Three sampling methods ran simultaneously in the same Costa Rican forest for 69 trap-years, collecting over 1.6 million individual insects [1]. When researchers tallied the results, 75 percent of parasitic wasp species appeared in only one of the three protocols [1]. Not three forests, one forest, three ways of looking. That collision of methods is the mechanism behind a new estimate that places Earth's insect diversity at 14 to 20 million species, more than double the 6 million figure that stood for four decades [1][2].
The research team studied Microgastrinae wasps at the Área de Conservación Guanacaste in northwestern Costa Rica using three distinct approaches: a core set of Malaise traps, a peripheral set of Malaise traps positioned differently in the landscape, and direct caterpillar collection followed by rearing to identify which wasp species emerged [1]. Malaise traps are tent-like structures that intercept flying insects [1]. The protocols were designed to sample the same taxonomic group in the same habitat, yet they functioned as nearly orthogonal apertures, each capturing a slice of reality the others missed [1].
The implication scales. If three methods in a single forest partition wasp diversity such that three-quarters of species are invisible to any two of them, then global sampling efforts, which rely heavily on a narrower set of standardized protocols, are missing most of what exists [1]. The previous estimate of 6 million insect species, established decades ago, did not account for this sampling blindness [1]. The new PNAS study extrapolates from the Costa Rican data to suggest 8 to 14 million additional insect species beyond those already known [1].
What we cannot see, we cannot protect
The gap between what exists and what science can document has direct consequences for conservation decisions in tropical forests and agricultural regions [1]. Species that never appear in standardized surveys cannot be included in biodiversity assessments that inform land-use policy [1]. When development projects or agricultural expansion require environmental impact studies, those assessments rely on existing species inventories, inventories that this research suggests may be missing three-quarters of the insects actually present in a given habitat [1].
The missing millions are not randomly distributed across insect biology. Most undiscovered species are almost certainly small, rare, and highly specialized [1]. Many will turn out to be parasitoids, insects that lay eggs inside other insects, with larvae that consume the host from within before emerging [1]. Microgastrinae wasps, the focus of the Costa Rican study, follow this pattern: they deposit eggs inside caterpillars, and the wasp larvae eat the caterpillar's insides before pupating [1]. These parasitoids regulate caterpillar populations that affect crop yields and forest health, yet their absence from species lists means their ecological roles remain unquantified in management plans [1].
Metamorphosis permits insects to occupy multiple spatial niches across a single lifetime [1]. A caterpillar in the forest canopy becomes an adult wasp in the understory. That life-history flexibility, combined with small body size, allows insect populations to persist in microhabitats that are spatially restricted and difficult for scientists to access [1]. Some species may exist only in the forest canopy, a zone that has been barely sampled [1]. Others are so rare or temporally restricted that even intensive trapping over years yields only a handful of individuals [1].
The estimate includes possibilities like 2,400 firefly species alone [1]. The actual number of described insect species is a fraction of the new estimate, only a small percentage of the projected 14 to 20 million have been formally documented [1]. The rate of taxonomic description is the constraint, not the rate at which insects exist [1].
Quantifying invisibility
The revision is not evidence that insect diversity is increasing. The number of species has not changed; the estimate changed because researchers quantified how much their sampling methods fail to overlap [1]. Each protocol is a narrow window. Insects exist in the gaps between windows, in life stages the traps do not target, in spatial strata the nets do not reach, in temporal windows the sampling schedule misses [1].
The study's co-authors include Robert Colwell, an entomologist at the Museum of Natural History at the University of Colorado and the University of Connecticut; Michael Sharkey, a University of Kentucky professor emeritus; and Laura Melissa Guzman, a Cornell University entomologist and biodiversity scientist [1]. Their work does not propose new insect biology. It exposes the geometry of how sampling protocols partition the organisms they are designed to detect [1].
The new number is more accurate, but most of those 14 to 20 million species will never be individually described [1]. The gap between existence and documentation widens faster than taxonomic work can close it [1]. At current rates of species description, roughly 20,000 new insect species per year, it would take between 700 and 1,000 years to catalog the full estimate, assuming no taxonomist retirements, no funding gaps, and no species extinctions [1]. The infrastructure does not exist to close that gap [1].
For researchers working in tropical conservation, the practical question becomes whether to wait for complete species inventories that will never arrive, or to make land-use decisions based on the partial data that standardized protocols can provide within project timelines [1]. The math suggests that hyperspecialized, rare insects are present in unsampled canopies and microhabitats [1]. What remains is determining whether knowing biodiversity means counting what we can catch, or modeling what we cannot [1].
