-8 Functional Plant Indicators - Naturally Open Ecosystems +8 Functional Plant Indicators - Naturally Open Ecosystems
Norwegian name: Planteindikatorer @@ -156,9 +156,9 @@
Functional plant indicators can be used to describe the functional signature of plant communities by calculating community-weighted means of plant indicator values for plant communities (Diekmann 2003). The functional signature of plant communities may be indicative of ecosystem identity, depending on which functional plant indicator we look at (cf. Töpper et al. 2018). For instance, using an indicator for moisture one would find a functional signature with higher moisture values for plant communities in mires compared to e.g. grasslands or forests. Deviations in the functional signature of such an indicator beyond a certain range of indicator values (as there of course is natural variation of functional signatures within an ecosystem type) may be related to a reduction in ecological condition. Here, we combine functional plant indicator data with field sampled plant community data from the Norwegian nature monitoring programs ANO (Tingstad et al. 2019) and GRUK (Evju et al. 2020) for naturally open ecosystems below tree line (abbreviated as ‘nat-open’ henceforth). We calculate the functional signature of plant communities in the monitored sites with respect to Grime’s CSR values, light, nitrogen, and soil disturbance. These functional signatures are then compared to reference distributions of functional signature, separately for each nat-open ecosystem type, calculated from ‘generalized species lists’ developed for ecosystem types in the Norwegian categorization system for eco-diversity (Halvorsen et al. 2020). These plant functional condition indicators are developed following the principles and technical protocol of the IBECA framework (Jakobsson et al. 2021, Töpper & Jakobsson 2021). Note that deviations from the reference may occur in both directions, e.g. the nitrogen signature from the testing data may be higher or lower than in the reference. Deviations in these two directions indicate very different environmental phenomena and thus have to be treated separately. Therefore, we develop two condition indicators for each functional plant indicator, a lower one and an upper one (see further down for more details).
-8.2 About the underlying data +8.2 About the underlying data
In the ‘functional plant indicator’ project for nat-open ecosystems, we use five sets of data for building indicators for ecological condition:
-
@@ -181,9 +181,9 @@
The Swedish plant indicator set published by Tyler et al. (2021) contains a large collection of plant indicators based on the Swedish flora, which is well representative of the Norwegian flora as well. From this set, we decided to include indicator data for light, moisture, pH, nitrogen, phosphorus, grazing_mowing, and soil disturbance for semi-natural ecosystems, as these are thought to be subject to potential change due to abandonment, drainage/flooding, pollution, and erosion.
Grime’s system of plant strategy scores (Grime 1974) comprises relative (too one another) scores for the competition-, stress-, and disturbance(“ruderality”)-related life strategy of plant species. In the analysis in this document, we use all three variables, C, S and R, as different pressures acting on the ecosystem might change every one of the strategies (e.g. alien species for competition, climate change for stress, land use change for ruderality).
-- Southern Norway: 61
- in every bootstrap iteration the abundance of the sampled species can be randomly changed by a limited amount if wished by introducing a re-sampling of abundance values from adjacent abundance steps with a certain probability (var.abun)
-8.2.1 Representativity in time and space +8.2.1 Representativity in time and space
For nat-open ecosystems, the ANO data in this analysis contain 143 plots in 52 sites, in principle distributed randomly across the country. As nat-open ecosystems occur more often in certain regions of Norway than in others, the amount of plots and sites is not equal among Norway’s five regions. The 143 plots are distributed across regions in the following way:
-
@@ -203,9 +203,9 @@
-8.2.2 Temporal coverage +8.2.2 Temporal coverage
The ANO evaluation data cover the first three years, 2019-2021, of the first 5-year-cycle in the ANO monitoring scheme. GRUK covers 2020-2022. Thus, there is no actual time series to these data, and the indicator evaluation does therefore not include any temporal analyses.
-8.4 Reference state and values +8.4 Reference state and values
--8.4.1 Reference state +8.4.1 Reference state
The reference state is defined via the functional signature of the generalized species lists for NiN ecosystem types (see also Töpper et al. 2018). For the nat-open ecosystem types these lists have been newly prepared by Evju et al. (2023). By bootstrapping the species lists (see details further below) and calculating community-weighted means of functional plant indicators for every re-sampled community, we describe the reference state as a distribution of indicator values for each respective plant functional indicator. These distributions are calculated for minor ecosystem types (“grunntyper” or “kartleggingsenheter” at a 1:5000 mapping scale) within the major ecosystem types (hovedtyper) in NiN. A more extensive discussion on the use of reference communities can be found in Jakobsson et al. (2020).
In this analysis, we derive scaling values from statistical (here, non-parametric) distributions (see Jakobsson et al. 2010). For each ecosystem sub-type and plant functional indicator, the reference value is defined as the median value of the indicator value distribution. As in most cases the distributions naturally are two-sided (but see the Heat-requirement indicator in the mountain assessment for an example of a one-sided functional plant indicator, Framstad et al. 2022), and deviation from the optimal state thus may occur in both direction (e.g. indicating too low or too high pH), we need to define two threshold values for good ecological condition as well as both a minimum and maximum value. In line with previous assessments of ecological condition for Norwegian forests and mountains, we define a lower and an upper threshold value via the 95% confidence interval of the reference distribution, i.e. its 0.025 and 0.975 quantiles. The minimum and maximum values are given by the minimum and maximum of the possible indicator values for each respective plant functional indicator. For details on the scaling principles in IBECA, please see Töpper & Jakobsson (2021).
-8.5 Uncertainties +8.5 Uncertainties
We can calculate a mean indicator value (after scaling) for every region (or any other delimited area of interest) as well as its corresponding standard error as a measure of spatial uncertainty for a geographical area.
-8.6 References +8.6 References
Diekmann, M. 2003. Species indicator values as an important tool in applied plant ecology - a review. Basic and Applied Ecology 4: 493-506, doi:10.1078/1439-1791-00185
Evju, M., Stabbetorp, O.E., Olsen, S.L., Bratli, H., Often, A. & Bakkestuen, V. 2020. Dry calcareous grasslands in the Oslofjord region. A test of monitoring protocols and results for 2020. NINA Report 1910. Norwegian Institute for Nature Research.
@@ -279,7 +279,7 @@(not shown here, but documented in the code)
leaving us with the monitoring data including plant indicators (ANO.sp.ind, GRUK.species.ind) and the reference data including plant indicators (NiN.natopen.cov)
-+head(ANO.sp.ind) #> Species art_dekning #> 1 Abies alba 0 @@ -536,7 +536,7 @@
Running the bootstraps
-+-colnames(NiN.natopen) # 1st column is the species # 6th-71st column is the abundances of sp in different ecosystem types @@ -555,7 +555,7 @@natopen.ref.cov[[i]][,j] <- v } }
+-head(natopen.ref.cov[[1]]) #> T2-C-1 T2-C-2 T2-C-3 T2-C-4 T2-C-5 #> 1 0.2845850 0.25000000 0.14655172 0.10347682 0.2346319 @@ -3232,7 +3232,7 @@#> 3rd Qu.:4.76404 3rd Qu.:4.6944 3rd Qu.:2.500 #> Max. :7.63252 Max. :8.8259 Max. :9.000 #> NA's :192 NA's :192
+head(natopen.ref.cov.val) #> N1 hoved grunn county region Ind Rv #> 1 natopen NA T2-C-1 all all CC1 0.2187255 @@ -3250,7 +3250,7 @@#> 6 0.39205879 1
Once test data (ANO, GRUK) and the scaling values from the reference data are in place, we can calculate community-weighted means (CWM) of the selected indicators for the ANO and GRUK community data and scale them against the scaling values from the reference distribution. Note that we scale each ANO/GRUK plot’s CWM against either the lower threshold value and the min value OR the upper threshold value and the max value based on whether the CWM is smaller or higher than the reference value. Since the scaled values for both sides range between 0 and 1, we generate separate lower and upper condition indicators for each functional plant indicator. An ANO/GRUK plot can only have a scaled value in either the lower or the upper indicator (the other one will be ‘NA’), except for the unlikely event that the CWM exactly matches the reference value, in which case both lower and upper indicator will receive a scaled indicator value of 1.
Here is the scaling function
-+#### scaled values #### r.s <- 1 # reference value @@ -3287,7 +3287,7 @@}
We then can prepare a list of data frames to hold the results and perform the scaling according to the principles described in NINA report 1967 (Töpper and Jakobsson 2021) This is done separately for ANO and ASO. First for ANO:
-+#### calculating scaled and non-truncated values for the indicators based on the dataset #### for (i in 1:nrow(ANO.natopen) ) { # @@ -3570,7 +3570,7 @@results.natopen.ANO[['2-sided']]$Soil_disturbance1[results.natopen.ANO[['2-sided']]$Soil_disturbance1>1] <- NA results.natopen.ANO[['2-sided']]$Soil_disturbance2[results.natopen.ANO[['2-sided']]$Soil_disturbance2>1] <- NA
and for GRUK:
-+-#### calculating scaled and non-truncated values for the indicators based on the dataset #### for (i in 1:nrow(GRUK.natopen) ) { # @@ -3804,7 +3804,7 @@results.natopen.GRUK[['2-sided']]$Soil_disturbance1[results.natopen.GRUK[['2-sided']]$Soil_disturbance1>1] <- NA results.natopen.GRUK[['2-sided']]$Soil_disturbance2[results.natopen.GRUK[['2-sided']]$Soil_disturbance2>1] <- NA
+-head(results.natopen.ANO[['2-sided']]) #> GlobalID #> 1 {A24A765B-A1B4-4AF7-A1B9-A4A3F61F6295} @@ -4276,7 +4276,7 @@8.7.1.2 Scaled value analyses
In order to visualize the results we need to rearrange the results-objects from wide to long format (note that there is both a lower and an upper condition indicator for each of the functional plant indicators).
-+#### plotting scaled values by main ecosystem type #### ## continuing with 2-sided res.natopen.ANO <- results.natopen.ANO[['2-sided']] @@ -4349,15 +4349,15 @@)
+-8.7.2 Ecosystem sub-types +8.7.2 Ecosystem sub-types
And we can show the resulting scaled values as Violin plots for each indicator and main ecosystem type
-The ANO results show a high frequency of too low CSR-R values (CSR-R1) and Soil disturbance while also showing signs of higher than expected competition (CSR-C2), which indicates that a number of naturally open areas are changing in their dominance structure and physical stability. The GRUK data show a generally large spread of scaled values, but the violin plots (the shapes indicate the relative amount of observations across the y-axis) suggest that there is a number of sites with too high nitrogen (Nitrogen2), and some sites with too much disturbance (CSR-R2) and light (Light2). Many GRUK sites are popular outdoor locations with the local population, which could increase disturbance and thus cause deviations towards high CSR-R values (CSR-R2). Increased nitrogen may be an effect of alien species, which the GRUK data records as a major pressure and problem for many monitoring sites.
+colnames(res.natopen.GRUK)[38] <- "fremmedartsdekning" ggplot(res.natopen.GRUK[res.natopen.GRUK$fp_ind=="Nitrogen2",], aes(x=fremmedartsdekning, y=scaled_value)) + geom_point() + @@ -4391,7 +4391,7 @@#> Projected CRS: ETRS89 / UTM zone 33N
-
For GRUK we can zoom in on the area around the Oslofjord…
+boks <- st_bbox(c(xmin = 10.3, xmax = 10.8, ymax = 59.95, ymin = 59.4), crs = st_crs(4326)) tm_shape(regnor,bbox=boks) + @@ -4418,7 +4418,7 @@(ii) both the ANO and the GRUK datasets have a nested structure
Therefore, we need to (i) use a beta-model, that (ii) can account for the nested structure of the data. Here, we apply the following function using either a glmmTMB null-model with a beta-distribution, logit link, and the nesting as a random intercept, or a simple betareg null-model with logit link if the nesting is not extensive enough for a mixed model.
-+-expit <- function(L) exp(L) / (1+exp(L)) # since the beta-models use a logit link, we need to calculate the estimates back to the identity scale @@ -4472,7 +4472,7 @@} }
+# we have to join the ANO and GRUK results spatial objects with the Norway and region mask res.natopen.ANO2 = st_join(res.natopen.ANO2, regnor, left = TRUE) res.natopen.GRUK2 = st_join(res.natopen.GRUK2, regnor, left = TRUE) @@ -4494,7 +4494,7 @@"0.7 - 0.8", "0.8 - 0.9", "0.9 - 1.0", "NA"), title = "index values")
-+# they seem to lie in water # all sites but the southernmost one are in Eastern Norway, the remaining one in Southern Norway summary(res.natopen.GRUK2[is.na(res.natopen.GRUK2$region),"y"]) @@ -4585,7 +4585,7 @@
-Mean index value by region for the lower CSR-R indicator (i.e. index shows deviations towards less ruderal species) from the ANO monitoring data. Numbers in grey fields show the number of observations in the respective region.
+# se tm_shape(regnor) + tm_polygons(col="RR1.ANO.reg.se", title="CSR-R (lower)", style="quantile", palette=(get_brewer_pal(palette="OrRd", n=5, plot=FALSE))) + @@ -4594,7 +4594,7 @@
Standard error to the mean index value by region for the lower CSR-R indicator from the ANO monitoring data. Numbers in grey fields show the number of observations in the respective region.
And here are the corresponding maps for the upper Nitrogen indicator in the GRUK data:
-+## scaled value maps for Nitrogen2 (upper indicator), ASO # mean tm_shape(regnor) + @@ -4603,7 +4603,7 @@
-Mean index value by region for the upper Nitrogen indicator (i.e. index shows deviations towards increased Nitrogen affinity in the plant community) from the GRUK monitoring data. Numbers in grey fields show the number of observations in the respective region.
+# se tm_shape(regnor) + tm_polygons(col="Nitrogen2.GRUK.reg.se", title="Nitrogen (upper), 2 SE", style="quantile", palette=(get_brewer_pal(palette="OrRd", n=5, plot=FALSE))) + @@ -4622,7 +4622,7 @@
We can also compare the unscaled values to the reference distribution in order to identify ecosystem types and functional plant indicators showing a deviation from the expectation. Since CSR-R for ANO and nitrogen for GRUK show some deviations, we exemplify this with these indicators for unscaled values.
CSR-R, ANO:
-+#summary(res.natopen.ANO[!is.na(res.natopen.ANO$original),]$kartleggingsenhet_1m2) #sort(unique(res.natopen.ANO$kartleggingsenhet_1m2)) # 8 NiN-types with data to plot @@ -4650,7 +4650,7 @@legend("topright", legend=c("reference","field data"), pch=c(NULL,1), lty=1, col=c("black","red"),bty="n", cex=1)
Nitrogen, GRUK
-
