Acoustic data are invaluable information sources for characterizing the distribution and abundance of mid-trophic-level organisms (micronekton). These organisms play a pivotal role in the ecosystem as prey of top predators and as predators of low-trophic-level organisms. Although shipboard-ADCP (acoustic Doppler current profiler) acoustic backscatter signal intensity cannot provide an absolute biomass estimate, it may be a useful proxy to investigate variability in the distribution and relative density of micro-nekton. This study used acoustic recordings data spread across 19 years (1999-2017) from 54 ADCP cruises in New Caledonia's subtropical EEZ (exclusive economic zone) to assess seasonal and interannual variabilities and spatial distribution of micro-nekton. The dataset was composed of two different ADCPs: 150 kHz for the first period, followed by 75 kHz for more recent years. We examined the 20-120 m averaged scattering layer. Using the few cruises with concurrent EK60 measurements, we proposed that the backscatter from the ADCPs and 70 kHz EK60 were sufficiently closely linked to allow the use of the backscatter signal from the ADCPs in a combined dataset over the full time series. We then designed a GAMM (generalized additive mixed model) model that takes into account the two ADCP devices as well as temporal variability. After accounting for the effect of the devices, we showed that the acoustic signal was mainly driven by diel vertical migration, season, year, and ENSO (El Nino-Southern Oscillation). In a second step, a consensus model between two statistical approaches (GAMM and SVM) (support vector machine) was constructed, linking the nighttime 20-120 m backscatter to the oceanographic and geographic environment. This model showed that sea surface temperature was the main factor driving backscatter variability in the EEZ, with intensified backscatter during the austral summer (December to May) in the northern part of the EEZ. We showed that acoustic density differed significantly, spatially and temporally from micronekton biomass predicted for the same period by the SEAPODYM-MTL (mid-trophic level) ecosystem model. The seasonal cycle given by ADCP data lagged behind the SEAPODYM-MTL seasonal cycle by around three months. Reasons to explain these differences and further needs in observation and modeling were explored in the discussion. In addition to providing new insights for micronekton dynamics in this EEZ (i.e., the science needed for ecosystem-based fisheries management), the data should help improve our ability to model this key trophic component.