Evaluation of urban wetland landscapes based on a comprehensive model — a comparative study of three urban wetlands in Hangzhou, China

Urban wetlands constitute an integral part of urban ecosystems. Unsurprisingly, wetland landscapes affect the frequency with which urban residents visit wetland areas. Objectively evaluating the beauty of landscapes and exploring the factors that influence their perceived beauty can inform urban wetland landscape design. This study selects three wetlands located in the core, fringe, and suburban areas of Hangzhou, respectively: Xixi wetland, Tongjian Lake wetland, and Qingshan Lake wetland. To explore the impact of ecological design techniques on the scenic beauty of wetlands, the selected wetlands were used to conduct field investigations. In addition, the wetland landscapes were submitted to public evaluations using scenic beauty estimation (SBE). At the same time, the factors affecting the beauty of the wetland landscape were quantified using the semantic differential (SD) method, and the impact paths of ecological design and plant diversity on scenic beauty were analyzed through the structural equation model (SEM). The results indicated that plant richness, cultural expression, color harmony, and coordination of landscape elements have both a significant and positive impact on the scenic beauty of urban wetlands. By way of contrast, habitat diversity had a significant yet negative impact. Across the three urban wetlands, ecological design techniques were found to significantly improve plant diversity and scenic beauty. This study demonstrated that the application of well-thought-out ecological design methods was conducive to the integration of environmental protection, landscape aesthetics, and the sustainable utilization of urban wetlands in the rapid urbanization process, providing a reference for urban wetland landscape conservation, planning, and design.


Introduction
Urban wetlands are an important part of urban ecosystems, performing valuable ecological service functions for urban residents, including water source conservation and water purification to improve water quality (Chen et al 2016, Mi 2022. In light of the increasing attention paid to the protection and construction of urban wetlands in recent years, some damaged urban wetlands have been repaired, thus improving the health of wetland landscapes and ecosystems (Salizzoni et al 2019. It should be stressed that the diverse habitats in urban wetlands are vital for the survival of animals and plants in urbanized areas (Hansson et al 2005, Murry 2019). Meanwhile, the capacity of urban wetlands to maintain local biodiversity has been further improved, most notably through habitat design, and appropriate man-made management and protection measures. At the same time, urban wetlands, as a common public open area, constitute important living environments and open, green spaces in which city residents can engage with nature (Zeng et al 2011). Though there is a clear desire amongst the public to visit wetlands, the beauty of the wetland landscape typically influences the public's intention to visit and the frequency with which they visit. In turn, this mediates the transmission of wetland ecological services to residents and influences the extent to which they participate in wetland protection. Therefore, to achieve sustainable development, it is first necessary to understand the relationship between ecological design and beauty in urban wetlands in the planning, design and management of processes.
Ecological design and restoration have been deployed as part of urban wetland protection and construction efforts (Semeraro et al 2015, Zhang and. For example, the application of various ecological technologies and methods and the use of ecological materials play an important role in improving urban wetland biodiversity (Salizzoni et al 2019, Xu et al 2019. The overarching goal of urban wetland protection and management is to synergistically improve ecological benefits and landscape effects. Notably, researchers have observed a link between the use of ecological design methods to improve biodiversity and public aesthetic preferences (Soliva and Hunziker 2009) and landscape design (Di Giulio et al 2009, Salizzoni et al 2019, such as how tourists are attracted to increased plant diversity (Zhang and Wang 2017). Additionally, the design of natural habitat methods helps to increase regional biodiversity (Kummerling and Muller 2012) and improve wetland landscape quality. Additionally, it should be highlighted that artificial wetland technology can maintain and improve biodiversity (Semeraro et al 2015).
The existing studies on wetlands focus on natural wetlands, with little work examining their urban counterparts. For instance, Zhao et al (2018) looked into the use of ecological technologies, to improve wetland water quality and landscapes on Taihu Sanshan Island, such as wetland ecological revetment, wetland plant optimization, and chain recovery. Similarly, Cai et al (2021) designs a concise ecological restoration plan for Zaozhadian wetland based on close-to-natural wetland restoration strategies. The gap in the research leaves open the question of which ecological design technique characteristics influence the perceived beauty of wetland landscapes in urban areas. Moreover, the interrelationships between ecological design, wetland landscapes, and biodiversity require clarification. By articulating answers to these questions, corresponding management strategies can be formulated and applied to improve the quality of urban wetland habitats, alleviating the tension between urban wetland protection and utilization in the process.
As vision dominates the psychological underpinnings of thought and feeling in human activities (Qiu and Gao 2017), the beauty of wetland landscapes directly affects residents' willingness to visit such areas and engage in protection activities. Since the 1960s, much theoretical and practical research has scientifically evaluated visual landscape environments (Brown andDaniel 1986, Gandy et al 2007). To date, two main landscape evaluation models have been used: an expert evaluation based on the objective attributes of the landscape and a public evaluation based on the viewer's perceptions. Among them, scenic beauty estimation (SBE) is a psychophysics-based landscape aesthetic quality evaluation method that investigates how people's subjective preferences change as landscape attributes change (Daniel 2001). To be precise, by studying the relationship between people's subjective preferences and landscape attributes through SBE, the optimal methods for the scientific planning and design of landscapes can be determined (Han and Dong 2015). However, due to the lack of correspondence between people's visual perceptions, the spatial characteristics of the landscape, and the characteristics of the design elements, correlation and restriction testing between the elements have yet to be carried out. As a result, the evaluation results offer less guidance when applied in a specific practice.
The semantic differential (SD) extracts landscape elements that can be employed as factors to evaluate beauty. Additionally, when combined with factor analysis, SD is able to extract independent factors, establish quantitative models, and evaluate landscape quality (Osgood andSuci 1957, Zhang 2004). Furthermore, when combined with SBE, it reflects the advantages and disadvantages of the landscape, thereby enhancing the objectivity of the results (Cao andZhang 2020, Sun et al 2021). It should be noted that the SD method typically determines the score based on the positive and negative poles of the 'adjective pair' (Cao and Zhang 2020); even though the subjectivity of 'adjective pair' scoring remains strong, including too many 'adjective pair' scoring items may lead the scorers to lose patience, which in turn impacts the objectivity of the score. In response, the expert evaluation method is used to carry out SD evaluation and identify a more objective, accurate, and effective factor evaluation method in relation to scenic beauty. Experts have previously used this method to determine and quantify the landscape factors whilst excluding the subjectivity of the subject and the non-authenticity of the data. Given that it generates more objective results, this approach is better suited to evaluating objective factors Tang 2015, Li et al 2020). In recent years, there has been a growing trend to evaluate scenic beauty using a combination of subjective and objective evaluation methods (Wang et al 2020).
This study uses the structural equation model (SEM) to further explore the relationship between ecological design, biodiversity, and scenic beauty in urban wetlands, whilst also clarifying how the three mutually interact. The SEM has specifically been used to establish, estimate, and test causal relationship models and process complex multivariate data (Li et al 2015). Notably, the SEM can study both observable and hidden variables (i.e., variables that cannot be directly observed) and the indirect effects between variables, whilst also providing measurement errors for each variable (Qu et al 2005). The utilization of the structural equation model provides a theoretical basis for the planning and design of urban wetland landscapes.
In the process of urbanization, urban sprawl subsumes natural land; over time, some wetlands in both ruralurban and rural areas have gradually evolved into urban wetlands. Due to their geographical location, functional positioning, and artificial interference intensity, wetlands located in urban-rural gradients differ from typical wetlands. On this basis, the relationship between landscape aesthetics and biodiversity may be mediated by wetland type. Hangzhou is located in eastern China with rich wetland resources. The area of Wetlands that more than 8 hm 2 is 114900 hm 2 , accounting for 7.0% of the total area of the region (Tian et al 2020). In recent years, a significant area of the natural wetlands has become surrounded by urban build-up, thus driving the expansion of urban wetlands. It must also be stressed that the planning and design of wetland areas have precipitated changes in the landscape, function, and biodiversity of some of Hangzhou's wetlands (Pan et al 2022). Understanding the differences in landscape and biodiversity in wetlands from different locations along the urban-rural gradient in this region can help reduce landscape homogeneity and promote urban wetland protection.
Based on the above, the present study selected three urban wetlands located in the urban core area, fringe area and suburban area of Hangzhou, respectively, to explore how the differentiation of wetland landscape and function, in conjunction with the sustainable utilization of wetlands, can be realized through ecological planning and design. The SBE and SD evaluation methods were combined to evaluate the beauty of wetland landscapes. Additionally, SEM was used to analyze its association with biodiversity and ecological design techniques. The present study aimed to achieve the following: (1) To understand the differences in the beauty of wetland landscapes in different urban areas; (2) To analyze the main factors influencing the beauty of wetland landscapes in different urban areas; (3) To discern the correlation between the beauty of urban wetland landscapes, plant diversity, and ecological design; (4) To articulate theories and scientific bases for the protection, planning, and design of urban wetland landscapes across different regions and functions.

Study area
Each of the representative urban lake wetlands in the urban core, fringe, and suburban areas are located within Hangzhou, Zhejiang Province (29°11′-30°33′ N, 118°21′-120°30 'E) (figure 1). Located in the urban core area, Xixi wetland is a fishing and farming region enjoying natural wetland resources and was also China's first national wetland park. Meanwhile, Tongjian Lake wetland is located at the edge of the city and is a manually excavated lake designed to control, drain, and store flood water, whilst improving the area's freshwater environment and landscape. Finally, Qingshan Lake wetland is located in suburban Hangzhou. Again, it is a large-scale artificial lake designed to assist with flood control and irrigation and improve the downstream water environment, whilst also offering outdoor sports facilities and water sightseeing.

Study framework
The present study developed a multidisciplinary research framework consisting of five steps (figure 2): (1) Photographs were taken and wetland landscape data was recorded through field surveys; (2) Relevant literature was gathered to determine the key factors affecting the landscape; (3) SBE was carried out based on the photographs; (4) SD evaluation was conducted based on key factors affecting the landscape (divided into plant diversity and ecological design); (5) The impact path of ecological design and plant diversity on SBE were assessed using the structural equation model (SEM) (to do so, steps 3 and 4 were synthesized).

Data generation
The SBE score was determined by both landscape's characteristics and the aesthetics of the selectors (Li et al 2007). As a unified landscape evaluation object, the photos taken of the three wetlands in the present study formed the basic data used to facilitate the quantitative beauty evaluation analysis. It should be pointed out that the photos provided a consistent basis for evaluation.
For each wetland, 30 samples were randomly selected, after which the landscape was photographed and recorded. To reduce the impact of the shooting belt on the landscape evaluation, parameters and limiting conditions were standardized when shooting: (1) The researcher shot the photos on Sunny days in November 2021 between 9:00-17:00; (2) The same camera was used to take the photos, fixed at a shooting height of 1.6 m; (3) All of the photographs were shot horizontally; (4) All of the photographs were taken by the same person to maintain a consistent photo composition and selection scale.
To ensure the landscape photos shot for the research accurately reflected the selected wetland landscape and increase their comparability (Holling 2001), four photos were taken at each sample point (to the north, east, south, and west of the centre sample point, respectively) (Tang, Yao (2020)). Due to terrain restrictions, it was not possible to take four photos of some sample points. In total, 320 photos were taken. The photos were screened as follows: (1) Any photos that were blurry, overexposed, or included vehicles or people were removed; (2) The most representative photo was selected for each sample point . Ultimately, 90 valid representative sample photos were picked out and numbered.
Once the photographs were taken, a survey sample plot with an area of 30 m × 30 m was set up at each sample site and a GPS device was used to record the coordinates of the four vertices of the sample plot. At the same time, the plant species, habitat types, community structure levels, number of ecological technologies, and hard landscape types were recorded. The coordinates of each quadrant were determined using Google Earth (https://www.google.com/Earth/). Furthermore, by using the hard landscape information recorded in the sample plot survey, the hard landscape area in each sample plot and its proportions were calculated using the Polygon tool.

Evaluation method
The group of evaluators was made up of both professional and non-professionals: of the 140 were in the professional group, 6 individuals were well-known scholars in this field and 134 landscape were design professionals, whilst the remaining 160 evaluators were non-professionals.
There was no significant difference between the indoor slide evaluation results and the on-site evaluation (Shuttleworth 1980). The 90 selected photos were used to create slideshows. These were played indoors and the evaluators were given time to carry out the SBE scoring. The 7-point beauty evaluation was used as the measurement standard (Habron 1998): in this evaluation scale, 'Love', 'Prefer', ' Like', 'Normal', 'Dislike', 'Hate' and 'Detest' had corresponding scores of 3, 2, 1, 0, −1, −2, −3, respectively. Prior to carrying out the scoring, the scoring purpose and scoring requirements were explained to the evaluators. Additionally, sample slides were shown to familiarize the raters with the process and allow them to establish their own standards prior to commencing with the scoring. For the official scoring, each slide was shown for 8 s (Chen et al 2014), during which time the raters recorded the scores for each landscape on the questionnaire in the order they were presented on the slides.
Although studies have found that different evaluation subjects exhibit a clear degree of consistency in terms of aesthetic attitudes (Buhyoff et al 1983, Xu andLi 2014), the evaluations based on the SD factor layer suggest that the perceptive abilities of the experts and students in the professional group are better than those of the general public (Brown andDaniel 1986, Clay andDanile 2000). Therefore, the SD evaluation of the subjective factor layer was scored by 30 individuals from the professional group, whilst the objective factor layer was quantitatively evaluated by 6 academics from the professional group. When evaluating the subjective factor layer SD, each slide was shown for 1 min (Sun et al 2021).

Beauty evaluation
To minimize the differences in aesthetic standards between different evaluators, a beauty standardization formula was used to standardize the evaluation scores and obtain the beauty value of the on-site landscape. The calculation formula is as follows: where, Z ij is the standardized value of the j-th judge's evaluation of the i-th sample landscape, R ij is the direct score of the j-th judge's evaluation of the i-th landscape and the average value of all the evaluation values of all the landscapes by the j-th judge, and S j is the standard deviation of the evaluation values of all the landscapes by the j-th judge, is the mean standardized score value of the i-th landscape and N j is the total number of judges.

Landscape factor evaluation
The relevant literature was referenced to determine the key factors influencing the perceived beauty of wetland landscapes (Liu and Fan 2013, Amici et al 2015, Schuepbach et al 2016. Subsequently, the Delphi method and factor screening were used to organize and summarize 15 evaluation factors from the perspective of aesthetics and ecological design. Subsequently, 5 subjective evaluation factors and 5 objective evaluation factors for the evaluation of wetland beauty were filtered out, graded, and assigned scores of 2, 1, 0,−1, and −2, respectively (table 1).

Structural equation model
The structural equation model was used to analyze the potential influence of urban wetland ecological design on wetland scenic beauty and plant diversity. The model framework was largely based on the existing academic literature, supplemented by our group's prior field investigation. Previous studies observed that ecological design can maintain and improve biodiversity (Semeraro et al 2015), whilst vegetation type diversity significantly and positively improves landscape visual environment quality (Akbar et al 2003, Zhang andWang 2017). Similarly, the plant diversity and ecological design characterization factors were also based on the prior field survey. It should be noted that SBE was included in the model as an interpreted variable. As ecological design and plant diversity have the potential to affect SBE, a hypothetical framework was constructed ( figure 3). To be precise, the Lavan package in R 3.4.1 was used to run the structural equation model, whilst the adequacy of the model and data was verified using the χ2 test and the root mean square error approximation (RMSEA).

Comparison of SBE
Of the 300 questionnaires distributed, 300 were returned (100% recovery rate). After invalid questionnaires with incomplete scores and questionnaires with exactly the same scores were removed, 290 valid questionnaires remained (97% efficiency).
Regarding the scenic beauty of the three urban wetlands ( figure 4, table 2), the evaluation results show that the SBE range of Qingshan Lake wetland was −0.76 ∼ 0.95, and there were 17 landscapes with SBE values 0. The SBE range of Xixi wetland was −1.15 ∼ 0.84, and there were also 17 landscapes with SBE values 0. The SBE range of Tongjian Lake wetland was −0.85 ∼ 0.87, and there were 12 landscapes with SBE values 0. Among the three urban wetlands, Qingshan Lake wetland achieved the highest scenic beauty score. The SEB value indicated that the respondents' overall evaluations of the landscape quality of Qingshan Lake wetland and Xixi wetland were positive, though they were notably less satisfied with the Tongjian Lake wetland landscape. The coefficients of variation for the CV values of the three wetland scenic beauty score values were very similar (table 2),  indicating that the overall quality of the landscape sample photos selected for the three wetlands was more appropriate. From these results, it can be concluded that there were no excellent or very poor landscapes.
There were 46 landscapes with SBE values 0 in the three wetlands, accounting for 51.1% of the total number of landscapes. This indicated that the respondents' overall preference for wetland landscapes was neutral. The variance analysis showed that there was no significant difference in scenic beauty between the three wetlands (P > 0.05).
Photos with high SBE scores were typically characterized by open water, clear reflections of plants, structures, and the sky, prominent close-ups, middle-views, and long-view levels, rich plant species and community structures, large areas of vegetation, and coordination of artificial landscape elements (figures 5(a)-(f)). Those photos with lower scores tended to feature insufficient landscape levels, single colors, simple community structures, poor water quality (in those landscapes with open water surfaces), exposed soil on the revetment, poor plant growth, and various other factors negatively affecting the landscape (figures 5(g)-(l)).
A cross-listing table was compiled using the photos from the three wetlands. Subsequently, a nonparametric chi-square test was performed using the respective scoring values of the professional group and the non-professional group. The results showed that the overall score of the non-professional group was higher and fluctuated relatively little, whilst the overall score of the professional group fluctuated a lot. It should be noted that most of the score values were lower than the scores of the non-professional group. It is interesting the noted that of the 90 photos, the scores of the professional group and the non-professional group were essentially the same for 81 of the photos. Notably, there were significant differences in the scores of 9 photos, all of which pertained to the landscape photo evaluation of Tongjian Lake wetland ( figure 6). This may be related to the fact that Tongjian Lake wetland, as a newly built urban wetland, has not yet achieved the expected design effect following completion, resulting in low scores from the expert group.

Evaluation of SD
The 10 landscape factors in the three urban wetlands were comprehensively scored and used to draw a box diagram ( figure 7). To compare the landscape factor differences that affect the beauty of various urban wetland types, we conducted a variance analysis of 10 landscape factors (table 3). Further multiple comparison tests were then conducted between the groups to identify the differences that affect the beauty of three different urban wetlands. The results showed that the scores for plant richness, appropriate spatial scale, and coordination of landscape elements in Xixi wetland were significantly higher than those in Qingshan Lake wetland and Tongjian Lake wetland. Meanwhile, the scores for plant community structure, ecological technology, and hard landscape proportion in Qingshan Lake wetland were significantly lower than those in the other two wetlands. There were no significant differences among the three urban wetlands in terms of habitat diversity, naturalness, harmony, and cultural expression.

Wetland landscape evaluation model
The evaluation model of wetland scenic beauty was constructed by taking the SBE standardized value of the scorers as the dependent variable and 10 evaluation factors as the independent variable. SPSS 26.0 was used to carry out backward stepwise regression and the relevant factors affecting the beauty of wetland landscape were determined. The regression coefficient of the wetland landscape evaluation model showed that only the 'color harmony' factor of Qingshan Lake wetland had a significant impact on beauty evaluation (table 4). Meanwhile, the 'plant richness' and 'color harmony' factors of Tongjian Lake wetland significantly impacted the scorers' aesthetic appreciation. Further, the following factors significantly impacted the perceived beauty of Xixi wetland: 'new biodiversity', 'habitat diversity', 'color harmony', 'harmony of landscape elements', 'cultural expression' and 'habitat diversity'. The tolerance of the three models was less than 1 and the variance inflation factor (VIF) value was less than 3. An expansion factor between 1-10 indicates that there is no multicollinearity relationship between the independent variables (Arriaza et al 2004, Yao et al 2012, Xie 2020. This indicates both that the dependent variables and independent variables in the evaluation model formed a significant linear relationship and also that the three regression models were effective.  Figure 5. Photos of the three urban wetlands with higher scenic beauty scores (a)-(f) and lower scenic beauty scores (g)-(l). (QSH: Qingshan Lake wetland; TJH: Tongjian Lake wetland; XX: Xixi wetland). The number represents the corresponding photo number, ranging from photos 1-90. Figure 6. Comparison of professional and non-professional groups' SBE across the three urban wetlands (the solid line and dotted line represent the respective scores of professional and non-professional groups; the numbers 1-30, 31-60, and 61-90 represent the landscape photos of Tongjian Lake wetland, Qingshan Lake wetland and Xixi wetland, respectively).

The relationship between ecological design, plant diversity, and SBE
Using the data collected from the three cases, the model constructed in this study was tested for χ2 (P > 0.05), in addition to conducting the root mean square error approximation (RMSEA < 0.05). Additionally, the model was simplified by removing those variables that had no significant impact. The results show that the model is in  good agreement with the experimental data. The structural equation model constructed from the data derived from the three urban wetlands ( figure 8(a)) shows that ecological design techniques had a direct positive impact on plant diversity and scenic beauty, though no significant relationship was observed between plant diversity and scenic beauty. To further explore how ecological design, plant diversity, and scenic beauty mutually influence each other in urban wetlands across different regions, three separate urban wetland structure equation models were constructed, of which only the structural equation model of Xixi wetland was constructed successfully. The model indicated that in Xixi wetland, ecological design techniques indirectly affect the beauty of the wetland landscape through plant diversity ( figure 8(b)).

Landscape evaluation of urban wetlands in different regions
The average beauty of urban wetland landscapes across the three different regions was similar, and there were no significant differences among the wetlands (table 2). The findings here were similar to the research results of Shen et al (2018) and , though the scores were lower than those observed by Zhang et al (2017) for Xiazhu Lake wetland park in Deqing. The disparity here may be related to the large fluctuation in the beauty value of the park.
Of the sampled areas, Xixi wetland possessed the most significant factors affecting the scenic beauty, while Qingshan Lake wetland had the least. As an urban wetland in the core area, Xixi wetland has been effectively managed and maintained for nearly 20 years. As a result of the passing of time, it is rich in plant species and its plant communities have largely stabilized (Yao et al 2021). Plant diversity has the most apparent impact on the scenic beauty of the wetland. The landscape design of the wetland retains the original fishing and farming style, whilst also incorporating elements such as patterns, symbols, and aspects of folk culture that reflect the broader regional culture. It should also be noted that the color changes and morphological expressions of artificial and natural landscape elements are highly coordinated, and there is an abundance of culturally expressive elements. These factors directly and significantly affected the scenic beauty of Xixi wetland. Regarding the beauty rating of the Xixi wetland landscape, it was also found that the beauty value of 17 samples was 0, and the overall public evaluation was positive.
The Tongjian Lake wetland provides an open space for leisure and entertainment for residents in the fringe area of the city and a large number of ornamental plants have been planted to form a colorful plant landscape effect. Therefore, the plant richness and color harmony of Tongjian Lake wetland were found to play a significant role in the beauty of the landscape. However, since Tongjian Lake wetland is a newly built wetland, it currently features an excessive number of artificial hard landscapes, decreasing the degree of integration with the environment. It can be assumed this is why only 12 samples received a beauty value 0 and the public's positive evaluation was low.
Qingshan Lake wetland is located in an urban-suburban area. In an effort to achieve differentiated development, a plant theme park mainly consists of Taxodium distichum var. imbricatum and a greenway around the lake. The landscape type is relatively homogenous compared to the other two wetlands. Moreover, the water surface is open and the proportion of vegetation is not high in the wetland. Therefore, among the ten factors, only the rich colors formed by artificial greenways and the forestry of T. distichum var. imbricatum significantly impact the beauty of the landscape. However, as a result of the combination of natural mountains and lakes and artificial landscapes in Qingshan Lake wetland, the public highly appraises its beauty (based on the present study's results). In the survey results, there were 17 samples of the wetland with beauty values 0, and the highest scoring samples were also in the Qingshan Lake wetland. It can be seen that the public's overall evaluation of the wetland landscape that integrates man-made and natural elements is also positive.
Through a comparison of three different types of wetlands in the city, the study respondents' preferences for the functional positioning of wetlands near natural ecosystems indicated that in addition to visual influence, landscape evaluation may also be affected by the interviewee's expected psychological positioning of the subject of evaluation.

Key factors affecting urban wetland landscapes
Color harmony is the landscape factor that significantly influenced the perceived scenic beauty in the three wetlands. In the landscape with the highest score in Xixi wetland, spring plants formed a unified green color, accompanied by dark gray buildings and light gray wooden plank roads to form a harmonious landscape color palette. This color harmony was reflected in the preeminence of green plant tones supplemented by structures (figure 5(e)); this observation was consistent with Zhou et al (2019) research suggesting that the larger the proportion of main colors, the higher the beauty value. In the landscape with the highest score, Tongjian Lake wetland, color harmony was mainly manifested in how the color of the sky is reflected by the large surface area of the water, contrasting with the warm colors of autumnal plants (figures 5(c) and (d)). Research from Arriaza et al (2004) concluded that a strong color contrast can significantly improve the quality of the visual environment of the landscape. In the high-scoring Qingshan Lake wetland landscape, the color harmony was reflected in the richness of colors of the autumn scenery of forestry from T. distichum var. imbricatum, which significantly impacted the landscape's perceived scenic beauty. However, as a result of the combination, the artificial colors on the overhead path acted as embellishments ( figure 5(b)). Plant landscape design theory posits that color richness is related to the seasonal phases of plants: in specific seasons, people will be more inclined to spaces with a sense of color (Tang, Yao (2020)). Regarding plant configuration, greater consideration can be given to the use of plants with seasonal performance, paying particular attention to their texture and color matching .
In the construction of wetland plant communities, plant richness plays an important role. A community landscape that lacks species diversity is not in line with the point of view of ecology, nor will it maximize its potential aesthetic value (Duan et al 2018). Diverse vegetation types have been found to have an important, positive effect on improving the quality of the visual environment of the landscape (Cook and Cable 1995, Akbar et al 2003, Yang et al 2009. The landscape photos of Xixi wetland with relatively high beauty values typically featured a wetland landscape with high plant richness and a large area of wetland plants (figures 5(e) and (f)). In the photos of Tongjian Lake wetland, the wetland landscape featuring a combination of natural native plants with artificially planted ornamental plants achieved the highest score and was more popular with the public (figures 5(c) and (d)). Meanwhile, there is less artificial interference in Qingshan Lake wetlands; as such, it has readily formed single-optimal communities and the vegetation type is relatively homogenous. At the same time, there were relatively few artificially planted plants visible in the pictures, thus limiting the plant richness factor effect (figures 5(a) and (b)).
The three landscape factors of cultural expression, coordination of landscape elements, and habitat diversity only played a significant role in the beauty of the landscape in Xixi wetland. By drawing on traditional fishing and farming culture and folklore, Xixi wetland has designed various cultural pavilions and landscape facilities. In the landscape photos of the wetland, patterns, symbols and structures that reflect cultural characteristics can be seen. Additionally, the beauty rating is also higher, which implies that the public may prefer designs that embody cultural expressions. Tongjian Lake wetland and Qingshan Lake wetland feature very few design aspects that reflect regional culture, such that cultural expression factors did not play a significant role in the beauty evaluation of the two wetlands' landscapes.
There are many decisive factors determining the coordination of landscape elements, such as the material, color, form, and style of artificial elements (i.e., roads, structures, and landscape sketches). Landscapes featuring natural materials, colors, and styles that are in harmony with the environment have been found to elicit higher scenic beauty values (Zhang et al 2011). This observation is similar to the evaluation of the scenic beauty of Xixi wetland in the present study and the finding that those landscapes with higher scores tended to have greater use of natural materials (figures 5(e) and (f)). The materials of the artificial landscape elements in the Tongjian Lake wetland were relatively new and the proportion of concrete materials was too high. Moreover, though it is designed to conform to navigation needs, the bridge body was too large and there was minimal coordination with the surrounding environment. Although the number of artificial landscape elements in Qingshan Lake wetland was small and relatively homogenous in type, this factor did not significantly affect the perceived scenic beauty of Qingshan Lake wetland.
In the case of Xixi wetland, it was found that habitat diversity negatively influenced the perceived beauty of the landscape: the higher the habitat diversity, the more negatively the respondents viewed the landscape. Although habitat diversity is conducive to improving biodiversity, with regard to the beauty of the landscape, too many habitats may produce a messy visual perception, thereby affecting the perceived beauty of the landscape. Therefore, in the design of urban wetlands, it is necessary to strike a balance between scenic beauty and habitat diversity protection.

The relationship between ecological design, plant diversity, and SBE
Previous studies have shown that the plant diversity and scenic beauty of urban wetlands can simultaneously be improved through appropriate ecological design (Soliva et al 2009). Broadly, the present study found that ecological design practices positively and significantly impact plant diversity and scenic beauty. Accordingly, ecological design, can not only promote wetland plant diversity, but also enhance wetland scenic beauty ( figure 8(a)). However, the impact path of ecological design on scenic beauty was not consistent across the sampled wetlands. For example, in Xixi wetland, the application of an ecological design method significantly affected plant diversity, thus improving the scenic beauty of the wetland ( figure 8(b)). However, the same is not true of Tongjian Lake wetland and Qingshan Lake wetland, where the relationship was not found to be significant. The reason for this result may be related to the time-delay effect of the impact of ecological design techniques on plant diversity. The structure of the artificially modified or reconstructed wetland ecosystem often takes time to achieve the intended design outcome. As a wetland with a long construction time, the plant diversity of Xixi wetland has had sufficient time to reflect the contribution of ecological design. However, Tongjian Lake wetland and Qingshan Lake wetland are newer, meaning they do not yet directly reflect the relationship between plant diversity and scenic beauty. As shown in figure 8(a), SBE was found to depend on the subjective attitude of the interviewees, which gave rise to a relatively high random error. Therefore, the relationship between ecological design and plant diversity construction model analysis may be the reason for the model's low R 2 .
Xixi wetland is an urban wetland built on the basis of primitive fishing and farming and natural wetland resources. Accordingly, wetland ecological protection and display and wetland viewing are its main functions. The planning and design apply more ecological design techniques, retaining the natural style of the wetland and incorporating a wide range of local plants (Yang and Li 2010). The plant community structure in the wetland is relatively stable. However, it should also be noted that Xixi wetland's management is highly tolerant of autogenous plants in its conservation and management and therefore will not remove these plants (Lu and Xu 2007). This has promoted the improvement of wetland plant diversity and created an urban wetland landscape that is satisfactory and pleasing to the public. Similarly, in their research on rain gardens, Yang et al (2014) proposed that landscape ecological design can improve plant diversity and significantly affect the visual quality of the landscape. Therefore, scientific and well-considered ecological planning and design can provide urban wetlands with a stable and rich plant landscape, thereby enhancing the beauty of the wetland landscape.
The improvement of plant diversity in urban wetlands is conducive to the restoration of wetland functions and enhances the stability of the ecosystem, whilst also providing living conditions for more wild animals in urban wetlands (Denny 1994). In addition, urban wetland plants shape the broader wetland landscape. Through the application of well-thought-out ecological design techniques in the planning and design of new urban wetlands, a beautiful wetland plant landscape featuring a diversity of plants can be created to enhance the public's love and preference for urban wetlands, so as to better cater to the ecological service function of urban wetlands. For existing urban wetlands, an appropriate design plan can be used to allow plants to move in, either naturally or artificially, thus increasing the diversity of wetland plants and further enhancing the beauty of the wetland landscape.

Suggestions for enhancing the beauty of urban wetland landscapes
Xixi wetland is an urban wetland in the urban core area that has adopted ecological protection as its main function. Its plant richness, color harmony, cultural expression, and coordination of landscape elements were each found to positively and significantly affect the perceived beauty of the landscape. It should be stressed that habitat diversity was observed to have the opposite effect. Therefore, in the planning and design of ecologically protected wetlands based on natural landscapes, ecological design techniques should be used to create diversified plant landscapes, whilst also paying close attention to the harmonious collocation of colors. At the same time, the proportions, materials, and forms of landscape element design should be coordinated and integrated into certain cultural design elements. Nevertheless, the types of habitats should be controlled so as not to create a cluttered scene. As a newly built wetland on the fringe of the city, the plant richness and color harmony of Tongjian Lake wetland can positively and significantly affect the beauty of the landscape. However, as it is a newly built park with many artificial landscapes, the degree of integration with the environment is not high, such that its overall beauty was the lowest. Therefore, in a newly built wetland park, a variety of plants can be increased artificially, and attention should be paid to the color matching of the four seasons of the plants. At the same time, the planning and design should retain the original vegetation of the site to the greatest extent possible, reduce excessive construction of the artificial landscapes, and pay attention to the coordination and integration design of the artificial landscape elements and the overall environment. Qingshan Lake wetland, a suburban wetland, utilizes artificial landscape design to increase the beauty of landscape colors, although the wetland landscape type is relatively homogenous. It is recommended to increase the diversity of plants, especially aquatic plants, in future landscape plant design, and strengthen the artificial management and conservation of wetlands.

Conclusion
The SBE model was used to evaluate the beauty of 90 landscape samples taken from three urban wetlands. The results showed that although the respondents' overall evaluations of the landscape environmental quality of Qingshan Lake wetland and Xixi wetland were positive and the evaluation of Tongjian Lake wetland was negative, there were no significant differences between the perceived beauty of the three wetland landscapes. Among the three urban wetlands, the factors affecting scenic beauty were the highest in Xixi wetland and the lowest in Qingshan Lake wetland. Further, the SBE models for each of the three wetlands found that the presence of high plant richness, harmonious colors, adequate cultural expression, and coordinated landscape elements helped to improve wetland beauty; contrastingly, too many habitat types negatively impacted wetland scenic beauty.
The scenic beauty of urban wetlands is affected by many factors. This study demonstrated that a wellthought-out ecological design can improve the plant diversity of urban wetland, thus affecting the landscape of urban wetlands. It is hoped it will provide inspiration for urban wetland planning and design. During the development and utilization of urban wetlands, ecological design techniques should be widely applied in design practice. Doing so is conducive to the integration of ecological protection measures, landscape aesthetics, and the sustainable utilization of urban wetlands in the process of rapid urbanization.