Protected Area Tourism in a Changing Climate: Will Visitation at US National Parks Warm Up or Overheat?

Abstract

Climate change will affect not only natural and cultural resources within protected areas but also tourism and visitation patterns. The U.S. National Park Service systematically collects data regarding its 270+ million annual recreation visits, and therefore provides an opportunity to examine how human visitation may respond to climate change from the tropics to the polar regions. To assess the relationship between climate and park visitation, we evaluated historical monthly mean air temperature and visitation data (1979-2013) at 340 parks and projected potential future visitation (2041-2060) based on two warming-climate scenarios and two visitation-growth scenarios. For the entire park system a third-order polynomial temperature model explained 69% of the variation in historical visitation trends. Visitation generally increased with increasing average monthly temperature, but decreased strongly with temperatures > 25°C. Linear to polynomial monthly temperature models also explained historical visitation at individual parks (R2 0.12-0.99, mean = 0.79, median = 0.87). Future visitation at almost all parks (95%) may change based on historical temperature, historical visitation, and future temperature projections. Warming-mediated increases in potential visitation are projected for most months in most parks (67-77% of months; range across future scenarios), resulting in future increases in total annual visits across the park system (8-23%) and expansion of the visitation season at individual parks (13-31 days). Although very warm months at some parks may see decreases in future visitation, this potential change represents a relatively small proportion of visitation across the national park system. A changing climate is likely to have cascading and complex effects on protected area visitation, management, and local economies. Results suggest that protected areas and neighboring communities that develop adaptation strategies for these changes may be able to both capitalize on opportunities and minimize detriment related to changing visitation.

Fig 1. Third-order polynomial glm relationship of historical (1979–2013) monthly mean temperature and monthly park visitation (proportion of annual).

Mean (black circles) and error bars (+/- twice the standard error) from observed data are based on 2.5°C bins. Model estimate (dark gray solid line) ± 1 standard error (light gray shaded area), R² = 0.69.

Fig 2. Historical (1979–2013) and potential future visitation (2041–2060) patterns in example parks.

a) Acadia National Park, Maine; b) Grand Teton National Park, Wyoming; c) Blue Ridge Parkway, North Carolina and Virginia; d) Arches National Park, Utah; e) Big Cypress National Preserve, Florida; and f) Pu‘uhonua o Hōnaunau NHP (National Historical Park, Hawai‘i; where a visitation-temperature relationship was not supported and historical mean monthly temperature [gray triangles and dashed line] is shown). Historical (black) error bars are +/- one standard error and future projection error bars show +/- one standard error of the prediction estimates. Solid red lines are for RCP 8.5 and solid orange lines are for RCP 4.5 low maximum growth projections, dotted lines of the same color and thickness show high maximum growth projections when different from low maximum growth.

Fig 3. Potential future total annual visitation (2041–2060) expressed as a percentage of historical values (1979–2013).

Future visitation for each park is based on modeled monthly visitation relationship to temperature. Future projections, a) RCP 4.5 low growth and b) RCP 8.5 high growth, were limited to parks with temperature as an explanatory variable in the best-fit model and an adjusted R² ≥ 0.5 (n = 282).

Fig 4. Potential changes in future visitation by climate scenario and season at U.S. national parks.

Future visitation (2041–2060) at each park is expressed as a percentage of historical values (1979–2013). Future projections were limited to parks with temperature as an explanatory variable in the best-fit model and an adjusted R² ≥ 0.5 (n = 282). Boxplots: thick vertical black line indicates median, the boxes are the interquartile range (IQR), and the whiskers extend 1.5 x IQR.

Fig 5. Potential future change in length (days) of the overall visitation season at parks (2041–2060 – 1979–2013).

Overall visitation season length was defined as beginning on the date when 10% of historical cumulative visitation was achieved and ending on the date when 10% of historical cumulative visitation remained for the year. Projections did not vary between low- and high-maximum growth models for each RCP. Boxplots: thick vertical black line indicates median, the boxes are the interquartile range (IQR), and the whiskers extend 1.5 x IQR.