Balancing Landscape and Purification in Urban Aquatic Horticulture: Selection Strategies Based on Public Perception

Abstract

In the face of the challenge of urban water resource degradation, green infrastructure construction has become a core strategy in modern urban water resource management. Urban aquatic horticulture (UAH), as an important component of this strategy, possesses the dual value of ecological purification and landscape aesthetics. However, its practical implementation is often constrained by public awareness and acceptance. This study aims to address the mismatch between the dual values of urban aquatic horticulture and public perception, and to develop an optimised plant selection strategy that integrates purification functions with public perception. Based on literature reviews, 18 images of aquatic plant landscapes showcasing different ornamental forms, species richness, and life types were created. A questionnaire survey was conducted on 320 participants to assess their perceptions of landscape aesthetic appeal and visual preferences, and a quantitative relationship model was established using multiple stepwise linear regression analysis. The public’s aesthetic perception of aquatic plant landscapes with different ornamental forms and species richness varies significantly, with flowering plant landscapes more likely to evoke aesthetic perception than non-flowering landscapes. The public’s visual preferences for landscape attributes significantly influence their aesthetic perception of aquatic plant landscapes. A multiple stepwise linear regression equation was established to model the relationship between the aesthetic perception of aquatic plant community landscapes and the public’s visual preferences for landscape attributes. There is no significant association between species richness and perceived landscape aesthetic appeal. The study developed an optimised selection strategy for aquatic plants that integrates purification functions with public perception, providing theoretical basis and practical guidance for the scientific configuration of aquatic horticultural systems in urban green infrastructure. In landscape design, flowering plants with ornamental value should be prioritised, with emphasis on landscape layers, colour, and spatial shaping to enhance public acceptance and promote the sustainable development of urban water resource management.

1. Introduction

Water resources from the foundation for urban development and play crucial roles in energy production, food supply, industrial manufacturing, and environmental quality maintenance [1,2]. However, extreme climate change has significantly affected aquatic environments and ecosystems, with frequent floods, droughts, and water resource degradation becoming increasingly prominent issues [3]. In response to these challenges, green infrastructure development has emerged as a core strategy in modern urban water resource management [4,5]. Urban aquatic horticulture (UAH), as an integral part of green infrastructure, is a comprehensive urban horticultural system that integrates aquatic plant cultivation, water body landscape design, and ecological function optimisation [6,7]. Compared with traditional terrestrial horticulture, UAH exhibits a unique dual-value coupling: first, it purifies urban water bodies through the physiological metabolic processes of aquatic plants [8,9]; second, it creates attractive urban water environments through carefully designed aquatic plant landscape compositions, thereby delivering substantial aesthetic value [10,11]. This dual-value coupling makes UAH indispensable for meeting the twin demands of improving water quality and enhancing landscape aesthetics [12].

With respect to ecological purification, UAH systems can effectively remove nitrogen and phosphorus, heavy-metal ions, and organic pollutants from water bodies through root absorption, stem-and-leaf filtration, and microbial synergy [13,14]. Studies have shown that, after optimised configuration, UAH systems can remove 60–90% of pollutants from urban runoff [15]. For example, Gao et al. reported that appropriately designed aquatic plant landscapes significantly enhance microplastic removal efficiency in urban constructed wetlands [16]. Regarding landscape aesthetics, UAH offers urban residents rich visual experiences and recreational spaces through plant combinations featuring diverse ornamental forms, colour schemes, and seasonal changes [17,18]. Despite this considerable ecological and aesthetic value, much existing research remains confined to laboratory studies or small-scale constructed wetlands. Practical application and promotion are still constrained by public awareness and acceptance [19,20]. Numerous studies indicate that the public’s understanding of UAH is often limited to the traditional “lotus pond” concept, with insufficient awareness of its ecological purification functions [21]. In addition, public aesthetic preferences for aquatic-plant landscapes directly influence decisions on the selection, design, and maintenance of UAH projects [22,23]. This cognitive bias leads to an imbalance in UAH development, where either “aesthetics are prioritised over functionality” or “functionality is prioritised over aesthetics,” hindering full realisation of its dual-value synergy [24].

Landscape aesthetics, as a crucial bridge between ecological functions and public perception, play a key role in UAH design [25,26]. Research indicates that the public’s aesthetic perception of aquatic plant landscapes is primarily influenced by factors such as ornamental form, colour coordination, spatial layering, and seasonal change [27,28]. Among these, flowering aquatic plants—with their vibrant colours and unique morphologies—often evoke stronger aesthetic resonance [29]. Species richness, an important indicator of ecological diversity, also shapes public landscape preferences and aesthetic perceptions [30,31]. Human perception of landscape environments is rooted in visual preferences, which are influenced by multiple dimensions, including consistency, complexity, legibility, and mystery [32,33]. In the context of UAH, public visual preferences not only relate to landscape aesthetic value but also directly affect the social acceptance and sustainability of horticultural projects [34,35]. Therefore, a deeper understanding of aesthetic cognitive differences among the public—together with a quantitative relationship between landscape aesthetic perception and visual preferences—is essential for optimising UAH design [36]. Current research predominantly addresses ecological functions at the technical level, whereas studies on public cognition and aesthetic preferences remain relatively limited [37,38]. A systematic exploration of the mismatch between UAH’s dual values and public perception is lacking, underscoring the need for plant selection strategies that balance ecological benefits with public acceptance [39,40].

Based on this context, the present study seeks to address the disconnect between purification functions and public perceptions in UAH. Unlike previous landscape-first studies [41], this paper adopts an integrated-first approach: combining the purification of polluted aquatic plants—e.g., removing pollutants from leachate or leaking sewage networks [42,43],—with public aesthetic perception to provide highly effective plant combinations suitable for small-scale targeted-treatment blocks. Given that research on pollutant adsorption by aquatic plants is mature, whereas practical urban horticulture focuses more on plant-landscape construction and sourcing ornamental plant material [44], and considering the long duration and high cost of field trials, virtual scenario generation is employed as a back-propagation strategy. The virtual assessment is expected to inspire real-world experiments and lay the groundwork for small-scale treatments (e.g., sewage-network leakage areas), laboratory studies, and artificial wetlands. By systematically analysing public perceptions of aesthetic differences in aquatic plant landscapes with varying ornamental forms and species richness, this study explores the quantitative relationship between landscape aesthetic perceptions and visual preferences and constructs an optimised plant selection strategy that combines purification functions with public perceptions. The findings aim to provide theoretical foundations and practical guidance for the scientific configuration of UAH systems and for future experimental studies. Specifically, the study addresses four core questions:

• Do public perceptions of UAH landscapes with different aesthetic forms and species richness vary?
• Which aquatic plant combination patterns are considered most consistent with the aesthetic requirements of UAH?
• What is the relationship between the perceived aesthetic appeal and visual preferences of aquatic plant landscapes with different aesthetic types?
• How can an optimised plant selection strategy for UAH be developed that integrates purification functions with public perceptions?

2. Materials and Methods

2.1. Materials

A field survey of aquatic plant landscapes in urban parks was conducted to diagnose existing problems and, in parallel, a literature review was undertaken to screen assemblages of widely adaptable aquatic plants with proven purification potential. This survey aimed to bridge the gap between theoretical objectives and practical species selection, thereby informing assemblage design. The study area—Fuzhou City, China—experiences a subtropical monsoon climate with annual precipitation of 1200–1600 mm and acidic lateritic soils (pH 4.5–6.5) typical of the southeastern coastal region. Aquatic plant landscapes in 16 municipal parks (Figure 1a) were examined, and 90 representative photographs were collected (Figure 1b), primarily depicting common species and planting configurations.

Table 1. Aquatic plant species commonly used in the creation of aquatic plant landscapes in Fuzhou city parks (life types, ornamental characteristics).

Rank	Species	Life Type	Ornamental Value	Florescence	Flower Colour	Plant Height/cm
1	Phragmites australis (Cav.) Trin. ex Steud.	Emergent plants	View flowers, view leaves	6–7	White-green, brown	100–200
2	Cyperus papyrus L.	Emergent plants	View leaves	8–11	lavender	100–300
3	Arundo donax var. versicolor (Mill.) Kunth	Emergent plants	View leaves	9–12	Off-white	150–200
4	Miscanthus sinensis ‘Gracillimus’	Emergent plants	View leaves	9–10	Pink, red, silvery white	100–200
5	Cyperus involucratus Rottb.	Emergent plants	View leaves	6–8	Lavender	60–150
6	Acorus calamus L.	Emergent plants	View leaves	6–9	Yellow, purple, red, white	30–80
7	Hymenocallis littoralis (Jacq.) Salisb.	Emergent plants	View leaves	6–7	White	30–70
8	Persicaria hydropiper (L.) Spach	Emergent plants	View leaves	5–9	White, light red	70–90
9	Schoenoplectus tabernaemontani (C. C. Gmel.) Palla	Emergent plants	View leaves	7–9	Onion brown, purple, brown	100–200
10	Typha orientalis C. Presl	Emergent plants	View leaves	6–7	Beige	150–250
11	Sagittaria trifolia subsp. leucopetala ‘Flore Pleno’	Emergent plants	View flowers, view leaves	8–10	White	70–120
12	Alisma plantago-aquatica L.	Emergent plants	View flowers, view leaves	5–10	White, pink	50–100
13	Juncus effusus L.	Emergent plants	View leaves	4–7	light green	35–100
14	Nelumbo nucifera Gaertn.	Emergent plants	View flowers	6–9	Red, pink, white, purple	35–50
15	Canna glauca L.	Emergent plants	View flowers	6–10	Red, bright yellow, red powder, orange yellow	60–150
16	Pontederia cordata L.	Emergent plants	View flowers, view leaves	5–10	Purple, light blue	50–150
17	Iris tectorum Maxim.	Emergent plants	View flowers, view leaves	6–7	Purple, blue, pink, white	20–50
18	Lythrum salicaria L.	Emergent plants	View flowers, view leaves	7–9	Light purple, red purple	30–100
19	Thalia dealbata Fraser	Emergent plants	View flowers, view leaves	4–10	Mauve	100–250
20	Pontederia korsakowii (Regel & Maack) M.Pell. and C.N.Horn	Free-leaved plants	View flowers, view leaves	7–8	Baby blue	50–90
21	Nymphaea tetragona Georgi	Free-leaved plants	View flowers	6–8	Red, White, Pink, Yellow	40–190
22	Victoria amazonica (Poepp.) Sowerby	Free-leaved plants	View flowers	7–8	White	5–10
23	Nymphoides peltata (S. G. Gmel.) Kuntze	Free-leaved plants	View flowers, view leaves	5–10	Golden yellow	3
24	Pontederia crassipes Mart.	Free-floating plants	View flowers, view leaves	7–10	Lavender	20–40
25	Pistia stratiotes L.	Free-floating plants	View leaves	-	Green	-
26	Ludwigia peploides subsp. stipulacea (Ohwi) Raven	Free-floating plants	View leaves	5–6	Golden yellow	-
27	Myriophyllum verticillatum L.	Submerge plants	View leaves	-	Jasmine	30–60
28	Hydrilla verticillata (L. f.) Royle	Submerge plants	View leaves	5–6	White	-
29	Vallisneria natans (Lour.) H. Hara	Submerge plants	View leaves	8–9	Green or dark purple red	40–80
30	Ottelia alismoides (L.) Pers.	Submerge plants	View leaves	7–9	White or light blue	-

lists the species most frequently employed in Fuzhou, including Canna glauca L., Pontederia cordata L., and Thalia dealbata Fraser, together with their growth habits and key ornamental traits. The survey revealed five recurrent issues:

• pronounced homogenization of aquatic plant landscapes;
• limited species and life-form diversity;
• inappropriate planting combinations;
• low ecological and aesthetic returns;
• minimal integration of aquatic plants with water pollution control.

Phytoplankton, floating-leaved plants, and emergent macrophytes constitute the most visible components of an aquatic plant landscape, whereas submerged species remain below the water surface and are therefore difficult to observe. Accordingly, this study centres on three functional groups—phytoplankton, floating-leaved plants, and emergent macrophytes. Because flowering and foliage traits dominate landscape design, photographic data were used to catalogue the aquatic species prevalent in the Fuzhou region and to analyses their ornamental attributes and life forms. From this catalogue, nine species with demonstrated purification capacity and stable use in urban parks were chosen (emergent plants: Canna glauca L. [45,46,47], Nelumbo nucifera Gaertn. [48,49], Cyperus papyrus L. [50], Arundo donax ‘Versicolor’ [51], Iris tectorum Maxim. [52], Thalia dealbata Fraser [53]; free-leaved/free-floating plants: Pistia stratiotes L. [54], Nymphoides peltata (S. G. Gmel.) Kuntze [55], Nymphaea tetragona Georgi [56]). Three ornamental categories were defined—flowering, foliage, and flowering + foliage.

Single-, double-, and triple-species assemblages were then constructed in accordance with landscape strata and life-form complementarity (

Table 2. Aquatic plant combinations. The groups are numbered as follows: F1: Canna glauca L. F2: Nelumbo nucifera Gaertn. F3: Canna glauca L. + Nelumbo nucifera Gaertn. F4: Canna glauca L. + Pistia stratiotes L. F5: Nelumbo nucifera Gaertn. + Pistia stratiotes L. F6: Canna glauca L. + Nelumbo nucifera Gaertn. + Pistia stratiotes L. L1: Cyperus papyrus L. L2: Arundo donax ‘Versicolor’. L3: Cyperus papyrus L. + Arundo donax ‘Versicolor’. L4: Cyperus papyrus L. + Nymphoides peltata (S. G. Gmel.) Kuntze. L5: Arundo donax ‘Versicolor’ + Nymphoides peltata (S. G. Gmel.) Kuntze. L6: Cyperus papyrus L. + Arundo donax ‘Versicolor’ + Nymphoides peltata (S. G. Gmel.) Kuntze. FL1: Iris tectorum Maxim. FL2: Thalia dealbata Fraser. FL3: Iris tectorum Maxim. + Thalia dealbata Fraser. FL4: Iris tectorum Maxim. + Nymphaea tetragona Georgi. FL5: Thalia dealbata Fraser + Nymphaea tetragona Georgi. FL6: Iris tectorum Maxim. + Thalia dealbata Fraser + Nymphaea tetragona Georgi.

Aquatic Plant Life Type	Viewing Type
	View Flowers (F)	View Leaves (L)	View Flowers, and Leaves (FL)
Emergent plants	Canna glauca L.	Cyperus papyrus L.	Iris tectorum Maxim.
	Nelumbo nucifera Gaertn.	Arundo donax ‘Versicolor’	Thalia dealbata Fraser
Emergent plants + Emergent plants	Canna glauca L. + Nelumbo nucifera Gaertn.	Cyperus papyrus L. + Arundo donax ‘Versicolor’	Iris tectorum Maxim. + Thalia dealbata Fraser
Emergent plants + Free-leaved plants/Free-floating plants	Canna glauca L. + Pistia stratiotes L.	Cyperus papyrus L. + Nymphoides peltata (S. G. Gmel.) Kuntze	Iris tectorum Maxim. + Nymphaea tetragona Georgi
	Nelumbo nucifera Gaertn. + Pistia stratiotes L.	Arundo donax ‘Versicolor’ + Nymphoides peltata (S. G. Gmel.) Kuntze	Thalia dealbata Fraser + Nymphaea tetragona Georgi
Emergent plants + Emergent plants + Free-leaved plants/Free-floating plants	Canna glauca L. + Nelumbo nucifera Gaertn. + Pistia stratiotes L.	Cyperus papyrus L. + Arundo donax ‘Versicolor’ + Nymphoides peltata (S. G. Gmel.) Kuntze	Iris tectorum Maxim. + Thalia dealbata Fraser + Nymphaea tetragona Georgi

). These assemblages were derived from the field survey taxa and cross-referenced with purification performance reported for artificial wetlands, notably high nitrogen and phosphorus removal efficiencies and enhanced visual layering (e.g., emergent for vertical structure and floating-leaved species for surface cover). This procedure produced three plant-configuration models comprising 18 experimental groups for subsequent virtual-simulation assessment.

Landscape photographs have long been used to quantify human perceptual responses [28]. In the present study, Lumion Pro v10.3.2 was employed to generate photorealistic renderings of 18 aquatic plant assemblages. The software accurately reproduces plant morphology and allows strict control over scene variables; its principal interface is displayed in Figure 2. A standard waterfront template was selected, and the specified aquatic plants were positioned identically across all scenes. The camera was set to Photo mode with a viewer height of 1.70 m and a focal length of 16.9 mm; additional rendering parameters were fine-tuned to maximise realism. All images were exported at a uniform resolution and aspect ratio, yielding 18 stimulus photographs (Figure 3). The substitution of photographic images for actual landscapes is well established in landscape perception research [57].

2.2. Questionnaire Design and Data Collection

2.2.1. Structure of the Questionnaire

The questionnaire comprised three sections, each aligned with the study’s goal of evaluating aquatic plant landscape aesthetics and visual preferences.

• Section 1 contained five items: age, education, occupation (specialty), and viewing preference (foliage vs. flowers)—to capture respondent demographics and general plantscape interests.
• Section 2 posed a single item requesting an overall aesthetic rating of aquatic plant landscapes.
• Section 3 formed the Visual Preference Survey. To meet the objectives of this research, six well-established visual landscape attributes were expanded to seven metrics: ornamental value, visual richness, hierarchical structure, colour, spatial quality, naturalness, and overall preference (Table 3).

These seven items were derived from a synthesis of botanical traits and traditional landscape assessment factors [58]. Each attribute underwent a multi-step validation to ensure scientific soundness and relevance to public preferences:

• candidate species were first matched with findings from aquatic plant purification studies and constructed wetland experiments to confirm ecological pertinence;
• attribute lists were then cross-checked against current experimental needs to guarantee coverage of features that embody purification–aesthetic synergy;
• finally, wording was refined to reflect the present study’s focus on purification–aesthetic harmonisation.

The resulting scale demonstrated excellent psychometric properties: exploratory factor analysis yielded KMO = 0.866 and Bartlett’s test p < 0.001; internal consistency was high (Cronbach’s α = 0.967)—well above the accepted threshold for reliability [59,60]—indicating that the instrument provides an adequate and coherent description of public landscape preferences and perceptions.

Table 3. Survey dimensions, indicators, and questions of the questionnaire.

Survey Dimension	Indicator (Abbr.)	Questionnaire Item	Selection Rationale
Plant landscape aesthetics	Landscape Beauty Perception (LBP)	“The plant landscape in the photograph is aesthetically pleasing.”	Positive scores typically arise from visual harmony, colour diversity, and overall attractiveness. LBP was retained because ornamental species used in aquatic plant purification studies have been shown to boost pollutant removal while simultaneously elevating public appreciation. However, the relative merits of monospecific versus mixed assemblages remain unresolved; a broader species palette is therefore needed to reconcile purification stability with visual quality [61].
Visual preference	Plant Landscape Preference (PLP)	“I find this plant landscape attractive.”	PLP captures the respondent’s overall liking, which integrates harmony and visual dynamism. The item allows direct comparison of monoculture and mixed planting, a key question in both purification and design research [62,63].
	Aquatic Plant Ornamental (APO)	“I think the plants in the picture are highly ornamental.”	High ornamental value—elegant form and seasonal change—is central to APO. The indicator links aesthetic appeal with documented differences in pollutant uptake among floral and foliar types, encouraging combinations that maximise both functions [64,65,66,67].
	Plant Landscape Visual Richness (PLVR)	“The botanical landscape offers a variety of visual experiences.”	Richness reflects species diversity and textural dynamics. Including PLVR addresses how mixed cultures influence water quality and viewer satisfaction, reinforcing the role of diversity in purification–aesthetic synergy [68,69]
	Plant Landscape Layers (PLL)	“The plant landscape has rich vertical layering.”	Vertical structure enhances both depth perception and hydraulic performance. PLL tests whether adding floating-leaved or submerged tiers improve microplastic removal and visual depth relative to single-layer plantings [16,70,71,72].
	Plant Landscape Colour (PLC)	“The colours of the plant landscape are attractive to me.”	Colours, especially from flowering and variegated foliage—correlates with plant health and nutrient uptake. PLC gauges the ability of varied colour schemes to balance purification efficiency with visual stimulation [73,74,75].
	Plant Landscape Space (PLS)	“The spatial arrangement of the planted landscape is appealing.”	Perceived openness or enclosure depends on layout density and surrounding context. PLS evaluates how mixed plantings, and gradient arrangements optimise both water coverage (for pollutant dispersion) and viewer comfort [16,41,76,77].
	Plant Landscape Naturalness (PLN)	“This botanical landscape feels close to nature.”	Natural integrates ecological authenticity with aesthetic realism. PLN was included to ensure that design solutions maintain wetland integrity while enhancing viewer connection; diverse combinations generally achieve higher naturalness than single-species displays [78,79].

Table 3. Survey dimensions, indicators, and questions of the questionnaire.

Survey Dimension	Indicator (Abbr.)	Questionnaire Item	Selection Rationale
Plant landscape aesthetics	Landscape Beauty Perception (LBP)	“The plant landscape in the photograph is aesthetically pleasing.”	Positive scores typically arise from visual harmony, colour diversity, and overall attractiveness. LBP was retained because ornamental species used in aquatic plant purification studies have been shown to boost pollutant removal while simultaneously elevating public appreciation. However, the relative merits of monospecific versus mixed assemblages remain unresolved; a broader species palette is therefore needed to reconcile purification stability with visual quality [61].
Visual preference	Plant Landscape Preference (PLP)	“I find this plant landscape attractive.”	PLP captures the respondent’s overall liking, which integrates harmony and visual dynamism. The item allows direct comparison of monoculture and mixed planting, a key question in both purification and design research [62,63].
	Aquatic Plant Ornamental (APO)	“I think the plants in the picture are highly ornamental.”	High ornamental value—elegant form and seasonal change—is central to APO. The indicator links aesthetic appeal with documented differences in pollutant uptake among floral and foliar types, encouraging combinations that maximise both functions [64,65,66,67].
	Plant Landscape Visual Richness (PLVR)	“The botanical landscape offers a variety of visual experiences.”	Richness reflects species diversity and textural dynamics. Including PLVR addresses how mixed cultures influence water quality and viewer satisfaction, reinforcing the role of diversity in purification–aesthetic synergy [68,69]
	Plant Landscape Layers (PLL)	“The plant landscape has rich vertical layering.”	Vertical structure enhances both depth perception and hydraulic performance. PLL tests whether adding floating-leaved or submerged tiers improve microplastic removal and visual depth relative to single-layer plantings [16,70,71,72].
	Plant Landscape Colour (PLC)	“The colours of the plant landscape are attractive to me.”	Colours, especially from flowering and variegated foliage—correlates with plant health and nutrient uptake. PLC gauges the ability of varied colour schemes to balance purification efficiency with visual stimulation [73,74,75].
	Plant Landscape Space (PLS)	“The spatial arrangement of the planted landscape is appealing.”	Perceived openness or enclosure depends on layout density and surrounding context. PLS evaluates how mixed plantings, and gradient arrangements optimise both water coverage (for pollutant dispersion) and viewer comfort [16,41,76,77].
	Plant Landscape Naturalness (PLN)	“This botanical landscape feels close to nature.”	Natural integrates ecological authenticity with aesthetic realism. PLN was included to ensure that design solutions maintain wetland integrity while enhancing viewer connection; diverse combinations generally achieve higher naturalness than single-species displays [78,79].

Participants rated the simulated UAH scenes according to their perceived beauty and visual preference [80]. Each scene was assessed with one LBP and seven visual preference items (PLP, APO, PLVR, PLL, PLC, PLS, PLN). Ratings were recorded on a 7-point Likert scale (1 = “not at all”, 7 = “very much”), ensuring that scores captured respondents’ immediate perceptual judgements. Before each rating block, a concise explanation of the attribute under evaluation was provided. The questionnaire, administered in Chinese, employed standardised terminology to maximise clarity and public acceptance. All items were mandatory; incomplete questionnaires could not be submitted.

2.2.2. Data Collection

An internet-based survey was administered via Questionnaire Star, a widely used Chinese platform that supports large-scale public polling and thereby avoids the higher costs of on-site data collection. Prior studies have confirmed the reliability of web surveys conducted through this service [81]. Although online sampling cannot fully control respondent composition, the resulting pool encompassed diverse demographic groups, and no significant between-group differences were detected in preliminary analyses [82]. A snowball sampling strategy was adopted: initial participants were asked to distribute the survey link through social media networks, and each was reminded not to complete the questionnaire more than once. Upon opening the survey, respondents saw the prompt: “Imagine you are physically present in the landscape shown; please rate it according to your personal perception.” Participants could revise any rating before submission, but the form could be sent only after all items were completed, preventing partial responses.

Data collection ran from October to November 2022. A total of 347 questionnaires were returned; after quality screening—(1) ≥8 identical consecutive ratings, indicating inattentiveness, and (2) completion time < 2 min—320 valid responses remained, in accordance with established SCI protocols for online survey validation.

2.3. Data Analysis

Scenic beauty evaluation (SBE) is a widely accepted method for measuring public aesthetic perception to visually respond to the public’s assessment of the beauty of aquatic plant landscapes. However, due to variations in public aesthetic evaluation criteria, the precision of the evaluation results may be compromised. Consequently, the questionnaires collected were initially standardised. The landscape beauty value of a single photograph is the average single photograph normalised to all landscape beauty scores; this method has been well-established in previous studies [83]. The following techniques are applied to standardise landscape views [84,85,86]:

$$Z_{ij}=(R_{ij}-{\overline{R}}_j)/S_j,$$

(1)

$${\overline{R}}_j={\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n{R}_{ij}},$$

(2)

where $Z_{ij}$ is the standardised landscape beauty score of the j participant for the Chapter i photo; $R_{ij}$ is the landscape beauty score of the j participant for the i photo; ${\overline{R}}_j$ is the average of the j participant’s scores of the beauty of all photos of the landscape; $S_j$ is the standard deviation of the j participant’s score of the beauty of all photos of the landscape.

Statistical analysis utilised EXCEL 2016, SPSS 25.0 software, and Python 3.13 for comprehensive visualisation analysis. The questionnaire’s dependability was first evaluated. Next, the beauty scores of different aquatic plant landscape combinations were determined. Then, multiple linear stepwise regression analysis (stepwise) was performed to determine whether the parameters conformed to normal distribution for landscape beauty perception and visual preference landscape attributes (aquatic plant ornamental, aquatic plant landscape preference, visual richness, landscape level, colour, space, and naturalness) of different aquatic plant landscape combinations. Python 3.13 was employed to generate interactive heatmaps, correlation matrices, and regression coefficient visualisations, enhancing the interpretability of complex relationships between variables across different plant combinations. When different ornamental forms were integrated with different species richness indicators of aquatic plant species, the values were determined by averaging the indicators for each category. The parameters follow a normal distribution, with a KMO value of 0.866. Bartlett’s sphericity test yielded a p-value of less than 0.001, indicating the data is appropriate for factor analysis. Furthermore, a Cronbach’s α of 0.967 demonstrates excellent internal consistency for the scale.

3. Results

3.1. Basic Description of the Participants’ Situation

An examination of the 320 valid questionnaires shows that 42% of respondents were male and 58% female. Educational attainment was distributed as follows: 45 respondents (14%) had a high-school diploma or below, 135 (42.2%) held a college degree, and 140 (43.8%) held (or were pursuing) a graduate degree. Age composition was 6 respondents (1.9%) under 18 years, 165 (51.6%) between 18 and 25 years, 77 (24.1%) between 26 and 35 years, 30 (9.4%) between 36 and 45 years, and 42 (13%) between 46 and 60 years. Majors or professions most frequently reported included animation design, atmospheric science, landscape architecture, environmental design, environmental engineering, and computer science. With respect to viewing preference, 216 participants (67.5%) favoured flowers, whereas 104 (32.5%) preferred foliage.

3.2. Analysis of Differences in Aesthetic Appreciation of Urban Aquatic Horticultural Landscapes

Based on the SBE results (Figure 4), the 18 aquatic plant landscape schemes differed markedly in ornamental form. As presented in Figure 4a, the five highest-rated combinations were all flowering types, with SBE scores F6 (0.487) > F3 (0.480) > F2 (0.267) > F1 (0.241) > F5 (0.165). At the opposite end, the lowest score was recorded by L5 (−0.367), followed by L1 (−0.349), L2 (−0.324), and L4 (−0.314); all four are foliage-based and thus carry negative SBE values. Figure 4b further shows that within the six top-scoring schemes, F6 and F3 occupy first and second place, both significantly higher than the remaining configurations.

Ornamental form exerts a consistent influence on public aesthetic perception. The boxplot in Figure 4c indicates that flowering schemes are generally positive (median about 0.25), foliage schemes negative (median about −0.25), and mixed schemes centred on zero (median near 0). Mean values in Figure 4d reinforce this trend: flowering, 0.29 ± 0.15 > mixed, 0.01 ± 0.13 > foliage, −0.29 ± 0.05 (p < 0.01).

Species richness affects perception in a more nuanced way. In both flowering and foliage sets, the two best-rated schemes share the structure “emergent + floating-leaf + free-floating > emergent + emergent.” The top combination, F6 (0.487), offers high ornamental value, greater richness, and diverse life forms. Yet richness is not strictly linear: the two-species mix F3 (0.480) nearly matches F6, and the Nelumbo nucifera Gaertn. monoculture F2 (0.267) also ranks highly. Species identity matters as well: adding Nymphoides peltata (S. G. Gmel.) Kuntze lowers the rating of variegated reed Arundo donax ‘Versicolor’ (L5 < L2), whereas introducing Nymphaea tetragona Georgi markedly boosts the Iris tectorum Maxim. (FL4 > FL1), as Figure 4a illustrates.

Overall, Figure 4 analyses supply clear, quantitative guidance for selecting urban aquatic–horticultural species and configurations. They confirm the pivotal role of flowering traits in elevating public aesthetic perception and unveil the complex interactions between species composition and perceived landscape quality.

3.3. Analysis of Perceptual Aesthetics and Visual Preference Attributes in Urban Aquatic Horticultural Landscapes

Regression modelling of the urban aquatic dataset (Figure 5) quantifies the relationship between LBP and specific visual preference attributes. As shown in Figure 5a, the adjusted R² values for the individual planting schemes range from 0.422 to 0.644; the Arundo donax ‘Versicolor’. monoculture L2 offers the greatest explanatory power (R² = 0.644), followed by Nelumbo nucifera Gaertn. + Pistia stratiotes L.F5 (R² = 0.607). When the schemes are pooled by ornamental type (Figure 5b), foliage combinations display the highest mean R² (0.572 ± 0.063), mixed combinations follow (0.562 ± 0.010), and flowering combinations have the lowest mean R² (0.501 ± 0.086); differences among the three groups, however, are not statistically significant.

Coefficient inspection (Figure 6a,b) shows that APO is the single most influential factor: it exhibits a highly significant positive effect (p < 0.01) in eight of the eleven models, with standardised coefficients between 0.191 and 0.521 (median = 0.297). Secondary predictors include PLVR, PLL, and PLC; the correlation matrix in Figure 7a reports their Pearson coefficients with LBP as 0.47, 0.42, and 0.39, respectively. The five best-performing schemes—F5, F2, L1, L2, and FL1 (Figure 7b)—all reach R² values between 0.562 and 0.607, indicating that their visual preferences can be forecast accurately with the chosen variables.

Patterns differ by ornamental category. Flowering schemes: the Nelumbo nucifera Gaertn. + Pistia stratiotes L. mix F5 fits best (R² = 0.607), with PLL, APO and PLVR all highly significant (p < 0.01); the Nelumbo nucifera Gaertn. monoculture F2 ranks second (R² = 0.561), where APO, PLP, PLC and PLVR are all highly significant (p < 0.01). Foliage schemes: the Arundo donax ‘Versicolor’ reed L2 performs best (R² = 0.644); APO, PLVR, PLC, and PLP each correlate positively with LBP at the p < 0.01 level and together explain about 57% of the variance. Mixed schemes: the Iris tectorum Maxim. FL1 shows moderate explanatory power (R² = 0.556); APO, PLP, and PLL remain highly significant (p < 0.01). In Figure 7a, APO correlates moderately with Plant Layout Structure (PLS) (r = 0.53), suggesting that ornamentality and spatial structure interact through mechanisms such as visual complexity or balance.

Taken together, the results identify APO as the fundamental determinant of LBP, yet the most accurate predictive models arise from tailored combinations of attributes. Foliage-oriented schemes yield more stable and predictable aesthetic outcomes, challenging the conventional assumption that flowering plants always dominate landscape appeal. These findings provide quantitative guidance for urban aquatic plant design and underscore the need for precise attribute configuration according to plant type and combination characteristics.

3.4. Analysis of the Relationship Between Perceptual Aesthetics and Visual Preferences in Different Types of Urban Aquatic Horticultural Landscapes

Figure 8 presents the regression results linking LBP to visual preference attributes for the three ornamental types. As shown in Figure 8a, foliage (L) combinations achieve the best goodness of fit (R² = 0.793, adjusted R² = 0.790), clearly outperforming mixed (FL) combinations (R² = 0.718, adjusted R² = 0.715) and flowering (F) combinations (R² = 0.666, adjusted R² = 0.661). Thus, the selected visual preference variables explain roughly 79% of the variance in the aesthetic perception of foliage landscapes.

The heat map in Figure 8b highlights distinct coefficient patterns across the three types. In foliage landscapes, APO shows the largest coefficient (0.521), followed by PLC (0.274). In flowering landscapes, the constant term (0.261) and PLVR share the highest magnitude. In mixed landscapes, the constant term (0.404) and PLP dominate. These differences indicate that the key drivers of LBP vary fundamentally by ornamental type.

Figure 8c lists coefficient values, significance levels, and regression equations for each type.

• Flowering landscapes: APO, PLN, PLP, and PLL each display highly significant positive correlations with LBP (p < 0.001), and PLVR is also significant (p < 0.05). The model explains 66.6% of the variance, yielding: LBP = 0.261 + 0.191 APO + 0.161 PLN + 0.180 PLP + 0.123 PLL + 0.261 PLVR.
• Foliage landscapes: the APO coefficient (0.521, p < 0.001) far exceeds all others, followed by PLC (0.274, p < 0.001) and PLN (0.161, p < 0.01). The constant term is negative (−0.140) and non-significant, suggesting that additional, unmeasured visual factors may be involved. The equation is: LBP = −0.140 + 0.521 APO + 0.274 PLC + 0.161 PLN + 0.114 PLL.
• Mixed landscapes: a more balanced influence pattern appears, and PLP surpasses APO for the first time (coefficients 0.295 and 0.272, respectively). The constant term (0.404) is highly significant (p < 0.001), implying that stable intrinsic factors also contribute. All predictors reach high significance (p < 0.001), giving: LBP = 0.404 + 0.295 PLP + 0.272 APO + 0.227 PLL + 0.158 PLC.

Overall, the results in Figure 8a–c demonstrate type-specific aesthetic mechanisms: foliage landscapes are chiefly driven by APO, flowering landscapes by PLVR together with several visual attributes, and mixed landscapes by PLP. These insights provide a quantitative basis for designing urban aquatic horticultural scenes tailored to each ornamental category.

3.5. Analysis of the Relationship Between Species Richness, Landscape Aesthetic Perception, and Visual Preferences in Urban Aquatic Horticulture

Figure 9 summarises the regression results for LBP versus visual preference attributes after grouping the schemes by species richness. As shown in Figure 9a, the two-species configuration (Combination Two) provides the best model fit (R² = 0.717, adjusted R² = 0.713), followed by the single-species configuration (Combination One) (R² = 0.684, adjusted R² = 0.679). The three-species configuration (Combination Three) shows a noticeably lower explanatory power (R² = 0.527, adjusted R² = 0.521). These results point to a non-linear relation between richness and predictive accuracy, with the two-species schemes performing best.

The heat map in Figure 9b visualises the coefficient patterns. For single-species schemes, the constant term is largest (0.481), followed by APO (0.297). In two-species schemes, APO (0.316) and PLC (0.274) predominate. In three-species schemes, an unusually large constant term (1.086) is observed, with PLP (0.304) and APO (0.268) next in importance. This shifting distribution suggests that the mechanism linking richness to perceived aesthetics changes as additional species are introduced.

Figure 9c lists coefficients, significance levels, and equations.

• Single-species landscapes: APO, PLVR, PLC, and PLL are all highly significant (p < 0.01), and PLN is significant (p < 0.05). The equation: LBP = 0.481 + 0.297 APO + 0.202 PLVR + 0.178 PLC + 0.147 PLL + 0.130 PLN, explains 68.4% of the variance, indicating a joint influence of intrinsic ornamentality and several layout features.
• Two-species landscapes: APO (0.316) and PLC (0.274) are the strongest, both highly significant (p < 0.01); the constant (0.240) and PLL (0.114) are significant at p < 0.05. The equation: LBP = 0.240 + 0.316 APO + 0.274 PLC + 0.161 PLN + 0.114 PLL, accounts for 71.7% of the variance, the highest among the three richness levels—showing that ornamentality and colour dominate aesthetic judgements in two-species mixes.
• Three-species landscapes: the constant term (1.086) is markedly high and highly significant (p < 0.01), implying strong effects from factors outside the model. PLP (0.304) exceeds APO (0.268) for the first time; both are highly significant (p < 0.01), while PLC (0.123) and PLL (0.133) are significant at p < 0.05. The equation: LBP = 1.086 + 0.304 PLP + 0.268 APO + 0.133 PLL + 0.123 PLC, which explains 52.7% of the variance, the lowest of the three groups.

Taken together, Figure 9a–c shows a systematic shift in dominant visual factors as richness increases:

• Single-species schemes rely mainly on APO plus several layout attributes.
• Two-species schemes are driven chiefly by APO and PLC.
• Three-species schemes shift to a PLP-centred model, with a growing role for variables outside the current framework.

These richness-specific patterns provide differentiated guidance for designing aquatic plant landscapes, suggesting that more complex mixes should prioritise public preference cues rather than depending solely on plant ornamental value.

4. Discussion

4.1. Key Findings and Theoretical Implications of the Study

This study quantifies the relationship between aesthetic perception of urban aquatic–horticultural landscapes and visual preference attributes, thus addressing the long-standing mismatch between the ecological purification role of aquatic plants and public aesthetic expectations. We show that aesthetic perception differs significantly by viewing form (flowering > mixed > foliage) and that species richness relates to landscape aesthetics in a non-linear manner. Moreover, APO—identified as the core visual preference attribute—exerts distinct influence patterns across viewing forms and richness levels. These results support the “feature-priority recognition” hypothesis in environmental aesthetics theory [87], which proposes that observers first attend to salient features (e.g., flowers) before evaluating overall composition. The findings also corroborate Kaplan’s “information-processing” theory [88]: moderately complex, two-species combinations deliver the highest predictive efficiency for public aesthetic judgments.

Among the 18 test landscapes, Canna glauca L. and Nelumbo nucifera Gaertn., whether alone or in combination, received the highest aesthetic ratings. This accords with Zhang et al.’s observation that culturally symbolic plants elevate perceived landscape quality [89]. The preference reflects not only the visual appeal of flowering traits but also the cultural symbolism embedded in certain species. Nelumbo nucifera Gaertn., for example, carries profound meaning in East Asian culture, which can amplify aesthetic perception via cultural memory [90]. Ren and Jing-Yuan [91] likewise report that landscape elements with strong cultural associations tend to secure higher public approval.

4.2. Cognitive Mechanisms and Applications of Aesthetic Perception in Different Types of Aquatic Plant Landscapes

4.2.1. Cognitive Mechanisms of Aesthetic Perception in Flowering Aquatic Plant Landscapes

Flowering aquatic plant landscapes obtained the highest public aesthetic ratings (mean SBE = 0.29 ± 0.15), echoing Todorova et al. [92] and Elsadek et al. [93], who reported strong positive emotional responses to floral scenes. The regression model (R² = 0.666) is

LBP = 0.261 + 0.191APO + 0.161PLN + 0.180PLP + 0.123PLL + 0.261PLVR.

This equation shows that aesthetic perception depends on a balanced interplay of several attributes, with PLVR and APO most influential. The multifactor pattern supports Tveit et al.’s “visual-diversity synergy” theory [26], which argues that landscape aesthetics emerge from combined visual cues. Flowering plants supply not only colour contrast but also dynamic seasonal change—an aspect that static evaluations can underestimate [94].

Practical implications: the Canna glauca L. + Nelumbo nucifera Gaertn. F3 and Canna glauca L. + Nelumbo nucifera Gaertn. + Pistia stratiotes L. F6 can serve as core flowering schemes. Design priorities should be (1) highlighting ornamental traits, (2) creating natural vertical transitions, and (3) introducing rhythmic variation in height, form and flowering sequence. Such measures can significantly raise public acceptance [95].

4.2.2. Cognitive Mechanisms of Aesthetic Perception in Foliage Aquatic Plant Landscapes

Although foliage landscapes received lower aesthetic scores (mean SBE = −0.29 ± 0.05), they produced the strongest model fit (R² = 0.793). The equation is

LBP = −0.140 + 0.521APO + 0.274PLC + 0.161PLN + 0.114PLL.

The dominant APO coefficient (0.521) indicates that plant-intrinsic aesthetics drive perception when flowers are absent, consistent with Zheng et al. [96]. The negative, non-significant intercept (−0.140) hints at unmodelled negative factors, such as the “jungle anxiety” effect described by Anand et al. [97].

Design recommendations: (1) select species with distinctive leaf form, texture or colour (e.g., the well-rated Arundo donax ‘Versicolor’, L2; SBE = −0.324); (2) enrich layout colour by combining diverse leaf hues; (3) enhance naturalness while maintaining “naturalness within order” [98]—a balance between tidy structure and spontaneous growth.

4.2.3. Cognitive Mechanisms of Aesthetic Perception in Mixed Aquatic Plant Landscapes

Mixed (flower + foliage) landscapes show intermediate aesthetic scores (mean SBE = 0.01 ± 0.13) and model fit (R² = 0.718). The regression equation is

LBP = 0.404 + 0.295PLP + 0.272APO + 0.227PLL + 0.158PLC.

Here, PLP surpasses APO as the primary driver, indicating that overall composition is judged more than individual plant traits. This finding supports Lindemann-Matthies et al.’s proposition that “composition outweighs elements” [29]. The sizeable intercept (0.404) suggests influence from stable external factors, aligning with Gobster et al.’s “meaning aesthetics” concept [99].

Design advice: for combinations such as FL4 (Iris tectorum Maxim. + Nymphaea tetragona Georgi), emphasise (1) clear stratification so flowers and foliage complement, not complete; (2) controlled visual complexity to avoid clutter; (3) strong colour contrast from the interaction of blooms and leaves. Note that adding or removing specific species can markedly shift public response [100]; for example, introducing Nymphaea tetragona Georgi substantially improved the rating of Iris tectorum Maxim.

4.3. The Complex Relationship Between Species Richness and Aesthetic Perception of Aquatic Plant Landscapes

4.3.1. Aesthetic Perception Characteristics of Single-Species Combinations

The single-species model (R² = 0.684) is

LBP = 0.481 + 0.297APO + 0.202PLVR + 0.178PLC + 0.147PLL + 0.130PLN.

The equation shows that aesthetic perception in single-species landscapes arises from a balanced interaction of visual attributes, with APO and PLVR playing leading roles. This supports Lindemann-Matthies and Bose’s findings [101] that intrinsic plant traits dominate evaluations in monocultures. The sizable constant (0.481) suggests a stable baseline appreciation, echoing the “understanding” dimension in Kaplan and Kaplan’s preference matrix [102], whereby simple scenes are cognitively processed with ease.

Within the single-species group, Nelumbo nucifera Gaertn. (F2) and Canna glauca L. (F1) score highest, whereas Cyperus papyrus L. (L1) and Arundo donax ‘Versicolor’ (L2) score lowest—consistent with Hoyle et al. [103], who note that distinctive features such as flowers boost ratings. Design implications are to: (1) exploit each species’ inherent ornamental appeal; (2) use mass plantings to generate rhythm and layering [104]; and (3) employ species like Nelumbo nucifera Gaertn. to create strong visual focus.

4.3.2. Aesthetic Perception Characteristics of Two-Species Combinations

Two-species schemes yield the best fit (R² = 0.717):

LBP = 0.240 + 0.316APO + 0.274PLC + 0.161PLN + 0.114PLL.

Here, APO and PLC dominate, indicating richer chromatic experiences—consistent with Kuper’s emphasis on colour in aesthetic judgments [105]. The strong fit supports Kaplan’s information-processing view [88] that moderate complexity offers optimal information without overload.

The Canna glauca L. + Nelumbo nucifera Gaertn. mix (F3, SBE = 0.480) outperforms many monocultures, substantiating Ode et al.’s “moderate-complexity principle” [106]. Practical guidance: (1) ensure complementary, not competing, visuals; (2) highlight colour contrasts; and (3) build clear spatial hierarchy [107]. Emergent + floating-leaf pairings (e.g., F3) work better than emergent + emergent mixes.

4.3.3. Aesthetic Perception Characteristics of Three-Species Combinations

The three-species model shows lower fit (R² = 0.527):

LBP = 1.086 + 0.304PLP + 0.268APO + 0.133PLL + 0.123PLC

The unusually large constant (1.086) and the fact that PLP exceeds APO indicate a qualitative shift. When complexity passes a threshold, evaluation shifts from element-based to holistic, as noted by Anna Jorgensen et al. [107]. The reduced fit (52.7%) implies additional, unmodelled variables, aligning with Kuper and Rob’s findings on complex scenes [108].

The Canna glauca L. + Nelumbo nu-cifera Gaertn. + Pistia stratiotes L. (F6, SBE = 0.487) tops this group, showing that success depends on complementary traits, not species count alone—supporting Kuper’s view [105]. Design recommendations: (1) build a clear high–medium–low hierarchy; (2) select species with complementary colour and form; (3) focus on overall scene quality rather than single plants; and (4) manage complexity to avoid clutter [109]. The emergent + floating leaf + free-floating arrangement (e.g., F6) delivers the most appealing visual experience.

4.4. Application Strategies for Urban Aquatic Horticultural Landscape Design

Based on the research findings, the following design strategies for urban aquatic horticulture are proposed.

Landscape configuration strategy based on ornamental form: In areas where visual aesthetics are paramount, flowering aquatic plants should predominate (F6). In sites that require more stable ecological performance, leaf-type or mixed-type plantings can be adopted to maintain year-round landscape stability. Zhang et al. [110] noted that seasonal variation significantly affects public evaluations of aquatic landscapes.

Landscape design strategy based on species richness: Choose a species-richness pattern that matches the design objective. High-quality single-species schemes (e.g., F2) suit small water bodies or focal spots. Two-species mixes (e.g., F3) offer the best visual experience in medium-sized areas. For large basins or ecological restoration zones, three-species combinations can increase biodiversity while still providing a strong visual effect [111].

Precision design strategy based on visual preference attributes: Highlight key visual preference attributes according to viewing type and species richness. Flowering landscapes should emphasise PLVR; foliage landscapes should stress APO and PLC; mixed landscapes should focus on PLP and PLL. As species richness rises, the design focus should shift from individual plant traits to overall composition [57].

Integration strategy for ecological function and aesthetic value: Select species with dual benefits—such as Canna glauca L. and Nelumbo nucifera Gaertn.—that combine strong purification capacity with high ornamental value. Poje et al. demonstrated that public aesthetic evaluations improve when people understand a plant’s ecological role. Educational interpretation and participatory design can increase awareness of the ecological value of aquatic plants and, in turn, boost acceptance [112].

Strategy for integrating cultural connotations and landscape experience: Use aquatic plants with cultural symbolism (e.g., Nymphaea tetragona Georgi) to enrich the cultural dimension of the landscape and enhance experiential quality through multisensory design. Wang et al. [113] found that plantings with cultural associations typically receive higher public ratings.

4.5. Implications for Future Experimental Research

By integrating purification performance with public aesthetic perception, this study offers a foundational framework for forthcoming field trials and laboratory experiments in urban aquatic horticulture. Unlike earlier “landscape-first” approaches [114], the present work follows an “integration-first” strategy, combining pollutant removal, ecological management, and visual appeal. The goal is to deliver efficient assemblage cues for targeted small-scale remediation, plant selection in artificial wetlands, and controlled laboratory studies [115]. Virtual scenes generated in Lumion 10 Pro served as a reverse-engineering tool: high-SBE combinations (e.g., F6: Canna glauca L. + Nelumbo nucifera Gaertn. + Pistia stratiotes L., SBE = 0.487) and regression modelling (R² = 0.422–0.793) supplied preliminary data on the synergy between aesthetics and purification. These outputs now inspire the next pilot study, which will expand the plant combination model to identify optimal mixes for specific pollutant profiles while maintaining high landscape appeal.

Field experiments: the findings can inform site-specific trials—such as micro-artificial wetlands—to test multi-pollutant removal under local climate and soil conditions, thereby broadening urban aquatic–horticultural applications.

Laboratory studies: begin with controlled single plant–pollutant tests, then scale up to multi-species combinations, addressing the current gap in integrated design research. This reverse pathway can shorten traditional lead times and lower costs, supporting sustainable urban designs that balance governance and aesthetics.

Plant selection through park surveys and literature validation ensured alignment with the dual aims of purification and ornamentation; all chosen species demonstrated practical adaptability and evidence-based effectiveness, validating the study’s methodology. The conclusions are transferable to cities sharing Fuzhou’s subtropical monsoon climate and green-space challenges—e.g., Guangzhou and Xiamen—providing scalable aquatic–horticultural solutions. Ultimately, the insights can extend to cross-cultural comparisons and dynamic modelling, enhancing the resilience of blue–green infrastructure.

5. Conclusions

5.1. Summary of Key Findings and Research Contributions

This study systematically analysed public aesthetic perceptions of aquatic plant landscapes with differing ornamental forms and species richness, and constructed an optimal selection strategy for urban aquatic horticulture that simultaneously addresses purification function and perceptual needs. A questionnaire completed by 320 participants covered 18 landscape combinations, confirming the dual value coupling of urban aquatic horticulture as a component of green infrastructure. Significant differences emerged in public aesthetic perception of landscapes featuring different ornamental forms (flowering, foliage, flowering + foliage) and specie richness (one, two or three species). Flowering landscapes elicited the strongest aesthetic response, underscoring the importance of ornamental morphology, whereas increasing species richness alone did not significantly enhance aesthetic appeal.

Stepwise linear regression produced quantitative models linking landscape aesthetic perception to visual preference attributes. Public preferences for the four classic attributes—coherence, complexity, legibility, and mystery—significantly influenced perceived appeal. The model therefore provides a scientific, quantifiable design tool. Recommended high-aesthetic combinations include the following:

• Canna glauca L. + Nelumbo nucifera Gaertn. + Pistia stratiotes L.;
• Canna glauca L. + Nelumbo nucifera Gaertn.;
• Nelumbo nucifera Gaertn.;
• Canna glauca L.;
• Nelumbo nucifera Gaertn. + Pistia stratiotes L.

5.2. Limitations and Future Directions

This work addresses the gap between purification efficacy and public perception, adopting an integrated priority approach: pollutant removal management is merged with aesthetic evaluation to identify efficient plant assemblages, especially for small-scale, targeted applications. Virtual scenes created in Lumion served as a reverse-engineering strategy, providing initial data to inform subsequent field and laboratory work. Nevertheless, several limitations remain.

• Sample bias: respondents were mainly urban residents. Broader, stratified sampling will be pursued in future studies.
• Virtual imagery: simulated scenes may not fully replicate real-world experience. Follow-up studies will compare virtual and actual settings to strengthen ecological validity.
• Seasonal effects: bloom period and leaf-colour changes were not modelled. Longitudinal work will incorporate multi-seasonal data.
• Temporal stability: no long-term follow-up was conducted. Future trials will reassess perceptions over time.
• Questionnaire scope: seven core questions limited coverage of perceptual dimensions. Additional items will be tested in later phases.

Site-specific soil factors were not included at this preliminary stage; they will be assessed during field trials once candidate combinations are validated.

Future research should: (1) broaden the species pool and explore more combinatorial patterns (e.g., tiered arrangements of emergent, floating-leaf and submerged plants); (2) integrate seasonal variability into dynamic evaluation models; (3) combine on-site assessment with advanced simulation; (4) examine interactions between aesthetic and ecological function perceptions; (5) conduct cross-cultural studies of aquatic landscape aesthetics; (6) optimise design guidelines based on evolving aesthetic data.

A deeper exploration of life-form combinations (emergent, floating-leaf, submerged, planktonic) will advance urban aquatic–horticultural design [116]. As a key element of blue infrastructure, such design must integrate aesthetic value, ecological performance, cultural meaning and public participation [117]. Through evidence-based plant selection and landscape configuration, purification function and aesthetic perception can be synergistically enhanced, promoting sustainable urban water management.