You don't give enough information. Somewhat above the lines you show are, on my Mac OS X 10.3 machine, are
Making `all' in directory /Users/rwcox/AFNI/src/netcdf-3.5.0/src/libsrc
cc -c -O -I. -DNDEBUG attr.c
cc -c -O -I. -DNDEBUG dim.c
cc -c -O -I. -DNDEBUG error.c
cc -c -O -I. -DNDEBUG -DVERSION=`cat ../VERSION` libvers.c
cc -c -O -I. -DNDEBUG nc.c
cc -c -O -I. -DNDEBUG ncio.c
cc -c -O -I. -DNDEBUG ncx.c
cc -c -O -I. -DNDEBUG putget.c
cc -c -O -I. -DNDEBUG string.c
cc -c -O -I. -DNDEBUG v1hpg.c
cc -c -O -I. -DNDEBUG v2i.c
cc -c -O -I. -DNDEBUG var.c
It sounds like the "error.c" file did not compile correctly on your system -- this is the file whose object error.o is renamed to netcdf_error.o for inclusion into the AFNI libmri.a library, which is the file apparently missing. You'll have to look backwards through the output of "make" to see if compilation does indeed fail for error.c. Are there any files named "netcdf_*.o" in your AFNI source directory?
Also, when you were trying to run AFNI from the binaries, did you set the environment variable AFNI_PLUGINPATH to point to the location where the AFNI binaries, plugins, and TTatlas+tlrc dataset are stored? If not, then plugins won't load and the atlas dataset won't be found.