Page 417 - Contributed Paper Session (CPS) - Volume 6
P. 417
CPS2007 Jai-Hua Yen et al.
estimator by the data without species identity error. In observed method, the
estimation of species richness used the traditional Chao2 estimator by the data
with species identity error. In adjusted method, the estimation of species
richness used the adjusted richness estimator by the data with species identity
error.
When species identity error occurs, the estimate of species richness by
observed method will be underestimated, which causes larger bias. The large
bias still exists even though the increase of the number of sampling units.
Since adjusted method slightly overestimated species richness when the
species identity error rate is large, it reduces a great quantity of bias. The
variation of observed method is lower, and it remains the same by different
species identity error rate. The adjusted method has a higher variation. When
species identity error rate is larger, the variation of adjusted method is larger.
By evaluating both bias and variation, the observed method has a larger RMSE
(Root Mean Square Error) due to its larger bias. The adjusted method has
about half RMSE of the observed method when the number of sampling unit
is large.
b. Extrapolation for Poisson Counts
The data set was collected of weed species from organic farmland located
at Soft Bridge county in the North of Taiwan. There are 12 transect lines with
length 20m each were conducted. Only the incidence (detection or non-
detection) of species in each transect line was recorded. Before richness
estimation, a subplot occupied by 40 known weed species was treated as the
testing of the degree of investigator's skill. Compare these 40 weed species
list with the inventories of the investigator, we have = 40, , = 35, and
,0 = 1. Therefore, we have the estimate of ̂= 0.14 and ̂= 0.82 based on
equations (1) and (2). Many of the misidentified species were misidentified as
species which did not exist in the plot. The summary of the frequency counts
of weed species is in Table 5. The result using our adjusted estimator is in Table
6. By simulation studies, the error rate is high in this case. Hence, the estimate
of species richness using row data directly underestimates and the adjusted
estimator should be applied to get the accurate estimate of species richness.
406 | I S I W S C 2 0 1 9