Changes in the seasonal cycle of phytoplankton are important indicators of coastal ecosystem responses to climate change and human activities. Based on historical observations from Jiaozhou Bay in 2005, 2010, 2015, and 2020, this study examined long-term variations in the seasonal cycle of phytoplankton biomass and their environmental drivers. The results showed that the seasonal distribution pattern of phytoplankton biomass in Jiaozhou Bay changed markedly during 2005~2020, shifting from a bimodal pattern with winter and summer peaks to a predominantly unimodal pattern with a summer-autumn peak. Winter peak values of net-collected phytoplankton abundance and chlorophyll a concentration decreased substantially, from 103.62×10⁶ cells/m³ and 15.29 μg/L in 2005 to 12.81×10⁶ cells/m³ and 1.13 μg/L in 2020, respectively. Community structure analysis indicated that diatoms consistently dominated the phytoplankton assemblage and that variation in diatom abundance was the primary driver of the observed shift in the seasonal cycle of phytoplankton biomass. In contrast, dinoflagellate abundance declined markedly after peaking in 2010. Correlation analyses suggested that increasing winter water temperature and decreasing phosphate concentration were the main factors underlying the attenuation and eventual disappearance of the winter phytoplankton biomass peak. These findings reveal the long-term evolution of phytoplankton seasonal structure in Jiaozhou Bay under the combined influence of climate change and nutrient adjustment, and offer scientific support for coastal ecosystem management.

There is a significant disconnection between the richness of data and the effectiveness of decision-making in current marine observation practices. This study argues that the root cause lies in the “ontological” bias and “methodological” mismatch in the logic that guides observation design. We first examine the “uniformity assumption” that views the ocean as a spatially homogeneous and temporally gradual system. Based on the “land-sea heterogeneity isomorphism” framework, we demonstrate that the ocean is essentially a heterogeneous landscape dominated by key dynamic processes, characterized by “resource-rich areas”, “risk-concentrated areas”, and “pulsatile events” that drive their changes. These local, transient, and nonlinear threshold processes often dominate the function and evolution of coastal and estuarine ecosystems. Traditional “mission-oriented” or “undifferentiated census” observations, which ignore this essence, lead to systematic omissions of core processes and key risk signals. Based on this, we propose a fundamental shift in the marine science observation paradigm: from “routine observation-driven alone” to “a combination of routine observation and scientific questions/management goal-driven”. We construct a logical framework of “scientific questions/management goals-key processes-observation requirements”, and systematically expound on three methodological breakthroughs to address the “fluid medium” characteristics of the ocean: building a “dynamic adaptive perception network” to replace fixed grids, establishing “full-chain process analysis” to replace isolated hypothesis testing, and achieving an intelligent closed loop of “real-time assimilation and predictive decision-making”. This study aims to provide a comprehensive theoretical pathway from ontological cognition to methodological practice to build the next generation of “precise, efficient, and applicable” marine science observation systems.