Predictive Modelling Software User Manual

5.5 Predicted/collected

Hide Navigation Pane

5.5 Predicted/collected

Previous topic Next topic No directory for this topic  

5.5 Predicted/collected

Previous topic Next topic Topic directory requires JavaScript JavaScript is required for the print function Mail us feedback on this topic!  

The predicted/collected data sheet provides an assessment of biological condition of a test site. The indices produced here are described in the following sections.

 

5.5.1 Predicted, expected and observed taxa

The first four columns of the predicted/collected data sheet give information on the predicted, expected and observed numbers of taxa.

 

NTE50

The Number of inverTebrate families Expected with greater than a 50% probability of occurrence, NTE50, is the sum of the probabilities of all the families predicted with greater than a 50% chance of occurrence

 

NTP50

NTP50 is a count of the Number of inverTebrate families Predicted with greater than a 50% probability of occurrence

 

NTC50

The inverTebrate families that were predicted above the threshold probability of 50% and which were also Collected at the test site are counted to form the observed (collected) value, NTC50.

 

OE50

The Observed to Expected ratio, OE50, is the ratio of the number of invertebrate families observed at a site (NTC50) to the number of families expected (NTE50) at that site. OE50 provides a measure of biological impairment at the test site.

The output of these four indices is shown in the screen shot below.

 

5.5.2 Predicted, expected and observed signal

In addition to calculating the expected number of taxa at a test site, AUSRIVAS also calculates the SIGNAL scores for a site. Calculation of the SIGNAL scores uses SIGNAL grades, an assigned value given to each invertebrate family that indicates their sensitivity to pollution. Columns 5 to 10 in the Predicted/Collected data sheet contain the calculated signal scores, each of which is explained below.

 

E50Signal

E50Signal is the expected signal score for taxa that have a probability of occurrence of greater than or equal to 50%. It is calculated by weighting the probability of occurrence of each predicted taxa (those taxa that have a probability of occurrence of greater than or equal to 50%) by the taxon's SIGNAL Grade, summing these and then dividing the total by the sum of the (unweighted) probabilities of occurrence.

 

clip0001

 

O50Signal

O50Signal is the observed signal score for taxa that have a probability of occurrence of greater than or equal to 50%. It is calculated by averaging the SIGNAL Grade's for all observed taxa with P(Taxa) >= 0.5.

 

clip0002

 

OE50Signal

The observed to expected SIGNAL ratio, OE50Signal, is the ratio of E50Signal to O50Signal.

E0Signal

E0Signal is calculated the same way as E50Signal, except all taxa that have a probability of occurrence of greater than 0% are included in the calculation.

 

clip0003

 

O0Signal

O0Signal is the observed signal score for taxa that have a probability of occurrence (P(Taxa)) of greater than 0%. It is calculated by averaging the SIGNAL Grade's for all observed taxa. O0Signal is equivalent to the ‘raw’ SIGNAL score (Chessman, 1995).

 

clip0004

 

OE0Signal

The observed to expected SIGNAL ratio, OE0Signal, is the ratio of E0Signal to O0Signal.

The calculated SIGNAL Scores are shown in the screen shot below.

 

5.5.3 The banding scheme

For each model the O/E taxa ratios are divided into bands representing different levels of impairment. Band X represents a richer invertebrate community than reference, band A is considered equivalent to reference; band B represents sites below reference condition; band C represents sites well below reference condition; and band D represents impoverished sites. The band for each site is listed in the last column on the Predicted/Collected data sheet, as shown in the screen shot above.

 

5.5.4 Further information

For further explanation of how indices and bands are calculated, refer to the AUSRIVAS Predictive Modelling Manual, particularly Chapter 3, 'The AUSRIVAS Models'.