Regeneration dynamics and diversity patterns across disturbance gradients in kaya Kauma Forest, coastal Kenya

• Regeneration density declines consistently from seedlings to juvenile trees; • Ontogenetic stage drives regeneration patterns more strongly than forest zone; • Species diversity remains high across edge–interior disturbance gradients; • Regeneration communities show weak zonal differentiation across forest zones; • Sacred forests maintain strong regeneration despite disturbance gradients; Understanding natural regeneration dynamics is essential for evaluating forest resilience, particularly in culturally protected coastal forests where disturbance gradients may influence recruitment processes. This study assessed regeneration density, diversity, and community structure across edge, intermediate, and core zones of Kaya Kauma Forest, Kenya, to evaluate regenerative capacity and demographic balance. Tree regeneration was surveyed in 5 × 5 m subplots nested within 20 × 20 m plots across three zones. All woody regenerants were identified and assigned to four height classes representing developmental stages. Regeneration density, seedling-to-sapling ratios, alpha diversity metrics, and dominant species patterns were quantified. Community composition was analyzed using Bray–Curtis dissimilarity, non-metric multidimensional scaling (NMDS), and PERMANOVA, while regeneration counts were compared using Kruskal–Wallis tests and generalized linear modelling. Regeneration density declined consistently from Class I to Class IV across all zones. Seedling-to-sapling ratios exceeded 2.0, indicating strong recruitment but reduced progression into later development stages. Alpha diversity remained high and similar among zones (mean Shannon index ≈ 3.7). Dominant species assemblages were largely shared across zones, and NMDS ordination showed substantial compositional overlap. PERMANOVA indicated weak zonal structuring of regeneration communities (R² = 0.076, p = 0.778), while modelling confirmed that regeneration abundance was driven primarily by height class rather than spatial position. These results indicate strong regenerative potential and compositional stability, although demographic filtering between early and advanced stages may influence long-term forest structure.