I really should have titled my project the “Multi-scale exploration of California Market Squid” to begin with, but c’est la vie! Throughout the quarter I’ve worked mostly in R, both to make interpolated spatial maps of my data as well as to run spatial analyses. Specifically, I’ve been interested in determining which oceanographic covariates are correlated with the abundances of squid at the fine scale as well as at a broader regional scale.

Fine scale exporation:

In order to explore this data, I started this quarter by plotting the distributions of market squid based on the yearly sampled (in June) fine scale spatial abundance data (see yearly plots below). For this class I chose to work with only positive abundance data (as absences may be pseudoabsences due to catchability issues and therefore potentially not good indications of poor quality habitat for the species), however I may re-introduce the 0 catch data in future analyses (with this caveat in mind).

In  the interpolated surface maps, warmer colors indicate regions of higher squid abundance and colder colors indicate regions of lower squid abundance on a yearly basis. To create thises maps, wrote a script in R to create an IDW interpolation surface of squid abundances per year with overlayed contours using R’s gstat and maptools libraries.

California market squid (catch per unit effort: #/km2 towed):

Additionally, I interpolated and contoured both the temperature and salinity data (in situ data) collected at each station in each year so as to have a better idea of the environmental context that this species was experiencing in each year.

Temperature @ 3m depth:

Salinity @ 3m depth:

While my initial and future plan is to look at the satellite data corresponding to each cruise (June of each year), I realized I was biting off more than I could chew this quarter by having this as a goal, as working with remotely sensed oceanographic data files (daily images for 14 years!) is remarkably difficult to access and process due to the size of the individual files as well as the multiple frequencies (wavelengths) that correspond to distinct oceanographic processes (SST, Chl, turbidity, etc). Before tackling the satellite data from a brute force kitchen sink fashion, I instead was eager to identify which in-situ environmental variable with a remotely sensed analog dataset (SST from MODIS-Aqua would be an analog to surface temperature measured with a CTD from the ship), showed strong relationships with abundance.

Below is are simple scatterplots of  the log abundances of california market squid with the points colored by the years in which they were sampled. The top row has market squid abundance in relationship to temperature, salinity and chlorophyll (all variables with a remotely sensed equivalent, or nearly), the bottom row has market squid abundance in relationship to density (function of temperature and salinity), oxygen and water column depth (all variables with no remotely sensed analog).

Unfortunately, there were no strong linear or non-linear correlations of abundance with any of the in-situ covariates (R2<0.01).

Broad scale exploration:

Next, I approached this dataset from broader perspective (zooming out if you will). I was interested in tracking the centers of gravity and isotropy (a measure of the non-uniformity of a spatial dataset along multiple axes) of the sampled population through time. The center of gravity statistic is estimated from the data through discrete summations over sample locations. Practically, from sample values z_i at locations x_i, with areas of influence s_i, we have: 

The inertia I indicates how dispersed the population is around its center of gravity, and is given as follows:

A more in depth discussion of this and other spatial indicators can be found in Wolliez et al. (2009). Notes on survey-based spatial indicators for monitoring fish populations. Aquat. Living Resour. 22, 155-164.

I used both of these calculate to determine the center of gravity and isotropy for market squid captured in June (and Sept, not shown on map) of each year. I then created a map with each year’s center of gravity of the market squid population in the sampled region (caveat: centers of gravity and isotropy are not for entire squid population but specifically for the data sampled off the coasts of OR and WA). There is a remarkably strong northwards shift in the centers of gravity over time for both the June data as well as the Sept. data. I haven’t gotten a chance to correlate this northwards shift in the centers of gravity with summarized environmental variables yet, but this is the plan.