Getting started with trps

Our Objectives

The purpose of this vignette is to learn how to estimate trophic position of a species using stable isotope data (δ13C and δ15N). We can estimate trophic position using a one source model based on equations from Post 2002.

Trophic Position Model

The equation for a one source model consists of the following:

$$ \text{Trophic Position} = \lambda + \frac{(\delta^{15}N_c - \delta^{15}N_b)}{\Delta N} $$

Where λ is the trophic position of the baseline (e.g., 2), δ15Nc is the δ15N of the consumer, δ15Nb is the mean δ15N of the baseline, and ΔN is the trophic enrichment factor (e.g., 3.4).

To use this model with a Bayesian framework, we need to rearrange this equation to the following:

δ15Nc = δ15Nb + ΔN × (Trophic Position − λ)

The function one_source_model() uses this rearranged equation.

Vignette structure

First we need to organize the data prior to running the model. To do this work we will use {dplyr} and {tidyr} but we could also use {data.table}.

When running the model we will use {trps} and {brms} and iterative processes provided by {purrr}.

Once we have run the model we will use {bayesplot} to assess models and then extract posterior draws using {tidybayes}. Posterior distributions will be plotted using {ggplot2} and {ggdist} with colours provided by {viridis}.

Load packages

First we load all the packages needed to carry out the analysis.

{
library(bayesplot)
library(brms)
library(dplyr)
library(ggplot2)
library(ggdist)
library(grid)
library(purrr)
library(tidybayes)
library(tidyr)
library(trps)
library(viridis)
}

Assess data

In {trps} we have several data sets, they include stable isotope data (δ13C and δ15N) for a consumer, lake trout (Salvelinus namaycush), a benthic baseline, amphipods, and a pelagic baseline, dreissenids, for an ecoregion in Lake Ontario.

Consumer data

We check out each data set with the first being the consumer.

consumer_iso
#> # A tibble: 30 × 4
#>    common_name ecoregion  d13c  d15n
#>    <fct>       <fct>     <dbl> <dbl>
#>  1 Lake Trout  Embayment -22.9  15.9
#>  2 Lake Trout  Embayment -22.5  16.2
#>  3 Lake Trout  Embayment -22.8  17.0
#>  4 Lake Trout  Embayment -22.3  16.6
#>  5 Lake Trout  Embayment -22.5  16.6
#>  6 Lake Trout  Embayment -22.3  16.6
#>  7 Lake Trout  Embayment -22.3  16.6
#>  8 Lake Trout  Embayment -22.5  16.2
#>  9 Lake Trout  Embayment -22.9  16.4
#> 10 Lake Trout  Embayment -22.3  16.3
#> # ℹ 20 more rows

We can see that this data set contains the common_name of the consumer , the ecoregion samples were collected from, and δ13C (d13c) and δ15N (d15n).

Baseline data

Next we check out the benthic baseline data set.

baseline_1_iso
#> # A tibble: 14 × 5
#>    common_name ecoregion d13c_b1 d15n_b1    id
#>    <fct>       <fct>       <dbl>   <dbl> <int>
#>  1 Amphipoda   Embayment   -26.2    8.44     1
#>  2 Amphipoda   Embayment   -26.6    8.77     2
#>  3 Amphipoda   Embayment   -26.0    8.05     3
#>  4 Amphipoda   Embayment   -22.1   13.6      4
#>  5 Amphipoda   Embayment   -24.3    6.99     5
#>  6 Amphipoda   Embayment   -22.1    7.95     6
#>  7 Amphipoda   Embayment   -24.7    7.37     7
#>  8 Amphipoda   Embayment   -26.6    6.93     8
#>  9 Amphipoda   Embayment   -24.6    6.97     9
#> 10 Amphipoda   Embayment   -22.1    7.95    10
#> 11 Amphipoda   Embayment   -24.7    7.37    11
#> 12 Amphipoda   Embayment   -22.1    7.95    12
#> 13 Amphipoda   Embayment   -24.7    7.37    13
#> 14 Amphipoda   Embayment   -26.9   10.2     14

We can see that this data set contains the common_name of the baseline, the ecoregion samples were collected from, and δ13C (d13c_b1) and δ15N (d15n_b1).

Organizing data

Now that we understand the data we need to combine both data sets to estimate trophic position for our consumer.

To do this we first need to make an id column in each data set, which will allow us to join them together. We first arrange() the data by ecoregion and common_name. Next we group_by() the same variables, and add id for each grouping using row_number(). Always ungroup() the data.frame after using group_by(). Lastly, we use dplyr::select() to rearrange the order of the columns.

Consumer data

Let’s first add id to consumer_iso data frame.

consumer_iso <- consumer_iso %>% 
  arrange(ecoregion, common_name) %>% 
  group_by(ecoregion, common_name) %>% 
  mutate(
    id = row_number()
  ) %>% 
  ungroup() %>% 
  dplyr::select(id, common_name:d15n)

Baseline data

Next let’s add id to baseline_1_iso data frame. For joining purposes we are going to drop common_name from this data frame.

baseline_1_iso <- baseline_1_iso %>% 
  arrange(ecoregion, common_name) %>% 
  group_by(ecoregion, common_name) %>% 
  mutate(
    id = row_number()
  ) %>% 
  ungroup() %>% 
  dplyr::select(id, ecoregion:d15n_b1)

Joining isotope data

Now that we have the consumer and baseline data sets in a consistent format we can join them by "id" and "ecoregion" using left_join() from {dplyr}.

combined_iso_os <- consumer_iso %>% 
  left_join(baseline_1_iso, by = c("id", "ecoregion"))

We can see that we have successfully combined our consumer and baseline data. We need to do one last thing prior to analyzing the data, and that is calculate the mean δ13C (c1) and δ15N (n1) for the baseline and add in the constant λ (l1) to our data frame. We do this by using groub_by() to group the data by our two groups, then using mutate() and mean() to calculate the mean values.

Important note, to run the model successfully, columns need to be named d15n, n1, and l1.

combined_iso_os <- combined_iso_os %>% 
  group_by(ecoregion, common_name) %>% 
  mutate(
    c1 = mean(d13c_b1, na.rm = TRUE),
    n1 = mean(d15n_b1, na.rm = TRUE),
    l1 = 2
  ) %>% 
  ungroup()

Let’s view our combined data.

combined_iso_os
#> # A tibble: 30 × 10
#>       id common_name ecoregion  d13c  d15n d13c_b1 d15n_b1    c1    n1    l1
#>    <int> <fct>       <fct>     <dbl> <dbl>   <dbl>   <dbl> <dbl> <dbl> <dbl>
#>  1     1 Lake Trout  Embayment -22.9  15.9   -26.2    8.44 -24.6  8.28     2
#>  2     2 Lake Trout  Embayment -22.5  16.2   -26.6    8.77 -24.6  8.28     2
#>  3     3 Lake Trout  Embayment -22.8  17.0   -26.0    8.05 -24.6  8.28     2
#>  4     4 Lake Trout  Embayment -22.3  16.6   -22.1   13.6  -24.6  8.28     2
#>  5     5 Lake Trout  Embayment -22.5  16.6   -24.3    6.99 -24.6  8.28     2
#>  6     6 Lake Trout  Embayment -22.3  16.6   -22.1    7.95 -24.6  8.28     2
#>  7     7 Lake Trout  Embayment -22.3  16.6   -24.7    7.37 -24.6  8.28     2
#>  8     8 Lake Trout  Embayment -22.5  16.2   -26.6    6.93 -24.6  8.28     2
#>  9     9 Lake Trout  Embayment -22.9  16.4   -24.6    6.97 -24.6  8.28     2
#> 10    10 Lake Trout  Embayment -22.3  16.3   -22.1    7.95 -24.6  8.28     2
#> # ℹ 20 more rows

It is now ready to be analyzed!

Bayesian Analysis

We can now estimate trophic position for lake trout in an ecoregion of Lake Ontario.

There are a few things to know about running a Bayesian analysis, I suggest reading these resources:

  1. Basics of Bayesian Statistics - Book
  2. General Introduction to brms
  3. Estimating non-linear models with brms
  4. Nonlinear modelling using nls nlme and brms
  5. Andrew Proctor’s - Module 6
  6. van de Schoot et al., 2021

Priors

Bayesian analyses rely on supplying uninformed or informed prior distributions for each parameter (coefficient; predictor) in the model. The default informed priors for a one source model are the following, ΔN assumes a normal distribution (dn; μ = 3.4; σ = 0.25), trophic position assumes a uniform distribution (lower bound = 2 and upper bound = 10), σ assumes a uniform distribution (lower bound = 0 and upper bound = 10), and if informed priors are desired for δ15Nb (n1; μ = 9; σ = 1), we can set the argument bp to TRUE in all one_source_ functions.

You can change these default priors using one_source_priors_params(), however, I would suggest becoming familiar with Bayesian analyses, your study species, and system prior to adjusting these values.

Model convergence

It is important to always run the model with at least 2 chains. If the model does not converge you can try to increase the following:

  1. The amount of samples that are burned-in (discarded; in brm() this can be controlled by the argument warmup)

  2. The number of iterative samples retained (in brm() this can be controlled by the argument iter).

  3. The number of samples drawn (in brm() this is controlled by the argument thin).

  4. The adapt_delta value using control = list(adapt_delta = 0.95).

When assessing the model we want R̂ to be 1 or within 0.05 of 1, which indicates that the variance among and within chains are equal (see {rstan} documentation on R̂), a high value for effective sample size (ESS), trace plots to look “grassy” or “caterpillar like,” and posterior distributions to look relatively normal.

Estimating trophic position

We will use functions from {trps} that drop into a {brms} model. These functions are one_source_model() which provides brm() the formula structure needed to run a one source model. Next brm() needs the structure of the priors which is supplied to the prior argument using one_source_priors(). Lastly, values for these priors are supplied through the stanvars argument using one_source_priors_params(). You can adjust the mean (μ), variance (σ), or upper and lower bounds (lb and ub) for each prior of the model using one_source_priors_params(), however, only adjust priors if you have a good grasp of Bayesian frameworks and your study system and species.

Model

Let’s run the model!

m <- brm(
  formula = one_source_model(),
  prior = one_source_priors(),
  stanvars = one_source_priors_params(),
  data = combined_iso_os,
  family = gaussian(),
  chains = 2,
  iter = 4000,
  warmup = 1000,
  cores = 4,
  seed = 4,
  control = list(adapt_delta = 0.95)
)
#> Compiling Stan program...
#> Start sampling

Model output

Let’s view the summary of the model.

m
#>  Family: gaussian 
#>   Links: mu = identity; sigma = identity 
#> Formula: d15n ~ n1 + dn * (tp - l1) 
#>          dn ~ 1
#>          tp ~ 1
#>    Data: combined_iso_os (Number of observations: 30) 
#>   Draws: 2 chains, each with iter = 4000; warmup = 1000; thin = 1;
#>          total post-warmup draws = 6000
#> 
#> Regression Coefficients:
#>              Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
#> dn_Intercept     3.37      0.25     2.89     3.86 1.00     1506     1878
#> tp_Intercept     4.54      0.20     4.21     4.96 1.00     1531     1891
#> 
#> Further Distributional Parameters:
#>       Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
#> sigma     0.61      0.09     0.47     0.81 1.00     2145     1903
#> 
#> Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS
#> and Tail_ESS are effective sample size measures, and Rhat is the potential
#> scale reduction factor on split chains (at convergence, Rhat = 1).

We can see that R̂ is 1 meaning that variance among and within chains are equal (see {rstan} documentation on R̂) and that ESS is quite large. Overall, this means the model is converging and fitting accordingly.

Trace plots

Let’s view trace plots and posterior distributions for the model.

plot(m)

We can see that the trace plots look “grassy” meaning the model is converging!

Predictive posterior check

We can check how well the model is predicting the δ15N of the consumer using pp_check() from {bayesplot}.

pp_check(m)
#> Using 10 posterior draws for ppc type 'dens_overlay' by default.

We can see that posteriors draws (yrep; light lines) are effectively modeling δ15N of the consumer ( y; dark line).

Posterior draws

Let’s again look at the summary output from the model.

m
#>  Family: gaussian 
#>   Links: mu = identity; sigma = identity 
#> Formula: d15n ~ n1 + dn * (tp - l1) 
#>          dn ~ 1
#>          tp ~ 1
#>    Data: combined_iso_os (Number of observations: 30) 
#>   Draws: 2 chains, each with iter = 4000; warmup = 1000; thin = 1;
#>          total post-warmup draws = 6000
#> 
#> Regression Coefficients:
#>              Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
#> dn_Intercept     3.37      0.25     2.89     3.86 1.00     1506     1878
#> tp_Intercept     4.54      0.20     4.21     4.96 1.00     1531     1891
#> 
#> Further Distributional Parameters:
#>       Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
#> sigma     0.61      0.09     0.47     0.81 1.00     2145     1903
#> 
#> Draws were sampled using sampling(NUTS). For each parameter, Bulk_ESS
#> and Tail_ESS are effective sample size measures, and Rhat is the potential
#> scale reduction factor on split chains (at convergence, Rhat = 1).

We can see that ΔN is estimated to be 3.37 with l-95% CI of 2.89, and u-95% CI of 3.86. If we move down to trophic position (tp) we see trophic position is estimated to be 4.54 with l-95% CI of 4.21, and u-95% CI of 4.96.

Extract posterior draws

We use functions from {tidybayes} to do this work. First we look at the the names of the variables we want to extract using get_variables().

get_variables(m)
#>  [1] "b_dn_Intercept" "b_tp_Intercept" "sigma"          "lprior"        
#>  [5] "lp__"           "accept_stat__"  "stepsize__"     "treedepth__"   
#>  [9] "n_leapfrog__"   "divergent__"    "energy__"

You will notice that "b_tp_Intercept" is the name of the variable that we are wanting to extract. We extract posterior draws using gather_draws(), and rename "b_tp_Intercept" to tp.

post_draws <- m %>% 
  gather_draws(b_tp_Intercept) %>% 
  mutate(
    ecoregion = "Embayment",
    common_name = "Lake Trout",
    .variable = "tp"
  ) %>% 
  dplyr::select(common_name, ecoregion, .chain:.value)

Let’s view the post_draws

post_draws
#> # A tibble: 6,000 × 7
#> # Groups:   .variable [1]
#>    common_name ecoregion .chain .iteration .draw .variable .value
#>    <chr>       <chr>      <int>      <int> <int> <chr>      <dbl>
#>  1 Lake Trout  Embayment      1          1     1 tp          4.23
#>  2 Lake Trout  Embayment      1          2     2 tp          4.26
#>  3 Lake Trout  Embayment      1          3     3 tp          4.39
#>  4 Lake Trout  Embayment      1          4     4 tp          4.44
#>  5 Lake Trout  Embayment      1          5     5 tp          4.33
#>  6 Lake Trout  Embayment      1          6     6 tp          4.74
#>  7 Lake Trout  Embayment      1          7     7 tp          5.06
#>  8 Lake Trout  Embayment      1          8     8 tp          4.40
#>  9 Lake Trout  Embayment      1          9     9 tp          4.42
#> 10 Lake Trout  Embayment      1         10    10 tp          4.30
#> # ℹ 5,990 more rows

We can see that this consists of seven variables:

  1. ecoregion
  2. common_name
  3. .chain
  4. .iteration (number of sample after burn-in)
  5. .draw (number of samples from iter)
  6. .variable (this will have different variables depending on what is supplied to gather_draws())
  7. .value (estimated value)

Extracting credible intervals

Considering we are likely using this information for a paper or presentation, it is nice to be able to report the median and credible intervals (e.g., equal-tailed intervals; ETI). We can extract and export these values using spread_draws() and median_qi from {tidybayes}.

We rename b_tp_Intercept to tp, add the grouping columns, round all columns that are numeric to two decimal points using mutate_if(), and rearrange the order of the columns using dplyr::select().

medians_ci <- m %>%
        spread_draws(b_tp_Intercept) %>%
        median_qi() %>% 
        rename(
          tp = b_tp_Intercept
        ) %>% 
  mutate(
    ecoregion = "Embayment", 
    common_name = "Lake Trout"
  ) %>% 
  mutate_if(is.numeric, round, digits = 2) %>% 
  dplyr::select(ecoregion, common_name, tp:.interval)

Let’s view the output.

medians_ci
#> # A tibble: 1 × 8
#>   ecoregion common_name    tp .lower .upper .width .point .interval
#>   <chr>     <chr>       <dbl>  <dbl>  <dbl>  <dbl> <chr>  <chr>    
#> 1 Embayment Lake Trout   4.53   4.21   4.96   0.95 median qi

I like to use {openxlsx} to export these values into a table that I can use for presentations and papers. For the vignette I am not going to demonstrate how to do this but please check out {openxlsx}.

Plotting posterior distributions – single species or group

Now that we have our posterior draws extracted we can plot them. To analyze a single species or group, I like using density plots.

Density plot

For this example we first plot the density for posterior draws using geom_density().

ggplot(data = post_draws, aes(x = .value)) + 
  geom_density() + 
  theme_bw(base_size = 15) +
  theme(
    panel.grid = element_blank()
  ) + 
  labs(
    x = "P(Trophic Position | X)", 
    y = "Density"
  )

Point interval

Next we plot it as a point interval plot using stat_pointinterval().

ggplot(data = post_draws, aes(y = .value, 
                              x = common_name)) + 
  stat_pointinterval() + 
  theme_bw(base_size = 15) +
  theme(
    panel.grid = element_blank()
  ) + 
  labs(
    x = "P(Trophic Position | X)", 
    y = "Density"
  )

Congratulations we have estimated the trophic position for Lake Trout!

I’ll demonstrate in another vignette how to run the model with an iterative process to produce estimates of trophic position for more than one group (e.g., comparing trophic position among species or in this case different ecoregions).