Aims. The aim of our study is to investigate the physical properties of the star-forming interstellar medium (ISM) in the Large Magellanic Cloud (LMC) by separating the origin of the emission lines spatially and spectrally. The LMC provides a unique local template to bridge studies in the Galaxy and high redshift galaxies because of its low metallicity and proximity, enabling us to study the detailed physics of the ISM in spatially resolved individual star-forming regions. Following Okada et al. (2015, Paper I), we investigate different phases of the ISM traced by carbon-bearing species in four star-forming regions in the LMC, and model the physical properties using the KOSMA-τ PDR model. Methods. We mapped 3-13 arcmin 2 areas in 30 Dor, N158, N160 and N159 along the molecular ridge of the LMC in [C ii] 158 µm with GREAT on board SOFIA. We also observed the same area with CO(2-1) to (6-5), 13 CO(2-1) and (3-2), [C i] 3 P 1 -3 P 0 and 3 P 2 -3 P 1 with APEX. For selected positions in N159 and 30 Dor, we observed [O i] 145 µm and [O i] 63 µm with upGREAT. All spectra are velocity resolved. Results. In all four star-forming regions, the line profiles of CO, 13 CO, and [C i] emission are similar, being reproduced by a combination of Gaussian profiles defined by CO(3-2), whereas [C ii] typically shows wider line profiles or an additional velocity component. At several positions in N159 and 30 Dor, we observed the velocity-resolved [O i] 145 µm and 63 µm lines for the first time. At some positions, the [O i] line profiles match those of CO, at other positions they are more similar to the [C ii] profiles. We interpret the different line profiles of CO, [C ii] and [O i] as contributions from spatially separated clouds and/or clouds in different physical phases, which give different line ratios depending on their physical properties. We model the emission from the CO, [C i], [C ii], and [O i] lines and the far-infrared continuum emission using the latest KOSMA-τ PDR model, which treats the dust-related physics consistently and computes the dust continuum SED together with the line emission of the chemical species. We find that the line and continuum emissions are not well-reproduced by a single clump ensemble. Toward the CO peak at N159 W, we propose a scenario that the CO, [C ii], and [O i] 63 µm emission are weaker than expected because of mutual shielding among clumps.