Hello. This is the intended behavior. dcm2niix tries to create consistent NIfTI files regardless of the different interpretations of the DICOM standard used by different vendors. It will fill the first 3 dimensions with spatial data, and will create 4D data for sequences where the 4th dimension is time (fMRI, ASL, resting state, etc) or gradient direction (diffusion data). However, the NIfTI standard does not really specify how other properties (echo time, trigger time, phase data, real data) should be stored. Therefore, dcm2niix saves each of these different properties as a separate NIfTI file, and uses the file name to distinguish them. 

While for your own work, it might be convenient if dcm2niix packed the echo time in a different dimension, this would not be the desired behavior for someone working with a multi-echo fMRI time series. 

I think either [url=https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL%20FAQ]FSLMerge[/url] or nii = NII_TOOL('cat3D', filenames) will concatenate your images. Alternatively, you might want to contact the developers of NiftyFit.