The genotypes and greenhouse conditions used in different experiments are summarized in Additional file 4. In all the experiments, seeds were transferred on rockwool cubes with regular supply of nutrient solution . Seedlings were transplanted on rockwool slabs at a density of 2.5 plants m-2 in a compartment of a multispan Venlo-type glasshouse or in an air conditioned glasshouse, Wageningen, The Netherlands. Supplemental lighting by high pressure sodium lamps (Philips, SON-T, 600 W) for 16 hours (from 06.00 to 22.00) provided a minimum photon flux density of 125 μmol m-2 s-1 at the crop level. The terminal flower was removed from all plants at anthesis to support vegetative growth.
Occurrence of parthenocarpy among C. annuum genotypes
C. annuum (Table, 1) genotypes were selected on the basis of their blocky appearance and seed number (Additional file 4: Exp 1). In total, 70-150 emasculations were performed in each genotype using 10 plants per genotype and fruit set was evaluated when fruits were ripe.
Genotype effect on number and weight of carpelloid structures
Five genotypes: Parco, California Wonder 100 (CW), Riesen v. Californien (RVC), Bruinsma Wonder (BW), and Line 3 were arranged in one row of 8 plants at two temperatures (20/18°C, 16/14°C D/N) (Additional file 4: Exp 2). Two treatments (induced pollination or prevent-pollination) were completely randomized within the row. Induced pollination was done by vibrating the stem two times per week. Prevent-pollination was done by applying lanolin paste on the stigma of the flowers . In genotypes Parco, RVC, CW and Line 3, flowers were given the treatments till three fruits per plant were obtained. In genotype BW, two flowers (one on main branch and one on a side branch) were treated at each of 20 nodes. Mature red fruit were harvested and their length, diameter and fruit fresh weights were recorded. Those seedless fruits that reached minimum of 50% of the weight of seeded fruit were considered as parthenocarpic and were used in our analysis while remaining were considered as knots. The number of seeds and number of carpelloid structures was counted in each fruit and each carpelloid structure was weighed.
Ovule development in C. annuum
Line 3 and BW were inbred lines with high and medium potential to set parthenocarpic fruits  while Orlando (OR) was a fourth-generation inbred line developed from Orlando-F1 (De Ruiter seeds) (Additional file 4: Exp 3). Flowers were collected at 3-4 days before balloon stage. Pericarp was removed and morphological analysis of ovule development was conducted in the laboratory by using a field-emission cryo-scanning electron microscopy (SEM) (Jeol 6300F), equipped with an Oxford CT 1500HF cyro-stage system .
Correlation of abnormal ovule development with reduced seed set and enhanced development of carpelloid structures
Two set of experiments were conducted (Additional file 4: Exp 4). In first experiment, genotypes Line 3 and OR were used to evaluate the occurrence of carpelloid structures. Flowers were tagged at 2 days before anthesis and allowed to pollinate naturally. Developing ovaries were harvested at 2 days of interval, dissected and evaluated for the presence of carpelloid structures by visual inspection. With the same set of genotypes, percentage of aberrant ovules and number of seeds was evaluated. Flowers (n = 6) were collected randomly at or around the anthesis stage. After removing the carpel, ovules were scraped smoothly in a water medium on a clean slide and the frequency of abnormal ovules was observed under optical microscope (Leitz Aristoplan). Seed set was counted when fruits reached the maturity (red) in both genotypes. In second experiment, genotypes Line 3 and BW were used. To evaluate the female fertility, plants were grown at day/night temperature of 14/16°C (low), 18.20°C (normal) and 22/24°C (high) and pollination was induced by vibrating the main stem two times per week. Number of seeds was counted when fruits reached the maturity (red) in both genotypes.
Pollen viability and germination
Pollen grains of BW were collected from normal temperature (20/20°C day/night) and low night temperature (20/10°C day/night) in morning time (9.00-10.00 PM) (Additional file 4: Exp 5). To test the pollen viability, hydrated pollens were dissolved in FDA solution  and scored under fluorescence microscope. Pollen which fluoresced brightly under fluorescence was scored as viable. For pollen germination, the hanging drop technique was employed following published procedures  with some modifications. A liquid medium containing 0.25 mM MES (pH5.9), 15% (w/v) PEG 4000, 2% (w/v) sucrose, 700 ppm Ca (NO3)2, 100 ppm H3BO3, 200 ppm MgSO4, 100 pm KNO3) was used. Germination was considered only when germinating tube was larger or equal to the size of the pollen. Viability and germination percentages were determined, using 10-12 replicates of about 20-40 selected grains.
Relation between parthenocarpy and carpelloid structures
Genotype BW with moderate potential for parthenocarpy was used in the experiment (Additional file 4: Exp 6). To obtain seeded fruits, flowers were tagged at anthesis and allowed to pollinate naturally. To obtained seedless fruit, flowers were emasculated two days before the expected date of anthesis and stigmas were cover with the lanolin paste or lanolin paste containing 0.05% 1-Naphthaleneacetic acid (NAA) or Gibberellic acid (GA3) . Fifteen plants per treatment were used. On each plant, two flowers (one on main branch and one on a side branch) were treated at each of 20 nodes. All the fruits were harvested at mature red stage and their length, diameter and fruit fresh weights were recorded. Criteria to define parthenocarpic fruit and knot were the same as mentioned earlier (Exp.2). The number of seeds and number of carpelloids structures was counted in each fruit and each carpelloids structure was weighed.
Inheritance of parthenocarpy and its relation with carpelloid structures
In order to understand the genetics of parthenocarpy and a possible association with carpelloids structures, genetics of both traits were evaluated in cross progenies of Line 3 (Additional file 4: Exp 7). Line 3 was used as a parthenocarpic parent (Pp) and Lamuyo B, ORF2#1, and Parco as non-parthenocarpic parents (Pn). F2 progenies were obtained for all three crosses, and also a backcross with Line 3 in the cross with Lamuyo B. The flowers (15-20) were emasculated prior to anthesis and tagged. All the fruits were harvested at the mature red stage. Length, diameter and fruit fresh weights were recorded for individual fruits. In each fruit, carpelloids structures were counted and mass were weighed. All three crossing population were evaluated and mono- or digenic-models were tested to understand the genetics behind parthenocarpy and carpelloids structures.
Sequence analysis of CaARF8
Young leaf material from Line 3, BW and OR (Additional file 4: Exp 8) was collected for DNA extraction. Primers for PCR amplification were designed against pepper, tomato and potato ARF8 EST sequences available from the Sol Genomics Network (SGN), http://solgenomics.net/ (Additional file 5). SEFA PCR was used to amplify non-transcribed regions. PCR products were cleaned with the Invitek MSB® Spin PCRapace. 120 ng of PCR product per reaction was sent with the appropriate sequencing primer (12 pmol) to ServiceXS, Leiden, The Netherlands. Resulting chromatograms were manually trimmed and checked for calling errors. Contigs were built by using Contig Express of the Invitrogen Vector NTI suite Version 10.4.
Experiments and their statistical treatment are listed in additional file 4. For experiment 3 and 4, one way analysis of variance (ANOVA) was used, and treatment effects were tested at 5% probability level using F-test. For experiment 5, the effect of each treatment on each genotype at each temperature was tested separately by using a one way analysis of variance (ANOVA). Mean separation was done by student's t-test. Data processing and statistical tests were carried out with SPSS 15.0. The inheritance of parthenocarpy was tested by using chi square distribution with 1 degree of freedom at the 0.05 level of significance to test the null hypothesis that parthenocarpy was controlled by a single recessive gene. Carpelloid inheritance was tested using a chi square distribution, with different mono- or digenic models.