<p>Real-life optimization problems, present in practically every domain, are usually large-scale and complex. Hence, they are solved using metaheuristic algorithms, suited for large and complex problems. The advancements in most areas have resulted in newer and more complex optimization problems. Hence, the development of novel metaheuristics has surged. In this paper, a systematic review has been conducted on the most recently introduced population-based metaheuristics. In-depth analysis of the algorithms has been conducted by considering their inspiration type, strengths, learning mechanisms, performance, and application domains. Their performance on several benchmark functions has been thoroughly analysed and compared with each other, and some statistical analyses in the form of Friedman and Wilcoxon tests have been conducted. Moreover, detailed discussions have been provided on their novelties, strengths, performance, application domains, and challenges. Based on the review, the most prominent type of recent algorithms is swarm intelligence, making up 48% of the algorithms reviewed. We have identified their notable strengths, such as having effective ways of balancing the exploration and exploitation phases, amongst others. Based on the performance analysis, we have deduced that most of the recent metaheuristics outperform previous ones, and have identified the highest-performing ones in solving benchmark functions. In terms of application domains, 63% of the applications have been to benchmark functions and structural engineering applications. Besides that, only four main application domains, including machine learning, power systems, wireless networks, and path planning, have been considered. Moreover, we have identified several of their challenges, including having difficulty in balancing the exploration and exploitation phases, parameter sensitivity, implementation complexity, and risk of local optima.</p>