Interacting effects of water availability, neighbourhood density and ectomycorrhizal fungi on the traits and growth of a dominant Dipterocarp

Abstract Southeast Asian forests are increasingly impacted by global change. Precipitation reduction is a key driver whose effects on tree performance are mediated by biotic contexts such as tree density and altered soil fungal communities. Understanding tree responses to these drivers is central to predicting forest dynamics. However, these drivers are often studied independently despite their potential for interactive effects. Trait‐based approaches that ignore these interactions by assuming static trait values may provide limited or biased insights into future forests. We manipulated three key variables in a glasshouse experiment: water availability (simulating decreasing precipitation) and two biotic contexts, neighbourhood density and soil ectomycorrhizal (EcM) fungal communities. We assessed their independent and interactive effects on the growth, traits and trait–growth relationships of Parashorea chinensis , a dominant Dipterocarp species in southeast Asian tropical rainforests. Our results demonstrate that, while water availability is the dominant driver of seedling growth, the impact is magnified by seedling density. Promotion of seedling growth by soil EcM fungi is observed under wet but not dry conditions. Seedling trait–growth relationships are highly context‐dependent, varying across treatments (i.e. water availability, neighbourhood density and fungicide). For instance, while root and stem traits exhibited weak or inconsistent relationships with growth, leaf traits such as leaf area and specific leaf area correlated strongly with growth across treatments. These findings indicate that predictions based upon static species‐level trait values will be highly uncertain, revealing that trait–performance relationships are more complex than assumed. Synthesis . Our results reveal that trait–growth relationships shift across water availability, neighbourhood density and soil biotas, challenging current vegetation models that rely on mean trait values. These findings highlight the need to move beyond single‐driver studies and static trait assumptions by integrating these factors to develop more robust predictions of future forest performance.