## Introduction

From a genetic point of view, grafting creates a chimera of a single individual made up of genetic material issued from two (or more) distinct genomes, with each genome maintaining its own distinct genetic identity throughout the life of the grafted organism. Typically, grafting associates two tissues from different individuals, the recipient and graft, which form vascular connections and survive as a genetic chimera due to a unique symbiotic relationship1. The partners in a graft are known as the scion and root in plants and the donor and recipient in pearl oysters. The use of grafting technology in both plant cultivation and pearl production provides a means of improving quality in these industries. Grafting is a well-established practice that has many horticultural and biological uses. The numerous applications of plant grafting include vegetative propagation, avoidance of juvenility, cultivar change, creation of unusual growth forms, repair, size control, biotic and abiotic stress resistance and physiological improvement1. Some studies have examined the contribution of the scion; for example, a scion of a red-flowering rose grafted on a white rose stock will continue to produce red roses1. Grafting is commonly employed in agronomy to indirectly manipulate scion phenotype2. As for the host organism, rootstocks are commonly used to propagate selected scions to improve fruit tree tolerance to environmental stress, and to control tree size3, 4. Roots anchor plants in the ground to provide water and nutrients from the soil and can serve as storage organs. The root system is a crucial component in coordinating response to a range of abiotic and biotic stressors5, 6. The influence of rootstocks on yield productivity is being increasingly recognized to be as important as grafted scions. In the plant system, the genetic characteristics of the root appear to drive vegetative growth and the overall growth of the grafted system depending on environmental conditions7.

In the pearl oyster grafting process (Pinctada sp.), similarities to the scion–root system were suggested by studies on the contributions of the recipient and donor to the expression of the overall phenotype (including cultured pearl quality traits). In Pinctada maxima, results indicated low donor-related heritability of pearl size8, and xenografts between P. maxima and P. margaritifera demonstrated that individual donors significantly influence the color and the surface complexion of the pearl9. In P. margaritifera, one study demonstrated the differential influence of individual donors with black or red outer shell phenotypes, combined with green or yellow inner shell phenotypes, on pearl darkness level, color categories and luster10. Another study examined the phenotypic correlation between recipient shell weight and size of pearls in P. fucata, revealing a limited positive relationship whereby recipient oysters with heavier shells produced bigger pearls11. These results indicate potential underlying genetic correlations between the recipient or donor growth and pearl size, whereby selection for faster growing oysters might further improve pearl size.

In order to understand the rootstock–scion interaction, some reviews have investigated physiological and biochemical aspects12, or hormonal signaling13. In pearl oysters, earlier studies have examined the influence of genetics, environmental factors and their interactions on pearl quantitative and qualitative traits14. A study in P. maxima indicated that characteristics of both donors and hosts were correlated with pearl traits15. In P. margaritifera, one such study demonstrated the influence of grow-out sites on cultured pearl quality traits over a broad geographic scale between archipelagos in the South Pacific16. Another analyzed environmental influence on pearl size parameters in relation to the recipient oyster biometric parameters17. Although it is clear from these studies that both donor and recipient oysters are involved in pearl formation in P. margaritifera, the relative contributions have not yet been clearly identified. Despite much study on pearl oysters, there remains a knowledge gap about donor size influence on pearl phenotype.

In the present study, the goal was to improve our understanding of the mechanisms driving the phenotypic variability arising from the recipient and the donor. The working hypothesis was that the recipient oyster confers quantitative traits to the pearl while the individual donor oyster drives qualitative traits of the pearl. In other words, everything related to growth and transfer of energy would be driven by the recipient oyster, and everything related to the quality of the nacrein deposit would be driven by the donor of the graft. Within this framework we defined four specific experiments to determine respective donor and recipient effects. (1) We sought to discern whether the donor or recipient was the main driver of pearl growth rate by comparing different recipient-donor phenotype combinations. (2) We evaluated drivers of the pearl phenotype by comparing effects of different recipient–donor phenotype combinations on nacre deposit quality. (3) We measured biomineralisation potential through a comparison of quantitative gene expression in the mantle graft (donor only effect) and the pearl sac (donor + recipient effect). For this approach, a representative panel of eight genes encoding proteins involved in aragonite biomineralisation (Pif-177, MSI60 and Perline), calcite biomineralisation (Aspein, Shematrin and Prismalin) or proteins implicated in both layers (Nacrein) was screened18,19,20. Finally (4), we followed the kinetics of the recipient–donor influence on pearl formation by monitoring pearl size and quality parameters over 12 months. This information provides basic knowledge to help us to understand the donor-recipient correlations and their relative contributions in pearl oyster grafts.

## Results

Overall, a total of 289 cultured pearls were harvested and analyzed. Donor pearl oysters were divided into two distinct groups according to their shell size (dorso-ventral measurement): small donor oysters, denoted “QL”, and large ones, denoted “TL”. The distribution of QL and TL oysters are shown in Table 1. Results are presented in four sections relating to 1) the influence of donor and 2) recipient shell size effect on cultured pearl size and cultured pearl quality, 3) biomineralisation capabilities and 4) pearl development kinetics.

### QL vs. TL and individual donor influence on pearl phenotype

At grafting time, the donor oyster width, height and thickness were recorded, showing average measurements of 48.6 ± 5.2 mm, 48.6 ± 5.2 mm and 12.7 ± 1.4 mm, respectively, for the QL donor group. The TL donor group was significantly larger for each parameter (p < 0.0001) with 79.8 ± 6.3 mm for shell height, 79.9 ± 6.9 mm for shell width and 21.0 ± 2 mm for shell thickness.

Comparing differences in the nacre produced between the QL and TL donor groups, we found that the average nacre weight and thickness of the QL donor group were 0.38 ± 0.22 g and 0.79 ± 0.04 mm, respectively (N = 142). In the TL donor group, the average nacre weight and thickness were 0.34 ± 0.20 g and 0.70 ± 0.04 mm (N = 147). No significant effect of donor oyster shell size phenotype was detected for either nacre weight or thickness (p = 0.058 and p = 0.065, respectively) (Fig. 1).

When effects of QL/TL donor phenotype were compared for subsequent pearl quality, however, no significant effect of donor oyster size group was observed for any pearl qualitative traits (darkness, luster, surface defects, grade, shape or presence of circle/s) except for color categories (Fig. 2), where the QL phenotype seems to give more pearls of “other” colors and less peacock pearls than the TL donor phenotype.

A significant individual donor effect was detected for nacre thickness (p = 0.045), but no significant individual donor effect was shown for nacre weight (p = 0.059). Pearls from donor “Q21” (N = 7 pearls) showed the greatest average nacre thickness (1.16 ± 0.50 mm) and nacre weight (0.566 ± 0.298 g) compared with donor “Q17”, which represented the minimum (N = 7 pearls) (0.45 ± 0.32 mm and 0.215 ± 0.141 g, respectively). The influence of donor size and the evolution of this relationship through time on qualitative pearl traits at harvest are illustrated in Figure 2.

A significant individual donor effect was detected for darkness level (p = 0.002), color category (p < 0.0001), cultured pearl luster (matte or shiny/glossy/highly glossy) (p = 0.0002) and the surface defect classes (p < 0.0001). However, no significant difference was recorded between the individual donors for the absence/presence of circles (p = 0.218) or for the three shape categories (p = 0.156).

### Recipient influence on pearl phenotype

Nacre thickness and weight were both significantly correlated with recipient oyster shell biometric parameters (weight, height, thickness and width), Spearman rho coefficients ranged from 0.51 to 0.65 and 0.49 to 0.62, respectively. The highest significant positive correlations were found for nacre thickness and pearl weight with recipient pearl oyster weight (Rho = 0.65 and Rho = 0.62, respectively, p < 0.0001). In addition, there were significant positive correlations of nacre thickness and weight with shell width (+0.51 and +0.49, respectively, p < 0.0001) (Fig. 3).

### Molecular phenotype

The relative expression levels of the eight biomineralisation genes analyzed in the donor oyster mantle graft tissue and pearl sacs from the QL and TL groups are shown in Fig. 4a. The comparison between the two groups of donors for the three aragonite-related genes, Pif-177, MSI60 and Perl1, showed significant higher gene expression levels in the TL group (p < 0.0001) in the graft tissue compared with the QL group. Shem5, Shem 9 and Calc1 also showed significant higher gene expression levels in the TL group (p < 0.0001) compared with the QL group.

All biometric parameters of the donor oysters were significantly correlated with the relative expression level of seven biomineralisation genes in the graft (p < 0.0001); only Prism was not correlated. Nacre thickness and pearl weight were not correlated with the relative expression level in the graft of eight biomineralisation genes (Fig. 3). Furthermore, the recipients shell height was correlated with the relative expression level of Perl1 in the pearl sac (Rho = 0.236, p = 0.0004 for shell height) (Fig. 3).

Among the eight candidate genes studied in the pearl sac at the three harvest times, the expressions of six were significantly affected by donor phenotype (Fig. 4b). MSI60 and Calc1 gene expression levels were significantly higher for “TL” phenotypes compared with “QL” phenotypes (p = 0.009 and p = 0.014 respectively). Prism, Asp, Shem5 and Shem9 showed significantly greater expression in “QL” phenotypes compared with “TL” ones (p = 0.043, p = 0.049, p = 0.004 and p = 0.009, respectively).

No significant differences in relative expression of the panel of genes in the graft were found for pearl quality traits. In the pearl sac, however, significant effects were observed for the majority of the panel of genes, depending on grade, level of surface defects, luster and color (Supplementary Table S1).

### Pearl development kinetics

Analyzing kinetics data over the 12 months of the study, we observed, as expected, a very highly significant difference in nacre thickness between harvest times (p < 0.0001). Mean nacre thicknesses at harvest times T1, T2, T3 and T4 were 0.02 ± 0.01, 0.08 ± 0.03, 0.09 ± 0.03 and 0.11 ± 0.03 cm, respectively. Nacre was significantly thinner at T1 than at T2, T3 and T4. T3 and T4 were not significantly different from one another. Pearl nacre weight was also highly significantly different between harvest times (p < 0.0001). Mean pearl weights at harvest times T1, T2, T3 and T4 were 0.12 ± 0.07, 0.35 ± 0.14, 0.44 ± 0.17 and 0.54 ± 0.20 g, respectively. T1 pearls were significantly lighter in weight than T2, T3 and T4 ones. As with nacre thickness, T3 and T4 were not significantly different from one another (Fig. 1). Measurements on recipient oysters showed highly significant differences between the four harvest times (p < 0.0001), as shown in Fig. 5. Mean height (dorso-ventral measurement ± SE) for recipient shell at T1, T2, T3 and T4 harvest times were 86.38 ± 5.79, 91.42 ± 7.38, 96.15 ± 7.27 and 102.07 ± 7.61 mm, respectively. Mean weight for recipient shell at T1, T2, T3 and T4 harvest times were 84.99 ± 19.34, 110.38 ± 21.61, 130.87 ± 24.75 and 144.57 ± 26.14 g, respectively.

Comparison between harvest times showed no significant difference between the proportions of the three shape categories (p = 0.215). After 3 months of culture, the pearl shapes were distributed as follows: Baroque 29.6%, N = 21; Round 52%, N = 37; Oval 18%, N = 13. After 12 months of culture, the distribution had remained similar: Baroque 38%, N = 26; Round 31%, N = 21; Oval 31%, N = 21.

Figure 2  shows the ratios of pearls with and without circles at each harvest time. A highly significant effect was recorded for the presence or absence of circles (p = 0.005) between each of the harvest times. The T3 and T4 harvest times yielded about three times more circled pearls than T1.

Data analysis showed a highly significant difference in pearl luster and grade between harvest times (p < 0.0001). T1 pearls were significantly more “matte” and less “glossy” than T2, T3 and T4 pearls. T1 had significantly less grade B pearls (12.7%) and more Reject pearls (43.7%) than the other harvest times T2 (34.2% and 13.9%, respectively), T3 (19.7% and 11.3%) and T4 (38.2% and 2.9%).

A highly significant difference in the darkness of pearls was observed between the different harvest times (p = 0.0004). T1 harvest time had significantly more lighter-toned pearls (32.4%) than the later harvest times T2 (7.6%), T3 (18.3%) and T4 (12.7%). T3 and T4 had significantly more dark pearls (21.1% and 17.7%, respectively) than T1 (7%) and T2 (7.6%) (Figure 2).

For cultured pearl surface defects, no significant harvest time effect was detected (p = 0.067) (Figure 2).

## Discussion

Our approach, based on studying the influence of donor and recipient phenotypes on resulting pearl and molecular phenotypes, has yielded important results for pearl oyster breeding strategy. Overall, the results show that it is mainly the recipients that influence the size of the pearl, while the individual donors influence pearl quality traits (except for pearl shape and the presence/absence of circles, which seem to be more influenced by graft surgical procedure).

In the present study, we showed that pearl size phenotype (nacre weight and thickness) was highly correlated with the biometric parameters of the recipient oyster, while those of the donor oyster (QL/TL groupings) had no impact. This positive correlation confirms results observed in P. fucata 11 and in P. maxima 15. Furthermore we observe a significant positive correlation between the expression of the aragonite gene in the pearl sac (Perl1) and one recipient oyster biometric parameter (shell height). This result suggests that the recipient oyster affects the activities of the biomineralising tissues, which are mainly dependent on calcium metabolism in the epithelial tissue of the pearl sac21. The recipient oyster regulates the metabolism of the pearl sac by supplying nutrients during the period of culture and thus regulates the expression of the biomineralisation genes, especially those implicated in the aragonite layers22. High biomineralisation capabilities may have contributed to a greater nacre deposition, as already observed in P. maxima 23,24,25 and P. fucata 26. The recipient oyster can be compared to the rootstock of a grafted plant, with recipient oyster and rootstock each corresponding to a grafted organism. In the plant system, rootstocks are the driving component of production systems and studies have reported that they can influence tree growth, yield and fruit size, weight, and rind thickness27,28,29,30,31,32,33. However, it is not fully understood how the rootstock controls scion growth. The grafted organism seems to have a key role in growth owing to its nutritive function. Interestingly, the same principle applies in the animal system we studied. The positive relationship between pearl size (nacre weight and thickness) and weight of the recipient is valuable information because it can serve as a basis for selective breeding strategy for fast growing recipient pearl oysters. For mollusk species, between 10% and 20% improvement in selected growth traits can be obtained per generation34, 35.

A strong individual donor effect was observed on pearl quality traits, such as color (darkness and visual color categories) and quality parameters (luster, surface defects and grade). Pearl quality traits are influenced by a number of highly variable factors, including genetics, the environment and their interaction, as well as surgical technique and graft hygiene36. A recent study in P. maxima found significant individual donor effects for color, nacre deposition, luster, shape and defects; however, each of these traits was grouped into broad categories in this previous study, making specific comparisons between the two studies difficult15. In plant models, research on rootstock–scion interaction demonstrated that yield and quality are conferred by the scion cultivar. Scion quality clearly affects final yield and fruit quality in grafted plants, but rootstock effects can also impact these characteristics4, 37.

Expression analysis in the pearl oyster graft, based on a panel of genes encoding proteins implicated in the shell biomineralisation process, indicated significant differences of gene expression level between the QL and TL donor groups for the three genes of the panel implicated in aragonite formation (Pif-177, MSI60, and Perl1). The protein genes Pif-177 and MSI60 regulate growth, nucleation and the organization of the aragonite crystal20, 22, 38. Perline is the protein equivalent of the N14 protein identified in P. maxima 39 and seems to be specifically involved in the formation of the nacreous layer and promotion of aragonite crystal nucleation23. Aragonite-related gene expression levels were significantly higher in the graft tissue from the TL group than from the QL group. Some other genes (Shematrin 5 and 9 and Nacrein) showed significantly higher expression levels in the graft originating from the TL group. Shematrin 5 and 9 are implicated in the calcite mineralisation process, in which they facilitate the formation of calcite prisms40. These results seem to show the opposite trend to the pearl phenotype observations, since the TL and QL groups did not differ in nacre weight and thickness. The significantly higher expression levels of some genes in the TL group correspond to the biometric parameters of the donor oyster but not to the quantitative traits of the pearl. A study in P. fucata examined the differences of gene expression in large and small oysters. RNA-seq analysis revealed that the number of genes up-regulated in small oysters was greater compared to large oysters and could be explained by catch-up growth41. In the pearl sac analyses, we observed a relationship between relative gene expression in the pearl sac and grade, surface defects, luster and pearl color. The role of the pearl sac in nacreous layer biomineralisation is thought to mirror that of the oyster mantle. The gene-expression patterns of shell matrix proteins in pearl sacs were measured relative to the biomineralisation process of the nacreous layer42. The genome of the individual donor oyster and its influence on the pearl biomineralisation process remain active in the pearl sac tissue during the entire pearl culture period right up until harvest time9, 43,44,45, confirming the influence of individual donors on quality pearl traits.

Our examination of pearl formation over time (12 months of rearing) clearly showed that nacreous deposition is not linear. By the half-way point of the experiment (6 months), 69.7% of the final nacre thickness and 65% of the nacre weight had already been reached. At this point, nacre deposition slowed down. No previous study has examined the deposition of the nacreous layer on the pearl over culture time. However, some studies did examine the beginning stage (35 days) of nacreous layer deposition during pearl formation in P. fucata 42, 46 and in P. margaritifera 47. These found that the first CaCO3 deposited on the nucleus was the aragonitic47 or calcitic46 prismatic layer found in the cross section of the pearl, and that the nacreous layer then formed on top of this prismatic layer. The non-linear nacreous deposit seems to be influenced by the growth of the recipient oyster (whose biometric parameters are correlated with nacre weight and thickness). Shell growth is influenced by two major factors: microalgal concentration and temperature, both of which regulate the expression of most of the shell matrix protein genes implicated in the biomineralisation process22. However, recipient growth had a linear pattern over the 12 months of culture. Thus, the growth of the recipient seems to be an important driving factor of pearl growth but not the only influence implicated. The high influence of the recipient and the weak influence of the donor on pearl size at harvest suggest that their respective roles are not totally revealed here.

In our experiment, we also recorded that luster, grade and darkness of pearls improved with duration of culture, but we also noted an increase in the proportion of circled pearls. According to Wada48, who focused on surface color and pigmentation, the quality of nacreous pearls is determined by the ratio of the thickness of the lower prismatic layer to that of the upper nacreous layer. It was therefore logical that we found an improvement of quality grade over time, because nacre deposition increased with time of culture49. Individual donor oysters influence quality traits such as color, luster, amount of surface defects and grade but do not affect shape or the presence of circles. The differences observed between quality traits underline the complexity of their genetic basis and that of the donor/recipient interaction.

The present study is the first to identify the relative impact of donor/recipient oyster size on pearl size and quality traits during culture time in the marine bivalve P. margaritifera. It was demonstrated that 1) pearl size characters (nacre weight and thickness) are influenced by the recipient oyster phenotype and 2) the major qualitative traits (color, darkness, luster, and incidence of surface defects) are influenced by individual donor oysters. The recipient oyster shell parameters are correlated with pearl size, with the largest recipient giving the largest pearls. Similarly to other examples in the living world, the recipient organism seems to have a more important role than expected. Donor/recipient combinations should therefore be selected carefully. For the pearl industry, the contribution of the recipient found here suggests it would be beneficial to develop breeding programs on growth of recipient oysters in order to obtain larger and thus more valuable pearls. In parallel, a breeding program could be developed for quality and color of donor oysters in order to obtain high quality pearls in desired colors. Furthermore breeding program could be based on pedigree. In fact, genealogy information is essential for genetic traceability. The importance of maintaining pedigree records could aim to increase effective population sizes and minimise inbreeding to ensure long-term genetic gain, viability of aquaculture breeding programs and productivity by the development for example of a set of multiplex PCRs as in C. gigas oyster50, 51 or in the abalone Haliotis midae 52. The present work could serve as a basis for future studies exploring the donor/recipient relationship in greater depth.

## Materials and Methods

### Animals

Wild Pinctada margaritifera were collected as spat using commercial collectors in the lagoon of Takaroa atoll (Tuamutu archipelago, French Polynesia) in March 2013. Nine months later (December 2013), 20 of these collectors were transferred to Ifremer facilities in Vairao on Tahiti Island (Society Archipelago, French Polynesia). Pearl oysters (N = 1535) were then removed from the collectors and divided into two distinct groups according to their shell size (dorso-ventral measurement): small “QL”, (N = 393; 2.5 cm in means) and large “TL” (N = 292; 4.7 cm in means). The distribution and phenotypes of QL and TL oysters are shown in Table 1 and Fig. 6. These oysters were then tied onto a CTN (Cord Technical Nakasai) rearing system53, where they were left to grow, protected using plastic mesh to prevent predation, until they reached a sufficient size to start the grafting experiment. These oysters were then transferred to Regahiga Pearl Farm (Mangareva island, Gambier archipelago, French Polynesia) to be used as donors. Prior to nucleus implantation and graft surgery, oysters from the QL and TL groups were collected from the rearing station, detached, washed with a high pressure spray (to remove epibiota) and stored until the grafting operation.

### Grafting procedure

All grafts were performed under standard production conditions by a single expert at the Regahiga Pearl Farm so as to minimize the grafter effect on pearl quality traits described in ref. 54. The recipient oysters were issued from local natural spat collection in the Mangareva lagoon. A total of 40 donors (20 of the TL phenotype and 20 of the QL phenotype) were used to perform 600 grafts (20 grafts per donor for TL and 10 grafts per donor for QL) over a 2-day period, using 1.8 BU nuclei (5.45 mm diameter; Imai Seikaku Co. Ltd., Japan) (Fig. 6). During the grafting process, 3 to 5 graft tissue pieces of each donor oyster were sampled, preserved in RNAlater® and stored at −80 °C for subsequent RNA extraction. Recipient oysters were examined 45 days post grafting to estimate nucleus retention and mortality rates, as described in ref. 55.

### Measurements of shell biometric parameters and pearl growth rate

Prior to the grafting operation, shell height, width and thickness of the 40 donor oysters were measured using Vernier calipers. All recipient oysters were washed with a high pressure spray to remove epibionts and the following biometric measurements were taken: shell height, width, thickness, and total weight (soft tissue parts + shells) (to 0.01 g) (Fig. 5)17.

Once harvested, cultured pearls were cleaned by ultrasonication in soapy water with a LEO 801 laboratory cleaner (2-L capacity, 80 W, 46 kHz); they were then rinsed in distilled water. Some keshi (small non-nucleated nacre deposits) were also harvested but these were not graded in the present study. The size of the cultured pearls was assessed by measuring nacre thickness and weight49.

### Cultured pearl quality parameter measurement

Surface defects, luster and grade category of the cultured pearls were evaluated. Cultured pearl shape was characterized in two ways: the presence/absence of circle/s and the shape category (“b” for baroque and semi baroque, “o” for oval and drop, “r” for round and semi round pearls). Two kinds of color evaluation were made on the cultured pearls: the darkness of color and the visually-perceived color category (peacock, green, grey, other). Cultured pearl grade was determined for each sample according to the official Tahitian classification (Journal Officiel 2001 n° 30, 26 July 2001) from the most to least valuable quality: A, B, C, D and Rejects (rebuts). Briefly, the five grades are mostly based on surface purity and luster, from A (cultured pearls showing no surface defects or small defects confined to less than 10% of their surface and having very good luster;) to D (cultured pearls showing many highly visible defects over more than two thirds of their surface and having poor luster) and Rejects (cultured pearls that have too many defects to be graded). The latter are consequently discarded and ultimately destroyed. Finally, surface defects and luster (components of cultured pearl grade) were determined separately so that they could be studied independently. Quality traits were evaluated as described in ref. 56. To ensure homogeneity in parameter assessment, all evaluations were made visually (without a jeweler’s loupe) by the same operators.

### Biomineralisation gene expression analysis

Oyster shell and cultured pearls are formed by biomineralisation activities in two distinct tissues: the mantle of the recipient and the pearl sac in the gonad, formed from mantle tissue from the donor57. Around the nucleus, a pearl sac is formed by proliferation of the outer mantle epithelial cells of the mantle graft which secretes successive nacre layers on the nucleus58, 59. Pearl sac consists of mucous cells containing large acidophilic granules and epidermal cells42 that secrete proteins resulting into culture pearl formation, a highly controlled biomineralisation process similar to the development of inner shell regulated by the mantle60. The description of pearl sac collection is explained in the next section. Similarly to other bivalves, the shell of pearl oyster P. margaritifera consists of two polymorphs of calcium carbonate: the inner nacreous layer, which is composed of aragonite, and the outer prismatic layer, which is made of calcite61,62,63. Shell formation is a highly controlled process involving multiple matrix proteins18, 38, 64. Recently, the number of genes identified as coding for molluscan shell matrix components has increased65,66,67,68,69. In order to identify the respective contributions of donor vs. recipient, we sampled tissues of the graft during the graft operation (i.e., donor contribution) and tissues of the pearl sac during harvest (i.e., recipient and donor contributions) to compare relative gene expression through screening aragonite-related genes (Pif-177, MSI60, Perline), calcite-related genes (Aspein, Shematrin, Prismalin) and one gene implicated in both layers (Nacrein) (Table 2). Two genes were used as housekeeping genes chosen based on their ubiquitous and constitutive expression pattern in P. margaritifera tissue: SAGE (SAGES: AGCCTAGTGTGGGGGTTGG/SAGER: ACAGCGATGTACCCATTTCC) (called REF in ref. 22) and GAPDH (GAPDHS: AGGCTTGATGACCACTGTCC/GAPDHR: AGCCATTCCCGTCAACTTC)70.

After removing the RNAlater by pipetting and absorption, total cellular RNA was extracted from the individual graft tissue (n = 40) or pearl sac samples (n = 240), using TRIzol® reagent (Life Technologies) according to the manufacturer’s recommendations. RNA was quantified using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Inc.). Total RNA of each individual was then treated with DNAse I using a DNA-free Kit (Ambion). First, strand cDNA was synthesized from 500 ng total RNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche) and a mix of poly (dT) and random hexamer primers. Real-Time PCR amplifications were carried out on a Roche Light Cycler® 480. The amplification reaction contained 5 μL LC 480 SYBR Green I Mast (Roche), 4 μL cDNA template, and 1 μL of primer (1 µM), in a final volume of 10 μL. Each run included a positive cDNA and a blank control for each primer pair. The run protocol was as follows: initial denaturation at 95 °C for 10 min followed by 40 cycles of denaturation at 95 °C for 30 s, annealing at 60 °C for 30 s and extension at 72 °C for 60 s. Lastly, the amplicon melting temperature curve was analyzed using a melting curve program: 45–95 °C with a heating rate of 0.1 °C s−1 and continuous fluorescence measurement. All measurements were made in duplicate and all analyses were based on the Ct values of the PCR products.

Relative gene expression levels were calculated using the delta–delta method, normalized with two reference genes, to compare the relative expression results71 as follows:

$${\rm{Relative}}\,{{\rm{expression}}}_{({\rm{target}}{\rm{gene}},{\rm{sample}}{\rm{x}})}={2}^{\hat{\vphantom{a}}-({\rm{\Delta }}{\rm{Ct}}{\rm{sample}},{\rm{sample}}{\rm{x}}-\Delta {\rm{Ct}}{\rm{calibrator}},{\rm{sample}}{\rm{x}})}={2}^{-{\rm{\Delta }}{\rm{\Delta }}{\rm{Ct}}}.$$

Here, the ΔCt calibrator represents the mean of the ΔCt values obtained for the tested gene. The delta threshold cycle (ΔCt) is calculated by the difference in Ct for the target and reference genes. The relative stability of the GAPDH and SAGE combination was confirmed using NormFinder72. PCR efficiency (E) was estimated for each primer pair by determining the slopes of standard curves obtained from serial dilution analysis of a cDNA to ensure that E ranged from 90 to 110%. The primers used for amplification are listed in Table 2.

### Recipient and donor influence over kinetic evolution

The experiment was monitored over time to evaluate changes in the recipient/donor influence as the graft became established within the recipient. Two recipient oysters were harvested for each donor after 3, 6, 9, and 12 months of culture. Biometric parameters were measured as described in the section ‘Measurements of shell biometric parameters and pearl growth rate’. Pearls and pearl sac tissue were collected at the same time. Measurement of pearl growth rate and associated quality trait phenotypes were measured as described in sections ‘Measurements of shell biometric parameters and pearl growth rate’ and ‘Cultured pearl quality parameter measurement’. At the time of pearl harvest and in order to minimise the contamination of recipient tissues, we first cut the gonads from the host oysters. We then removed the gonad tissue with a surgical blade until only a thin (<0.5 mm) layer of tissue surrounding the pearls remained. At this point, only the pearl sac and the pearl remain. Next, we made an incision in the pearl sac, removed the pearl, and transferred the pearl sac into a 2.0 ml tube with RNAlater® until RNA extraction44 and placed the pearl in a numbered box. While our technique does not completely remove the potential for gonad tissue contamination within pearl sac samples, it does significantly reduce the possibility for gonad contaminations. In addition, the potential for contamination is further reduced because the gonad itself cannot mineralise and thus poses very little risk, as confirmed by Wang et al.73 who found MSI60 was not expressed within the gonads of P. fucata. A total of 80 pearl sacs were sampled every 3 months, giving a total of 240 pearl sac tissue samples over 9 months (tissues from the last point at 12 months were of too poor a quality to be used in the analyses). Biomineralisation potential for 40 grafts and 217 pearl sacs that contained pearls was screened via gene expression as described in section ‘Biomineralisation gene expression analysis’ (Fig. 6).

### Statistical analysis

The normality of the data distribution and homogeneity of variance were tested for pearl size, donor, recipient oyster biometric parameters and relative gene expression data using the Shapiro‒Wilk test and Bartlett’s test. When necessary, transformations were used to adjust data to this distribution (logarithm or square roots). Group donor size (QL vs. TL) and time harvest were treated as fixed variables for statistical analysis, with donor treated as a random variable unless otherwise stated. A linear mixed effects model was used to assess the effect of the fixed and random variables on nacre weight, thickness and relative gene expression, and a generalized linear mixed effects model (family = binomial) was used for the presence or absence of circles and luster, both using packages lme4 and nlme74 with the R© software. Qualitative classes based on cultured pearl surface defects, grade and darkness level were re-encoded into ordered categorical response variable. Scores from 0 to 4 were attributed to the different classes from the least to the most valuable. Analyses of ordered categorical response variables (pearl darkness, surface defects and grade) were analyzed using cumulative link mixed models implemented in the ordinal package75 with the R© software. The significance of fixed variables was tested by comparing the fit of an all-inclusive model with that of a reduced model for each variable using a fisher test or chi-squared test. For the cultured pearl “color categories” and shape categories, donor and time of harvest effect were compared using χ2 tests.

Spearman’s correlation coefficient was used to evaluate the correlations between nacre weight and thickness, and the four biometric measures of the recipient and donor oysters, and gene expression. We have adjusted the p-value with Bonferroni correction (level of significance = α/c, in which α = 0.05, and c = number of comparisons performed) to compare pairwise correlation in multiple comparison. The gene expression values of each graft donor and pearl sac were associated with the corresponding nacre weight and thickness of each pearl. The statistics and the visualization for the correlation matrix (Fig. 3) were done with the R© software Hmisc and Corrplot package using the spearman method76, 77.

In all cases, the differences were considered statistically significant when p values were lower than 0.05. Statistical analyses were performed using XLSTAT (version 2009.4.02) and R© software (version 3.2.1).