Counting What Lives in a Place

Someone wants to know what species live in a given area — a forest patch, a farm, a stretch of riverbank. That question sits behind conservation planning, environmental clearance, restoration monitoring and corporate biodiversity reporting.

Several methods exist and each sees a different slice:

• Camera traps catch ground-dwelling mammals well and miss small, arboreal and nocturnal-silent species
• Acoustic recorders catch birds, bats, frogs and insects that call, and miss everything that does not
• Environmental DNA detects traces of many species from a water or soil sample, but needs a laboratory, sequencing, and a reference database that is incomplete for Indian species
• Transect and plot surveys are reliable and expensive in expert time
• Citizen records such as bird and nature-observation platforms are large and heavily biased toward attractive species near roads

Each method gives you a list that reflects the method as much as the place. Combining them is not addition, because each one detects different species with different probability, so you cannot simply pool the lists and call it a survey.

Because it is a genuine open problem in ecology, not an engineering oversight. Estimating how many species are present when your detection is imperfect and varies by method is a live statistical research area. Anyone promising a complete answer is overselling.

• Branch — CS and Electronics, with a strong pull for anyone interested in ecology or statistics.
• Technical skills — Statistical estimation under imperfect detection, combining heterogeneous data sources, species classification from images or audio, sensor deployment, data visualisation.
• Other skills — Reading ecological literature, stating uncertainty, resisting the urge to claim completeness.