"Race" is not a social construction...but "ethnicity" is.

Clarification: Obviously if you are comparing the Japanese and Swedish ethnic groups you would not think they are social constructions, but, the key here is that the ethnic groups correlate to racial differences. My point is that though broad racial clusters seem to exist, ethnic differences are far less clear cut. While the predictive power of race as a level of classification is quite often rather high, that of ethnicity is far less clear cut when the racial component is extracted. There are haplotype discontinuities between neighboring groups (Lithuanians vs. Poles on the TAT-C), but they are far less substantial than the differences between races, and often individuals within each group have a difficult time telling one from another by inspection alone.

The implication is that many ethnic rivalries are, as many people would assume, constructs of culture and history rather than literal blood feuds. Since most wars have been between neighbors, they have been intraracial rather than interracial, and the putative "Race War" to come is a future imagining, not a reflection of past Darwinian history.

Addendum: Leo the Syrian was the emperor who defended Constantinople against the Arabs in the early 8th century, just to show you how cultural-religious divides were much more salient on the ground than racial differences because conflict is more likely between neighbors. The modern age of transcontinental migration has resulted in a terminology of racial conflict & tension that often gets confounded with the more ancient one of ethnic conflict, resulting in confusions. A classic example is the pattern of some Leftists of trying to depict the Palestinians as "people of color" and Israeli Jews as white, when only 50% of Israeli Jews are of European origin, and a substantial portion of the remainder are Yemeni or Ethiopian, ergo, no less "people of color" than Palestinian Arabs.