Chapter 7 Calculate Mussel Metrics

This chapter will describe how to calculate metrics of habitat quality for each sampled site. Each of these metrics will be used as presence data as input to create a set of Maxent habitat suitability models. The metrics calculated in this section were taken from the Upper Mississippi River Mussel Community Assessment Tool (MCAT) (Dunn, Zigler, and Newton 2016).

HREP Mussel Modeling Workflow, Calculate MCAT Metrics Step.

Figure 7.1: HREP Mussel Modeling Workflow, Calculate MCAT Metrics Step.

7.1 MCAT Metrics

The purpose of this step is to describe the MCAT metrics used in this study. The following MCAT metrics were selected for inclusion in the model:

  • percent listed species - Percent of listed threatened or endangered species (federal or bordering states) is a measure of sensitive species.
  • percent tolerant - Percent of tolerant species (Amblema plicata, Quadrula quadrula, and Obliquaria reflexa) is a measure of a highly disturbed mussel assemblage (i.e., dominated by species tolerant of unstable substrates, silt accumulation, low current velocities, and fluctuating flow conditions).
  • percent tribe Lampsilini - Percent of assemblage that falls within tribe Lampsilini is a measure of species composition, life history, and behavioral characteristics.
  • percent juveniles - Percent of mussels <= 5 years-old is a measure of recruitment into an assemblage over the past five years.
  • percent >= 15 yrs - Percent of mussels >= 15 years-old is a measure of age distribution in an assemblage.
  • abundance - A measure of abundance calculated as the density (no./\(m^2\)).
  • species evenness - Species evenness represents the dominance of an assemblage by a few species using Pielou’s evenness index (range 0 to 1).
  • tribe evenness - Tribe evenness represents the dominance of a particular taxonomic group using Pielou’s evenness index (range 0 to 1).
  • ES_100 - The expected number of species with a sample size of 100 estimated by rarefaction is a measure of a healthy mussel assemblage.

7.2 Include Live Mussels Only Prior to Metrics Calculations

Create a definition query on the Cordova_all_individuals feature class where `NumberLive > 0.

7.3 Create Percent Listed

The purpose of this step is to calculate the percent listed MCAT metric for each sampled site.

  • Determine which mussels in the Cordova_all_individuals feature class are listed and set the flag field. Use the list below to select each of the following species and set the flag Listed field equal to one.

UMR Listed Species

  • Cumberlandia monodonta
  • Elliptio crassidens
  • Elliptio dilatata
  • Ellipsaria lineolata
  • Fusconia ebena
  • Lampsilis higginsii
  • Lampsilis teres teres
  • Lampsilis teres anodontoides
  • Ligumia recta
  • Plethobasus cyphyus
  • Potamilus capax

  • Simpsonaias ambigua

  • Dissolve the Cordova_all_individuals feature class by SAMPLE_ID. Add SUM_NumberLive and SUM_Listed fields in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it PercentListed.
  • Open the PercentListed attribute table. Add a field titled perc_listed Type = double.

  • Use the Field Calculator tool to calculate the percent of individual listed species per sampled quadrat using the expression ([SUM_Listed] / [SUM_NumberLive]) * 100.

7.4 Create Percent Tolerant

The purpose of this step is to calculate the percent tolerant MCAT metric for each sampled site.

  • Determine which mussels in the Cordova_all_individuals feature class are considered tolerant and set the flag field. Use the list below to select each of the following species and set the flag Tolerant field equal to one.

Tolerant Species

  • Amblema plicata
  • Obliquaria reflexa

  • Quadrula quadrula

  • Dissolve the Cordova_all_individuals feature class by SAMPLE_ID. Add SUM_NumberLive and SUM_Tolerant fields in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it PercentTolerant.
  • Open the PercentTolerant attribute table. Add a field titled perc_tolerant Type = double.

  • Use the Field Calculator tool to calculate the percent of individual tolerant species per sampled quadrat using the expression ([SUM_Tolerant] / [SUM_NumberLive]) * 100.

7.5 Create Percent Tribe Lampsilini

The purpose of this step is to calculate the percent tribe Lampsilini MCAT metric for each sampled site.

  • Determine which mussels in the Cordova_all_individuals feature class are in the Lampsilini tribe and set the flag field. Use the list below to select each of the following species and set the flag Lampsilini field equal to one.

Lampsilini Tribe

  • Actinonaias ligamentina
  • Ellipsaria lineolata
  • Lampsilis cardium
  • Lampsilis higginsii
  • Lampsilis teres (teres)
  • Leptodea fragilis
  • Ligumia recta
  • Obliquaria reflexa
  • Obovaria olivaria
  • Potamilus alatus
  • Potamilus capax
  • Potamilus ohiensis
  • Toxolasma parvum
  • Truncilla donaciformis

  • Truncilla truncata

  • Dissolve the Cordova_all_individuals feature class by SAMPLE_ID. Add SUM_NumberLive and SUM_Lampsilini fields in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it PercentLampsilini.
  • Open the PercentLampsilini attribute table. Add a field titled perc_lampsilini Type = double.

  • Use the Field Calculator tool to calculate the percent of individual tolerant species per sampled quadrat using the expression ([SUM_Lampsilini] / [SUM_NumberLive]) * 100.

7.6 Create Percent Juveniles

The purpose of this step is to calculate the percent juveniles MCAT metric for each sampled site.

  • Determine which mussels in the Cordova_all_individuals feature class are <= 5 years-old. Sort the Age field in ascending order to select mussels aged 5 and under. Use the Field Calculator to set the flag Juveniles field equal to one.
  • Dissolve the Cordova_all_individuals feature class by SAMPLE_ID. Add SUM_NumberLive and SUM_Juveniles fields in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it PercentJuveniles.
  • Open the PercentJuveniles attribute table. Add a field titled perc_juveniles Type = double.
  • Use the Field Calculator tool to calculate the percent of individual tolerant species per sampled quadrat using the expression ([SUM_Juveniles] / [SUM_NumberLive]) * 100.

7.7 Create Percent Over 15 Years

The purpose of this step is to calculate the percent over 15 years old MCAT metric for each sampled site.

  • Determine which mussels in the Cordova_all_individuals feature class are >= 15 years-old. Sort the Age field in ascending order to select mussels aged 15 and over. Use the Field Calculator to set the flag Over_15yrs field equal to one.
  • Dissolve the Cordova_all_individuals feature class by SAMPLE_ID. Add SUM_NumberLive and SUM_Over_15yrs fields in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it PercentOver15yrs.
  • Open the PercentOver15yrs attribute table. Add a field titled perc_over_15yrs Type = double.
  • Use the Field Calculator tool to calculate the percent of individual tolerant species per sampled quadrat using the expression ([SUM_Over_15yrs] / [SUM_NumberLive]) * 100.

7.8 Create Abundance

The purpose of this step is to calculate the Abundance MCAT metric for each sampled site.

  • For the abundance metric, dissolve Cordova_all_individuals by SAMPLE_ID. Add ‘SUM_NumberLive’ field in the Statistics section of the tool. Uncheck create multipart features. Set Output Feature Class to the working .gdb and name it Abundance.
# Import Abundance from geodatabase
abundance_sp <- arc2sp("//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/Abundance")

# Convert sp object to a data frame
abundance <- abundance_sp@data

# Remove OBJECTID field
abundance <- abundance[,-1]
# Convert from num. of mussels per 0.25 sq m to num. per 1.0 square m
abundance$num_sq_m <- abundance$SUM_NumberLive * 4

# Calculate the quantiles
q <- quantile(abundance$num_sq_m)
q
##   0%  25%  50%  75% 100% 
##    4    4    8   20  180
# Merge the abundance scores back onto the sp object
sample_abundance <- sp::merge(x = abundance_sp,
                              y = abundance,
                              by.x = "SampleID", by.y = "SampleID")
# Export to geodatabase
sp2arc(sample_abundance, fc_path = "//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/sample_abundance")

7.9 Create Species Evenness

The purpose of this step is to calculate the Pielou’s evenness index (range 0-1), estimated at the species level for each sampled site.

  • Create a table of SampleID by species by abundance (number live). Use the Dissolve tool:
    • Input Features: Cordova_all_individuals
    • Output feature class: Cordova_sampleid_species
    • Dissolve Fields: SampleID, EName
    • Statistics Fields: NumberLive, SUM
    • Create Multipart: unchecked

Ensure that the Cordova_sampleid_species feature class is of type Point Features and not Multipoint Features. If multipoint, then use the Multipart to Singlepart tool to convert from multipoint to point feature type.

# Import ArcGIS feature class into R
cordova_sampleid_species <- arc2sp("//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/Cordova_sampleid_species")

# Convert sp object to a data frame
sample_species <- cordova_sampleid_species@data

# Remove OBJECTID field
sample_species <- sample_species[,-1]

# Convert to vegan community data matrix-like format using labdsv::matrify
sample_species_matrix <- labdsv::matrify(sample_species)
# Calculate the Shannon-Weaver diversity index
shannon_diversity <- vegan::diversity(sample_species_matrix, index = "shannon")

# Calculate the number species per site (see ?vegan::diversity)
species_number <- vegan::specnumber(sample_species_matrix)

# Use the following equation to calculate Pielou's evenness
# See the vegan Diversity Vignette for details
pielou_evenness <- shannon_diversity/log(species_number)

# Set NaN pielou_evenness values to zero
pielou_evenness <- ifelse(is.nan(pielou_evenness), 0, pielou_evenness)
# Create a data frame of results
pielou <- data.frame(sampleid = names(pielou_evenness), 
                     shannon_diversity, 
                     species_number, 
                     pielou_evenness)
# Merge the Pielou scores back onto the sp object
sample_species_pielou <- sp::merge(x = cordova_sampleid_species,
                                   y = pielou,
                                   by.x = "SampleID", by.y = "sampleid")
# Export to geodatabase
sp2arc(sample_species_pielou, fc_path = "//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/sample_species_pielou")
  • Create a table of SampleID by species Pielou evenness score. Use the Dissolve tool:
    • Input Features: sample_species_pielou
    • Output feature class: species_pielou
    • Dissolve Fields: SampleID
    • Statistics Fields: shannon_diversity, MEAN species_number, MEAN pielou_evenness, MEAN
    • Create Multipart: unchecked

7.10 Create Tribe Eveness

The purpose of this step is to calculate the Pielou’s evenness index (range 0-1), estimated at the tribe level. (will probably need to add a new column for tribe; dissolve on this with sample ID)

  • In the feature class Cordova_all_individuals, create a new text variable named tribe.
  • Use the Field Calculator tool on the tribe field to calculate its value using the following Python expression: !Ename!.split()[0]
  • Create a table of SampleID by tribe by abundance (number live). Use the Dissolve tool:
    • Input Features: Cordova_all_individuals
    • Output feature class: Cordova_sampleid_tribe
    • Dissolve Fields: SampleID, tribe
    • Statistics Fields: NumberLive, SUM
    • Create Multipart: unchecked
# Import ArcGIS feature class into R
cordova_sampleid_tribe <- arc2sp("//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/Cordova_sampleid_tribe")

# Convert sp object to a data frame
sample_tribe <- cordova_sampleid_tribe@data

# Remove OBJECTID field
sample_tribe <- sample_tribe[,-1]

# Convert to vegan community data matrix-like format using labdsv::matrify
sample_tribe_matrix <- labdsv::matrify(sample_tribe)
# Calculate the Shannon-Weaver diversity index
tribe_shannon_diversity <- vegan::diversity(sample_tribe_matrix, index = "shannon")

# Calculate the number species per site (see ?vegan::diversity)
tribe_number <- vegan::specnumber(sample_tribe_matrix)

# Use the following equation to calculate Pielou's evenness
# See the vegan Diversity Vignette for details
tribe_pielou_evenness <- tribe_shannon_diversity/log(tribe_number)

# Set NaN pielou_evenness values to zero
tribe_pielou_evenness <- ifelse(is.nan(tribe_pielou_evenness), 0, 
                                tribe_pielou_evenness)
# Create a data frame of results
tribe_pielou <- data.frame(sampleid = names(tribe_pielou_evenness), 
                           tribe_shannon_diversity, 
                           tribe_number, 
                           tribe_pielou_evenness)
# Merge the Pielou scores back onto the sp object
sample_tribe_pielou <- sp::merge(x = cordova_sampleid_tribe,
                                 y = tribe_pielou,
                                 by.x = "SampleID", by.y = "sampleid")
# Export to geodatabase
sp2arc(sample_tribe_pielou, fc_path = "//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/sample_tribe_pielou")
  • Create a table of SampleID by tribe Pielou evenness score. Use the Dissolve tool:
    • Input Features: sample_tribe_pielou
    • Output feature class: tribe_pielou
    • Dissolve Fields: SampleID
    • Statistics Fields: tribe_shannon_diversity, MEAN tribe_number, MEAN tribe_pielou_evenness, MEAN
    • Create Multipart: unchecked

7.11 Create ES 100

The purpose of this step is to calculate the ES 100 MCAT metric for each sampled site. ES_100 is the expected number of species with a sample size of 100 estimated by rarefaction based on random resampling of the data (Dunn, Zigler, and Newton 2016).

ES_100-species richness estimated by rarefaction; caveat: sites need to be compared based on an equal sample size because # of species and # of individuals sampled are large correlated. Not sure how best to approach this one; I defer to you!

# Uses the sample_species_matrix calculated from the species evenness section

# Calculate rarefaction using a sample size of 100
rarefy_es_100 <- vegan::rarefy(x = sample_species_matrix, 
                               sample = 100)
## Warning in vegan::rarefy(x = sample_species_matrix, sample = 100):
## requested 'sample' was larger than smallest site maximum (0)
# Calculate rarefaction using a reasonable sample size
rarefy_rowmedian <- vegan::rarefy(x = sample_species_matrix,
                                  sample = median(rowSums(sample_species_matrix)))
## Warning in vegan::rarefy(x = sample_species_matrix, sample =
## median(rowSums(sample_species_matrix))): requested 'sample' was larger than
## smallest site maximum (0)
# Create a data frame of results
rarefy_species <- data.frame(sampleid = names(rarefy_es_100), 
                             rarefy_es_100,
                             rarefy_rowmedian,
                             shannon = diversity(sample_species_matrix),
                             specnumber = specnumber(sample_species_matrix))
# Merge the rarefaction scores back onto the sp object
rarefy_sampleid_species <- sp::merge(x = cordova_sampleid_species,
                                     y = rarefy_species,
                                     by.x = "SampleID", by.y = "sampleid")
# Export to geodatabase
sp2arc(rarefy_sampleid_species, fc_path = "//mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/rarefy_sampleid_species")
  • Create a table of SampleID by tribe Pielou evenness score. Use the Dissolve tool:
    • Input Features: rarefy_sampleid_species
    • Output feature class: rarefy_samples
    • Dissolve Fields: SampleID
    • Statistics Fields: SUM_NumberLive, MEAN rarefy_es_100, MEAN rarefy_rowmedian, MEAN shannon, MEAN specnumber, MEAN
    • Create Multipart: unchecked

7.12 MCAT Metric Site Level Thresholds

The descriptive statistics presented in the previous section were used to select MCAT metrics thresholds.

Table 7.1: Translating MCAT Metric Thresholds from the Bed to the Site Scale.
Bed Scale
Site Scale
Metrics Poor Fair Good Site Threshold No. of Sites
% listed <0.6 0.6-3.6 >3.6 >3 34
% tolerant >62.7 38.3-62.7 <38.3 <40 72
% lampsilini <17.2 or >56.4 <17.2-34.7 or >39.5-56.4 >34.7-39.5 >40 93
% juveniles <19.8 <19.8-49.3 >49.3 >50 72
% >= 15 years <0.8 or >16.0 >5.6-16.0 >2.4-5.6 >5* 13
abundance <8 8-13 >13 >13 62
species evenness <0.665 0.665-0.780 >0.780 >0.7 98
tribe evenness <0.719 0.719-0.823 >0.823 >0.8 86
ES 100 <11.5 11.5-15.7 >15.7 >3* 49
* The site threshold for this metric is lower than the bed threshold due to the lower number of individuals of each species found at each sampled site.

7.13 Calculate MCAT Site Score

The purpose of this section is to calculate an MCAT Site score. This score is composed of the number of MCAT metrics met for each site. A site that meets none of the MCAT metrics receives a score of zero while a site that meets all of the scores receives a score of nine.

Frequency of MCAT Site Scores.

Figure 7.2: Frequency of MCAT Site Scores.

7.14 Export to Maxent “SWD” format

The purpose of this step is to assign the ADH predictor variables (and optionally the wind/wave variables) to each MCAT metric. The Maxent documentation refers to this format as the “Samples With Data” (SWD) format because it contains the samples (i.e., sample identifier, latitude, longitude) with the environmental predictor variables (e.g., q5_velocity, q5_depth, q5_slope, etc.).

  • In the samples_mcat feature class, add 3 fields named:

  • species Type = string and Length = 20
  • longitude Type = double

  • latitude Type = double

  • Use the Field Calculator tool to populate the species field using the following expression: species = "mcat".
  • Use the Calculate Geometry tool to populate the longitude and latitude fields, ensuring that the coordinate system is set to NAD 1983 2011 US Feet and StatePlane Illinois West FIPS 1202 and the units are set to US feet.

  • Use the Extract Multi Values to Points (spatial analyst) tool to write the raster values (q5_velocity, q5_depth, q5_slope, q5_ss, q5_reynolds, q5_froude, q95_velocity, q95_depth, q95_slope, q95_ss, q95_reynolds, q95_froude) to each point in the samples_mcat feature class and the background feature class (Section 5.2 of this document).

    • Input point features: //mvrdfs/EGIS/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Data/SB_Mussels.gdb/SB_ILStatePlaneW/samples_mcat
    • Input rasters: q5_velocity, q5_depth, q5_slope, q5_ss, q5_reynolds, q5_froude, q95_velocity, q95_depth, q95_slope, q95_ss, q95_reynolds, q95_froude
    • Repeat the previous step for background

  • Ensure all layers have same spatial reference before running the tool.

  • Select only records from the samples_cat feature class where the mcat_site_score >= 4. Create a definition query using the following expression: mcat_site_score >= 4

  • Export these selected records to a .csv files named mcat.csv to the following folder: //mvrdfs/egis/Work/EMP/HREP_Projects/SteamboatSlough/Mussels/SteamboatMusselModel/Maxent/existing_condition/input_swd

  • Repeat the Extract Multi Values to Points step for the background feature class to add environmetal predictor to the background points.
  • Export the background attribute table as a .csv file to the input_swd folder listed above.

  • Ensure the ObjectID field is deleted in Excel prior to use in Maxent.

References

Dunn, Heidi, Steve Zigler, and Teresa Newton. 2016. “Validation of a Mussel Community Assessment Tool for the Upper Mississippi River System.” 2014 MCA2. Rock Island, IL: U.S. Army Corps of Engineers, Rock Island District.