LETTER Why are birds eggs speckled?

Ecology Letters, (25) 8: 5 3 doi:./j.46-248.25.86.x LETTER Why are birds eggs speckled? Andrew G. Gosler, * James P. Higham and S. James Reynolds 2 Edward Grey Institute of Field Ornithology, Department of Zoology, South Parks Road, Oxford OX 3PS, UK 2 School of Biosciences, University of Birmingham, Edgbaston, Birmingham B5 2TT, UK *Correspondence: E-mail: andrew.gosler@zoo.ox.ac.uk Present address: James P. Higham, Centre for Research in Evolutionary Anthropology, Roehampton University, Holybourne Avenue, London SW5 4JD, UK Abstract Birds are unique in laying eggs with pigmented shells, but for most species (e.g. most passerines, which lay white eggs speckled with reddish spots of protoporphyrin) the pigmentsõ function is unknown. We studied a bird population at a geologically variable site, and considered a hitherto untested hypothesis: that protoporphyrin pigments might compensate for reduced eggshell-thickness (caused partly by calcium deficiency), which is known to reduce eggshell-strength and increase eggshell-permeability. We found that pigment spots specifically demarcated thinner areas of shell, with darker spots marking yet thinner shell than paler spots. Variation in pigmentation was thus associated with variation in shell thickness both within and between clutches, so accounting for the eggshell s characteristic spot patterns. Geological variability at this site has resulted in a great range of calcium availability and, as predicted by the hypothesis, variation in calcium availability was found to affect between-clutch variation in both eggshell-mass (+) and pigmentation characteristics ()). We suggest a physiological mechanism and some important implications of these findings. Keywords Calcium, eggshell-function, eggshell-pigmentation, great tit, signalling. Ecology Letters (25) 8: 5 3 Although evidence from a few taxa supports a signalling role for the pigments on avian eggshells, either in crypsis (effectively a dishonest ÔsignalÕ) from predators (Bakken et al. 978; Blanco & Bertellotti 22; Sanchez et al. 24), for eggpattern mimicry in the case of brood-parasites (Davies & Brooke 989a,b) or in signalling female condition in the case of biliverdin-based pigments (Moreno & Osorno 23; Moreno et al. 24), explanations based on a signalling function cannot account for the eggshell-pigmentation of most species (Newton 896; Lack 968; Verbeek 99; Weidinger 2). First, all the species in most non-passerine orders lay white, essentially unpigmented, eggs. Secondly, most small passerines lay white eggs speckled with reddish spots, typically forming a ÔcoronaÕ ring around the broad end of the egg (Lack 968; Davies & Brooke 989a,b); the inadequacy of these spots to function in crypsis, etc. has been debated without conclusion at least since 838 (Hewitson, cited by Newton 896). This ubiquitous, but hitherto unexplained, patterning is the subject of this paper. Here we present the first evidence that, rather than a signal function, these pigments compensate for eggshell thinning, caused by structural variation in the shell and by calcium deficiency (suggested by Solomon 987, 997), the effects of which are critical for the egg s viability (Graveland et al. 994; Perrins 996; Graveland & Berends 997; Graveland & Drent 997). Together, our results present a wholly new explanation for eggshell patterning, related to eggshell function. Although not restricted to the Passeriformes, small, asymmetrically speckled eggs (the maculated egg) are ubiquitous in that order (which comprises c. 6% of extant bird species), being represented in all 22 passerine families of the Holarctic (Sibley & Monroe 99), and frequently in the Neotropics, Australasia and elsewhere. Furthermore, although across all avian taxa, hole-nesting species tend, more than open-nesting species, to lay unpigmented eggs (Lack 968), maculated eggs occur in both open-nesting and hole-nesting species; they occur in cuckoo-hosts (whose eggs might be under selection through egg-pattern mimicry by cuckoos, but see Davies & Brooke 989a) and in species not parasitized by cuckoos, and across species ranging widely in terms of habitat and diet. Hence they are not restricted to any particular ecological or behavioural grouping. The great tit, Parus major, exemplifies the problem. Its maculated eggs are typical of small passerines (Fig. ) and, during laying, the female usually covers the eggs with nest material, which conceals them effectively, and, together with its hole-nesting habit, obviates the need for crypsis (Lack 968). The great tit is not a host to the European cuckoo, Cuculus canorus, and ejects

6 A. G. Gosler et al. Figure Pigment variation within and between clutches of great tit eggs. Great tits lay a clutch of 5 2 eggs, each c..2 2. g. The photograph shows the first (top row), ÔmiddleÕ (centre row) and last (bottom row) eggs from each of five clutches (columns). Note the increase in spot intensity and size between first and last eggs. neither real nor dummy, conspecific nor heterospecific eggs from its nest (Pettifor et al. 988; Davies & Brooke 989a; Hansen & Slagsvold 24); it will even incubate eggs of its own and other species already in a nest that it takes over (Gosler 993). Thus there is no evidence for a signal function for the spots on great tit eggs. The spots of the maculated egg consist chiefly of protoporphyrins (Kennedy & Vevers 976), made in excess by birds in the biosynthesis of blood haem (Burley & Vadhera 989). Protoporphyrin pigment occurs at various depths within (Mikhailov 997), and is thought to be integral to, the eggshell (Solomon 987, 997). In addition to their relative abundance in avian blood, and their being phneutral, protoporphyrins have two important properties in the present context. First their structure resembles that of certain solid-state lubricants used in engineering (Solomon 987, 997), hence it has been suggested that they might act as lubricants (shock-absorbers) within the shell matrix (Solomon 987, 997). Secondly, eggshell protoporphyrins reflect strongly in the infra-red (Bakken et al. 978), perhaps producing cold spots under incubation and reducing water loss. The rate of water-loss from the egg (typically c. 8% throughout incubation) is critical for normal embryonic development (Rahn & Ar 974). In small passerines, eggshell structure is considered ÔincompleteÕ. It lacks the vertical crystal layer (external zone) of larger species, and is composed almost entirely of a highly vesiculated squamatic zone (Mikhailov 997). Eggshell thickness is the principal determinant of shell strength (Tyler 969; Ar et al. 979), and of the rate of water loss (equivalent to mass loss, Ar et al. 974) during incubation; it also reflects calcium availability to the female during egg formation (Ar et al. 979; Graveland et al. 994). Furthermore, calcium, which must be sought daily for egg production (Graveland & Berends 997), is a critical limiting resource for shell formation in small birds (Graveland et al. 994; Perrins 996; Graveland & Berends 997; Graveland & Drent 997). Small snails are the main dietary source of calcium for great tits (Graveland et al. 994; Graveland & Berends 997; Graveland & Drent 997). Thus if pigments are an adaptive part of the eggshell s structure as suggested above, we can make three predictions: () we should see relationships between pigmentation and eggshell thickness; (2) if calcium is scarce, we should also see an effect of calcium availability on pigmentation and eggshell thickness; and (3) if variation in egg-shape is associated with variation in the distribution of mass in the shell, we should also see covariation between pigmentation and egg-shape. None of these relationships would be predicted by any signalling hypothesis. We studied the eggs of great tits using nestboxes in Wytham Woods near Oxford, UK (Gosler 993). We worked on the north slope of Wytham Hill (see Materials and methods), an area of just -km 2 across which, owing to varying geology, soil-calcium (SC) varies massively [63 23 mg ( g) ) soil] (Farmer 995), and where recent studies also indicate a strong correlation between snail abundance and SC (Jubb 25). Nestboxes were associated with their nearest SC values, and each egg assessed visually for two principal components of pigmentation: ÔdarknessÕ and ÔspreadÕ (defined in Materials and methods and in Gosler et al. 2) in selected clutches. Pigment ÔdarknessÕ is known to be heritable on the female line (Gosler et al. 2). However, further analysis showed that pigment ÔspreadÕ was not heritable (see Materials and methods). We predicted that if pigments served a structural function, variation in pigment darkness and/or spread should relate (both within and between clutches) to eggshell mass and/or thickness, and (between clutches) to SC. We therefore studied these relationships while accounting for other likely factors by using multivariate modelling. We also studied the rate of water-loss from the eggs in relation to their pigmentation, but will report those findings separately.

Why are birds eggs speckled? 7 MATERIALS AND METHODS Study population, site and soil surveys The Wytham great tit population has been studied since 947 (Gosler 993). Female great tits typically build a nest of moss and hair in April, and lay one egg each day until clutch completion, whereupon incubation, again by the female, starts and lasts 3 days. The birdsõ use of nestboxes allows close examination of the clutch and breeding females, which are individually identified by numbered BTO rings (Gosler 993). For ethical reasons, females are not trapped (or identified) until the chicks are at least a week old. Thus it is not possible to identify females associated with any clutch or brood deserted before this. Identified females were aged, and standard biometrics recorded (Gosler 993; Gosler et al. 2); each female is represented only once in any analysis. Wytham Woods (36 ha, 5 47 N, 9 W) stand on a sandstone hill, capped with limestone, rising to 64 m above sea-level (a.s.l.), from the clays of the Thames floodplain (6 m a.s.l.). In 974, members of the Commonwealth Forestry Institute, Oxford University, undertook a detailed soil survey of Wytham Woods. Samples were taken from each of three quadrates within alternate m 2 ( ha 2 ) of the OS grid in a chess-board pattern (Dawkins & Field 978; Farmer 995). Soil resampling in 99 showed that these data were still representative (Farmer 995). We used the mean calcium values for each -ha 2 of that survey. Our study area contains 29 tit nestboxes out of c. in the whole Wytham Woods estate. Nestboxes were associated with their nearest SC values by taking the maximum calcium value of the four or five (as available) -ha 2 nearest to the nestbox square. This was preferred to the mean because we assumed that birds would not search randomly for calcium. For analysis here, SC values were normalized by Log transformation and referred to as Ôsoil-calciumÕ throughout the text. Our study area is oak woodland (Quercus robur, quercetum, Whitbread & Kirby 992) within which four clear types are recognized [(a) ancient semi-natural woodland with Acer campestris; or (b) with A. pseudoplatanus; (c) regenerated woodpasture; (d) C 2th plantation (Gosler 99)]. These were factored into statistical models as the variable ÔhabitatÕ (H). Territory size (TS), included in initial models (but never a significant effect), was calculated from a year-by-year GIS interpolation of tessellated polygons based on the distribution of nestboxes occupied by great tits. Recording eggshell pigmentation Pigment intensity (I: scored in.5 increments from for palest spots to 5 for the darkest), distribution (D: scored in.5 increments from for > 9% of spots concentrated at one end, to 5 for an even spot distribution) and spot-size (S: scored in.5 increments from for small spots, to 3 for large spots) were recorded (see also Gosler et al. 2), and the principal components pc and pc2 calculated from the correlation matrix of I, D and S. All eggs were assessed by the same observer (AGG). Under this system, pure white eggs are scored as zero for I, D and S. However, as such values may be misleading for D and S (e.g. if pigment is absent it cannot have a distribution), no pure white clutches were used in the present study (see Gosler et al. 2). As pc shows a strong positive loading of spot intensity (I ) and size (S) (eigenvectors.622 and.562, respectively) and negative loading of spot distribution (D: ).546) it is referred to as the ÔdarknessÕ (PD) of the egg. pc expresses 56.8% of the total variation in I, D and S. Thus increasing pc represents increasing overall darkness of the eggshell s pigment. pc2 expresses a further 24.% of the total variation, and reflects a strong loading of spot distribution (D:.743) and size (S:.668), but with little expression of spot intensity (I:.5). It was therefore taken to represent the ÔspreadÕ (PS) of maculation. Low values of pc2 represent pigment concentrated with smaller spots at one end, becoming more evenly dispersed with larger spots over the egg with increasing pc2. Pigment ÔdarknessÕ is known to be heritable (by mother/daughter correlation, Gosler et al. 2). However, re-examination of the data used for that heritability analysis, which did not consider pc2, found no such correlation for pc2 (r 67 ¼.32, P ¼.676 n.s.). Thus while the ÔspreadÕ of maculation expresses a quarter of the variation, independent of ÔdarknessÕ, this is not of genetic origin. Egg sampling In 22, 3 clutches (in which every egg was pigment-scored for I, D and S) were selected covering a range of SC. These nests were visited daily during egg-laying and the day s egg lightly numbered (egg-number) with a felt-tip pen so identifying the clutch laying sequence (LS) to relate to eggshell mass, pigmentation and rates of water-loss (reported separately). Five of these clutches were collected (see acknowledgements for licence details) for analysis before incubation. The eggs (n ¼ 4) were measured (length, maximum breadth to. mm) and the breadth/length ratio (B/L), volume and surface-area (ESA) calculated (Hoyt 979). They were emptied and their shells reduced to ash in a Carbolite furnace at 8 C for 7 h, so vaporizing any organic matter (e.g. membrane) adhering to the shells; the ash was weighed to. g on a Sartorius R6 D electronic balance and expressed as g mm )2 ash/surfacearea (EAM). Four other clutches were excluded as they were deserted during egg-laying or incubation. This desertion rate (4/3) did not differ significantly from that of the whole study population in that year (25/38, v 2 ¼.2, n.s.). In 23, 45 eggs were collected representing the first, middle

8 A. G. Gosler et al. Eggshell thickness (mm) Pigmented shell thickness (mm).85.8.75.7.65.6.9.85.8.75.7.65 C S W F Egg region [third to fifth depending on clutch-size (CS)] and last eggs from each of 5 clutches, measured and ashed as in 22. In 24, 36 unincubated eggs (six clutches, laying-sequence unknown) were collected to measure shell thickness (to. mm) in different regions of the egg, using a Mitutoyo micrometre modified specifically for the purpose (Scharlemann 2). The eggs were emptied, and shell membranes removed. The thickness of unpigmented shell was measured within four regions: the ÔcrownÕ (flattened area over air-cell), ÔshoulderÕ (air-cell to broadest part of shell, including any C S W.6.6.65.7.75.8.85.9 Unpigmented shell thickness (mm) F CP CP7 E3 E43B E53 W5 Figure 2 Upper: Eggshell thickness was determined from eggs collected in 24 (36 eggs, six clutches, laying-sequence unknown). Variation in unpigmented-shell thickness (mm) across the egg is highly significant (ClutchID: F 5,52 ¼ 3.25 P <., region: F 3,52 ¼ 73.5 P <.); egg regions: C, crown; S, shoulder; W, Waist; and F, Foot. Lower : eggshell thickness (mm) inside (y) a pigment spot was regressed (y ¼ ).8 +.9298x) on thickness just outside the pigment spot (x). The line of unity (y ¼ x) is also shown. Eggshell thickness at a pigment spot is significantly thinner across spots, eggs and even clutches (t 5paired ¼ ).63, P <.; t 22paired ¼ )2.22, P <.; t 5paired ¼ )7.94, P ¼., respectively). The different symbols represent different clutches. ÔcoronaÕ ring), ÔwaistÕ (conical region) and ÔfootÕ (pointed end) (Fig. 2). The thickness of pigmented shell (i.e. within a pigment spot) was also measured in a subset of eggs (6 spots, 23 eggs, six clutches) in which spot size was large enough to permit this, together with that of the unpigmented shell immediately adjacent (c. mm from the spot) to it within the same egg region. Statistical treatment Statistical analysis used multivariate modelling. As CS and lay-date (LD: April ¼ ) are important to great tit reproductive ecology (Gosler 993; Perrins 996), these were incorporated initially in all models. Also, as eggs from the same clutch are not statistically independent, data were analysed using either General Linear Models with ÔClutch IDÕ (MINITAB V3) fitted as a factor, or by Generalized Linear Mixed Models using Restricted Maximum Likelihood (REML in Genstat Release 8) with Clutch ID and Year as random effects. In the latter case, for fixed effects, d.f. of the reported Wald Statistic W is unless stated otherwise, and effect direction is indicated + or ). The models reported are the minimum adequate models (MAM) resulting from a step-down model simplification procedure. REML Full Models were used with the following response variables (see above for abbreviations of variables): EAM, ESA, B/L, PD, PS, and fixed effects: FA, LS, LS 2, CS, LD, H, SC, TS, ALT (also EAM, ESA and B/L if not response variable). Abbreviations of factors not given above: FA, female-age (first-year or older), ALT, nestbox altitude (m). RESULTS Relationships between pigmentation and eggshell thickness In the eggs collected in 24, the thickness of unpigmented eggshell varied across the different regions of the egg (Fig. 2). Within eggs, the tiny foot region had the thinnest shell, but the unpigmented shell at the shoulder, the region of the ÔcoronaÕ ring, was thinner than that at either the crown or waist. In all regions, pigmented shell was consistently, and significantly thinner [by 7.8 ± 7.28 (SD) % on average, max 46.4%) than adjacent unpigmented shell (Fig. 2). Across spots, eggs and even clutches, the shell at pigment spots averaged thinner (Fig. 2). We found no difference between clutches or eggs in the mean difference in thickness between pigmented and unpigmented shell [ClutchID F 5, ¼.37, P ¼.867; egg (ClutchID) F 7,5 ¼.52, P ¼.936]. However, the mean difference between pigmented and unpigmented eggshell increased significantly with size (surface area) of the egg (ClutchID: F 5,5 ¼.8 n.s., ESA: F,5 ¼ 8.26 P ¼.2). Further-

Why are birds eggs speckled? 9 Shoulder shell thickness variation.2.8.6.4.2..8.6.4.2.5.6.6.7.7.8.8.9 Mean shoulder shell-thickness (mm) Crown / shoulder thickness.4.3.2..9.68.7.72.74.76.78.8.82 Egg breadth / length 3 Pigment spread (pc2) 3 2 2 3.5.6.7.8 Mean shoulder shell-thickness (mm) CP CP7 E3 E43B E53 W5 Pigment dakness (pc) 2 2 3 4.2.4.6.8..2 Eggshell thickness difference (mm) Figure 3 Top left: in unpigmented shell, the variation (c.v.) in thickness within eggs increases as mean shell-thickness declines (ClutchID: F 5,3 ¼.27 n.s., mean S: F,3 ¼ 6.9, P <.); this was also true of the standard deviation (mean S: F,3 ¼ 7.6, P ¼.), i.e. thinner shell is more patchy. Below left : Pigment ÔspreadÕ is intimately linked with shell thickness at the shoulder (ClutchID: F 5,3 ¼ 4., P ¼.6, mean S: F,3 ¼ 9.2, P ¼.). Top right across all eggs crown/shoulder ratio declined with increasing B/L ratio (r 4 ¼ ).423, P ¼.8). Although not significant when controlling for nest effects (ClutchID: F 5,3 ¼. n.s., crown/shoulder: F,3 ¼.32 n.s.), the clutch means (shown as +) of C/S ratio and B/L ratio were correlated significantly (r 4 ¼ ).87, P ¼.28). Note that in these first three graphs, shell-thickness refers to unpigmented shell, indicating that pigment does not itself reduce shell-thickness. Below right : Pigment ÔdarknessÕ is linked with the difference in shell thickness at the spot (ClutchID: F 5,5 ¼ 5.76, P ¼.4, PD: F,5 ¼.6 P ¼.5). Different symbols (all as lower left figure) denote different nests (clutches). more, we found that this mean difference in thickness between pigmented and unpigmented shell predicted the darkness (PD) of the pigment spots, but not the spread (Fig. 3), and consequently, pigment darkness was correlated with egg size (ClutchID: F 5,3 ¼ 3.62, P ¼., ESA: F,3 ¼ 4.95 P ¼.34), although not in the subset of eggs in which the thickness of pigmented shell was measured (ClutchID: F 5,5 ¼ 4.9, P ¼.4, ESA: F,5 ¼.2 n.s.). In effect, larger eggs had a less consistent or more ÔpatchyÕ eggshell thickness, and this was reflected in darker pigmentation. We found evidence that the distribution of mass in the eggshell, as indicated by the ratio of crown to shoulder shell thickness (C:S), changed with egg shape, being closer to unity in the most spherical eggs (Fig. 3). Variation in the crown/shoulder shell-thickness ratio largely reflected changes in the shoulder shell thickness (between C:S shell and mean S: r 36 ¼ ).552, P <.) rather than the crown shell thickness (between C:S shell and mean C: r 36 ¼.27 n.s.). For unpigmented eggshell, the variation (both SD and c.v.) in shell thickness within eggs declined with increasing mean (unpigmented) shell-thickness (Fig. 3), indicating that as it became thinner, the shell thickness became more patchy. Pigment ÔspreadÕ (but not ÔdarknessÕ) was strongly correlated with the unpigmented-shell thickness at the egg s shoulder (ClutchID: F 5,3 ¼ 4., P ¼.6, mean S: F,3 ¼ 9.2, P ¼., Fig. 3), less so with the crown (ClutchID: F 5,3 ¼.76, n.s., mean C: F,3 ¼ 7.36, P ¼.), but not at its waist (ClutchID: F 5,3 ¼ 2.76, P ¼.36, mean W: F,3 ¼ 3.7 n.s.) or foot (ClutchID: F 5,3 ¼ 3.7, P ¼.23, mean F: F,3 ¼ 3.85, n.s.). So pigment ÔspreadÕ was linked to shell thickness especially in the region of the ÔcoronaÕ ring at the shoulder and crown. Pigment also became more concentrated towards the crown (low pc2) as the ratio between crown and shoulder shellthickness increased (ClutchID: F 5,3 ¼ 3.8, P ¼.8, crown/shoulder: F,3 ¼ 5.3, P ¼.28), indicating that the specific pattern of localized shell-thinning varied with egg shape (see above and Table ).

A. G. Gosler et al. Table Minimum adequate model (MAM) significant results from REML models derived from a step-down model simplification procedure (see Materials and methods for full initial models, which included ClutchID and year as random effects) Dependent variable SC LS LS 2 CS B/L H FA ALT EAM EAM (n ¼ 85) 42.69 (<.) 2.75 (<.) 5.83 (.6) 4.39 (<.) ESA (n ¼ 9) )5.98 (.4) 5.5 (.9) )7.36 (.) B/L (n ¼ 9) ).7 (.) 7.42 (.6) )3.94 (.47) 2.3 (<.) 4.25 (<.) PD (n ¼ 69) )38.7 (<.) 86.89 (<.) 6.34 (.2) 7.35 (.7) PD* (n ¼ 9) (ex FA) )2.4 (<.) 97.44 (<.) 8.45 (<.) PS (n ¼ 9) 4.39 (.36) 3.6 (.27) 4.33 (.38) 5.82 (.6) The dependent variable is shown left, fixed effects along the top row, and Wald statistic/d.f. with P-value in brackets within the cells. The direction of the effect is indicated by the ÔsignÕ of the Wald statistic. D.f. is in all cases except H, where it is 3. Variables are: SC, soil calcium; LS, egg no. in laying sequence; LS 2, egg no. 2 ; CS, clutch-size; B/L, egg breadth/length; H, habitat; FA, female age; ALT, nestbox altitude; EAM, eggshell ash mass/surface area; ESA, egg surface area; PD, pigment darkness; PS, pigment spread. *The effect for PD was reassessed without female age (ex FA in table) which was only known for a subset of clutches: the result was found to be robust. Sources of variation in eggshell characteristics The results of REML analyses are presented in Table, and represented in part in Figs 4 and 5. These analyses effectively identified both within- and between-clutch effects on shell characteristics, including pigmentation. Eggshell ash-mass (both the original mass and scaled by egg-size as surface area: EAM) was strongly influenced by soil calcium (Fig. 4), habitat and position in the laying sequence, and showed a strong quadratic relationship with the latter (Fig. 5). This unexpected relationship, with ashmass initially increasing and then decreasing through the clutch, was robust between clutches and years. None of these factors (SC, H, LS) influenced egg-size (surface area), which instead declined with CS and nestbox altitude, and increased with egg shape (B/L ratio), with larger eggs being more spherical. Egg-shape itself showed a quadratic relationship with laying sequence, with eggs becoming more spherical towards the middle of the clutch, mirroring the change in eggshell mass with laying sequence. It was also influenced strongly by soil calcium, CS and habitat. Pigment darkness increased strongly with declining soil calcium, and with increasing position in the laying sequence; it was influenced by habitat, and younger females laid significantly darker eggs. Finally, pigment spread was influenced by eggshell ash-mass and egg-shape, but also by habitat and altitude. We comment on these specific results in the discussion, but would note in summary here that, consistent with the structural-function hypothesis, we found strong effects of SC (and other environmental variables) on protoporphyrin pigmentation, eggshell-ash mass and egg-shape, and that egg shape and size, eggshell thickness and ash mass also affected pigmentation. DISCUSSION From the Ôstructural-function hypothesisõ we predicted relationships between pigmentation, eggshell structure, egg shape and calcium availability that would not be predicted by a signalling function. Our predictions were well supported by data. Within eggs, thinner shell was marked by the addition of pigment, and the darkness of pigment indicated the degree of thinning. Consistent with the superficial structure of small eggs (Mikhailov 997) we found that eggshell thickness became ÔpatchyÕ as it decreased (Fig. 3); and consistent with the limiting nature of calcium as a resource for breeding passerines (Graveland et al. 994; Perrins 996; Graveland & Berends 997; Graveland & Drent 997), we found that calcium availability predicted both eggshell thickness and pigmentation between clutches. The association of pigment with thinner-shelled areas, and of pigment darkness with the degree of thinning, strongly

Why are birds eggs speckled?. E49A E28 E8.9 E29A E44 W6 4 Figure 4 The relationship between eggshell mass per unit surface area (g mm )2 ) and soil calcium in 85 great tit eggs collected in 22 and 23, and reduced to ash in a Carbolite furnace at 8 C for 7 h, so vaporizing any organic matter (e.g. membrane) adhering to the shells. Although ashing may not have removed all traces of pigment (by mass), as darker (more heavily pigmented) shells occurred in lower-calcium areas (see text), they cannot account for the strong positive relationship that was found here (see text). Ash mass (g mm 2 ).8.7.6.5 2 2.5 3 3.5 4 4.5 Log max calc (mg g soil) W5 W2 2 W2 W 5 CP3 E62E E62 E92 E7 CP2 CP4 CP7 E4C E43D Eggshell ash mass (g mm 2 ).84.82.8.78.76.74.72 Pigment pc 2.5.5.5.5.7 2 3 4 5 6 7 8 9 2 2 2 3 4 5 6 7 8 9 2 Egg number Egg number.8.5.79 Figure 5 Variation in eggshell characteristics with laying sequence. Top left: eggshell ashmass/esa (g mm )2 ); below left: egg breadth/ length ratio; top right: pigment darkness (pc); below right: pigment spread (pc2). Trend lines (calculated as regressions through the means) are shown as indicators where laying sequence, and its squared term, was a significant effect (see Table ). Egg B/L raito.78.77.76.75.74.73 2 3 4 5 6 7 8 9 2 Egg number Pigment pc 2.5.5.5 2 3 4 5 6 7 8 9 2 Egg number supports the view that protoporphyrins form an integral part of the shell matrix, as suggested by Solomon (987, 997), perhaps sharing the same protein carrier as calcium ions to cross the cell membranes of the shell gland (Solomon 987). Our data suggest that at the cellular level protoporphyrin is transported when calcium is scarce.

2 A. G. Gosler et al. The pattern of change in eggshell mass through the laying sequence was unexpected, but the robustness of this result between clutches and years gives confidence in the finding. We suggest two non-exclusive hypotheses: (a) varying resource allocation through the clutch by the female; (b) foraging-mediated calcium acquisition. Under hypothesis ÔaÕ the female would allocate calcium differentially through the clutch, perhaps as an adaptive response to varying egg requirements. The finding that eggs were smaller in larger clutches may indicate that some resource allocation occurs. Under hypothesis ÔbÕ females switch to foraging for calcium sources around the start of egg-laying, and improve in their ability to find such foods until either local resource depletion or a systematic change in female prey-selection starts to reduce the availability of calcium-sources. Although more research is required to resolve this, whatever the explanation, superimposed on it is a local environmental effect of calcium availability, since eggshells were generally thicker on higher-calcium soils (Table, Fig. 4). We also found a correlation between (soil) calcium availability and the shape of eggs, being more spherical on lower-calcium soils. This is interesting because a sphere (maximum B/L) is both the strongest shape for an egg, and the most conservative in the use of shell materials in that it minimizes the surface area: volume ratio. Changes in egg-shape were reflected in the relative thickness of the shell and thus in pigmentation. Calcium availability was not the only environmental influence on egg characteristics, since both altitude and habitat showed independent effects. Although altitude might indicate a temperature effect, it is more likely that both these variables act via their effects on food availability (e.g. snails) to the female. The fact that the linear increase through the clutch in pigment darkness, which indicates eggshell thinning, was not predicted by observed changes in eggshell mass, egg size or shape, implies that yet another, unmeasured, determinant such as a physiological programme operates through the clutch. Irrespective of pigmentation, eggshells were thinner around the shoulder of the egg. That this effect correlated with egg shape suggests that it may be a design constraint since it occurred in all eggs. However, it may also be adaptive since it is in this region that hatching (pipping) occurs. In this case, the pigment ÔcoronaÕ that typically occurs in this region might serve opposing functions depending on the source of pressure, since the very ÔlubricatingÕ properties of protoporphyrin that strengthen the shell under compression from without, should also weaken it under tension from within. The three properties of protoporphyrins: their abundance, infra-red reflectance, and as a semi-crystalline lubricant, would appear well-suited as a substitute for calcium to strengthen (and reduce the porosity of) thinner areas of shell. Thus the visible ÔpatternÕ of spots is best considered as a map of the deficiencies and porosity of the eggshell, and the intraclutch variation in pigmentation reflects changes in eggshell characteristics through the clutch. Whilst it is implausible that such a system might have evolved only in P. major, or indeed only in the paridae and their close relatives, we must nevertheless exercise caution in assuming that the system suggested here offers the only possible explanation of maculation in the eggs of birds, or indeed small passerines. Further research is needed to assess the generality of our findings. Nevertheless, some speculation may be offered here on the evolution of such a system. Although most nonpasserines lay unpigmented eggs, in a systematic survey of avian eggshell pigments, Kennedy & Vevers (976) noted the presence of protoporphyrins in trace amount in every non-passerine case that they examined. Thus it appears that small passerines may have adapted a pre-existing condition in evolving the system suggested by our observations. The fact that non-passerines themselves did not do this may be a matter of scale. Although there are some notable exceptions (e.g. trochilidae), on average, small passerines, which must seek calcium daily for eggshell formation (Graveland & Berends 997), are smaller than non-passerines, which can draw more extensively on skeletal sources of calcium for eggshell formation (Sugiyama & Kusuhara 2). The earlier failure to identify the purpose of these spots resulted from the assumption of a visual-signal function (Newton 896; Lack 968). However, it should be noted that although interspecific signalling could not account for the pigment patterns seen on these eggs, since speckling conveys information on the shell s structure, it has potential as an intraspecific signal of eggshell quality. The new paradigm described here has many implications for both pure and applied research. For example, it offers the possibility of a visual, non-destructive, bio-assay of eggshell thickness, which is known to be sensitive to diet and environmental pollutants (Furness 993; Scharlemann 2). Finally, we should state that however compelling a case is made by our data for the structural function hypothesis, it remains a correlational study. Further insight into the mechanism of eggshell pigmentation must be gained through, for example, the experimental manipulation of calcium availability to the birds. ACKNOWLEDGEMENTS This work was carried out under English Nature licences to AGG: 22: 22828 and 2258; 23: 23589, 2384 and 2349; 24: 242296. We are very grateful to Steve Wyatt, Earth Sciences, University of Oxford, who ashed the collected eggs, and Jörn Scharlemann, Rhys Green and Mark Adams (Natural History Museum, Tring) who lent us the eggshell micrometre. Robin McCleery made the CFI calcium data available to us. We are especially grateful to Ben Sheldon, Dany Garant, Tom Pizzari and three

Why are birds eggs speckled? 3 anonymous referees for commenting wisely on earlier drafts, and Teddy Wilkin for the territory-size data. J.P.H. was supported by a BBSRC studentship, AGG by the Edward Grey Institute, University of Oxford, and SJR by the School of Biosciences, University of Birmingham. REFERENCES Ar, A., Paganelli, C.V., Reeves, R.B., Greene, D.G. & Rahn, H. (974). The avian egg: water vapour conductance, shell thickness, and functional pore area. Condor, 76, 53 58. Ar, A., Rahn, H. & Paganelli, C.V. (979). The avian egg: mass and strength. Condor, 8, 32 327. Bakken, G.S., Vanderbilt, V.C., Buttemer, W.A. & Dawson, W.R. (978). Avian eggs: thermoregulatory value of very high nearinfrared reflectance. Science, 2, 32 323. Blanco, G. & Bertellotti, M. (22). Differential predation by mammals and birds: implications for egg-colour polymorphism in a nomadic breeding seabird. Biol. J. Linn. Soc., 75, 37 46. Burley, R.W. & Vadhera, D.V. (989). The Avian Egg. New York: John Wiley. Davies, N.B. & Brooke, M.D. (989a). An experimental-study of co-evolution between the cuckoo, Cuculus canorus, and its hosts.. Host egg discrimination. J. Anim. Ecol., 58, 27 224. Davies, N.B. & Brooke, M.D. (989b). An experimental-study of co-evolution between the cuckoo, Cuculus canorus, and its hosts. 2. Host egg markings, chick discrimination and general discussion. J. Anim. Ecol., 58, 225 236. Dawkins, H.C. & Field, D.R.B. (978). C.F.I. Occasional Paper No.. A Long-Term Surveillance System for British Woodland Vegetation. Commonwealth Forestry Institute, University of Oxford, Oxford. Farmer, A.M. (995). Soil chemistry change in a lowland English deciduous woodland. Water, Air and Soil Pollution, 85, 677 682. Furness, R.W. (993). Birds as monitors of pollutants. In Birds as Monitors of Environmental Change (eds Furness, R.W. & Greenwood, J.J.D.). London, Chapman & Hall, pp. 86 43. Gosler, A.G. (99). The birds of Wytham an historical survey. Fritillary,, 29 74. Gosler, A.G. (993). The Great Tit. Hamlyn Species Guides. Paul Hamlyn, London. Gosler, A.G., Barnett, P.R. & Reynolds, S.J. (2). Inheritance and variation in eggshell patterning in the great tit Parus major. Proc. R. Soc. Lond. Ser. B-Biol. Sci., 267, 2469 2473. Graveland, J. & Berends, A.E. (997). Timing of the calcium intake and effect of calcium deficiency on behaviour and egg laying in captive great tits, Parus major. Physiol. Zool., 7, 74 84. Graveland, J. & Drent, R.H. (997). Calcium availability limits breeding success of passerines on poor soils. J. Anim. Ecol., 66, 279 288. Graveland, J., van der Wal, R., van Balen, J.H. & van Noordwijk, A.J. (994). Poor reproduction in forest passerines from decline of snail abundance on acidified soils. Nature, 368, 446 448. Hansen, B.T. & Slagsvold, T. (24). Early learning affects social dominance: interspecifically cross-fostered tits become subdominant. Behav. Ecol., 5, 262 268. Hoyt, D.F. (979). Practical methods of estimating volume and fresh weight of bird eggs. Auk, 96, 73 77. Jubb, M. (25). Snail ecology in relation to geology and other environmental variables in Wytham Wood. B.A. (Hons.) Dissertation, Oxford University, Oxford. Kennedy, G.Y. & Vevers, H.G. (976). A survey of avian eggshell pigments. Comp. Biochem. Physiol., 55B, 7 23. Lack, D. (968). Ecological Adaptations for Breeding in Birds. Methuen, London. Mikhailov, K.E. (997). Avian Eggshells: An Atlas of Scanning Electron Micrographs. BOC Occasional Publications: 3 British OrnithologistsÕ Club, Tring, UK. Moreno, J. & Osorno, J.L. (23). ÔAvian egg colour and sexual selection: does eggshell pigmentation reflect female condition and genetic quality?õ. Ecol. Lett., 6, 83 86. Moreno, J., Osorno, J.L., Morales, J., Merino, S. & Tomás, G. (24). Egg colouration and male parental effort in the pied flycatcher Ficedula hypoleuca. J. Avian Biol., 35, 3 34. Newton, A.V. (896). A Dictionary of Birds. A & C Black, London. Perrins, C.M. (996). Eggs, egg formation and the timing of breeding. Ibis, 38, 2 5. Pettifor, R.A., Perrins, C.M. & McCleery, R.H. (988). Individual optimization of clutch size in great tits. Nature, 326, 6 62. Rahn, H. & Ar, A. (974). The avian egg: incubation time and water loss. Condor, 76, 47 52. Sanchez, J.M., Corbacho, C., del Viejo, A.M. & Parejo, D. (24). Colony-site tenacity and egg color crypsis in the Gull-billed Tern. Waterbirds, 27, 2 3. Scharlemann, J.P.W. (2). Eggshell thickness of some birds is declining. Nature First, 24, 6. Sibley, C.G. & Monroe, B.L. Jr (99). Distribution and Taxonomy of the Birds of the World. New Haven, CT: Yale University Press. Solomon, S.E. (987). Egg shell pigmentation. In Egg Quality Current Problems and Recent Advances (eds R.G. Wells & C.G. Belyarin). Butterworths, London, pp. 47 57. Solomon, S.E. (997). Egg and Eggshell Quality. Iowa State University Press, Ames, IA. Sugiyama, T. & S. Kusuhara (2). Avian calcium metabolism and bone function. Asian Australas. J. Anim. Sci., 4, 82 9. Tyler, C. (969). Avian egg shells: their structure and characteristics. In: International Review of General and Experimental Zoology (eds Felts, W.J.L. & Harrison, R.J.). New York, Academic Press. Verbeek, N.A.M. (99). Differential predation on eggs in clutches of Northwestern Crows the importance of egg color. Condor, 92, 695 7. Weidinger, K. (2). Does egg colour affect predation rate on open passerine nests?. Behav. Ecol. Sociobiol., 49, 456 464. Whitbread, A. & Kirby, K. (992). Summary of National Vegetation Classification Woodland Descriptions (UK nature conservation, No. 4) NCC, Peterborough, UK. Editor, Marcel M. Lambrechts Manuscript received 28 April 25 First decision made 6 June 25 Manuscript accepted July 25