observations.Rmd

---
title: "Observations"
output: github_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

For the [Jellyscape](https://github.com/BigelowLab/jellyscape) project observations are drawn from a recent release of unstaged [ECOMON](https://www.fisheries.noaa.gov/about/northeast-fisheries-science-centern) data. We choose the following groups to belong to the gelatinous "jellyscape" although the list is easily modified.

  + *Siphonophores* (siph)
  + *Hydromedusea* (hydrom)
  + *Coelenterates* (coel)
  + *Ctenophores* (ctenop)
  + *Salps* (salps)

These we can read from the [ecomon package](https://github.com/BigelowLab/ecomon) and aggregate these species into a count per 10m^2 or count per 100m^3. 

# The observation data
```{r read_ecomon, message=FALSE}
source("setup.R")
x = read_ecomon_spp(form = "sf")



```

For most analyses here we want the same in long form.

```{r long_form}
long = ecomon_to_long(x) |>
  glimpse()
```

# Distribution through time

```{r plot_timeseries, warning=FALSE}
ggplot(data = long,
       mapping = aes(x = date, y = value)) + 
  geom_point(alpha = 0.1) + 
  geom_smooth(method = "loess", formula = "y~x") + 
  scale_y_continuous(trans='log10') + 
  facet_wrap(~name)
```

Let's look at coverage by year - note we are counting occurences (absences dropped) which results in the exclusion of `Hydromedusea`.

```{r coverage_by_year, warning=FALSE, message=FALSE}
long = mutate(long, year = as.numeric(format(date, "%Y")), 
              month = as.numeric(format(date, "%m")))

ggplot(data = filter(long, value > 0),
       mapping = aes(year)) +
  geom_histogram() + 
  facet_wrap(~name)
```

And alternative is to look at the observations by month.
```{r coverage_by_month, warning=FALSE, message=FALSE}
ggplot(data = filter(long, value > 0),
       mapping = aes(month)) +
  geom_histogram(stat = "count") + 
  scale_x_continuous(breaks = 1:12,
                     labels = function(x) substring(month.abb[x],1,1)) +
  facet_wrap(~name)
```

# Distribution through space

We can read the same data in a spatial form and plot that. Once again, we exclude sample locations where there are abscences.


```{r as_spatial}
long = ecomon_as_sf(filter(long, value > 0)) 
coast = ne_coastline(scale = "medium", returnclass = "sf") |>
  st_crop(long)

ggplot() +
  geom_sf(data = long, shape = ".", alpha = 0.3) + 
  geom_sf(data = coast, color = "orange") +
  facet_wrap(~name)
```