Skip to content

Commit

Permalink
analyse aantal individuen
Browse files Browse the repository at this point in the history
  • Loading branch information
hansvancalster committed Jan 14, 2025
1 parent 79bf0c3 commit 6df0c07
Showing 1 changed file with 130 additions and 0 deletions.
130 changes: 130 additions & 0 deletions source/data_analysis_spring/data_analysis_spring_2023.Rmd
Original file line number Diff line number Diff line change
Expand Up @@ -696,6 +696,136 @@ marginaleffects::plot_predictions(
ylab("Probability of having zero species")
```

## Aantal individuen

### Met honingbijen

```{r}
m_n_ind <- glmmTMB(
n_ind ~
(location_code
+ maand
+ method_cd) * taxgroup,
ziformula = ~taxgroup,
family = "nbinom1",
na.action = na.fail,
data = data_sp_rich_spring
)
```

Verschillende distributies geprobeerd, maar deze modellen fitten niet goed.
Wellicht omwille van de honingbijen die soms in uitzonderlijk hoge aantallen gevangen worden.

```{r check_n_ind, fig.height=12, fig.cap = "Visuele controle van verschillende modelaannames."}
performance::check_model(m_n_ind)
performance::check_overdispersion(m_n_ind)
```

```{r, fig.cap = "Verwacht aantal soorten in een pantrap-set of transectsegment van 50 m per maand, soortgroep en SPRING methode op basis van ons model."}
marginaleffects::plot_predictions(
m_n_ind,
condition = c("maand", "method_cd", "taxgroup"),
type = "response",
vcov = TRUE
)
```

```{r, fig.cap = "Kans op een nulwaarneming in een pantrap-set of transectsegment van 50 m per soortgroep op basis van ons model."}
marginaleffects::plot_predictions(
m_n_ind,
condition = c("taxgroup"),
type = "zprob",
vcov = TRUE
) +
ylab("Probability of having zero individuals")
```

### Zonder honingbijen

```{r zonder-honingbij}
apoidea_zonder_honingbij <- apoidea %>%
filter(
species_nm != "Apis mellifera Linnaeus, 1758"
)
data_zonder_honingbij <- apoidea_zonder_honingbij |>
mutate(taxgroup = "Apoidea") |>
bind_rows(
syrphidae |>
mutate(taxgroup = "Syrphidae")
) |>
filter(time_series == 0) |>
mutate(
maand = as.factor(month(date_b))
) |>
group_by(
sample_code, location_code, method_combi,
method_cd, spring_code, uv, level, maand, taxgroup
) |>
summarise(
n_species = n_distinct(species_nm, na.rm = TRUE),
n_ind = sum(no_ind),
.groups = "drop"
)
```


Hier fit een negatief-binomiaal model zonder zero-inflation goed.

```{r}
m_n_ind_zh <- glmmTMB(
n_ind ~
(location_code
+ maand
+ method_combi) * taxgroup,
family = "nbinom2",
na.action = na.fail,
data = data_zonder_honingbij
)
```

```{r check_n_ind_zh, fig.height=12, fig.cap = "Visuele controle van verschillende modelaannames."}
performance::check_model(m_n_ind_zh)
performance::check_overdispersion(m_n_ind_zh)
performance::check_predictions(m_n_ind_zh)
```

```{r, fig.cap = "Verwacht aantal individuen (zonder honingbijen) in een pantrap-set of transectsegment van 50 m per soortgroep en methode op basis van ons model."}
marginaleffects::plot_predictions(
m_n_ind_zh,
condition = c("method_combi", "taxgroup"),
type = "response",
vcov = TRUE,
draw = FALSE
) |>
left_join(
data_sp_rich |>
distinct(method_cd, spring_code, uv, level, method_combi, taxgroup)
) |>
as_tibble() |>
ggplot() +
geom_pointrange(
aes(
x = method_cd, y = estimate, ymin = conf.low, ymax = conf.high,
colour = method_combi
),
position = position_dodge(width = 0.5)
) +
labs(y = "Number of individuals (without Apis mellifera)",
x = "",
colour = "") +
facet_wrap(~taxgroup, scales = "free")
```


```{r pwc-method-taxgroup-n-ind-zh}
emmeans::emmeans(m_n_ind_zh, pairwise ~ method_combi * taxgroup)
```



## Alpha biodiversiteit

De Shannon index wordt als volgt berekend:
Expand Down

0 comments on commit 6df0c07

Please sign in to comment.