This thesis investigates the origin and diversification of plastids in complex algae through the lens of nucleus-encoded plastid-targeted gene mosaicism. While plastid acquisition is classically explained by eukaryote-to-eukaryote endosymbiosis, kleptoplasty is gaining increasing attention as a plausible contributor to photosynthesis spread among eukaryotes, yet its potential role remains largely unexplored at the genomic scale. By combining a controlled phylogenomic framework with "PAGTI", an automated pipeline for taxonomy-aware and paraphyly-aware gene transfer inference from single-gene trees, this work quantifies donor diversity across multiple algal lineages using α-diversity indices. The results reveal that plastid-related gene mosaicism is structured rather than random and that donor α-diversity captures biologically meaningful differences among lineages, including unexpectedly high diversity in a genuine endosymbiotic system, i.e., chlorarachniophytes. Case studies in euglenids suggest a functional distinction between an early regime associated with plastid exploitation and maintenance, and a later regime linked to host metabolic integration, consistent with a kleptoplasty-driven pre-adaptive phase preceding permanent plastid acquisition. In peridinin-containing dinoflagellates, parallels with the euglenid system, including a three-membrane-bound plastid, combined with the independent diversification of at least three dinoflagellate subgroups toward kleptoplasty, raise questions about a possible kleptoplastic origin of the peridinin plastid itself, although a functional signature remains to be identified. This work proposes a reusable, reproducible framework linking phylogenomic transfer inference, donor diversity quantification, and functional interpretation to explore the plurality of evolutionary routes underlying plastid acquisition in complex algae.