler, A. G. 2008. Eggshell pigmentation indicates pesticide supplement dietary calcium with bone calcium, which can contamination. Á J. Appl. Ecol. 45: 133Á140.
be stored throughout the year in anticipation of egg laying.
Kilner, R. M. 2006. The evolution of egg colour and patterning in A study on a larger-bodied species, the sparrowhawk Accipiter nisus, showed that speckling was indeed related Ligon, J. D., and Husar, S. L. 1974. Notes on the behavioral to eggshell thinning, but that the effect was mediated ecology of Couch’s Mexican jay. Á Auk 91: 841Á843.
through environmental DDT, which blocks uptake of Madge, S. and Burn, H. 1994. Crows and jays. Á Princeton Univ.
calcium by the shell gland (Jagannath et al. 2008). While McCormack, J. E. and Brown, J. L. 2008. Mexican jay this study supports the idea that speckling can provide (Aphelocoma ultramarina). Á In: Poole, A. (ed.). The birds of structural support to calcium-deficient eggs even in a raptor, North America Online. Cornell Lab of Ornithol., Ithaca.
it did not test whether environmental calcium variation was Http://bna.birds.cornell.edu/bna/species/118.
an important factor or if large-bodied birds instead draw on Mikhailov, K. E. 1997. Avian eggshells: an atlas of scanning medullary sources of calcium during egg-laying.
electron micrographs. Á Brit. Ornithol. Club, UK.
Considering that nearly all New World corvids have Miller, A. H. 1955. The avifauna of the Sierra del Carmen of speckled eggs (Madge and Burn 1994), including congeners Coahuila, Mexico. Á Condor 57: 154Á178.
Pahl, R., Winkler, D. W., Graveland, J. and Batterman, B. W.
of the scrub-jay complex, speckled eggs in the couchii 1997. Songbirds do not create long-term stores of calcium in subspecies of Mexican jays could represent the retention of their legs prior to laying: results from high-resolution radio- an ancestral state that currently has no adaptive value.
graphy. Á Proc. R. Soc. B 264: 239Á244.
Before accepting this explanation, however, the adaptive Solomon, S. E. 1997. Egg and eggshell quality. Á Iowa State Univ.
value of speckling in this species should probably now be StataCorp. 2003. Stata statistical software. Á StataCorp LP, Underwood, T. J. and Sealy, S. G. 2002. Adaptive significance of egg coloration. Á In: Deeming, D. C. (ed.). Avian incubation, Tinbergen, N., Broekhuysen, G. J., Feekes, F., Houghton, behaviour, environment and evolution. Oxford Univ. Press, J. C. W., Kruuk, H. and Szulc, E. 1962. Egg shell removal by the black-headed gull, Larus ridibundus L.: a behaviour Williams, R. L. 2000. A note on robust variance estimation for component of camouflage. Á Behaviour 19: 74Á117.
cluster-correlated data. Á Biometrics 56: 645Á646.

Source: http://www.environment.ucla.edu/ctr/research/Ecol-Evol/Berg-J-Avian-Biol-Test-of-adaptive-2009.PDF