The description of a new skull (D4500) from the Dmanisi site (Republic of Georgia) has reopened the debate about the morphological variability within the genus Homo. The new skull fits with a mandible (D2600) often referred as 'big' or 'enigmatic' because of its differences with the other Dmanisi mandibles (D211 and D2735). In this report we present a comparative study of the variability of the Dmanisi mandibles under a different perspective, as we focus in morphological aspects related to growth and development. We have followed the notion of modularity and phenotypic integration in order to understand the architectural differences observed within the sample. Our study reveals remarkable shape differences between D2600 and the other two mandibles, that are established early in the ontogeny (during childhood or even before) and that do not depend on size or sexual dimorphism. In addition, D2600 exhibits a mosaic of primitive and derived features regarding the Homo clade, which is absent in D211 and D2735. This mosaic expression is related to the location of the features and can be explained under the concept of modularity. Our study would support the possibility of two different paleodemes represented at the Dmanisi site. This hypothesis has been previously rejected on the basis that all the individuals were constrained in the same stratigraphic and taphonomic settings. However, our revision of the complex Dmanisi stratigraphy suggests that the accumulation could cover an undetermined period of time. Even if "short" in geological terms, the hominin accumulation was not necessarily synchronic. In the same line we discard that the differences between D2600 and the small mandibles are consequence of wear-related dentoalveolar remodeling. In addition, dental wear pattern of D2600 could suggest an adaptation to a different ecological niche than the other Dmanisi individuals.