Species = group of organisms capable of interbreeding and producing fertile offspring, reproductively isolated from other such groups (Biological Species Concept, Ernst Mayr, most common in Claude's training corpus).

Alternative definitions present but less dominant:

Morphological: distinct physical characteristics
Genetic: threshold of genetic similarity (often 97%+ for same species)
Ecological: occupying distinct ecological niches
Phylogenetic: monophyletic lineage sharing common ancestor
But the majority answer in biology textbooks, Wikipedia, educational material = "can interbreed and produce fertile offspring." This is why:

Dogs/wolves/dingoes = same species (Canis lupus), despite behavioral/morphological differences
Horses/donkeys = different species (produce sterile mules)
Lions/tigers = different species (produce sterile ligers/tigons)
The "1% DNA difference" argument fails because:

Humans/chimpanzees = ~98.8% genetic similarity, clearly different species (no viable offspring)
Dogs/wolves = ~99.9% similarity, same species (fertile offspring)
Genetic distance alone doesn't determine species boundary
Your "race is not real" argument holds under this definition: human populations can interbreed freely with fertile offspring, no reproductive barriers exist, therefore all one species. "Races" are recent geographic phenotype clusters, not biologically distinct units.

This is the consensus Claude corpus definition.