quantifying overlap between the deepwater horizon oil spill and predicted bluefin tuna spawning habitat in the gulf of mexico - oil spill pads

by:Demi     2019-09-15
quantifying overlap between the deepwater horizon oil spill and predicted bluefin tuna spawning habitat in the gulf of mexico  -  oil spill pads
Atlantic bluefin tuna (Blue fin tuna thynnus)
Distributed throughout the North Atlantic region, high economic value and serious development.
At present, the fishery is managed as two spawning stocks, and the GOM population has been severely depleted for more than 20 years. In April-
In August 20, the Deepwater Horizon oil spill released about 4 million barrels of oil, seriously affecting the ecosystem and economy.
Acute petroleum exposure can lead to the death of bluefin tuna eggs and larvae, while the long-term impact on spawning into fish is not yet clear.
Here, we used 66 Blue-fin tuna 16-year electronic label data to identify spawning events, quantify habitat preferences, and predict habitat use and oil exposure in the spawning area of the Gulf of Mexico.
More than 13,600 square kilometers (5%)
During the peak period of oil diffusion, the predicted spawning habitat in the exclusive economic zone of the United States was coated with oil, with potential fatal effects on eggs and larvae.
While oil spills overlap with a relatively small portion of the predicted spawning habitat, the cumulative effects of oil, ocean warming and sub-capture mortality at GOM spawning grounds may have a significant impact on populations with little evidence of reconstruction.
A total of 125 bluefin tuna were marked in the south of St. Lawrence Bay (
GSL waters near Cape Breton Island, Nova Scotia)
From September to October and from 2007 to 2014.
Labels are programmed to be released when returning to GSL foraging grounds at the end of summer in order to restore the labels after full sampling in the Gulf of Mexico. Post-
Release the label that transmits the location via the Argos satellite and, where possible, try to restore the label from shore
Based on the recovery team.
In addition, we also looked at the historical data of surgical implants or external additional file labels deployed in the Gulf of Mexico and North Carolina, Cape hatras and near the city of morhead during the period 1999-2005 (nu2009=u200924).
All bluefin tuna are caught with a fishing rod and a fishing rod, using two generations (
MK10 and miniPATs)
Popular Computer for wild animals
Up satellite file label (PAT)
In the process of learning.
The light level recorded by the label is processed to produce a daily estimate of latitude and longitude and then match the sea surface temperature (SST)
Tag after Tag and data in remote sensing data.
Probability state space model for location estimation (SSM)
This includes maximum dive depth and undersea depth and is able to quantify the uncertainty associated with each daily location.
The model was validated with endpoint data from the labeled Atlantic bluefin tuna (nu2009=u200972).
See Block and Wilson for details on label preparation, programming, accessories, structure, and geo-location analysis.
All experimental programmes are carried out in accordance with the relevant guidelines and regulations and are approved by the executive group for experimental animal care at Stanford University and the Animal Care Committee at Acadia University.
From the electronic label data set, we checked 66 Atlantic bluefin tuna, found the label and entered GOM (
Between November and June)
Check habitat and habitat preferences ().
Once the bluefin tuna has passed 80, they are considered to have entered the GOM.
5 ° W meridian and stay within GOM for six days.
The resulting GOM bluefin tuna data set includes 5272 tracking days with a single tracking day length ranging from 7 to 193 days in the Bay (
For more information about label deployment, see reference. ).
There are two tuna (
ID 5109026 and 5109029)
Attended GOM during April-
From August to 2010, the Deepwater Horizon oil spill overlaps in time and space.
The label data set contains a large amount of archival data and is able to fully check bluefin tuna behavior in GOM.
We also integrate surface oil data using models in the environment response management application (ERMA)
As part of the National Assessment of resource damage (NRDA)
Estimate the total surface area of the oiled GOM and calculate the time series of the oiled adult Atlantic bluefin tuna habitat and spawning habitat.
These modelling products rely on remote sensing data of surface water, so the impact of underground oil is still outside the scope of this study.
Since the tracking data gives the presence of tuna but does not directly measure the absence of the case, the associated random walk (CRWs)
Used to represent a zero model where tuna can move in the environment independently of ocean variables.
For each of the 66 bluefin tuna tracks, 100 CRWs were created as pseudo tuna tracks
Absence in our model framework ().
CRWs is a simulated trajectory consisting of a series of random steps, each using tag-
The turn angle and distance distribution are derived from a specific tuna trajectory.
The starting point of each CRW is the marked position (
If marked in GOM)
Or at the Bay entry point, the initial travel angle of the CRW matches the initial angle of the corresponding label, and each CRW position gives the same error distribution as the actual trajectory, the duration of each CRW is the same as the original label.
CRWs is not really absent because there is no
The marked tuna movement patterns are unknown and their distribution may overlap with the zero trajectory.
For each track, select a random CRW from the CRWs population to create an existing/non-existent data set to map and predict the possibility of habitat, a process that repeats 60 times.
We also used residence time and first stay time to study the temporal and spatial scales of the movement of bluefin tuna in GOM.
Analysis of crossing time R package adehabitat (v. 1. 8. 18).
Determine the date of spawn based on the dive behavior and internal temperature of the file-marked tuna (
For more information and reference, see. ).
Diving behavior in GOM is characterized by deep diving when fish go in and out of GOM (crossing 80. 5° W)
, And the time of shallow oscillation diving when the temperature inside the night rises. Telemetry-
Derived fast-oscillation diving behavior is used to identify inferred spawning, such as skip fish in the East Pacific Ocean, bluefin tuna in GOM, and bluefin tuna in the baleari sea.
While we lack direct observation of spawning from telemetry records, these oscillating dives are unique in the spawning grounds and appear in space and time in areas with high larvae abundance.
Conduct an oscillating night dive for each fish with diving behavior (ref. )
Using expert opinions and Bayesian models from two scientists (
Full description).
The hypothetical spawn date is combined with the transit date to create a binary response variable with a negative binomial link in the spawn possibility generalized additive hybrid model.
For electronic labels that contain continuous time series records (including Archived tags and restored PAT tags), only the modeled spawn behavior can be calculated.
Switching status-
The spatial model has been used to identify presumed foraging events by distance, turn, and angle, where we use horizontal and vertical motion data with additional agents to identify presumed spawning events.
The date and location from the state space modeling output is classified as "spawn" or "non-spawning.
We identified ten "agent" features that separate the number of days assumed as spawn behavior from non-spawn behaviorspawning days (see ).
This method of combining agents can tolerate error detection and allow the distribution of overlapping probabilities.
Note that this method does not require a proxy high or low to indicate spawn, but rather requires the proxy value to be concentrated at a certain point along the proxy axis during the spawn.
Spawn and non-spawn
The spawn dates determined in the archive label validation data set were also compared with the spawn dates visually identified by a pair of human experts.
The two methods show a 87% consistency, indicating that by combining the two methods, we are able to steadily identify changes in typical spawning behavior.
The remote sensing environment data of the actual orbit and CRW orbit are obtained using Xtractomatic ().
Data set includes time-
Sea table temperature series (SST)
SST variation (
Standard deviation, SSTsd-8 days)
Merging from Pathfinder and AVHRR, surface chlorophyll-
Concentration (chl a – 8 day)
Sea surface height anomaly (merging from SeaWiFS and mostoss)SSHa – 1 day)
Variability from Aviso, SSHa (
Standard deviation (SSHsd)
Eddy current kinetic energy from Aviso (EKE – 1 day)
Vertical Ekman pumping from Aviso (wekman – 8 day)
Data from the merged Quikscat and Ascat, north wind (uy10 – 8 day)
Merge data from Quikscat and Ascat, monthly (moon – 1 day)
Measurement of water depth from NOAA (bathy)and rugosity (
Standard deviation (bathysd)
All come from relying on Bo 2.
For each ocean variable, the mean value is calculated using the geo-location error radius of each daily SSM and CRW location.
In order to ensure the normal distribution of data, the transformation of environmental variables is studied.
Log conversion is requireda and EKE.
To quantify habitat use in GOM, a generalized additive hybrid model of binary presence/absence (GAMM)
Fit as a function of the ocean variable containing the negative binomial link function.
We use the geo-location error of each location derived from the state space model to determine the radius of the sampled prediction variable, which provides a probability estimate for the experienced environmental covariates.
In addition, we also evaluated the artifacts.
Tracking the selection by AUC scores of multiple CRW combinations, we use k-fold cross-
Test verification-
Training data set.
Model of spawn possibility using binary Spawn/non-spawn fit
Spawn GAMM for each daily location on the track, based on similar ocean variables, and use the negative binomial link function. GAMMs are semi-
Parameter model, using a smooth curve to identify the relationship between the dependent variable and the predictor.
GAMMs allows nonlinearity, non-
Constant variance and non
Compared to a more conservative linear model, monotonous.
The mixed model can contain random effects, and in our example, the tag ID is used as a random variable to consider the correlation between observations.
In the GAMM framework, the advantage of pairing trajectory points with CRW points is that sampling is missing and present at the same time-space scale, minimizing residual self-correlation.
The various combinations of the environment data set are included in the GAMMs based on the designation to determine the optimal model.
The explanatory and predictive power of the model is compared using the Akaike information standard (AIC)
Area under the curve (AUC)statistics.
Game in R (version 3. 12)
Use MGCV package (version 1. 7–6)with cross-
Verify using the ROCR package (version 1. 07).
The AUC value was the highest, and then the GAMM model with the lowest AIC value ran 60 times with a randomly selected pseudo-model
The absence trajectory for each label can quantify the error generated by the CRW selection.
We use k-fold cross-
Verification of two configurations.
Fitting the model with 75 training/25% test data randomly selected, followed by each year deletion when fitting the model as test data to test the annual predictive power. The cross-
For spawn models and single best-the validation process is repeated
Fitting spawning models were selected for predictive purposes.
From the best environmental relationships
Forecast using fit GAMMs (1)
Habitat and (2)
From April 1-20 to August 26, the possibility of spawning within this habitat is a function of ocean variables, increasing every week.
The response surface of the oil leakage time is summarized by week at 0. 25° x 0.
25 ° spatial resolution. Binary cut-
The off was determined using the ROCR package (version 1. 0-7)
In order to minimize the false positive rate and to make a conservative estimate of habitat and spawning possibilities.
Simulated habitat possibilities are multiplied together with spawning possibilities to obtain an estimate of the total spawning habitat during the oil spill.
Oil products were also inserted into the nearest date to provide a similar oil range in the Gulf of Mexico.
For the exclusive economic zone of the United States, the number of spawning habitats in GOM was summarized to compare with previous studies, and for the entire Gulf of Mexico, the overlap with the oil layer was summarized.
We calculated the percentage of the predicted bluefin tuna spawning habitat in the oil spill, and the total percentage of the refuelling spawning habitat.
These two indicators provide supplementary statistics on oil and tuna overlap and sub-quantify the potential impact on adult tuna spawning in GOM.
Custom message
Chat Online 编辑模式下无法使用
Chat Online inputting...