An adaptive and joint carrier recovery scheme is proposed based on adaptive Kalman filter (AKF) for 16 quadrature amplitude modulation signals. The tuning parameter Q is adaptively adjusted according to the innovation vector under dynamic operating condition. Consequently, the proposed scheme can simultaneously estimate frequency offset (FO) and phase noise with excellent performance in estimation accuracy, estimation range and tracking capability as well as linewidth tolerance, especially in dynamic frequency offset scenarios. The proposed scheme is also demonstrated in 14GS/s dual-polarization 16 quadrature amplitude modulation experimental systems. Compared with conventional extended Kalman filter (EKF) and QPSK partitioning, the proposed scheme offers better estimation accuracy and adaptive configuration capability of the tuning parameter.