Potential areas for improvement to NatureServe CCVI algorithm

[see24]

• Method for combining BC and D index values
• Conflating neutral and unknown factors
• Use of extreme values for HTN and PTN which may be inaccurate with broad range polygons. Could be better to use some # or proportion of cells to get an extreme value less vulnerable to outliers. Similar to Culp et al 2017

• 37

• Using the average of two boxes checked to get score when there is uncertainty. Would it be better to use the most common index value from the monte carlo? Would need to highlight uncertainty more.
• Time frame of the assessment. Currently assumes 2050. Should any of the thresholds be adjusted if a different time frame is used? Should the thresholds change depending how far into the future we are projecting?

There is a limitation of the index when only one or two D factors are scored because it prevents the overall index from being EV unless the D index is at least HV which only happens if the score for the D section is >= 4 but if you only answered 2 of the Qs that is out of a possible total of 6 rather than the possible total of the whole section which is 12. EG yellow billed loon in USA which had BC score 11.8 and D score 3. The threshold for BC score to have EV index is 10 but he D score gives a D index of MV which pulls down the combined score to HV. We ignored the instruction that said not to score D3 if D2 had a score of 3 but if we had followed it it would have been completely impossible for any species to score EV. Not having data for the other 2 D questions brought down our index values. This is a general problem for the index not differentiating between questions left blank and marked as neutral which both get score of 0 but it is more pronounced for the D section questions.

[see24]

Consider using the percent change in CMD as opposed to future - normal? See https://www.sciencedirect.com/science/article/pii/S2405880721000054 which used percent change in moisture metric

[see24]

Consider option to set a min max elevation and use DEM that comes with climate data to limit the elevation where climate data is used from. Would help with issue of overly broad range polygon being used for historical niches

[see24]

Also see #11

[see24]

Also should consider ignoring the rule where D3 is set to Neutral if D2 is Greatly increases. This is supposed to prevent "double counting" future range model results that range will be completely lost in assessment area. But it means that eg Species 2 with Increases for both D2 and D3 will have a higher score (4) than Species 1 with Greatly increases for D2 which would also mean Greatly increases for D3 except that it is set to 0 leading to a D section score of (3). If D3 was not scored, and removed from the total possible score then this would be ok eg Species 1 would score 3/3 and Species 2 would score 4/6. So this could be fixed by a general strategy for fixing the section D scoring (see above) another option would just be to change it to a reduction of the max D3 score to reduce how extreme the effect is of having both at GI eg make the rule that if D2 is GI D3 is Inc = score of 5 in total. Or just ignore this rule altogether.