ICPRB 1 Conference Line 8662993188 Code 2679856222 Adobe Connect httpsepawebconferencingacmscomstreamhealthworkgroup Meeting Website httpwwwchesapeakebaynetcalendarevent23800 ID: 789035
Download The PPT/PDF document "TAG Meeting 2 Zachary M. Smith" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
TAG Meeting 2
Zachary M. SmithICPRB
1
Conference Line
:
866-299-3188
Code
:
267-985-6222
Adobe Connect
:
https://epawebconferencing.acms.com/streamhealthworkgroup/
Meeting Website
:
http://www.chesapeakebay.net/calendar/event/23800
Slide2Objective
Update and refine the 2011 Chessie BIBI for Chesapeake Bay Program assessment purposes.2
Slide3Outline
Section 1: Master Taxa List
Section 2: Site ClassificationSection 3: RarefactionSection 4: Metric Sensitivity
Section 5: Redundancy Analysis
Section 6: Scoring Protocol
Section 7: Index Resolution
3
Slide4Master Taxa List
TaxaFreshwater Benthic Macroinvertebrates collected in the Chesapeake Bay watershed
222 of families (151 insect families)760 of genera (621 insect genera)
11 taxonomic ranks included (ITIS Link)
Phylum, subphylum, class, subclass, order, suborder, family, subfamily, tribe, genus, and species
4
Slide5Master Taxa List
Common AttributesMunicipal Waste Tolerance
ValuesFunctional
Feeding Groups (FFG)
Habits
EPA Attributes
(
USGS 2006
,
EPA 2012
)
Respiratory Type
Voltinism
Specific Stressors
5
Slide6Master Taxa List
Tolerance ValuesMean tolerance values rounded to the nearest integer
NAs not includedFFGs and HabitsThe most frequent occurring category
If max frequencies were equal the categories were concatenated (e.g. SH, CG or CL, CB)
6
Slide7Tolerance Values (TV)
TV
Frequency
0
35
1
81
2
138
3
108
4
189
5
184
6
262
7
137
8
150
9311029n =1,344
Source%All Taxa% Order% Family% GenusEDAS_MD_TV80018RBP_MID_ATLANTIC_MACS_TV90417EDAS_NC_TV90120EDAS_DE_TV90019RBP_MIDWEST_OH_TV100212DC_TV11121619EDAS_RBP_TV110023EDAS_ITIS_TV1101420EDAS_KY_TV120027EDAS_PA_TV130129RBP_SOUTHEAST_NC_TV160119EDAS_WAB_TV170137EDAS_TV_FINAL2003832EDAS_WV_TV2003632EDAS_FAM_TV2003731EPA_TV36355464NYSDEC_TV53235456BIBI_TV69426979
7
Slide8Functional Feeding Groups (FFG)
FFG
Frequency
CF,SC
1
CF,SH
1
CG,CF,PR,PH
1
CG,PR,PH
1
CG,SC,PR
1
PR,PA
1
SC,PR
1
SH,OM
1
SH,SC
1
CG,CF,PR2CG,SH2PR,PH2SH,PR2
CG,PR3PA3PH4CG,CF5SC,OM5PR,OM7CG,SC10CG,OM12OM16SH135CF145SC172PR423CG443Source% All Taxa% Order% Family% GenusDC_FFG1141419EDAS_KY_FFG130029EDAS_ITIS_FFG1502625EDAS_FINAL1904130EDAS_WV_FFG2103835RBP_P_FFG32213143EPA_FFG41446373NYSDEC_FFG52125455BIBI_FFG725275838
Slide9Habits
HABIT
Frequency
BU,SK
1
CB,BU,SP
1
CB,CN,BU
1
SW,BU
1
CB,BU
2
CB,CN,SP
2
CB,SK
2
SW,SP
2
CB,CN
4
CB,SW4CN,BU5CB,SP7CN,SP7
SW,CN8BU,SP12SK15SW106CB151BU183CN232SP254Source% All Taxa% Order% Family% GenusEDAS_KY_HABIT4009DC_HABIT901218RBP_P_HABIT1541722EDAS__FINAL160235EDAS_WV_HABIT2103434HABIT_PRIM_ABBREV3803071EPA_HABIT41256274BIBI_HABIT512565829
Slide10Unclassified Taxa
Taxa without an assigned attributeFind the percentage of taxa in each sample that are unclassifiedIf
≥ 10% of the taxa are unclassified the sample is removed from the metric calculation
10
Slide11Master Taxa List
Any further comments or question regarding the master taxa list?Do you approve of the averaging tolerance values?
Do you approve of the concatenating FFGs and habits?
Are there any other tables that could be appended?
11
Slide12BIBI Workflow
12
Slide13Site Classification
Classes in each bioregion (7)
Reference
Near Reference
Minor Degradation
Moderate Degradation
Severe Degradation
Mixed
13
Slide14Set reference standards
Standardize the impact of degradationSix Habitat Parameters∑
x < 16 = 1
∑ x
<
5 = 2
Specific
Conductivity
x > 500 = 1
x >
750
= 2
x > 1,000 = 3
pH
x < 6 or x > 8 = 1
x < 5 or x > 9 = 2
Site Classification
14
Slide15Sum the degradation scores
Rate on an even scaleReference: x
= 0Near:
1 ≤ x < 3
Minor Degradation:
3
≤ x < 6
Moderate Degradation:
6
≤ x < 9
Severe Degradation:
x
≥ 9
Site Classification
15
Slide16500
750
1,000
0
1
2
3
5
16
2
1
0
5
6
8
2
1
0
1
9
2
Degradation Score
Specific ConductivityRabid Habitat AssessmentpH16
Slide17Class Gradient
100
50
0
IBI SCORE
REF
NEAR
MIN
MOD
SEV
17
Slide18Biological Condition Gradient
1
4
2
3
5
6
Stressor Gradient
Biological Condition
18
Slide19Biological Condition Gradient
1
4
2
3
5
6
Stressor Gradient
Biological Condition
Reference
Near Ref.
Minor Deg.
Moderate Deg.
Severe Deg.
19
Slide20Metric Calculation
Metrics78 metrics from the literature
Percentage of each taxon> 1,000 metrics testedMost metrics are useless
New Metric
Common
Metric
Bioregion
% Heptageniidae
% Ephemeroptera
Piedmont
%
Systellognatha
% Plecoptera
Ridges
%
Furcatergalia
% Ephemeroptera
Ridges
20
Slide21Rarefaction
Sampling without replacementrrarefy function in the Vegan package
“The random
rarefaction is made without replacement so that the variance of rarefied communities is rather related to rarefaction
proportion
than to
the
size of the sample
.”
-
Oksanen
and O’Hara
21
Slide22Frequency
Community Composition
Rarefaction
22
Slide23Family
Count
% Abundance
Rarefied % Abundance
1
EPHEMERELLIDAE
488
33.38
33
2
HYDROPSYCHIDAE
285
19.49
19
3
CHIRONOMIDAE
142
9.71
10
4
TAENIOPTERYGIDAE
1087.3975BAETIDAE94
6.4366LEPIDOSTOMATIDAE714.8657LEPTOPHLEBIIDAE654.4548TIPULIDAE654.4549HEPTAGENIIDAE543.69410CHLOROPERLIDAE191.30111ISONYCHIIDAE181.23112SPERCHONTIDAE181.23113PHILOPOTAMIDAE130.89114POLYCENTROPODIDAE130.89115EMPIDIDAE90.62116PERLODIDAE910010017CAPNIIDAE618HYDROPTILIDAE619GLOSSOSOMATIDAE420LUMBRICULIDAE421ATHERICIDAE222PERLIDAE223PSEPHENIDAE2
24
LEUCTRIDAE
1
25
ODONTOCERIDAE
1
26
PLANARIIDAE
1
27
RHYACOPHILIDAE
1
28
SIMULIIDAE
1
23
Slide24Min = 8 taxa
Median = 15 taxa
Max = 23 taxa
24
Slide25Min = 1.56
Median = 2.12Max = 2.60
25
Slide26Min = 48%
Median = 69%Max = 90%True = 75%
n = 1052nr
= 100
26
Slide2727
Slide28Rarefaction Discussion
28
Slide29Metric Sensitivity
Objective: Identify metrics that best reflect a disturbance gradient
Barbour et al. 1996
Quantile Threshold Confusion Matrix Accuracy (CMA)
29
Slide30Barbour et al. 1996
3
points
2 points
1 point
0 points
Reference
Degraded
30
Slide31Quantile Threshold
Frequency
Metric Value
2
5%
x 100
31
Slide32Confusion Matrix Accuracy (CMA)
Frequency
Metric Value
15%
32
Slide33Confusion Matrix
Predicted Class
Actual Class
Reference
Degraded
Reference
True
Reference
(
TR)
False
Degraded
(
FD)
Degraded
False
Reference (FR)
True
Degraded
(
TD) 33
Slide34Metric Sensitivity Discussion
34
Slide3535
Slide36Redundancy Analysis
Spearman Correlation (
≤ -0.7
or ≥ 0.7
)
Two methods:
Wilcoxon Rank Sum Test
Pairwise comparison between
Reference
and
Severe Degradation
The metric with the smaller p-value is held for further review
CMA Pairwise Gradient
Class 1
Class 2
CMA
Reference
Near Reference
73
Near Reference
Minor
Degradation64Minor DegradationModerate Degradation
57Moderate DegradationSevere Degradation77Severe DegradationReference92Mean72.636
Slide37Redundancy Analysis
Spearman Correlation (
≤ -0.7
or ≥ 0.7
)
Two methods:
Wilcoxon Rank Sum Test
Pairwise comparison between Reference and Severe Degradation
The metric with the smaller p-value is held for further review
CMA Pairwise Gradient
Class 1
Class 2
CMA
Reference
Near Reference
73
Near Reference
Minor
Degradation
64
Minor
DegradationModerate Degradation57Moderate DegradationSevere Degradation77
Severe DegradationReference92Mean72.637
Slide38Redundancy Analysis Discussion
38
Slide3939
Slide40Scoring Metrics
Categorical ScoringThree categories: 1, 3, and 5
Metric decreases with degradationx > 50% = 5
25% ≤ x ≤ 50% =
3
x
< 25% =
1
Metric
increases
with degradation
x
<
50% =
5
50%
≤ x ≤
75%
=
3
x > 75%
= 1
FrequencyMetric Value15340
Slide41Gradient Scoring
Reference Gradient vs. All GradientDecreasex ≤
5% = 0
x ≥ 95%
= 100
5% < x < 95%
Increase
x ≤
5%
=
100
x ≥
95%
=
0
5% < x < 95%
* 100
* 100
41
Slide42Frequency
Metric Value
0
100
1 - 99
Frequency
Metric Value
0
100
1 - 99
Reference Gradient
All Gradient
42
Slide43Threshold Gradient
Similar to the previous gradient approachesDecrease
x ≥
50%
=
100
x
≤
threshold%
=
0
threshold%
< x <
50%
Increase
x ≤
50%
= 100
x ≥
threshold
% = 050% < x < threshold% * 100 * 100 43
Slide44Frequency
Metric Value
0
100
1 - 99
Threshold %
50%
Threshold Gradient
44
Slide45Threshold Gradient
80
4
0
100
60
0
IBI SCORE
REF
NEAR
MIN
MOD
SEV
20
45
Slide46Scoring Protocol Discussion
46
Slide47Metric Sensitivity Approach
Barbour et al. 1996
Confusion Matrix Accuracy (CMA)
Quantile Threshold
Categorical
Ref. Gradient
All Gradient
Scoring Approach
Threshold Gradient
Redundancy Analysis Approach
Wilcoxon Rank Sum Test
Pairwise Confusion Matrix Accuracy (CMA)
47
Slide48Metric Sensitivity Approach
Barbour et al. 1996
Confusion Matrix Accuracy (CMA)
Quantile Threshold
Categorical
Ref. Gradient
All Gradient
Scoring Approach
Threshold Gradient
Redundancy Analysis Approach
Wilcoxon Rank Sum Test
Pairwise Confusion Matrix Accuracy (CMA)
48
Slide4949
Slide50Metric Sensitivity Approach
Barbour et al. 1996
Confusion Matrix Accuracy (CMA)
Quantile Threshold
Categorical
Ref. Gradient
All Gradient
Scoring Approach
Threshold Gradient
Redundancy Analysis Approach
Wilcoxon Rank Sum Test
Pairwise Confusion Matrix Accuracy (CMA)
50
Slide5151
Slide52Metric Sensitivity Approach
Barbour et al. 1996
Confusion Matrix Accuracy (CMA)
Quantile Threshold
Categorical
Ref. Gradient
All Gradient
Scoring Approach
Threshold Gradient
Redundancy Analysis Approach
Wilcoxon Rank Sum Test
Pairwise Confusion Matrix Accuracy (CMA)
52
Slide5353
Slide5443
1,073
445
49
6,663
415
n =
54
Slide5543
1,073
445
49
6,663
415
n =
55
Slide5656
43
1,073
445
49
6,663
415
n =
Slide5757
Slide5858
CMA Ref. Threshold
CMA Wilcoxon
DE Wilcoxon
Barbour Wilcoxon
1
Menhinicks
% Retreat_Trichop
% Hyrop_EPT
% Retreat_Trichop
2
HBI
%
Mod_Tol
HBI
% Mod_Tol
3
% EPT_Hydro_Baetid
% EPT
% EPT
% EPT4% Corydalidae% Corydalidae % Corydalidae % Corydalidae5% Heptageniidae% Pisciforma
% EPT_Rich_No_Tol% Pisciforma6% Isonychiidae% Isonychiidae% Isonychiidae% Isonychiidae7% Pterygota% Non_Insect% Non_Insect% Non_Insect8% Anisoptera% Odonata% Limestone% Trichop_No_Tol9% Perlidae% Systellognatha% Unionoida% Sytellognatha“Best” Metrics Piedmont
Slide5959
CMA Ref. Threshold
CMA Wilcoxon
DE Wilcoxon
Barbour Wilcoxon
1
Menhinicks
%
Retreat_Trichop
% Hyrop_EPT
%
Retreat_Trichop
2
HBI
%
Mod_Tol
HBI
%
Mod_Tol
3
% EPT_Hydro_Baetid% EPT% EPT% EPT
4% Corydalidae% Corydalidae % Corydalidae % Corydalidae5% Heptageniidae% Pisciforma% EPT_Rich_No_Tol% Pisciforma6% Isonychiidae% Isonychiidae% Isonychiidae% Isonychiidae7% Pterygota% Non_Insect% Non_Insect% Non_Insect8% Anisoptera% Odonata% Limestone% Trichop_No_Tol9% Perlidae% Systellognatha% Unionoida% Sytellognatha“Best” Metrics Piedmont
Slide6043
1,073
445
49
6,663
415
n =
60
Slide6161
43
1,073
445
49
6,663
415
n =
Slide6262
43
1,073
445
49
6,663
415
n =
Slide6363
43
1,073
445
49
6,663
415
n =
Slide6464
43
1,073
445
49
6,663
415
n =
Slide65Binary Sensitivity
65
Slide6666
Slide6767
Slide6868
Slide6969
Binary
Threshold CMA
Slide70Index Resolution Type
Low ResolutionOrder
Level IndexVolunteer programs and in situ identificationModerate Resolution
Family
Level Index
Studies with limited funding and/or minimal macroinvertebrate identification experience.
70
Slide71Index Ranks
High ResolutionGenus
Level Index
Studies with extensive funding and macroinvertebrate identification experience.Secondary indices identifying specific stressors
71
Slide72Metric Sensitivity Approach
Barbour et al. 1996
Confusion Matrix Accuracy (CMA)
Quantile Threshold
Categorical
Ref. Gradient
All Gradient
Scoring Approach
Threshold Gradient
Redundancy Analysis Approach
Wilcoxon Rank Sum Test
Pairwise Confusion Matrix Accuracy (CMA)
72
Slide73Order
Family
Genus
73
Slide74Final Remarks
Test for agency differences within each bioregionSampling period differencesDifferences in taxa identified
Deal with unidentified taxa and unclassified taxaMissing tolerance values, FFG, or habits
Compare the 2011 indices to the 2016 indices
74