Those maps can be very misleading, but the folks who put them together can only work with the published information they get. Some of which is crap. So garbage in, garbage out. If you zoom on those maps, Texas is uniformly pale green. Which means that some manual or report mentions the plant as being native to Texas but without designating where. So the whole state gets pale green, even if the plant was only found once and might not even still be present. But if you zoom to Michigan, you get a single county in dark green, which is where the plant is supposed to have been found. Which is awfully strange, this plant is supposedly native in Texas and Michigan but no place in between? Consider those maps as just giving a very general idea of the range of a particular species.

Expanding on that a bit. A native plant is one native to the United States. Period. The USDA maps can't distinguish between a native occurrence and an occurrence outside of the plant's native/natural range. So I would read the map for that plant as "native to the lower 48, has been recorded from Texas, maybe throughout the state, maybe just a single location, but without specific county data. Also recorded from one county in Michigan." You are left to work out the details.

in Texas

Note that the color changes when you are looking at the county outline maps versus the state outline maps. The lighter green on the county level maps means that there is no county level data available for that species. So even though all the counties of a particular state are colored in, it doesn't mean that there was an observed occurrence in any particular county.

The NRCS maps are compiled from many different sources, especially state-level "flora of" programs. Missouri's program is excellent (based out of the Missouri Botanical Garden in St. Louis) but there's a lot of variety state-to-state. Click on the "Data Source and Documentation" link to the left of the map. My guess is that a lot of this data was in state databases that were dumped, fed into a conversion program, and then loaded into the NRCS database. Each state would have different info available and so the conversions may or may not be accurate. Some records may have been checked by hand after that conversion, others not.

In the case of Ipomoea x multifida (which is I think the NRCS map you're looking at) I'm pretty sure that the plant (which is supposedly a hybrid of I. coccinea and I. quamoclit) is not native to Michigan, and probably not to Florida or Texas either. I'd guess that the state data may have just indicated "species present in state" and perhaps NRCS coded that as "native" during the conversion.

Personally I prefer BONAP for maps like these, but they don't always include hybrids. Here's the genus Ipomoea:

http://bonap.net/NAPA/TaxonMaps/Genus/County/Ipomoea

By the way here's the record in the current Michican Flora database, which mentions the one occurrence in Jackson County, MI (that's the one coded in green on the NRCS map):

http://michiganflora.net/species.aspx?id=862

Note that the record clearly indicates that it's not native. So -- either the conversion to the NRCS database wasn't correct, or else the Michigan Flora record has been updated with additional info since 1985.

Those maps are great, but they are incomplete. I myself have provided information to this group. They had-this is just one example-shown tamarack-an extremely common tree in this state, as not occurring in Shawano County (pronounced shaw-no). Yet, on the way to my land one county further up, I drive through numerous tamarack swamps on the way up. And these are in Shawano County. So I contacted the USDA people responsible for this map. They are going to want GPS coordinates, but I didn't have that. So what I did is gave them the name of the river watershed drainage this was in, the name of some highway coordinates that would put you there, and they have changed their map relative to this species.

These people are good, but as has been said, they have not personally walked the entire nation, taking notes of what plant grows where. So some anomalies can occur. And you can be a part of fixing these issues! I've also called in an herbaceous species or two.

Speaking of which, there's a page on the NRCS site about submitting data:

http://plants.usda.gov/du/DistributionUpdate.html

Also there is a FAQ about what they mean by "native":

http://plants.usda.gov/faq.html#native