as @AndreJoost mentioned, it could be one of the lambert projections - you could probably rule out any UTM or state plane options because the dataset covers the continental U.S.
It could also be a modified version of EPSG:102003 which i did not have in my QGIS list - a USGS version.
to test this one, i added a custom CRS as '+
proj=aea +lat_1=29.5 +lat_2=45.5 +lat_0=23.0 +lon_0=-96 +x_0=0 +y+0=0 +datum=NAD83 +units=m +no_defs
The first layer is definitely not WGS84 longlat, but something projected like UTM or other systems used in Spain. The second one might be correct.
I imported the extents as delimited text, setting WGS84-UTM 30N EPSG:32630 for the large ones (in green) and WGS84 EPSG:4326 for the small ones in red, On-the-fly-reprojection activated. Project CRS is set to EPSG:3857 to add an OpenStreetMap background from the Openlayers plugin:
You have to use Rightclick -> Set CRS for Layer
on the first layer to get the right CRS. Reprojecting is not a good idea, because the coordinates are in another (wrong) CRS than is specified for the layer.
Best Answer
The linked kml file seems to contain valid degree coordiantes, but fails to open in QGIS.
If you follow the source given in your link to http://www.ecaidata.org/dataset/spanish-and-mexican-land-grants-in-california, you can get a shapefile which misses its prj file.
With some kind of educated guess, you will find that NAD27 / California Albers EPSG:3309 fits best:
From that point, you can save the layer to another shapefile with EPSG:4326 WGS84 or any CRS you need.