When converting a PDF to a lower-resolution PDF file with Ghostscript, the background black colour is applied to the text and icons.
Download original PDF file
My code:
gs -q -dNOPAUSE -dBATCH -dSAFER -sDEVICE=pdfwrite -dPDFSETTINGS=/default -dDownsampleColorImages=true -dColorImageResolution=72 -dColorImageDownsampleThreshold=1.0 -dDownsampleGrayImages=true -dGrayImageResolution=72 -dGrayImageDownsampleThreshold=1.0 -dDownsampleMonoImages=true -dMonoImageResolution=72 -dMonoImageDownsampleThreshold=1.0 -dHaveTransparency=false -sProcessColorModel=DeviceRGB -sColorConversionStrategy=RGB -sColorConversionStrategyForImages=RGB -dTransferFunctionInfo=/Apply -sOutputFile=output.pdf input.pdf

If we simply isolate one image and apply the switch between transparency=false and transparency=true we can see the difference that will affect PDF images with transparency. This is common to many PDF handlers since PDF defines transparency as levels of soft Mask and full transparency is stored as if image has a black value!
The way colours are stored in a PDF is as a bitmap so take this simplest of samples
If we split the colour into 3 parts for each row we can see the text pattern.
Note that the bottom row which has a first RGB colour as transparent white is the same on the left as the right which looks black (actually neither is black or white simply 3 x null but here as a space character).
When we look at the transparency component it may look like this.