3.1. Sociodemographic characteristics and social preferences
A representative sample (n= 200) of participants between 18 and 70 years old was registered, 662 geographical points were obtained, of which 267 correspond to places of Positive Value and 381 to places of Negative Value. The participation of those surveyed registered 44% women and 53% men. The age groups with the highest participation are those between 40 - 65 years old (39.5%) and 25 - 40 years old (4.5%. The main relationship of proximity with the rivers is Residence (49%), followed by Leisure (26%). Regarding the activities that are carried out on the margins, the majority of the participants answered "Walking" (61%), and "Exercising and riding a bicycle" (17.91%). Regarding the landscape of the rivers, 47% of the participants qualified the landscape of the Malacatos River as "Bad" and (43.5%) as "Fair" the landscape of the Zamora River (See Figure 2). Finally, 80% of those surveyed stated that the current state of urban rivers negatively affects their quality of life.
3.2. Spatial distribution and model results
In the first phase, we obtain the distribution of social values based on the locations mapped by the respondents, most of them spread along the two water axes. In this regard, the results of the nearest neighbour spatial statistics, generated by the SolVES model, show that the geographical distribution of these points was not random, since statistically significant grouping patterns are identified, given that all R values are <1 with very negative Z-scores (Brown et al., 2002).
Regarding the AUC, to measure the performance and predictive capacity, the model yielded values >0.9 for most cases, which indicates that it has a good fit for the study area, in addition to the fact that the AUC Test indicates that the model has a useful predictive capacity to transfer social values to other environments (Sherrouse et al., 2014). The results showed that 10 models are transferable, wich would be used in future research to obtain the negative and positive landscape preferences in similar river cities.
Finally, the Maximum Value Index (Max-VI) scores for the two subgroups ranged from 5 to 10. A higher Max-VI indicates stronger interest. In this case, the highest indices are found within the NSV, with “Unpleasantness” being the highest (Max-VI 10), and it also registers the largest number of mapped points (n= 121), which suggests that it is the most important SV for the respondents.
To identify the order of preference and importance of the VS, the indicators with a small R relationship, a large negative Z score, an AUC => 0.9 and the highest Max-VI were considered. The results are summarized in Table 3, for each of the 10 SV types of ES. The respective number of mapped points is also included.
Table 3. Results of statistical values of the SolVES model, R-Value (R < 1), Z Score, R-value and Z score, Training AUC, Test AUC, and Maximum Value Index.
Social Values
|
Count #
|
Nearest Neighbour Analysis
|
AUC
|
Max-VI
|
|
R-Ratio
|
Z-Score
|
Training
|
Test
|
|
PSV
|
1 Aesthetic
|
76
|
0.32
|
-11.3
|
0.9
|
0.85
|
6
|
|
2 Learning
|
40
|
0.53
|
-5.7
|
0.93
|
0.82
|
7
|
|
3 Life sustaining
|
46
|
0.49
|
-6.7
|
0.87
|
0.77
|
5
|
|
4 Recreation
|
55
|
0.42
|
-8.2
|
0.93
|
0.73
|
5
|
|
5 Therapeutic
|
50
|
0.38
|
-8.4
|
0.89
|
0.87
|
5
|
|
NSV
|
6 Flood threat
|
79
|
0.47
|
-9
|
0.95
|
0.98
|
8
|
|
7 Unpleasantness
|
121
|
0.44
|
-11.7
|
0.93
|
0.96
|
10
|
|
8 Unsafe, delinquency & harassment
|
63
|
0.49
|
-7.7
|
0.96
|
0.95
|
8
|
|
9 Little aesthetic value & lack of vegetation
|
45
|
0.53
|
-6
|
0.96
|
0.94
|
7
|
|
10 Poor infrastructure & inaccessible
|
73
|
0.44
|
-9.1
|
0.95
|
0.94
|
9
|
|
Boldface values indicate better results. Abbreviations: Positive Social Values (PSV), Negative Social Values (NSV), Area Under the Curve (AUC), Value Index (VI).
We identify that for the P-SV the classification in descending order is Learning, Aesthetic, Therapeutic, Recreation and Life-Sustaining. In the case of the N-SVs, the descending order is Unpleasantness, Poor Infrastructure & Inaccessible, Flood Threat, Unsafe, Delinquency & Harassment and Little Aesthetic Value & Lack of Vegetation. In addition, it can be seen that the highest social value indices are found in the N-SV group.
3.3. Landscape metrics
To interpret the relative importance and relationship of the biophysical variables used in the model, the percentage of contribution and the percentage of importance calculated by Maxent were considered (Table 4). The DTAG variable was the most significant contributor, with a percentage contribution between 34 - 63% and with the importance of permutation of 26% - 57%, being in both cases the highest values for all the models, in fact in similar investigations social values: Aesthetic, Life-sustaining and Recreation register a greater scope together with urban green spaces (Sun et al., 2019; van Riper et al., 2012).
For "Poor infrastructure & inaccessible" and "Flood threat" 40% and 37% respectively in permutation importance were obtained with the ELEV variable and "Poor infrastructure & inaccessible" with 31% in permutation importance with the SLOPE variable (Table 4). Therefore, the variables LANDFORM and LULC register the least participation.
Table 4. Summary of the environmental variable percentage contribution (Con) and the importance of the permutation (Imp) for each SV.
S O C I A L V A L U E S
|
ELEV
|
LANDFORM
|
LULC
|
SLOPE
|
DTGA
|
%
Con
|
% Imp
|
% Con
|
% Imp
|
% Con
|
% Imp
|
% Con
|
% Imp
|
% Con
|
% Imp
|
PSV
|
1 Aesthetic
|
8
|
33
|
1
|
3
|
5
|
6
|
25
|
11
|
60
|
48
|
2 Learning
|
7
|
19
|
20
|
16
|
7
|
11
|
8
|
7
|
59
|
47
|
3 Life sustaining
|
8
|
18
|
15
|
12
|
7
|
13
|
12
|
12
|
58
|
45
|
4 Recreation
|
7
|
14
|
8
|
11
|
7
|
9
|
15
|
10
|
63
|
57
|
5 Therapeutic
|
3
|
9
|
7
|
16
|
14
|
18
|
17
|
6
|
59
|
50
|
NSV
|
6 Flood threat
|
16
|
17
|
4
|
4
|
6
|
4
|
27
|
22
|
47
|
52
|
7 Unpleasantness
|
12
|
37
|
4
|
4
|
4
|
1
|
35
|
31
|
45
|
26
|
8 Unsafe, delinquency & harassment
|
21
|
40
|
5
|
7
|
4
|
1
|
21
|
16
|
48
|
35
|
9 Little aesthetic value & lack of vegetation
|
16
|
36
|
8
|
6
|
3
|
1
|
22
|
16
|
51
|
41
|
10 Poor infrastructure & inaccessible
|
25
|
34
|
15
|
7
|
2
|
1
|
24
|
23
|
34
|
35
|
Abbreviations: Positive Social Values (PSV), Negative Social Values (NSV), Contribution (Con), Importance (Imp), Elevation (ELEV), Land Use and Land Cover (LULC), horizontal Distance To Green Areas (DTGA).
The value maps generated by SolVES (Fig. 4 - 5) provide a spatial representation of the calculated Max-VI for the PSVs and NSVs, indicating the range and extent of the mapped and classified value index zones. Warm colours represent high VI values.
3.4. Positive Social Values (P-SV) and Negative Social Values (N-SV) Maps
The cartographic results for the five types of PSV of the ES, in general, presented a distributed and extensive distribution along the stream, reaching groups in the urban centre, in the periphery and rural areas, in an approximate extension of 13 km, aspect that is reflected in the delimitation of the study area.
The "El Carmen" sector, a peripheral neighbourhood, considered territoriality close to the rural sector, appears in most PSV maps (Fig 1), mainly in Aesthetic and Learning. The results show that it corresponds to a place of interest for the respondents, which could be due to the natural conditions that it still preserves, because it has two tributaries of the Zamora River, a greater vegetation cover, in addition to being one of the access roads to the Podocarpus Protected Area, since in the urban context the open areas of the fluvial landscape are valuable (Deason et al., 2010; Garcia & Pargament, 2015). It is also worth mentioning that for this locality there were no NSV groups.
Relatively similar distributions were found for Aesthetic, Therapeutic and Life-Sustaining, while they are different for Recreation and Learning the latter corresponds to the highest PSV (7/10) (Table 3). An unusual result presented in similar studies (Sherrouse et al., 2014; Sun et al., 2019; van Riper et al., 2012) where Learning registers low VI, which causality it not to be considered, although it represents one of the valuable contributions of the landscape for which it should be considered to provide well-being to the inhabitants.
The areas with the highest rating appeared around the main green areas of the city: Linear Park "Zamora Huayco", Linear Park "La Tebaida", Jipiro Recreational Park and Zoo, since they offer recreational opportunities and less anthropic conditions in the fluvial landscape, conditions that sustain the sense of place (Gobster et al., 2007; Martín-López et al., 2012).
As shown in Table 4, the variable related to distance to green areas (DTGA) was the most influential for all models. Other research, even though they did not use the metric in the development of the model, found that the highest-scoring SVs were around urban green spaces (van Riper et al., 2012).
These findings demonstrate that mapped points outside the city centre tend to record the positive aspects of the landscape and its ES. In contrast, the NSVs (Fig 5) were clustered in a smaller section corresponding to the urban center
The center corresponds to the area with the highest urban density, where the river landscape is more visible and accessible. It coincides with the confluence of both rivers, very close to the architectural landmark "Puerta de la Ciudad" (Fig 1). In this area of the city, the Malacatos River is channeled, offering a homogeneous and artificial landscape. The Zamora River lacks appropriate pedestrian infrastructure (Fig 2), an aspect that is evidenced in the Poor Infrastructure & Inaccessible map (Fig 5). For all these reasons, the urban center obtained a high representation in all the models, since it evokes feelings of concern, interest and attachment (McCormick et al., 2015).
Regarding the sites where the spatial distribution of places with a high positive and high negative value coincides, it could be because they are locations which the respondents are familiar with, visit often, with high historical and heritage value, or that in general evoke a complexity of activities; these results are consistent with similar studies (Baumeister et al., 2022; Campagne et al., 2018; Plieninger et al., 2013; Rodríguez-Morales et al., 2020). Therefore, the improvement actions should be considered as community concentration sites.
Small spatial clusters of the NSV group were found in peripheral areas to the north of the city, specifically in "Sauces-Norte", near the Zoo. In this locality, thanks of the Zamora River are used as waste and debris dumps, a situation that affects social value and even conflicts with the positive values of the landscape mentioned above, that is, they affect the use of these areas (Özgüner et al., 2012; Weber & Ringold, 2015).
In the Unpleasantness map, we notice how the intensity and grouping outline the rivers in the most consolidated area of the city and around the small parks and squares present there. This negative assessment, corresponding to the SV with the highest index (Max-VI 10), could be due to factors such as lack of vegetation, pollution (Campagne et al., 2018; Larson et al., 2016; Özgüner et al., 2012), bad odors (Hands & Brown, 2002), garbage (Asakawa et al., 2004; Fish et al., 2016), insecurity, and is also related to the rating given by the respondents regarding the current state of the Malacatos River and the Zamora River, "bad" and "regular" respectively, which indicates how the fluvial landscape is perceived by its inhabitants. In addition, the results reveal that the respondents show a greater interest in mapping negative evaluations than positive ones, which makes sense as it is an anthropized ecosystem (Campagne et al., 2018).
The Flood Threat was also an SV of interest, as it could be influenced by recent events such as extreme rains and floods that caused significant damage to the city, and it is also an aspect commonly considered in riverfront regeneration plans (Seidl & Stauffacher, 2013). With these results, we can obtain the perception of risk from the local community.
In NSVs such as Unsafe, Delinquency & Harassment, Little Aesthetic Value & Lack of Vegetation, Poor Infrastructure & Inaccessible, we can identify the relationship between the degraded state of rivers and the perception of crime, areas where the inhabitants want a greater presence of vegetation, as well as the need for pedestrian infrastructure along the riverbanks. These places of conflict with little or no social value have the potential to be transferred as improvements preferences by the community, considering that they could inform about the impact and effects that landscape degradation has on social well-being (Lyytimäki & Sipilä, 2009; Özgüner et al., 2012), in addition to working as an indicator of the loss of SE (Chapin III et al., 2000).
In general, the socio-ecological relationships identified showed that there were preferences and patterns of defined groups, regarding the locations of high and low value.
A variable score was also seen within each group (PSV and NSV), suggesting that the community perceives and experiences the landscape in a complex way and for different purposes (Chapin III et al., 2000; Ives et al., 2017; Larson et al., 2016). Condition that corresponds to the holistic nature of the cultural values of the landscape, since they are made up of a series of meanings, cognitive and affective dimensions (Martín-López et al., 2014; Milcu et al., 2013).