Accurate modelling of distribution transformers (TXs) is crucial to identify their operating characteristics across several power system applications. Consequently, this paper employs an improved version of the artificial ecosystem optimizer (called IAEO) in parameters estimation of distribution TXs with different four sizes (i.e. 4, 15, 112.5 and 167 kVA ratings). Comparison with well-known literature optimizers, including genetic algorithm, particle swarm optimizer, coyote optimization algorithm, artificial hummingbird optimizer, and others, validates the performance of the proposed IAEO. The IAEO demonstrates its superiority by getting the lowest possible value of the sum of absolute errors (SAEs) between measured and calculated values, which serves as the objective function (OF) to be optimized. Moreover, three additional optimizers are employed and compared to IAEO for all study cases: firefly algorithm, political optimizer, and exponential distribution optimizer. It is found that IAEO attains the minimum SAEs values of 1.12e-5 and 0.0322, outperforming the best competitors for 4 kVA and 15 kVA TXs, respectively. Furthermore, IAEO accurately captures the steady state fingerprint of all studied TXs in terms of efficiency and voltage regulation (VR). This way, the peak efficiency occurs at 36.2% loading in 112.5 kVA TX while the negative VR may reach -8% when the 167 kVA TX is loaded with its rated leading power factor. Finally, all executed optimizers are analyzed using several statistical indices, including t-test, where the proposed IAEO gets the smoothest and fastest OF minimization trend.