To improve energy efficiency, total network scalability and data aggregation in Wireless Sensor Networks (WSNs), sensor nodes are often grouped into disjoint and mostly non-overlapping subsets called clusters. To provide an energy-efficient system by clustering, there are three main challenges. First is to find the optimum number of nodes in a specific cluster, second is to find the optimum number of clusters in the network and the third is to find the optimum position of Cluster Head (CH) in a specific cluster. Selecting an optimum number of clusters in WSNs provide greater improvement in terms of system scalability, energy efficiency, collision reduction, network lifetime, latency, and efficient routing backbone in the network. Selection of optimal number of clusters in WSNs is affected by level at which WSNs is modeled viz. Radio Energy Model Level, Network Model Level and Clustering Level. The objective of this paper is to present a state-of-the-art survey of distinct analytical methods used to calculate the optimum number of clusters, and its time-line comparative analysis based on network type, mathematical formula for an optimal number of clusters, base station positioning, energy model, strengths, weaknesses and applications of WSNs. We have also discussed the impact of different parameters on selecting the optimal number of clusters in WSNs.