3.4 Figures
Each modelled species gets its own chapter with results. All results are display in a graphical format.
3.4.1 Estimated number of birds
The results start with a figure showing the estimated number of birds an observer would encounter at an average point on an average site during the first period. The line displays the point estimate for each year. This is the most likely value for the average number of birds. The three ribbons display the uncertainty around this point estimate. They are, from small/dark to wide/light, the 30%, 60% and 90% credible intervals. These numbers in the figure are always based on the non-linear model (1.1). The caption indicates whether the model is non-linear and how strong the linear trend is.
3.4.2 Indices
An index is a change compared to a baseline. This baseline is typically the estimate for some reference winter. E.g we use 2014 as a baseline and compare 2015 with 2013 or 2016 with 2014. However we cannot use the figure with 2014 as baseline to compare 2015 with 2016. For that we need a figure with either 2015 or 2016 as baseline. To facilitate any pairwise comparison among years, we display one figure for every winter using that winter as baseline.
3.4.3 Index raster
Currently a separate index figure for each reference winter is doable since the data contains only 5 winters. The number of index figures will grow over the years, making it harder to interpret them. The third plot summarises the information on a raster. The x axis holds the winter we want to interpret. The y axis holds the reference winter The dots given the relative change from the baseline (y axis) to the other winter (x axis). Their colour indicates the strength of the change. Stronger changes have darker dots, white dots indicate no change. Red dots indicate a decrease from the baseline, blue dots an increase. A baseline with all red (blue) dots indicates the winter with the largest (smallest) numbers. The shape of the dots indicates the classification of the effect. Informative dots (significant or non-significant but stable) get solid shapes.
L
a linear model is best, NL
a non-linear model is best. NL?
a non-linear is possibly better.++
+
+
+
+
+
+
+
+
+
+
+
~
~
~
~
~
~
-
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?+
?-
?-
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?