1. Zooplankton surveys
The Interagency Ecological Program (IEP) is a consortium of State
and federal agencies that has been conducting cooperative ecological
investigations since the 1970s. The IEP runs over twenty long-term
monitoring surveys on biological components of the Upper San
Francisco Estuary. Surveys monitor phytoplankton, zooplankton,
benthic invertebrates, water quality, and many types of fish.
Several fish surveys sample zooplankton concurrently, and
information on zooplankton species composition and abundance can be
coupled with fish diet studies. The IEP long-term surveys that
monitor zooplankton are the Environmental Monitoring Program (EMP;
also known as the IEP zooplankton study), 20-mm Survey (20mm), Fall
Midwater Trawl (FMWT), Summer Townet Survey (STN), the Yolo Bypass
Fisheries Monitoring Survey (not included in this integrated
dataset), and the Fish Restoration Program (FRP). An overview of
these programs is provided in study_metadata.csv. Locations of fixed
sampling locations for surveys with fixed sampling designs are
provided in stations.csv, and coordinates for every sample are
provided in environment.csv. Zooplankton surveys sample 3 different
size classes of zooplankton, by towing nets with different sized
mesh (or in one case a pump that pumps water into a microzooplankton
net). Typically, the zooplankton targeted by these different
sampling instruments are:
Every IEP survey that collects zooplankton samples with a
mesozooplankton net, which targets adult copepods and cladocerans,
because these taxa are believed to comprise the majority of
zooplankton in juvenile and adult planktivorous fish diets. Some
surveys also sample with micro- or macro-zooplankton nets.
1.1. Environmental Monitoring Program
The Environmental Monitoring Program (EMP) Zooplankton Study (also
known as the IEP Zooplankton Study) began in 1972 in order to assess
trends in fish food resources ranging from San Pablo Bay to the east
Delta, as well as to detect and assess the impacts of recently
introduced zooplankton species on native species. The study is
mandated by Water Right Decision 1641 for operation of the State
Water Project and Central Valley Project (SWRCB 2000). The EMP study
is conducted by the California Department of Fish and Wildlife
(CDFW), California Department of Water Resources (DWR), and the US
Bureau of Reclamation (Reclamation) and currently samples 17 fixed
stations and 2 floating entrapment zone (EZ) stations (locations
where the bottom salinity is 2 and 6 PPT). There are also 3
additional stations located in Carquinez Strait and San Pablo Bay,
which are sampled during periods of high outflow and low salinity.
Historically (prior to 1995) the survey sampled at a much larger
number of stations.
Zooplankton for EMP are sampled in 3 different size ranges:
microzooplankton are sampled using a pump, mesozooplankton are
sampled using a modified Clarke-Bumpus net, and macrozooplankton are
sampled using a mysid net. All EMP zooplankton are collected monthly
at fixed stations year-round in open channels at high slack tide and
preserved in 10% formalin dyed with rose bengal. Macrozooplankton
and Mesozooplankton are collected using 10-minute oblique tows with
a 124 cm long net with 505 µm mesh, and a 73 cm long net with 160 µm
mesh, respectively. Prior to 1974, macrozooplankton were sampled
with a 930 µm mesh net. Microzooplankton are collected with a Teel
marine pump while the intake hose is raised through the water column
and pumped into a net with 43 µm mesh. Pump samples collected
approximately 1.5 - 1.9 L from 1972 – 2007, and 75 L from 2008 -
present, measured by a digital flowmeter connected to the hose.
Microzooplankton samples are passed through a 154 µm mesh sieve
nested on top of a 43 µm mesh sieve in the lab, and only the smaller
size fraction that passes through the larger sieve and is retained
on the smaller sieve is counted. Lengths are recorded for
macrozooplankton (but not included in this dataset), and biomass is
estimated by length-weight equations for macrozooplankton and by
average values for mesozooplankton and microzooplankton
(biomass_mesomicro.csv). Recorded environmental variables for all
samples include time, depth, surface and bottom conductivity,
surface temperature, Secchi depth, and chlorophyll-a.
More information on EMP and its methods can be found on the EMP
zooplankton study website
(https://wildlife.ca.gov/Conservation/Delta/Zooplankton-Study) or
zooplankton data publication (Barros 2022), or environmental data
publication (Battey and Perry 2022).
1.2. 20mm Survey
The 20-mm survey was initiated in 1995 by the California Department
of Fish and Wildlife to monitor postlarval-juvenile Delta Smelt
(Hypomesus transpacificus) distribution, abundance, and timing
throughout their historical spring range in the Delta. The survey is
mandated under the Endangered Species Act Biological Opinion for
operation of the State and Central Valley water projects (USFWS
2019). 20-mm refers to the length of the fish targeted by the net.
Zooplankton samples are collected concurrently with fish samples to
monitor Delta Smelt food supply. Between 41 and 55 stations have
been sampled each year since the survey began.
Zooplankton are sampled twice per month between March and July at
fixed stations in open channels. Mesozooplankton are sampled using
10-minute stepped-oblique tows with a 73 cm long 160 µm mesh
modified Clarke-Bumpus net. The net is attached to the top of the
20-mm net frame and a flowmeter is mounted in the mouth. Samples are
preserved in 10% formalin. Lengths are not recorded, and biomass is
estimated by literature values. Recorded environmental variables
include times, tidal stage, depth, surface and bottom conductivity,
surface temperature, Secchi depth, and turbidity.
More information on the 20-mm survey and its methods can be found on
the 20-mm survey website (California Department of Fish and Wildlife
2021).
1.3. Fall Midwater Trawl
The Fall Midwater Trawl (FMWT) was initiated by the California
Department of Fish and Wildlife in 1967 in order to determine the
relative abundance and distribution of age-0 Striped Bass (Morone
saxatilis), but the data has also been used for other upper estuary
pelagic fish species, including Delta Smelt (Hypomesus
transpacificus), Longfin Smelt (Spirinchus thaleichthys), American
Shad (Alosa sapidissima), Splittail (Pogonichthys macrolepidotus),
and Threadfin Shad (Dorosoma petenense). The FMWT is currently
mandated by the 2019 Delta Smelt Biological Opinion for the
coordinated operation of the Central Valley Project and State Water
Project (USFWS 2019). The FMWT samples 122 stations each month from
September to December ranging from San Pablo Bay to Stockton, Hood,
and The Sacramento Deep Water Ship Channel. FMWT samples both
macrozooplankton and mesozooplankton at a subset of these stations
since 2011, with some pilot studies in earlier years.
Zooplankton samples are collected along with the fish trawl at fixed
stations in open channels using 10-minute oblique tows.
Macrozooplankton are sampled using a 124 cm long net with 505 µm
mesh, while mesozooplankton is sampled using a 73 cm long modified
Clark-Bumpus net with 160 µm mesh. For both zooplankton sizes,
samples are preserved in 10% formalin dyed with rose bengal. Lengths
are recorded for macrozooplankton but not mesozooplankton, biomass
is estimated for both as in EMP. Recorded environmental variables
include time, tidal stage, depth, surface and bottom conductivity,
surface and bottom temperature, Secchi depth, Microcystis presence,
and turbidity.
More information on FMWT and its methods can be found on the FMWT
data publication (Burdi et al. 2022).
1.4. Summer Townet Survey
The Summer Townet Survey (STN) was initiated by the California
Department of Fish and Wildlife in 1959 in order to determine the
relative abundance and distribution of upper estuary pelagic
species, namely age-0 Striped Bass (Morone saxatilis). As with the
FMWT, the STN is currently mandated by the 2019 Delta Smelt
Biological Opinion (USFWS 2019) and began in response to the
development of the Central Valley Project pumping plants. The Summer
Townet Survey collects mesozooplankton samples from 32 historic
stations and 8 supplemental stations ranging from San Pablo Bay to
Rio Vista, Stockton, Cache Slough, and the Deep-Water Ship Channel.
Zooplankton monitoring began in 2005 with samples collected every 2
weeks between June and August.
STN samples only mesozooplankton during their fish trawl with a net
attached to the townet frame. Zooplankton samples are collected
during 1 of the fish tows at each fixed station in open channels
using 10-minute oblique tows. Mesozooplankton are sampled using a 73
cm long modified Clark-Bumpus net with 160 µm mesh and preserved in
10% formalin dyed with rose bengal. Biomass is estimated and
recorded environmental variables include time, tidal stage, depth,
surface and bottom conductivity, surface and bottom temperature,
Secchi depth, Microcystis presence, and turbidity.
More information on STN and its methods can be found on the STN data
publication (Burdi et al. 2022).
1.5. Fish Restoration Program
The Fish Restoration Program (FRP) is devoted to restoring 8,000
acres of tidal habitat in the Delta and Suisun Marsh to provide
Delta Smelt habitat and 800 acres of low salinity habitat to benefit
Longfin Smelt. These restoration projects are pursuant to
requirements in the 2019 Biological Opinions for state and federal
water project operations (USFWS 2019). The FRP Monitoring Team
monitors fish and their food resources (including zooplankton)
within these restored wetlands in order to better understand the
benefits of the restored habitats to native fish species. The FRP
Monitoring Team surveys zooplankton in shallow waters, generally
near tidal marshes or sites that will soon be converted to tidal
marsh. The FRP has worked closely with some other IEP surveys to
compare zooplankton communities in shallow water to the open-water
channel samples collected by the long-term surveys (Contreras et al.
2018).
Zooplankton are sampled annually to monthly between March and
December beginning in 2015. Samples are taken from haphazardly
selected locations within fixed sites at restored and existing
wetlands and adjacent open-water areas across the Delta and Suisun
Marsh. Macrozooplankton are collected with 10-minute horizontal
surface tows using a 0.4 m x 0.4 m mouth net (500 µm mesh size).
Mesozooplankton are collected with 5-minute surface tows using a
14.6 cm diameter net (150 µm mesh size). A flowmeter is attached to
the net for both zooplankton size collections. Samples are preserved
in 70% ethanol with rose bengal. Lengths are recorded for
macrozooplankton but not mesozooplankton, biomass is estimated by
literature values for both. Recorded environmental variables include
time, tidal stage, surface conductivity, surface temperature, Secchi
depth, turbidity, Microcystis, pH, chlorophyll, and dissolved
oxygen.
More information on FRP and its methods can be found on the FRP data
publication (California Department of Fish and Wildlife et al.
2019).
1.6. Yolo Bypass Fish Monitoring Program
The Yolo Bypass Fish Monitoring Program (YBFMP) was initiated by the
California Department of Water Resources in 1998 (zooplankton in
1999) to collect baseline data on lower trophic levels
(phytoplankton, zooplankton, and insect drift), juvenile and adult
fish, hydrology, and water quality parameters. The baseline data are
intended to be used to evaluate the success of restoration projects
and more broadly understand the ecological functions of the Yolo
Bypass. Zooplankton samples are collected from 2 sites, one in the
Yolo Bypass (he base of the Toe Drain) and one in the Sacramento
River (Sherwood Harbor). In 1999 and 2000, samples were collected
for a couple of months each year, and then from 2001-2010 samples
were collected in roughly winter/spring. Since 2011, samples are
collected twice monthly during most of the year, and weekly when the
bypass is inundated. Samples are collected during ebb tides. In the
Yolo Bypass, nets are deployed from the rotary screw trap that is
anchored in the middle of the Toe Drain Channel. In the Sacramento
River, nets are deployed from a dock when flows are higher than
about 2 fps, or from a boat moving 2-3 mph when flows are lower.
YBFMP samples with micro (50 μm) and meso (150 μm) zooplankton nets.
Both nets have a 0.50 m diameter mouth and are 2 m long with a
polyethylene cod-end jar. The micro net is held just below the
surface for 2 minutes and the meso net is held just below the
surface for 5 minutes, although tow times may be shorter with debris
interference. Samples are washed in deionized water then preserved
in 10% formalin with Rose Bengal. Before counting, samples are
rinsed through sieves (106 μm for the meso sample and 45 μm for the
micro sample) to remove large debris, then samples are transferred
to 8% Lugol’s Iodine solution.
More information on YBFMP and its methods can be found in the YBFMP
data publication (Interagency Ecological Program et al. 2021).
1.7. Directed Outflow Project Lower Trophic Study
The Directed Outflow Project was initiated by the United States
Bureau of Reclamation in 2017 to evaluate the effectiveness of fall
outflow actions on Delta Smelt. Field sampling is paired with the
United States Fish and Wildlife Service Enhanced Delta Smelt
Monitoring. Three random sites are chosen per region (Suisun Bay,
Suisun Marsh, the Lower Sacramento River, the Cache Slough Complex,
and the Sacramento Deep Water Ship Channel) and time period using a
Generalized Random Tessellation Stratified design. Samples were
collected once every two weeks in 2017 and weekly thereafter.
Sampling is conducted in the fall and, starting in 2019, spring and
summer seasons have also been sampled. In 2017, stations were
sampled in 3 additional regions: West of the Benicia Bridge, Lower
San Joaquin, and Upper Sac River. At each station, sample collection
is attempted at both shoal (less than or equal to 10 feet deep) and
channel (greater than 10 feet deep) habitat. Channels are sampled at
the surface (0.5 m deep) and, if deeper than 20 feet, also at the
bottom 1/2 to 1/3 of the water column. In 2017, oblique tows were
also conducted to compare with the horizontal tows.
DOP samples with meso (150 μm) and macro (500 μm) zooplankton nets
that are each 2 m in length. In 2017, both nets were 50 cm in
diameter but starting in 2018 a smaller 20 cm diameter net was used
for the mesozooplankton samples. Samples are stored in 10% formalin.
2. Descriptions of common methods
2.1. Mesh sizes
Nets/sieves typically sample zooplankton species whose smallest
dimension is larger than the mesh size, but may also capture some
organisms smaller than the mesh size (which are under-sampled since
some of these smaller plankters are washed through the mesh).
Furthermore, organisms significantly larger than the net mesh may be
able to avoid the net and thereby evade capture.
Since the meso- and micro-zooplankton data overlap in sampled taxa,
we investigated sampling biases of these 2 mesh sizes by comparing
taxa counted in both. We used EMP data, filtered to include only
stations and dates when both meso- and micro-zooplankton samples
were collected.
For each taxon (or life stage) represented in both datasets, we
compared the total summed catch per unit effort (individuals per m3
of water sampled; CPUE) from the mesozooplankton net (153 µm mesh,
net) and the microzooplankton (43 µm mesh, pump) to assess where
each method may be under-sampling. In almost all cases, the two
methods had drastically different total CPUEs, with the
microzooplankton (pump) sample collecting substantially more
individuals (19 out of 23 taxa). The mesozooplankton (net) sample
was only better at capturing Cirripedia larvae, Cyclopoida adults,
and Oithona similis adults. The two catches were very similar for
Cyclopoida juveniles (mesozooplankton/net captured 80% of the catch
of microzooplankton/pump). Using this information, we developed a
list of taxa and life stages under sampled by each method (excluding
only Cyclopoida juveniles since the catch was so close), included as
undersampled.csv. These under sampled plankton are retained in the
integrated zooplankton dataset but can be flagged and removed using
undersampled.csv.
It is important to note that, prior to counting in the lab, the EMP
microzooplankton (pump) samples are passed through a 154 µm sieve in
lab and the larger size fraction is counted separately (under the
assumption that those individuals are better sampled by the
mesozooplankton/net sample). Thus, some of the under sampling of
larger taxa by the microzooplankton (pump) samples may be an
artifact of this lab methodology rather than an effect of the net
mesh size. Therefore, these results may not apply to other
zooplankton studies.
2.2. Tow duration and tow type
2.2.1. Stepped-Oblique net tow
The most commonly used sampling technique employed by IEP’s
long-term monitoring surveys is the stepped-oblique net tow. In this
method, the zooplankton net is attached to a metal sled. This sled
may be solely used for meso and macro-zooplankton (as in EMP and
FMWT), or it may be attached to a larger fish sampling net (as in
20mm and STN). The sled is deployed off the stern or side of a boat
using a winch or a-frame with a cable attached to a winch. The cable
is spooled out to a standardized length based on the depth of the
water. The boat proceeds at slow speeds while a specified amount of
cable is slowly drawn in at specified time intervals following a tow
schedule. As the cable is drawn in, the sled rises through the water
in a stepwise fashion, sampling each strata of the water column.
2.2.2. Horizontal net tow
In a horizontal net tow, the net is held at a constant depth while
the boat proceeds forward at slow speeds. FRP, YBFMP, and DOP uses
horizontal tows in which the net is held just below the surface of
the water. DOP also conducts horizontal tows at the bottom of the
water column (in sites deeper than 20 feet).
2.2.3. Stationary sampling
In some stations, FRP samples by holding the zooplankton net in a
constant position and allowing the current to flow through the net
for a pre-defined period of time (IEP Tidal Wetlands Monitoring
Project Work Team 2017). YBFMP also uses this approach within the
Yolo Bypass and in the Sacramento River when flows are high enough.
This works best when sampling from shore or a stable structure, to
attach the net to, and is most commonly used on ebb tides to sample
water flowing out of a wetland.
2.2.4. Pump
Pumps are used by EMP for sampling smaller zooplankton (rotifers,
copepod nauplii, etc.; Hennessy 2019). Pumps are advantageous for
microzooplankton because the filtered volume is easier to measure,
and net clogging is easier to monitor. However, larger organisms can
escape the narrow mouth of a pump intake (Harris et al. 2000).
2.3. Measurement of environmental variables
2.3.1. Salinity and temperature
Specific conductivity and temperature are measured by all surveys
using YSI probes. Surface measurements are taken in the upper 1 m of
the water, while EMP, FMWT, STN, 20mm, and DOP also collect
conductivity measurements at the bottom of the water column. For
this dataset, we have converted conductivity to salinity using the
ec2pss function from the wql package (Jassby et al. 2017) for the R
statistical programming language (R Core Team 2023). This function
converts electrical conductivity to salinity using the Practical
Salinity Scale 1978 for salinities between 2 and 42 (Fofonoff and
Millard Jr 1983) and the extension of the Practical Salinity Scale
(Hill et al. 1986) for salinities below 2.
2.3.2. Turbidity and Secchi depth
All surveys measure Secchi depth, the depth at which a black and
white disk is no longer visible from the surface. Secchi depth is
recorded from the shaded side of the boat by an observer not wearing
sunglasses. Secchi depth is inversely related to turbidity, which is
measured by some surveys (20mm, FMWT, STN, FRP, YBFMP, DOP) using
YSI or Hach turbidity meters starting in more recent years.
2.4. Target organisms identified
Depending on the goals of the study, some surveys will enumerate
different organisms than others, and identify them to a different
level of taxonomic resolution. For example, FMWT macrozooplankton
samples are only processed for mysids and amphipods. Other
invertebrates (insects, isopods, etc.), are not counted. FRP
macrozooplankton samples are processed for all macrozooplankton and
micronekton, however insects are only identified to the family
level, whereas mysids are identified to species. Comparing these 2
data sets requires understanding and accounting for these
differences to avoid erroneously believing that FMWT samples had
lower diversity than FRP samples.
2.5. Subsampling methods in the lab
Due to the patchy distribution of zooplankton in the water column,
most surveys collect relatively large samples and process a randomly
selected subsample of the original sample. The accuracy of an
abundance estimate based on a sample is directly related to the
number of organisms counted, assuming they are randomly distributed
with a Poisson distribution (Harris et al. 2000). Therefore, the
size of the original sample and proportion of the sample enumerated
will determine accuracy of any derived abundance estimates. If one
program collects significantly larger samples or enumerates a higher
number of individuals in its subsample, comparing abundance
estimates between the two surveys could be confounded by their
differing accuracies. In addition, differences in subsampling method
can impact precision of an estimate (Guelpen et al. 1982). For these
surveys, subsampling is conducted with 1-ml pipetted aliquots for
micro- and mesozooplankton, and divider trays for macrozooplankton.
2.5.1. Aliquots
Mesozooplankton samples are typically sampled with a micropipette
(for specifics, see (Fujimura et al. 2017; Hennessy 2019). The
sample is first diluted to achieve a zooplankton concentration of
between 200 and 400 organisms ml-1. The sample is then mixed in a
beaker and the taxonomist withdraws a 1-ml subsample with a
micropipette and places the subsample on a gridded Sedgewick-Rafter
glass slide, or a Ward counting wheel. The organisms are then
identified under a microscope. Subsamples are processed until a
target is reached, but these targets have changed over time. For
EMP, FMWT, STN, and 20mm, the target was 200 total organisms from
1972 to 2003, 6% of the total dilution volume from 2004 to 2005, and
from 2006 to present organisms were counted until 6% of the dilution
volume had been processed and at least 5 and no more than 20 1-ml
subsamples were processed (Fujimura et al. 2017; Hennessy 2019).
Under current methods and at the target concentration (200-400
organisms per ml), this results in at least 1,000 organisms and a
maximum of 8,000 organisms counted per sample. When samples contain
debris or detritus, dilution volume is often increased to enable
staff to see all the organisms on a slide clearly, which results in
lower total organism counts. FRP processes a minimum of five 1-ml
subsamples until 400 organisms are counted, or 20 ml total,
depending on which occurs first. DOP processes 6 percent of the
diluted sample volume with a minimum of five 1 mL subsamples and a
maximum of 20 (2017) or 10 (after 2017) subsamples are processed.
Between 1,000 organisms to 5,000 organisms are identified and
counted per sample. YBFMP counts 300 organisms per sample in
microzooplankton samples and 200-250 organisms per sample in
mesozooplankton samples.
2.5.2. Divider trays
In the divider-tray method, the macrozooplankton sample is uniformly
spread across a plastic tray and a 4-quadrant divider is then
dropped on top of the tray. Technicians then enumerate only the
invertebrates in the lower right-hand corner (or random position) of
the tray. For very heavy samples, this procedure may be repeated so
that a 1/16th or a 1/64th fraction of the original sample is
enumerated (for specifics, see IEP Tidal Wetlands Monitoring Project
Work Team 2017; Hennessy 2019). This technique is simple to conduct;
however, it relies on the sample being randomly distributed in the
tray. Organisms and detritus may also be stuck under the dividers
when they are placed in the tray. From 1972 to 1984, these surveys
targeted a minimum count of 220 total organisms before subsampling
is completed. From 1984 to present, 400 total organisms were
targeted (IEP Tidal Wetlands Monitoring Project Work Team 2017;
Hennessy 2019), which gives a precision of +/- 10 % (Harris et al.
2000). For DOP, a minimum of 400 non-gravid mysids, 100 gravid
mysids, and 250 amphipods are targeted for total counts per species.
2.6. Calculations
2.6.1. Count per unit effort (CPUE)
CPUE calculations are based on the volume of water sampled. Most IEP
surveys estimate volume using a flowmeter in the center of the net
mouth (model 2030R, General Oceanics, Inc, Miami Florida). The
volume of water sampled is calculated from flowmeter counts, a meter
constant, and the net mouth area. For EMP microzooplankton samples,
the volume of the water pumped into the net is measured directly
using a GPI inline digital flowmeter (Great Plaines Industries, Inc,
Sparta, NJ) near the output end of the hose where water enters the
net for filtration.
2.6.2. Biomass
Meso- and microzooplankton biomass is most frequently calculated
based on average weights derived from literature values. These
calculations apply a single value for mg C per individual for all
individuals of a given life stage (Culver et al. 1985; Kimmerer et
al. 2011). There are no existing biomass values for many species, so
related species must be used.
For mysids collected by EMP and FMWT, the first 100 individuals are
also measured to the nearest mm. Biomass is than calculated based on
length-weight regressions. Length-weight regressions provide a
somewhat more accurate estimate of total biomass, however the extent
to which a given individual fits the regression will vary based on
sex, reproductive state, health, and time of year. Mysid length data
and conversion equations are not currently included in the
integrated dataset but should be included in future revisions.
We have compiled updated micro- and mesozooplankton biomass
conversions from the literature into biomass_mesomicro.csv. All
species and taxonomic groups are not covered, reflecting gaps in the
literature, but these conversion values provide a starting point for
researchers interested in estimating zooplankton biomass.
3. Data integration methods
Data integration was completed in R version 4.2.2 (R Core Team
2023). All code is available in the R package zooper version v2.5.0
(Bashevkin et al. 2023) and in the R script “Data_processing.R.”
First, we created lookup tables to assist with the data integration.
The locations of fixed sampling locations were compiled into
stations.csv. Taxonomic classifications were compiled into
taxonomy.csv, while the taxonomic resolution of each source dataset
and the dates this resolution changed or species were introduced
were compiled into taxa_lists.csv.
Datasets were downloaded from their sources online and reformatted
for consistency by converting species codes to scientific names,
renaming column names, converting units, and pivoting all datasets
to the “long” format (where each row contains just one CPUE value
for each taxon and sample). In some datasets, CPUE was reported as 0
in years before the taxa was counted at that taxonomic level. These
values should have been “NA” because the abundances of those taxa
were unknown before they were counted. To resolve this issue, in
years when taxa were not counted, we replaced those incorrect 0s
with “NA” values. However, non-native species were left with 0 CPUE
before their known introduction year.
The consistent datasets were then bound together by column name. All
environmental parameters were not measured by all datasets and those
gaps are represented in the combined datasets with “NA” values. To
reduce data duplication and file size, the combined dataset was then
split into sample-level data (environment.csv; sampling location,
date, environmental parameters, etc.) and zooplankton catch data
(zooplankton.csv), each retaining the column “SampleID” as a key to
rejoin them. The taxonomic resolution of each source dataset is
unaltered and thus variable across surveys within the integrated
dataset. Information on the taxonomic resolution of source datasets
can be found in taxa_lists.csv.
3.1. Resolving differences in taxonomic resolution: methods used to
create zooplankton_community.csv
The approaches described in this section were not applied to the
YBFMP survey due to inconsistencies in taxonomy and life stage
differentiation over time and with the other surveys. Differences in
taxonomic resolution among studies could result in misleading
findings from a community-level analysis of the integrated dataset.
For example, EMP and FMWT lump all members of the genus Tortanus
together and count them within the category Tortanus spp., while
20mm separates and counts Tortanus discaudatus, Tortanus
dextrilobatus, and other Tortanus (Tortanus spp.). A naïve analysis
would conclude that Tortanus discaudatus and Tortanus dextrilobatus
only appear at the 20mm sampling locations while the lumped Tortanus
spp. category is much more prevalent at EMP and FMWT sites. However,
these results would be due solely to differences in taxonomic
resolution among surveys. To resolve this issue, we developed a
method to standardize taxonomic resolution to ensure data are
comparable across surveys.
To start, we find all taxa that are not counted in every survey
(Tortanus discaudatus and Tortanus dextrilobatus in the example
above). Then, we sum these taxa up to a higher taxonomic level that
is counted in all surveys (Tortanus spp.) and remove the lower taxa
that have been summed (Tortanus discaudatus and Tortanus
dextrilobatus) to prevent double counting. Now, all surveys have
categories (Tortanus spp.) that represent the same set of taxa (all
copepods in the genus Tortanus). Any taxa that are not represented
at a higher taxonomic level in all surveys are removed from the
dataset. These removed taxa are less-commonly counted taxa such as
Annelida, Nematoda, or Insecta. This process considers each life
stage of a taxa separately and is applied separately to each size
class, so taxa lists for microzooplankton samples are only compared
to taxa lists for other microzooplankton samples (and the same for
meso- and macrozooplankton). This solution has been applied to the
zooplankton_community.csv table, which also has been merged with the
environment.csv, taxonomy.csv, and undersampled.csv tables so that
it is an analysis-ready dataset with all taxon- and sample-level
information.
3.2. Other considerations and features of the zooper package
The approaches described in this section were not applied to the
YBFMP survey due to the amount of inconsistencies in taxonomy and
life stage differentiation over time. The taxonomic resolution for
many surveys (all except FRP) has changed over time. In some cases,
recently introduced taxa were added after their introduction, but in
other cases taxa formerly identified e.g., to the genus level were
subsequently identified to the species level. Analyses of community
change over time must take these changes in taxonomic resolution
over time into account to prevent a naïve analysis from discovering
increasing diversity over time that is solely attributed to changes
in methods. The zooper R package (Bashevkin et al. 2023) can correct
for changes in taxonomic resolution over time by reducing the
taxonomic resolution of the dataset to its lowest resolution at any
point in time. However, this would exclude introduced species from
analyses, so the package allows users to input a time-lag for
introduced species. If surveys began counting an introduced species
within a defined period of years (the time-lag) after its
introduction, that species is retained in the time-corrected
dataset.
In addition to the incorporation of a fix for changing taxonomic
resolution over time, the zooper R package (and its associated
interactive point-and-click shiny application:
https://deltascience.shinyapps.io/ZoopSynth/) have a number of other
options to customize your zooplankton dataset. They allow users to
filter the data by date, salinity, temperature, survey, size class,
or sampling location. The taxonomic resolution fixes are then
applied on the filtered dataset. This ensures the fewest possible
alterations to the data are made. Lastly, the R package and shiny
application also have an alternative solution for resolving
differences in taxonomic resolution among studies. For users
interested in querying all available data on certain taxa, the
package will return all data on your chosen taxa along with summed
categories representing higher taxonomic levels that are comparable
across surveys. Unlike the process used to create the
zooplankton_community.csv file described above, this method does not
remove lower taxonomic categories that are members of summed groups,
so plankton are double counted. Thus, outputs with this option
selected should not be used for multivariate or community-level
analyses. Users interested in using these advanced options to return
a more customizable dataset are encouraged to produce their dataset
with the zooper R package or shiny application, instead of using
zooplankton_community.csv.
4. References
Barros, A. E. 2022. Interagency Ecological Program Zooplankton
Study. ver 11. Environmental Data Initiative.
doi:10.6073/pasta/51bfce9bc26011095d8b99b3a2aee1b8
Bashevkin, S. M., R. Hartman, K. Alstad, and C. Pien. 2023. zooper:
an R package to download and integrate zooplankton datasets from the
Upper San Francisco Estuary. v2.5.0. Zenodo.
doi:10.5281/zenodo.7641064
Battey, M., and S. Perry. 2022. Interagency Ecological Program:
Discrete water quality monitoring in the Sacramento-San Joaquin
Bay-Delta, collected by the Environmental Monitoring Program,
1975-2021. ver 7. Environmental Data Initiative.
doi:10.6073/pasta/3b12b205c1b53290ff555abd809edfcf
Burdi, C. E., S. M. Breining-Aday, and S. B. Slater. 2022.
Interagency Ecological Program: Zooplankton and water quality data
in the San Francisco Estuary collected by the Summer Townet and Fall
Midwater Trawl monitoring programs. Environmental Data Initiative.
doi:10.6073/pasta/00db7da0265df448f167c823fb063a9a
California Department of Fish and Wildlife. 2021. CDFW 20-mm Survey
meso-zooplankton data.
https://filelib.wildlife.ca.gov/Public/Delta%20Smelt.
California Department of Fish and Wildlife, R. Hartman, S. Sherman,
D. Contreras, and D. Ellis. 2019. Fish catch, invertebrate catch,
and water quality data from the Sacramento-San Joaquin Delta
collected by the Fish Restoration Monitoring Program, 2015-2017.
Environmental Data Initiative.
doi:10.6073/PASTA/86810E72766AD19FCCB1B9DD3955BDF8
Contreras, D., R. Hartman, and S. Sherman. 2018. Fish sampling gear
comparisons in littoral and channel habitats. IEP Newsletter 32:
32–47.
Culver, D. A., M. M. Boucherle, D. J. Bean, and J. W. Fletcher.
1985. Biomass of freshwater crustacean zooplankton from
length–weight regressions. Canadian Journal of Fisheries and Aquatic
Sciences 42: 1380–1390. doi:10.1139/f85-173
Fofonoff, N. P., and R. C. Millard Jr. 1983. Algorithms for the
computation of fundamental properties of seawater. UNESCO Technical
Papers in Marine Science 44.
Fujimura, B., L. Damon, and T. Tempel. 2017. 20mm survey metadata
1995-2016. California Department of Fish and Wildlife.
Guelpen, V. L., D. F. Markle, and D. J. Duggan. 1982. An evaluation
of accuracy, precision, and speed of several zooplankton subsampling
techniques. Journal Du Conseil International Pour L’Exploration De
La Mer 40: 226–236.
Harris, R. P., P. H. Wiebe, J. Lenz, H. R. Skjoldal, and M. Huntley,
eds. 2000. ICES zooplankton methodology manual, Academic Press.
Hennessy, A. 2019. Zooplankton meta data, IEP Bay-Delta Monitoring
and Analysis Section, Department of Water Resources, Sacramento, CA.
Hill, K., T. Dauphinee, and D. Woods. 1986. The extension of the
Practical Salinity Scale 1978 to low salinities. IEEE Journal of
Oceanic Engineering 11: 109–112.
IEP Tidal Wetlands Monitoring Project Work Team. 2017. Tidal wetland
monitoring framework for the upper San Francisco Estuary: Standard
Operating Procedures.
Interagency Ecological Program, C. L. Pien, J. B. Adams, and N.
Kwan. 2021. Interagency Ecological
Jassby, A. D., J. E. Cloern, and J. Stachelek. 2017. wql: Exploring
Water Quality Monitoring Data.
Kimmerer, W., T. Ignoffo, and L. Sullivan. 2011. Length, weight,
carbon, and nitrogen content of common copepods in the San Francisco
Estuary. Romberg Tiburon Center, San Francisco State University.
R Core Team. 2023. R: A Language and Environment for Statistical
Computing, R Foundation for Statistical Computing.
Schultz, A. A., and A. J. Kalmbach. 2023. Directed Outflow Project
Lower Trophic Study.
Environmental Data Initiative.
doi:10.6073/pasta/73b4acc0e1a72cb31d95ba96a4071fbc
SWRCB. 2000. Revised Water Right Decision 1641. In the Matter of
Implementation of Water Quality Objectives for the San Francisco
Bay/Sacramento-San Joaquin Delta Estuary; A Petition to Change
Points of Diversion of the Central Valley Project and the State
Water Project in the Southern Delta, and A Petition to Change Places
of Use and Purposes of Use of the Central Valley Project, State
Water Resources Control Board, California Environmental Protection
Agency.
USFWS. 2019. Biological Opinion for the Reinitiation of Consultation
of the Coordinated Operations of the Central Valley Project and
State Water Project, U.S. Fish and Wildlife Service.
