Flavonoid content is one of the critical quality indicators for peanut seed. Near-infrared spectroscopy (NIR) is an effective method for rapid detection of flavonoid content in peanut. However, the differences of testa color may affect the accuracy of the detection results. Therefore, the construction of NIR prediction models for peanuts with red and black testa can provide a guarantee for efficient and rapid detection of flavonoid content in special peanut kernels.

In this study, 232 peanut germplasms with different testa colors were selected as materials, including 108 peanut with red testa and 124 peanut with black testa. The total flavonoid content was determined by aluminum chloride chromogenic method, with rutin serving as the standard (RT: rutin). Using the Swedish Broadcom DA7250 Diode Array Analyzer for spectral acquisition, within a scanning spectral range of 950-1 650 nm. Employing the Unscrambler X10.4 modeling software, various calibration models were established through both single and composite processing, utilizing diverse derivative and scattering spectral preprocessing methods, based on full-band partial least squares (PLS) modeling. By comparing the correlation coefficients and errors among different models, the optimal processing method was selected to establish a prediction model for flavonoid content in both red and black peanut kernels. For model external validation, materials were derived from a recombinant inbred line population derived from the parents of Silihong and Jinonghei 3, with 30 lines with red testa and 30 lines with black testa each undergoing external cross-validation.

The results showed that flavonoid content of peanut with red testa was between 60.33-122.49 mg RT/100 g, with an average of 94.34 mg RT/100 g. The flavonoid content of peanut with black testa was between 64.98-121.55 mg RT/100 g, with an average of 95.59 mg RT/100 g. The best spectral pretreatment method of the peanut with red testa prediction model was “Derivative Savitzky-Golay+ SNV+Detrend”, yielding a correction correlation coefficient (Rc) of 0.9022, a root means square error of cross validation (RMSECV) of 1.9101, a prediction correlation coefficient (Rp) of 0.9021, and a root mean square error of prediction (RMSEP) of 1.9606 mg RT/100 g. The external validation correlation coefficient (R²) was 0.923, with a prediction model deviation range of -4.86-8.47 mg/100 g. The best spectral pre-treatment method for the peanut with black testa prediction model was “Derivative Savitzky-Golay+SNV+Deresolve”, resulting in an Rc of 0.9521, an RMSECV of 1.6978, the correlation coefficient (Rp) of the peanut with black testa prediction model was 0.915, and RMSEP of 2.292 mg RT/100 g, the correlation coefficient R² of external verification was 0.907, with a prediction model deviation range of -4.56-2.87 mg/100 g. Cross-validation was carried out with non-corresponding color models, and the correlation coefficient was between 0.0015-0.0975.

The testa color strongly affected the accuracy of detection, and the near-infrared prediction models constructed in this study are suitable for the detection of flavonoid content in peanuts with red and black testa, which provide an important selection method for breeding characteristic peanuts with high flavonoids.

Peanut (Arachis hypogaea L.) is one of the important vegetable oil and cash crop. High yield is always the predominant objective in peanut breeding and determined by seed number per unit area and seed weight. Seed number per unit area is produced by planting density per unit area×number of pods per plant×number of seeds per pod. Therefore, the genetic dissection of the number of seeds per pod is helpful to explore the gene/locus related to this trait, which provides an important theoretical basis for the molecular breeding of yield in peanut.

A RIL population, derived from Silihong×Jinonghei 3, were planted at Qingyuan experimental station of Hebei Agricultural University in Baoding city, Hebei province in 2018(E1) and 2020(E2). Phenotypic values of traits associated with the number of one seed pods per plant, two seeds pods per plant and multiple pods per plant were investigated at harvest stage. By using the genetic linkage map constructed by laboratory of Peanut innovation team, Hebei Agricultural University and software of QTL Icimapping V4.1(the Inclusive composite interval Mapping (ICIM)), QTL mapping for the number of seeds per pod was carried out under two environments.

The results showed that the rates of one seed pods per plant and two seeds pods per plant were normal distribution, while the rate of multiple pods per plant was skewed normal distribution. A total of 11 QTLs were detected for the three traits, which could explain the phenotypic variation of 4.66%-22.34% and the additive effects of -9.35-9.42. Among of them, 5 QTLs for the rate of multiple pods per plant with explained 3.19% to 22.34% of phenotypic variation were obtained. The additive effect of one QTL from Jinonghei 3 was negative (-4.77), while the additive effect of the other four QTLs from Silihong was positive (3.59-9.42). Two QTLs for the rate of one seed pods per plant were mapped with explained 4.97%-6.43% of phenotypic variation. The additive effects of the two QTLs from Jinonghei 3 were negative (-4.45 and -4.54). Four QTLs for the rate of two seed pods per plant were located with explained 4.97%-6.43% of phenotypic variation. The additive effects of the four QTLs from Jinonghei 3 were negative (-9.35--3.84). Among of these QTLs, 6 QTLs were major QTLs, of which qRMSPA05 was repeatedly detected, and the heritable phenotypic variation was 16.58%-17.34%, and the additive effect was 7.69-8.12.

Six major QTLs and one major stable QTL for multiple pods per plant were identified, which will be helpful for improving the yield traits in peanut. The results can be used as important candidate segments for genetic improvement, and molecular marker assisted selection and fine mapping.