Is maternal thyroid hormone deposition subject to a trade-off between self and egg because of iodine? An experimental study in rock pigeon

Maternal hormones constitute a key signalling pathway for mothers to shape offspring phenotype and fitness. Thyroid hormones (THs; triiodothyronine, T3 and thyroxine, T4) are metabolic hormones known to play crucial roles in embryonic development and survival in all vertebrates. During early developmental stages, embryos exclusively rely on the exposure to maternal THs, and maternal hypothyroidism can cause severe embryonic maldevelopment. The TH molecule includes iodine, an element that cannot be synthesised by the organism. Therefore, TH production may become costly when environmental iodine availability is low. This may yield a trade-off for breeding females between allocating the hormones to self or to their eggs, potentially to the extent that it even influences the number of laid eggs. In this study, we investigated whether low dietary iodine may limit TH production and transfer to the eggs in a captive population of Rock pigeons (Columba livia). We provided breeding females with an iodine-restricted (I- diet) or iodine-supplemented diet (I+ diet) and measured the resulting circulating and yolk iodine and TH concentrations and the number of eggs laid. Our iodine-restricted diet successfully decreased both circulating and yolk iodine concentrations compared to the supplemented diet, but not circulating or yolk THs. This indicates that mothers may not be able to independently regulate hormone exposure for self and their embryos. However, egg production was clearly reduced in the I- group, with fewer females laying eggs. This result shows that restricted availability of iodine does induce a cost in terms of egg production. Whether females reduced egg production to preserve THs for themselves or to prevent embryos from exposure to low iodine and/or THs is as yet unclear.


Introduction 24
Non-genetic inheritance is defined as the transmission of information between generations 25 beyond coding genes (Danchin et al., 2011). Parental effects are included in this non-genetic 26 inheritance, may be considered adaptive (Moore et al., 2019;Mousseau and Fox, 1998;Yin et 27 al., 2019), and parental effect of maternal origin, i.e. maternal effects, have received 28 increasing attention since the 1990's (Bernardo, 1996;Mousseau and Fox, 1998). Hormones 29 of maternal origin can be transferred to the offspring and constitute a potential pathway for 30 mothers to influence their offspring's phenotype (Groothuis et al., 2019). Hormone allocation 31 to offspring could be costly for mothers as it could induce a trade-off between allocating 32 hormones to their own metabolism or to their offspring's. Steroid hormones, the most studied 33 hormones in the context of hormone-mediated maternal effects, may not be that costly to 34 produce as they are derived from cholesterol, which is abundant in the organism (Groothuis 35 and von Engelhardt, 2005). On the other hand, thyroid hormones (THs) may be considered 36 costly, as their molecular structure includes iodine, a trace element that cannot be synthesised 37 by organisms and must therefore be found in the environment. 38 Thyroid hormones are metabolic hormones that are present in two main forms: 39 thyroxine (T 4 ) that contains four atoms of iodine, and triiodothyronine (T 3 ) that contains three 40 atoms of iodine. Iodine is concentrated into the thyroid gland and incorporated into tyrosines 41 that will be combined to form T 4 and T 3 (McNabb and Darras, 2015). The thyroid gland 42 produces mostly T 4 and lesser amounts of T 3 . In the peripheral tissues (e.g. liver, kidney, 43 muscle), T 3 is mostly obtained from T 4 via removal of an iodine atom by deiodinase enzymes 44 (McNabb and Darras, 2015). TH action mainly depends on TH receptors that have a greater 45 affinity for T 3 than for T 4 (Zoeller et al., 2007). This is why T 4 is mostly seen as a precursor of 46 T 3 , the biologically active form. THs play important roles in growth, reproduction, 47

Experimental design 123
Iodine restricted and supplemented diet 124 We provided the experimental birds with either an iodine restricted (I-, n = 19 pairs) or an 125 iodine supplemented (I+, n = 19 pairs) diet until all eggs and blood samples were collected. 126 Egg collection was ended around three weeks after the initiation of second clutches, leading to 127 a total of approx. 10 weeks (see below for more details). The restricted diet contained 0.06 mg 128 of iodine/kg of food (Altromin™ C1042) and the supplemented diet was the same food 129 supplemented with 3 mg of iodine/kg by the manufacturer. Therefore, both diets had exactly 130 the same composition of all essential micro-and macronutrients except iodine. The restricted 131 treatment corresponds to about 10% of the iodine content in the standard pigeon diet (0.65 132 mg/kg), and approximately 20% of the minimum iodine requirement for ring doves (0.30 mg 133 I/kg) according to Spear and Moon (1985). In addition, this restricted treatment corresponds 134 to a low iodine treatment (0.05 mg/kg) used in a previous experiment on Japanese quails 135 (McNabb et al., 1985a) that induced a significant decrease in circulating and yolk iodine. The 136 supplemented treatment (3 mg/kg) corresponds to five times the concentration of iodine in the 137 standard pigeon diet. In McNabb and colleagues (1985a), the maximal dietary iodine (1.2 mg 138 I/kg feed) was ca. eight times the sufficient iodine concentration required for Japanese quails 139 (0.15 mg I/ kg feed) and the authors observed no detrimental effects of this high dose. 140 Therefore, we expected no detrimental effect of our supplemented treatment either. Food, 141 water, and grit were provided ad libitum throughout the experiment. We collected both eggs 142 from first and second clutches of females fed with restricted or supplemented iodine diets. We 143 also collected blood samples after clutch completion from the two experimental groups (I-and 144 I+). The second set of samples (i.e. eggs and blood) was collected to test for the effect of 145 exposure duration of the treatment (see timeline below).

Timeline of the experiment 147
Nest boxes were opened and nesting material was provided two weeks after the experimental 148 diet was introduced to stimulate egg laying. Egg laying usually starts within a week of nest-149 box opening. Based on Newcomer (1978), who fed hatchling chicken with low iodine diet 150 (0.07 mg I/kg feed), we could expect thyroid iodine content to be the lowest from 10 days 151 onwards after introducing the experimental diet. The first eggs (i.e. from the first 152 experimental clutches) were collected 3 weeks after the introduction of the experimental diet 153 and were collected over 12 days. On average, the eggs from 1 st clutches were laid 26.4 days 154 (SD = 2.9) days after the onset of the experimental diet. Freshly laid eggs were collected, 155 replaced by dummy eggs to avoid nest desertion. Second clutches were initiated by removing 156 dummy eggs approximately 2 weeks after the completion of the first clutch (i.e., 5 weeks after 157 the start of the experimental diet), and eggs collected over a period of 18 days. On average, 158 the eggs from 2 nd clutches were laid 53.5 days (SD = 3.3) after the onset of the experimental 159 diet. We also collected some late first clutches (on average 52.6 (SD = 6.5) days after the 160 onset of the experimental diet). 161 Egg and blood sample collection 162 Table 1 summarises the number of samples collected. Eggs and blood samples were collected 163 in the exact same manner in the first and second clutches. Freshly laid eggs were collected 164 and were stored in a -20°C freezer. Not all females laid complete clutches of 2 eggs, and 165 several females did not lay an egg at all. Females were captured during incubation in the nest 166 boxes, and blood samples (ca. 400 µl) were taken from the brachial vein after clutch 167 completion (average (SD) = 4 (4.7) days after clutch completion, range = 0-21 days). 168 Unfortunately, we could not blood sample the females that did not lay eggs, as this would 169 have caused serious disturbance to all the birds in the same aviaries as we had to catch them 170 by hand netting in the large aviary. Half of the blood sample (ca. 200 µl) was taken with heparinised capillaries for plasma extraction (for TH analyses) and stored on ice until 172 centrifugation. The other half of the sample was taken with a sterile 1 ml syringe (BD 173 Plastipak ™) and let to coagulate for 30 min at room temperature before centrifugation for 174 serum extraction (for iodine analyses). Previous studies measured iodine in serum samples 175 (McNabb et al., 1985a;McNichols and McNabb, 1987), therefore we decided to measure 176 iodine in the serum for comparable results. Whole blood samples were centrifuged at 3 500 177 RPM (ca. 1164 G-force) for 5 min to separate the plasma from red blood cells (RBCs), and at 178 5 000 RPM (ca. 2376 G-force) for 6 min to separate the serum from RBCs. After separation, 179 all samples (plasma, serum and RBCs) were stored in a -80°C freezer for analyses of THs and 180 iodine. 181

Hormone and iodine analyses in plasma and yolk samples 182
Eggs were thawed, yolks separated, homogenised in MilliQ water (1:1) and a small sample 183 (ca. 50 mg) was used for TH analysis. Yolk and plasma THs were analysed using nano-LC-184 MS/MS, following . TH concentrations, corrected for extraction 185 efficiency, are expressed as pg/mg yolk or pg/ml plasma. 186 Yolk and serum iodine (ICP-MS, LOD of 3 ng/g of yolk and 1.5 ng/ml of serum) 187 analyses were conducted by Vitas Analytical Services (Oslo, Norway). Yolk iodine was 188 measured in a sample of ca. 1 g of yolk, and serum iodine was measured in a sample of ca. 0.2 189 ml of serum. 190

General information 192
Data were analysed with the software R version 4.0.2 (R Core Team, 2020). To test for the 193 effect of iodine restriction on egg laying, we compared the number of females that laid first 194 clutches in both groups, and the total number of eggs laid in first clutches with two Pearson's 195 chi-squared tests. The rest were fitted with linear mixed models (LMMs) using the R package lme4 (Bates et al., 2015) and p-values of the predictors were calculated by model comparison 197 using Kenward-Roger approximation with the package pbkrtest (Halekoh and Højsgaard, 198 2014). The response variables were plasma THs concentrations (T 3 , T 4 ), serum iodine, and 199 concentrations of yolk THs and yolk iodine. Relevant interactions between predictors were 200 tested by adding them one-by-one in the models with main effects only. Post-hoc tests of 201 interactions were performed with the package phia (de Rosario-Martinez, 2015). Model 202 residuals were inspected for normality and homogeneity with the package DHARMa with 203 1,000 simulations (Hartig, 2020). When either of the assumptions was violated, the response 204 was ln-transformed (see Tables) and in these cases the model residuals showed the required 205 distributions. Estimated marginal means (EMMs) were calculated from the models using the 206 package emmeans (Lenth, 2019). When the response was transformed, the EMMs were 207 calculated on the back-transformed data. 208 Although the treatment started for all females on the same date, each female, at the 209 time of egg laying, was exposed to the experimental diet for different durations because of the 210 varying laying dates between females. This may influence the effects of iodine manipulation 211 on circulating and yolk iodine and THs. However, because the second clutches were laid after 212 a longer exposure to the treatments than were first clutches (average exposure duration (SD) 213 2 nd clutches = 53.5 (3.3) days; 1 st clutches = 30.9 (10.6) days), clutch order and exposure 214 duration (i.e. the number of days between the onset of the experimental diet and laying date) 215 were confounded. Therefore, we used two different models, one with exposure duration and 216 another one with clutch order. We controlled for egg order (i.e. first or second egg in a clutch) 217 in our models for yolk THs initially since a previous study in the rock pigeons showed a non-218 significant trend for higher yolk T 3 concentrations in the second eggs (Hsu et al., 2016). 219 However, we detected no such effect in our models (all F < 0.57, all p-values > 0.45) and thus 220 egg order was excluded from the final models.

Model specification 222
Circulating iodine (ln-transformed) and T 3 and T 4 concentrations were analysed by fitting an 223 LMM with treatment (I-or I+), exposure duration, completeness of a clutch as a categorical 224 variable (complete or incomplete, i.e., to further test for the effect of number of eggs laid), 225 and the two-way interactions between treatment and exposure duration or completeness as 226 fixed factors. Female identity (for iodine and T 4 , but not T 3 because of singularity: variance 227 estimate collapsed to 0) and hormone extraction batch (for T 3 and T 4 ) were added as random 228 intercepts. 229 Yolk components (iodine, T 3 , T 4 ) were analysed in two different sets of models. The 230 first set of models, for yolk iodine only, compared untreated eggs (collected before the start of 231 the treatment) to the eggs collected in the two treatments (I+ and I-). This way, we could test 232 whether yolk iodine differed between untreated and experimental eggs. Here we only used 233 experimental first clutches as only one clutch of eggs per untreated female was collected. The 234 model only included treatment (a three-level categorical predictor: untreated, I+ and I-) as the 235 predictor. 236 The second set of models tested the effect of iodine treatment, exposure duration to the 237 treatment or clutch order on yolk iodine and THs. Here we included both 1 st and 2 nd clutches 238 from iodine treatments, but no eggs from untreated females. Yolk iodine (ln-transformed) was 239 first analysed in a LMM that included treatment as a categorical variable, exposure duration 240 (days since the start of the experiment), completeness of a clutch (complete or incomplete) 241 and the two-way interactions between treatment and exposure duration or completeness, and 242 female identity as a random intercept. This LMM somewhat violated the assumption of 243 homogeneity of variances between the groups because of the larger variance in yolk iodine in 244 the I+ group. Nevertheless, such a violation should not undermine our results as a recent 245 paper demonstrated that LMMs are fairly robust against violations of distributional 246 assumptions (Schielzeth et al., 2020). Yolk iodine was also analysed in a second model that 247 included treatment, clutch order (1 st or 2 nd clutch, categorical variable) and their interaction as 248 the predictors. Yolk T 3 , T 4 (ln-transformed) were analysed using the same models (with 249 exposure duration or clutch order) as for yolk iodine. Hormone extraction batch was added as 250 a random intercept for yolk T 3 and T 4 . 251

Circulating iodine and TH concentrations 253
In line with our expectations, there was a clear effect of iodine treatment on circulating iodine 254 concentrations: serum iodine was about four times lower in the I-group than in the I+ group 255 (raw data average (SE), I-= 11.1 (1.2) ng/ml serum, I+ = 44.0 (4.7) ng/ml serum; Table 2, 256 Fig. 1). The effects of clutch completeness, exposure duration and their interactions with 257 treatment on serum iodine were statistically unclear (Table 2). 258 Plasma T 3 was not affected by supplementation or restriction of iodine, nor by 259 exposure duration (Table 2; Fig. 2A). There was an almost statistically significant interaction 260 between iodine treatment and exposure duration of the treatment on plasma T 4 ( Table 2) Fig. 2B). Yet, the 264 large confidence intervals warrant due caution in interpreting this trend. There were no clear 265 effects of clutch completeness and its interaction with treatment on plasma THs (Table 2). 266

Discussion 295
In this study we tested whether dietary iodine limits mothers' circulating TH concentration, 296 TH transfer to the yolk, and egg production. To our knowledge, our study is the first to 297 investigate the potential trade-off between circulating and yolk THs induced by low dietary 298 iodine. We found that fewer females laid first clutches under the iodine-restricted diet 299 compared to the females under the iodine-supplemented diet, resulting in a lower total number 300 of eggs laid. We found that the iodine restricted diet decreased circulating and yolk iodine 301 levels, though circulating and yolk THs were unaffected. Longer exposure to restricted iodine 302 had no clear effect on circulating or yolk iodine and THs. Finally, we observed a slight 303 increase in plasma T 4 and in yolk T 3 across time that was unrelated to the dietary iodine and is 304 likely explained by seasonal changes or clutch order effects. Yet, because exposure duration to 305 the treatment and clutch order are partly confounded, our experimental design does not allow 306 us to fully disentangle both variables. 307

Does restricted iodine induce a cost and a trade-off between circulating and yolk THs? 308
Our iodine-restricted diet successfully decreased circulating iodine concentrations compared 309 with the supplemented diet. Despite this effect, we observed no differences in circulating TH 310 concentrations. This is consistent with a previous study that showed that Japanese quails 311 under limiting iodine availability maintained normal circulating THs concentrations (McNabb 312 et al., 1985a). However, a similar study on ring doves found a decrease in circulating T 4 313 concentrations with no changes in circulating T 3 , suggesting increased peripheral conversion 314 from T 4 to T 3 to maintain normal T 3 levels (McNichols and McNabb, 1987). The causes for 315 discrepancies between our study and that of the dove study (McNichols and McNabb, 1987) 316 are not clear. One potential explanation is that, in our study, we could only sample the females 317 that laid eggs and thus apparently managed to maintain normal circulating THs despite 318 restricted iodine whereas those that did no lay eggs may have suffered from low circulating 319 TH concentrations. 320 In the yolk, like in the circulation, restricted dietary iodine decreased yolk iodine but 321 not yolk TH concentration, in contrast to our predictions. The result of decreased yolk iodine 322 is in line with a previous study on quails, which found that mothers fed with low dietary 323 iodine produced eggs with low iodine but did not report yolk TH concentrations (McNabb et 324 al., 1985a). Low egg iodine concentration in turn disturbs thyroid function in embryos and 325 hatchlings (McNabb et al., 1985b;Stallard and McNabb, 1990). Circulating TH 326 concentrations of embryos, however, were not affected by low egg iodine (McNabb et al., 327 1985b;Stallard and McNabb, 1990). 328 Contrary to previous studies that manipulated dietary iodine, we found that limited 329 iodine availability hampered egg production, with 40% fewer females producing eggs in the I-330 group compared to the I+ group. However, females that managed to maintain normal 331 circulating THs were also able to lay eggs with normal yolk THs, similar to the study by 332 McNabb and colleagues (1985a). At the moment it is unclear why some females were affected 333 and others not, but a potential explanation may be for example individual differences in the 334 ability to store iodine. Two other studies found that administration of methimazole, a TH-335 production inhibitor, ceased egg laying in Japanese quails (Wilson and McNabb, 1997), and 336 reduced egg production in chickens (Van Herck et al., 2013). These results suggest that our 337 restricted diet might have induced hypothyroidism in some females, thus preventing them 338 from laying eggs. 339 Therefore, we did not show evidence for a cost of restricted iodine in females that 340 managed to lay eggs. Those females did not appear to face any trade-off between allocating 341 iodine and THs to either self or their eggs. Yet, 40% of the females under the restricted diet 342 paid a cost in terms of egg production. Whether this effect is due to limited production of THs is as yet unclear, but these females may have faced a trade-off between maintaining normal 344 circulating THs or yolk TH deposition. 345

Is there a regulatory mechanism to cope with the cost of restricted iodine and its associated 346
trade-off? 347 The absence of an effect of restricted iodine on circulating and yolk THs suggest that mothers 348 are not able to regulate yolk TH deposition independently from their own circulating THs, as 349 recently proposed by Sarraude et al. (2020b). This is contradictory to previous studies 350 showing evidence of independent regulation (Van Herck et al., 2013;Wilson and McNabb, 351 1997). However, the latter two studies induced supraphysiological hypo-or hyperthyroidism 352 to the birds, which may explain such discrepancies. 353 Interestingly, we found that our restricted diet reduced egg production. As discussed 354 above, this may be due to a hypothyroid condition that prevented females from laying eggs. 355 This may have evolved to protect embryos from exposure to too low iodine and/or TH 356 concentrations. In breeding hens, restricted dietary iodine can decrease egg hatchability 357 (Rogler et al., 1959;Rogler et al., 1961b) and retard embryonic development (McNabb et al., 358 1985b;Rogler et al., 1959;Rogler et al., 1961b), and can induce thyroid gland hypertrophy in 359 embryos and hatchlings (Rogler et al., 1961a;McNabb et al., 1985b; but see Stallard & 360 McNabb, 1990).This may suggest that females may be able to regulate egg production when 361 iodine availability is low. Overall, our results suggest that mothers in the restricted group 362 appear to prioritise self-maintenance and offspring quality over offspring quantity. 363

Restricted iodine and trade-offs in wild populations 364
Our low-iodine diet (0.06 mg I/kg food) is comparable to what birds may sometimes 365 experience in the wild. Although relevant data are scarce, estimates of iodine content in food 366 items such as barley and maize grains, wheat, or rye is highly variable, ranging from 0.06 to 367 0.4 mg I/ kg (Anke, 2004). Insectivorous species may also encounter iodine deficiency as the 368 iodine content in insects vary from <0.10 up to 0.30 mg I/kg (Anke, 2004). As such low 369 iodine availability can also be found in the wild, it is therefore relevant to study whether 370 mothers may face trade-offs in iodine or TH allocation during the breeding season, or whether 371 it influences egg laying itself. Yet, our study did not show evidence for the existence of a 372 trade-off between circulating and yolk THs when environmental iodine is limited. 373 Nevertheless, mothers may face a trade-off between allocating resources to themselves or 374 producing eggs of sufficient quality. 375 376 In conclusion, we found that restricted dietary iodine did not decrease circulating or 377 yolk THs despite reduced circulating and yolk iodine. Nevertheless, we found evidence that 378 restricted availability of iodine induces a cost on egg production. Thus, mothers may not be 379 able to regulate yolk TH transfer, but may be able to regulate egg production when facing 380 limited iodine. Our results also indicate that females under limited iodine availability may 381 prioritise their own metabolism over reproduction, or avoid exposing their offspring to 382 detrimentally low iodine and/or THs. These explanations serve as interesting hypotheses for 383 future research to further explore the consequences of limited iodine in wild populations. 6 females could not be captured and sampled (1 in I-and 5 in I+ group). LMM on plasma T 4 included batch of hormone extraction and female identity as random intercepts. LMM on plasma T 3 only included batch as the random intercept. LMM on serum iodine only included female identity as the random intercept. P-values and estimates (SE) for main effects were calculated from a model without the interaction. Interactions were added one by one and tested by comparison with the model without any interactions. Yolk iodine and T 4 were ln-transformed to achieve homogeneity of the model residuals. Ndf = 1. See Table 1 for sample sizes. Two different LMMs (separated by the dashed line) were computed, one with exposure duration, and the other one with clutch order. The coefficient for treatment is only presented in the first model as there were no differences with the second model. LMMs on yolk THs included batch and female identity as random intercepts. LMM on yolk iodine only included female identity as the random intercept. P-values and estimates (SE) for main effects were calculated from a model without the interaction. Interactions were added one by one and tested by comparison with the model without any interactions. Yolk iodine and T 4 were lntransformed to achieve homogeneity of the model residuals. Ndf = 1. See Table 1 for sample sizes. Figure 1: Circulating iodine in rock pigeon females treated with an I-or I+ diet. Black lines and shadow areas represent average and s.e.m values within each group, and grey dashed lines connect blood samples from the same females. Some females were only captured once, hence not all dots are connected. See Table 1 for sample sizes.

Figures
Figure 2: Circulating T 3 (A) and T 4 (B) in rock pigeon females treated with an I-or I+ diet. Red dots refer to blood samples after collected the first clutches and blue dots refer to blood samples collected after the second clutches. Black lines and shadow areas represent average and s.e.m values within each group, and grey dashed lines connect blood samples from the same females. Some females were only captured once, hence not all dots are connected. See Table 1 for sample sizes.  Table 1 for sample sizes. Figure 4: Yolk iodine in eggs from 1 st and 2 nd clutches laid by rock pigeon females treated with an I-or I+ diet. Eggs from the same female and the same clutch were averaged. Red dots refer to eggs from the first clutches and blue dots refer to eggs from the second clutches. Black lines and shadow areas represent average and s.e.m values within each group, and grey dashed lines connect eggs from the same females. Some females did not lay two clutches, hence not all dots are connected. See Table 1 for sample sizes.  Table 1 for sample sizes.

List of symbols and abbreviations
I: iodine