3.1. Ecosystem Service () and Diversity (H) Variables
The results of the ecosystem service assessment are shown in
Table 3. According to the consulted experts, the agricultural ecosystem service is provided by herbaceous crops, woody crops, greenhouses, and heterogeneous lands. Livestock grazing would be generated by woody crops, heterogeneous lands, and grassland. The microclimate regulation service is generated by woody crops, heterogeneous lands, bush, forest, bush-and-forest, and water. The environmental education service is provided by heterogeneous lands, grassland, bush, forest, bush-and-forest, and water. Finally, the tourism service is generated by heterogeneous lands, forest, bush-and-forest, and water.
Therefore, the provisioning, regulating and cultural ESs (i.e., variables ) are generated by the combination of land-uses that generate the individual ESs belonging to each category, i.e., the provisioning ESs considered are generated by herbaceous crops, woody crops, greenhouses, heterogeneous lands, and grassland; while the cultural ESs are generated by heterogeneous lands, grassland, bush, forest, bush and forest and water.
The results of this assessment were used to compute the proportion of cell that generates each individual ES given the distribution of land-uses in each cell of the grid.
Figure 3 shows six maps in which the spatial distribution of the ecosystem service variables (
) and Shannon’s index (
H) is represented at the grid scale. These maps can be analyzed according to the geomorphological units shown in
Figure 1.
The Baetic Depression shows high values for the Agriculture ES, while Livestock grazing and microclimatic regulation ESs take high values just in the eastern region. Eastern Baetic Depression is largely covered by olive groves, which explains the generation of provisioning ESs and the microclimatic regulation (due to carbon sequestration). Cultural ESs show low values in the entire area. On the other hand, the diversity index also shows lower values in this region.
The Sierra Morena mountain range shows high values of regulating and cultural ESs, with the provisioning ES taking lower values, except for the northernmost region. The border between Sierra Morena and the Baetic depression is clearly visible. Regarding the diversity index, cells lying in this region seem to take slightly higher values than those falling in the Baetic Depression.
The Baetic systems present higher variability than the aforementioned geomorphological units. Regarding variables
and
, the higher values correspond to cells located at a lower elevation, which are closer to the Baetic Depression, with woody crops (particularly olive groves) and rain-fed herbaceous crops being the main land-uses. With respect to the
variable, most of the Baetic Systems take high values, with the exception of the low-lying areas, especially the easternmost part, where herbaceous crops dominate. Regarding the cultural ecosystem services and the diversity index, higher values generally correspond to cells lying in protected areas (
Figure 1).
Finally, the Littoral presents low values of , with the exception of a region dominated by greenhouses (the so-called “sea of plastic” of Almeria, located in the Southeast). and , except for the westernmost part, which corresponds with the Doñana National and Natural Parks, also present low values. On the other hand, the westernmost part of the Littoral (i.e., the Atlantic Littoral) excels at having higher terrestrial vertebrate diversity index.
3.2. Description of the Protected Areas
The proportion of protected area (PA) that generates each individual ecosystem service (
) given the distribution of land-uses was obtained. On the other hand, the average value of the Shannon’s diversity index (
H) within each PA was computed.
Table 4 shows the value of these variables per PA. The results show that
Sierras Subbéticas Natural Park (number 11) presents the highest value of
and
, with 44.3% and 52.1%, respectively, while
Bahía de Cádiz Natural Park (number 23) shows the lowest value of these 2 variables, with 0.3% and 1%, respectively. Regarding the
variable,
Sierra de Hornachuelos Natural Park (number 3) presents the highest value (97.5%), while
Del Estrecho Natural Park (number 25) shows the lowest value (67.9%). Concerning the
variable,
Sierra de Andújar Natural Park (number 5) presents the highest value (97.8%), while
Sierras Subbéticas Natural Park (number 11) shows the lowest (61.2%). With respect to the
variable,
Bahía de Cádiz Natural Park (number 23) presents the highest value (92.4%), while
Cabo de Gata - Níjar Natural Park (number 26) shows the lowest (7%). Finally,
Bahía de Cádiz Natural Park (number 23) presents the highest value of
H (1.541), while
Sierra María - Los Vélez shows the lowest value (1.022).
Figure 4 shows the boxplots of the values presented in
Table 4. Variable
has a minimum value of 0.003, maximum of 0.443, median of 0.136, mean of 0.167 and standard deviation of 0.135. Variable
has a minimum value of 0.010, maximum of 0.521, median of 0.187, mean of 0.2117 and standard deviation of 0.134. Variable
has a minimum value of 0.679, maximum of 0.975, median of 0.910, mean of 0.884 and standard deviation of 0.075. Variable
has a minimum value of 0.612, maximum of 0.978, median of 0.926, mean of 0.903 and standard deviation of 0.073. Variable
has a minimum value of 0.070, maximum of 0.924, median of 0.732, mean of 0.689 and standard deviation of 0.196. Finally, variable
H has a minimum value of 1.022, maximum of 1.541, median of 1.351, mean of 1.344 and standard deviation of 0.106. Regarding the ecosystem service variables, it is noticeable that the provisioning ESs have the lowest values, while
,
have the lowest variation.
In order to provide an aggregate value for the different ES categories, the individual ESs belonging to the same category were combined.
Figure 5 displays the union of the provisioning, regulating, cultural ESs, as well as the average of the
H variable per PA. The provisioning ES ranges from 0.010 to 0.526, with the
Sierras Subbéticas Natural Park (number 11) presenting the highest number. It is noticeable that three PAs located in the Sierra Morena mountain range present values above 40%, in particular, those corresponding to numbers 1, 2 and 4. The regulating ES coincides with the
since we did not consider more regulating ecosystem services. The lowest values correspond to PAs located in the Littoral. The cultural ES ranges from 0.612 to 0.978, with the
Sierras Subbéticas Natural Park (number 11) presenting the lowest number and the
Sierra de Andújar Natural Park (number 5) the highest value. In general, all PAs present fairly high values, except for the provisioning case, where values are lower and the standard deviation is higher, i.e., the variability is higher.
3.3. Cluster Analysis
The 26 protected areas were grouped according to their landscape. In order to do that, the percentage of each landscape unit (LSU) within each PA was computed and only the most frequent LSU in each PA was used for the cluster analysis.
The Elbow method was used to determine the optimal number of clusters before carrying out the cluster analysis.
Figure 6 shows the plot of the within-cluster sum of square, i.e., the variance within the groups, for each value of
k from 1 to 10. According to the method used, the optimal number of clusters is 6.
As the data were standardized before running the cluster analysis (i.e., transformed to variables with mean 0 and variation 1, whose values are called z-scores), the cluster means are also standardized and represent standard deviations from the mean. For instance, a z-score of 0 indicates that the value of the original variable coincides with the mean, whereas negative z-scores correspond with values of the variable lower than the mean and positive z-scores with values higher than the mean.
The cluster analysis revealed some characteristics of the six groups.
Table 5 shows the cluster means of each variable (LSU) used to run the cluster analysis and
Figure 7 depicts the membership of the PAs to the different clusters. The first cluster is composed of five PAs and is characterized by landscapes dominated by scrubland-with-forest (cluster mean = 0.86) and scrubland (cluster mean = 1.31). The second cluster only has 1 PA, corresponding with Cabo de Gata Natural Park, which is characterized by volcanic landscapes (cluster mean = 4.9). The third cluster is composed of four PAs (all of them located in the West coast of the study area) and is characterized by natural marshes (cluster mean = 2.13), salty marshes (cluster mean = 1.28) and dunes and sandbanks (cluster mean = 1.94). The fourth cluster comprises three PAs (all of them located in the Sierra Morena Mountain range) and is characterized by the dominance of dehesa (cluster mean = 2.58) and scrubland (cluster mean = 0.66). The fifth cluster is the largest, comprising seven PAs, and is characterized by limestone formations (cluster mean = 1.29) and scrubland with forest (cluster mean = 0.52). The sixth cluster contains six PAs, characterized by pines and conifers (cluster mean = 1.37) and naked soil and snowfield (cluster mean = 0.68). To summarize, we can say that cluster 1 corresponds to regions dominated by scrubland; cluster 2 to volcanic lands; cluster 3 to marshes; cluster 4 to dehesas; cluster 5 to limestone areas; and cluster 6 to regions dominated by pines and conifers.
Regarding the remainder LSU (which were not used to run the cluster analysis because they are not dominant in the PA) it is noticeable that olive groves present an over-average mean z-score in clusters 4 and 5 (0.402 and 0.589, respectively), with the other clusters being below the mean (negative z-score).
Table 6 shows the weighted average of each variable (
,
H) within each cluster, as well as the proportional area of each PA over the total area of PAs belonging to a given cluster. It can be noted that, on average, the area that generates the
,
and
ecosystem services is greater in PAs belonging to cluster 4, where the dehesa landscape dominates. In the case of the provisioning ESs, cluster 4 takes relatively high values (0.412 and 0.436) in comparison to the second best cluster (cluster 5), which takes 0.199 for
and 0.257 for
. On the other hand, the
and
ESs are more extended in PAs belonging to cluster 1, where scrubland landscapes dominate. Finally, the diversity of taxonomic vertebrate classes shows very similar values in all clusters, with cluster 2 presenting the highest value.