WESPEN Online Order Form print this page
US Fish & Wildlife Service FieldNotes

Identification of Individual Cultured Delta Smelt Using Visual and Automated Analysis of Natural Marks

Region 8, September 30, 2016
Fig.1 Dorsal head area of cultured adult Delta Smelt
Fig.1 Dorsal head area of cultured adult Delta Smelt - Photo Credit: n/a
Fig.2 Dorsal head area of wild adult Delta Smelt
Fig.2 Dorsal head area of wild adult Delta Smelt - Photo Credit: n/a

By Gonzalo Castillo

Natural external marks such as spots and scars have been used to identify individual organisms including fish and marine mammals. Because artificial tags can reduce survival and increase tag shedding in small fishes, natural marks could provide an alternative identification method for individual fish.

In cooperation with the UC Davis Fish Conservation and Culture Lab (FCCL) and U.S. Bureau of Reclamation, Lodi Fish and Wildlife Service fish biologist, Gonzalo Castillo, will present preliminary results on the feasibility of using external natural marks to identify cultured Delta Smelt (2016 Bay Delta Science Conference).

Researchers selected the dorsal head area (DHA) of Delta Smelt, where external pigmentation is particularly abundant (Fig. 1). To verify the usefulness of natural marks, each fish was tagged with a unique alpha numeric tag. Researchers evaluated the potential of the DHA as a unique natural mark using visual (naked eye) and an automated image recognition program (TinEye’s Match Engine API).

Three photo sessions were conducted beginning nearly 244 days post-hatch in two fish groups over two periods (January-May 2013; n= 299 and October-April 2014-15; n= 560).

The group in the last period included low and high light treatments to evaluate the influence of ambient light on natural marks. A digital camera with a 100 mm f/2.8 macro-lens was used to acquire DHA photos. Preliminary findings showed that shorter intervals between photo sessions (nearly 56 days in 2013 and 90 days in 2014-15) resulted in higher percent of correct matching for both visual recognition (100% in 2013 and 70-100 % in 2014-15) and automated recognition (59-89 % in 2013 and 19-33 % in 2014-15).

Pigmentation generally became less apparent under the high light treatment, leading to a marginally lower percent of automated recognition relative to fish under low light. Unlike visual recognition for the 2014-15 sessions, the percent of correct visual matching in the 2013 sessions remained constant (100%) as a function of the number of days between sessions, suggesting a smaller change in pigmentation during the maturation and spawning periods (winter-spring) than before maturation (fall-winter).

These results suggest natural marks may be more reliable to track cultured Delta Smelt at the adult stage than from the sub-adult to adult stage. Researchers also plan to conduct a preliminary evaluation of natural marks on wild adult Delta Smelt collected and photographed by Lodi FWS staff during fisheries monitoring surveys (Fig. 2).

Contact Info: Gonzalo Castillo, 209-334-2968 x418, gonzalo_castillo@fws.gov