Examining the Influence of Lake Water Quality on Visitor Sentiment: Empirical Findings from 85 US Lakes

Abstract

Cultural ecosystem services (CESs) are the intangible benefits that humans receive from nature, resulting from the interplay between biotic nature and abiotic nature (Gray, 2011). And very often, the intangible dimensions hold greater significance for individuals compared to material benefits (Kobryn et al., 2018). Lakes are highly valued for CESs because they provide opportunities for recreational activities, aesthetic experiences, spiritual enrichment, and reflection (Blicharska et al., 2017). Lake CESs are intricately linked to water quality, where sedimentation, nutrients, and bacteria affect people's perceptions and interactions with the water body (Keeler et al., 2012). However, our understanding of lake water quality's influence on CESs remains limited, hindering effective management strategies to enhance the economic and social values of lakes. Water-related CESs research is also constrained at broader scales (e.g., national scale) due to measurement challenges, undermining research generalizability. We hypothesized a connection between lake water quality and CESs, with water quality influencing CESs through recreational behaviors and visitor sentiment. To test this, we coupled big data from Google Review and National Lake Assessment to investigate water quality effects on CESs and visitor sentiment across 85 U.S. lake-front public parks. We implemented five analytical steps: (1) Collecting geographic data of 85 public-owned lakes and identifying lake-front parks using Google Place API, (2) Obtaining ~40,000 relevant text reviews with Python Crawler, (3) Filtering Google reviews related to lake CESs through content analysis with keyword labels, (4) Deriving positive, neutral, and negative sentiments from lake-related user-generated content using Google Cloud Natural Language API, and (5) Identifying associations between key water quality indicators and CESs-related visitor sentiment with a Bayesian multinomial model. Our results showed that CESs provided by lakes were significantly affected by multiple water quality indicators. Higher chlorophyll-a concentration and APHA color value increased negative sentiments toward lake CESs, while higher total phosphorus unexpectedly associated with increased positive CESs-related visitor experiences, possibly due to higher productivity. Our findings have implications for lake management, advocating targeted planning to promote harmony between human well-being and ecological health in lake ecosystems. Moreover, analysis of user-generated text contents for sentiment evaluation can yield valuable information regarding user satisfaction and memories of scenes, and thereby facilitate lake management practices aimed at fostering sustainable local economic development.