Large dinosaurs get studied because they survive as fossils. Bones mineralize more completely at larger sizes. Skeletons resist crushing. Museums display them. The result is a systematic bias: the fossil record over-represents large-bodied species, and paleontologists over-study what the fossil record over-represents. When a lineage appears to lack diversity at small body sizes, the absence may reflect preservation and attention, not biology.
Foskeia pelendonum (Verdú et al., Papers in Palaeontology, 2026) is a chicken-sized ornithopod from 120-million-year-old deposits in Burgos, Spain. Five individuals were recovered from red-clay floodplain sediments. The animals were fully mature adults, not juveniles of a larger species. Their skulls are described as “weird and hyper-derived” — highly specialized dentition and cranial architecture that distinguishes them from every other known ornithopod. Phylogenetic analysis places them as sister to the Australian Muttaburrasaurus within Rhabdodontomorpha, filling a 70-million-year gap in a lineage previously known only from much later and much larger representatives.
The finding that matters is not the gap itself but what occupied it. Rhabdodontomorphs were thought to be a late, modestly specialized group. Foskeia shows the lineage was experimenting with radical cranial specialization at miniature body sizes 70 million years before the forms we knew. The evolutionary innovation was not missing — it was invisible, hidden by the size-dependent filter that determines what enters the fossil record and what gets studied once it does.
The general principle: when a detection method is biased along a dimension correlated with the property of interest, absence of evidence is not evidence of absence — it is evidence of the bias. The fossil record's body-size filter doesn't just reduce the sample. It systematically removes the cases most likely to show that a lineage's evolutionary range was wider than the surviving record suggests. The emptiest parts of a biased dataset are the most informative about what the bias is hiding.