Performance Objective 



1. Provide estimates of important measurement-related errors associated with the indicator, and 

 compare them to established performance criteria for the proposed monitoring framework. 



Several different types of errors can affect either the measurement data or the development of indicator 

 values from measurement data (Table 4-14). Measurement-related errors of field collection data, in terms of 

 number of species collected, species composition, and number of individuals cannot be estimated directly for 

 the indicator by collecting replicate samples during a single visit to a site (Table 4-9, Fore et al. 1994). In 

 terms of repeatability, other published studies may not be applicable to the entire mid-Atlantic highlands 

 region or to the proposed monitoring framework. Measurement-related errors can be treated as a part of 

 temporal indicator variability (Guideline 9) and can be indirectly evaluated as part of "within-year extraneous 

 variance." The other critical source of error in measurement data is incorrect identifications offish species. 

 Various means of controlling this source of error have been presented previously, including the collection and 

 confirmation of voucher specimens (Table 4-7), using personnel experienced in fish identification (Table 4-8) 

 and additional training in field identification of regional fishes (Table 4-10). 



Table 4-14. Potential sources of measurement error 



Data Collection 



Poor repeatability in number of species collected, species composition, and counts 



• Cannot be estimated directly using replicate samples 



• Controlled by standardized protocols and methods 



Incorrect identification of species by field crews or data recording errors 



• Performance objective is < 1 0% errors 



• Controlled through training, field audits and performance checks, and confirmation 

 of voucher specimens 



See Table 4-9 for additional details 



A more quantitative evaluation of errors related to identification of fish species by field crews was derived from 

 3 years of sampling for the MAHA study (Fig. 4-5). Five types of error are investigated. Transcription errors 

 occur when the wrong species code is recorded on the field data form. The remaining four relate to actual 

 errors in taxonomy. They include a cumulative estimate of errors for all species, errors specific to two groups 

 of fishes that are difficult to identify to species in the field (sculpins, genus Cottus, and a cyprinid genus 

 Nocomis), and errors at the genus level. Over the 3-year period, improvements were made to field data 

 forms, crew training, and the procedure for collecting voucher specimens. Performance improved each year, 

 with the virtual elimination of transcription errors, misidentification of sculpins, and misidentifications at the 

 genus level (which are potentially more serious than species level identification in terms of the impact on 

 metric responses). The remaining error levels for overall species misidentifications and identification of Nocomis 

 species declined to well below the performance objective initially established (< 10%; Table 4-14). 



4-25 



