Recreational activities in protected areas have been increasing and have created the need to improve understanding of the impacts and management (Hammitt et al., 2015; Chrisfield et al., 2012; Monz et al., 2013). The trampling of vegetation and soil by hikers (Cole, 1989; Bright, 1986) is often a cause of land degradation in national parks. Recreational trails are often a source of negative impacts on the persistence of threatened, endangered, rare and keystone species (Ballantyne and Pickering, 2015). Trampling, especially in tundra ecosystems, may lead to altered environmental conditions, including decreased infiltration capacity and nutrient cycles in soils, and more extreme temperatures at the soil surface (Chrisfield et al., 2012). To date, large amounts of research are focused on the impact of visitors on soil and vegetation including monitoring and modeling (Dixon et al., 2004; Farell and Marion, 2001; Monti and MacKintosh, 1979; Godefroid and Koedam, 2004; Özcan et al., 2013). A variety of efficient methods for evaluating trails and their resource conditions, especially in sensitive and vulnerable areas (alpine environment), have been developed and described in the literature (Jewell and Hammitt, 2000; Hawes et al., 2006; Ólafsdótirr and Runnström, 2013; Tomczyk and Ewertowski, 2011; Brevik and Fenton, 2012). A review by Marion and Leung (2001) concluded that the point sampling method provides accurate and precise measures of trail characteristics that are continuous or frequent (e.g., tread width). Ground-based surveys are fairly accurate (with GPS), use existing staff and resources and provide immediate results. However, there are also some limitations of point sampling techniques – e.g., time consumption (Hill and Pickering, 2009).

Parks and protected areas are often set aside for conservation and recreational purposes, and have become some of the most sought-after vacation areas in the world, creating conflicts between conservation and recreation. In the US, National Park Service (NPS) units receive approximately 280 million visitors per year (IRMA, 2014). Couple this extensive visitation with the mission of the NPS, which is to protect and preserve both natural and cultural resources while providing for the freest opportunities for public enjoyment and recreation, and conflict between conservation issues and visitor use occur. Striking a balance between these competing goals often forces land managers to make compromises between impacts from visitation and protection of resources.

Parks apply a wide range of tools and techniques to manage impacts from visitor use. By providing a network of formal trails, protected areas can limit negative trampling impacts and prevent widespread degradation that would be caused by a less structured pattern of visitor activity and traffic (Marion et al., 2011). To balance resource protection and visitor experience, several frameworks have been developed to guide management decisions (Manning, 1999). These frameworks use numerical standards for biophysical or social condition indicators and set limits to define the critical threshold between an acceptable and unacceptable change in resources and social conditions (Kim and Shelby, 2006). Baseline data and future monitoring can also be used to compare past conditions with future conditions. If actual conditions are above quantitatively defined standards, managers can effectively deal with these factors to improve or stabilize the conditions. Such visitor impact monitoring programs can provide managers reliable information necessary to evaluate resource protection policies, trends, strategies and measures (Vistad, 2003). However, many authors have stated that the impacts of visitors on the ecological conditions of an area are influenced more by visitor behavior, park infrastructure and the resilience of soil and vegetation, and are less related to overall use levels (McCool and Lime, 2001). For example sustainable usage levels depend on a range of factors, including extent of trail hardening and frequency of trail maintenance (Washburn, 1982).

To better understand use and associated resource impacts, a visitor and trail monitoring program needs a diverse set of indicators that evaluate changes over time (Leung and Marion, 2000). Most commonly used trail indicators include the number, length and density of visitor-created trails, along with tread. Soil loss, the most ecologically significant trail impact, is less common, though it can be efficiently determined by measuring maximum incision or cross-sectional area at points along the trail (Olive and Marion, 2009). Other problems include visitors participating in a variety of recreation activities (hiking, camping, horseback riding), each of which contributes a unique impact on natural resources (vegetation, soil, water, wildlife). Some authors have compared and assessed the impacts of different recreation activities (hiking, mountain biking, horse riding) on vegetation and soils (e.g., Pickering et al., 2010; Wilson and Seney, 1994). There is limited research on the ecological impacts of tourism and recreation in some parts of the world (Barros et al., 2015; Zdruli, 2014; Ibáñez et al., 2015). Existing studies document a range of impacts on vegetation, birds and mammals, including changes in plant species richness, composition and vegetation cover and the tolerance of wildlife to visitor use. Comparable studies, especially in high alpine environments, are needed to predict the effects of topographic and climatic extremes (Nepal, 2003).

Conducting formal trail surveys provides information for a number of important management questions and decisions, though it is commonly overlooked due to funding constraints. Information about trail conditions can be used to inform the public about trail status, justify staffing and financing, evaluate the acceptability of existing resource conditions, understand relationships between trail impacts and the controlling mechanism, identify and select appropriate management actions and determine the effectiveness of implemented actions. This paper presents research and assessment of impacts on the trail network of the Rocky Mountain National Park (RMNP) study area to understand how abiotic factors such as grade, elevation, surface type and trail slope alignment can influence trail conditions. We also want to understand how visitation type (e.g., people vs. horses) and level of use can impact trails. Finally, our last goal is to determine which factors are prevailing and what connection between factors exist. This would help managers reduce the effects of visitor use on natural resources of the park.

Rocky Mountain National Park (RMNP) is located in northern Colorado (USA), comprises an area of 1075 km2 and provides access to wild places for visitors to recreate and experience solitude and outstanding beauty (Fig. 1). The elevation range within the park spans from 2316 to 4346 m, which creates a highly complex and steep topographic gradient, allowing for diverse vegetation communities. The underlying geology of this mountain is also highly complex, though it is primarily granitic. Severe climatic conditions and thin soils have created a fragile environment at higher elevations throughout the park (alpine tundra encompasses one-third of the park area) that is neither resistant nor resilient to human use (RMNP, 2013). The historical annual temperature average (1900–2002) is 1.5 ∘C, and the annual average precipitation for the same period is 400 mm yr-1. Over the past several decades temperatures have been increasing +0.8 ∘ century-1 (Gonzalez, 2012) and precipitation patterns have been highly variable, with increased drought years followed by extreme rain events and record snowpacks, causing varying degrees of freeze/thaw actions and greater spring runoff events.

Yearly visitation over the past decade has hovered around 3 million visitors a year, with the total number of recreation visitors in 2012 being 3.3 million. The busiest tourist season is the summer months (June–August), but in recent years, the heaviest visitation days have occurred during the weekends of late summer and early fall due to the elk rut and foliage change. Overnight as well as day use has steadily increased over the past several decades, resulting in more impacts from visitation (RMNP, 2001).

Monitoring visitor use focused on vegetation and soil impacts is important in alpine areas, climbing areas and riparian areas where the information can help with determining thresholds of degradation (NPS, 2010). The loss of soil and vegetation from high use and unacceptable behavior of visitors are a principal concern. Besides educating visitors about principles (e.g., staying on the trail) and monitoring visitor use numbers, the results of this research can help inform park managers.

Early studies focused on the impact of visitors on natural ecosystems in RMNP (e.g., Willard and Marr, 1963) and stated the need to develop a system of evaluating day use destination sites, document trends in day use, develop guidelines, install flip signs, voluntary permits or a self-registration system and set concrete use limits (e.g., for parking). Later works are devoted especially to the monitoring of trail impacts (Summer, 1986; Benninger-Truax et al., 1992; KellerLynn, 2006; Pettebone et al., 2009).

Approximately 571 km of hiking trails provides visitors with recreation opportunities throughout the park (RMNP, 2013; Fig. 1). Some of the current trail system evolved from game trails used by Native Americans, then explorers and herders and was finally adopted by the National Park Service. A lot of the trails were built or improved by the Civilian Conservation Corps (CCC) during the 1930s. These trails span the entire elevation gradient, running across valley bottoms and ridgetops. RMNP is divided into ten planning units based on similar physiographic features and visitor use patterns (RMNP, 2000). Evaluated trails are situated in several planning units (Fig. 1; Front Range, Longs Peak, Wild Basin, Roaring River, Trail Ridge). Each planning unit is specified with trail project priorities (safety of visitors, mitigation of resource damage) and cost estimates. Since 2008 there have been new Federal Trail Data Standards, which include four fundamental concepts that are cornerstones of effective trail planning and management (trail type, trail class, managed use, designed use). Although not entirely new, these interagency concepts provide an integrated means to consistently record and communicate the intended design and management guidelines for trail design, construction, maintenance and use.

During August 2013, we applied impact assessment procedures to eight formal and informal trails (56 km) within RMNP. They represent a subset of the entire trail system and were selected because they provide a unique look at the variation of impacts along an elevation gradient and visitor use gradient, while representing the greatest possible spatial extent of RMNP. Some of the trails (or sections of trails) are used not only by hikers but also by other user groups such as equestrians (about 80 % of the total trails maintained in the park are open to commercial and private stock use). Four trails were evaluated on the north side of the park: Saddle trail (SDDL), Ute trail West (UTEW), Ute trail East (UTEE) and Mount Ida trail (IDAM), three on the south side of the park: Flattop Mountain trail (FLTM), Boulder Field trail (BLDF) and Thunder Lake trail (THLA). Also one short section of an informal trail, Old Fire trail (OFIR), was measured with detailed sampling (30.5 m interval) – see Table 1a, b and Fig. 2.

National Park Service units are charged with providing opportunities for recreation along with the protection and preservation of natural and cultural resources and ecological processes. This research provides information on the impacts of visitor use to trails and which abiotic factors are the most influential on trail conditions. This type of information can serve as the basis for the management of visitors. This research used a variety of trail inventory and impact indicators to understand trail conditions, while also providing a baseline to assess future trail conditions against, as it serves as data for the current condition of trails (see example in Fig. 6). These data could also be used for the evaluation of trail condition trends over time which allows for more informed management decisions to be made in the future.

During the initial literature review, we found many studies related to trail impacts monitoring and trail indicators. Dixon et al. (2004) proposed the use of two trail indicators – track depth and track width – to understand trail conditions. Their analysis revealed that track depth and rates of erosion are strongly influenced by track type and to a lesser extent by usage, while track width is influenced mainly by usage and track bogginess. Slope of the path and the number of visitors were two main factors explaining trail width and depth in other studies (Selkimaki and Mola-Yudego, 2011). Tomczyk and Ewertowski (2013) discovered that no connection was demonstrated between amount (number of visitors) or type of trail use and the amount of soil loss or deposition a trail underwent. A study by Jubenville and O'Sullivan (1987) concluded that vegetation type and slope gradient to trail erosion explained not much of variance in soil loss (could be explained by trail design and permafrost in Alaska). Nepal (2003) found that trails tend to be more degraded at higher altitude and on steep gradients, along with a strong positive correlation between trail degradation and frequencies of visitor use. Nepal and Nepal (2004) also found a strong correlation between visitor use and trail degradation. However, locational and environmental factors are equally important variables. The study concludes that more systematic and experimental studies are needed that can make a clear distinction between human-induced trail damage and the effects of natural factors. Trail grade and trail slope alignment angle, which often impact trail width and soil loss, were the two most important inventory indicators (Dissmeyer and Foster, 1984; Aust et al., 2004) assessed in the survey. Trails located in flatter terrain can be susceptible to widening and muddiness problems due to drainage issues. Fall-aligned trails are of particular concern due to their increased potential for erosion. This study found that trail alignment tends to be more influential on soil loss than the predominant type of visitor use (e.g., horse vs. hiker traffic) or number of users. We also assumed that soil loss increases exponentially with steeper trail grade, though the natural rockiness of RMNP's trail treads and stonework in our case probably limit erosion and help sustain steeper trail sections. Soil loss, attributable to several causal factors, was assessed for the trails using three measures: mean trail depth (7.1 cm), maximum incision (19.0 cm) and cross-sectional area (444.5 cm2). Relational analyses for soil loss revealed that level of trail use and trail grade had the most influence, however dependence with trail slope alignment angle was not significant as other studies found (e.g., Wimpey and Marion, 2010). Ólafsdóttir and Runnström (2013) discovered, after analyzing several physical properties, that only elevation has a clear relationship with hiking trail condition in their study sites. Severe conditions never apply to a whole trail, suggesting that trail conditions are a function of trampling magnitude and local physical properties. Hence, when maintaining hiking trails in vulnerable environments, a holistic understanding of the effects of trampling is critical for trail sustainability.

When comparing two types of recreational visitor use (hikers and horseback riders), our results indicated that medians were greater for trail width, maximum incisions and soil loss in case of trails that allowed horseback riders. This shows that horse use within the park generally increases impacts on the trail system when looking at specific indicators in specific locations. It is compatible with results of other studies. Pack animals, according to Barros and Pickering (2015), caused more damage than hikers to the alpine meadow and their impacts were apparent at a lower level of use than for hikers. Horse traffic also consistently made more sediment available for erosion from llamas, hikers or no traffic (Deluca at al., 1998). It is also important to notice that horse riding trails can promote exotic plant species, many of which are not native to the area, which may lead to changes in the structure of vegetation communities (Törn et al., 2009).

Fall-aligned trails, with steeper grades, frequently require significant investments in rockwork and maintenance to keep them sustainable and to keep them from widening. This is especially true in areas prone to freeze/thaw and water runoff. Trail width can also be influenced by many other variables including use level, visitor behavior, trail grade, landform grade, trail ruggedness and trail borders. Wimpey and Marion (2010) found the relationship of trail width was most associated with trail and landform grade and trail slope alignment since steeper grade restricted lateral dispersion of hikers. Our results confirm that trail width is predominantly a function of use level. Mean trail width is a relatively wide 89.9 cm, though many trails are purposefully designed with wider widths to support heavy visitor use. A trail width difference with a mean of 22.6 cm indicates that the formal trails are generally wider than intended. Other important factors we found were the behavior of visitors and absence of trail borders. Trails without borders will lead to further widening, since visitors have a hard time discerning where the trail is located. The ruggedness of a trails' tread can also encourage the widening of trails, since hikers often look for easier passage to avoid these areas which are often along trail sides. To address these issues, managers can manipulate the level of trail use, create trail borders or educate visitors on how to decrease their impact on trails. These solutions are easily implemented and relatively cost-effective. An obvious solution for managers to prevent soil loss would be to control use levels, though this is often not popular with visitors and does not act as part of the park's mission to allow access. A second option would be to relocate trails located in areas highly prone to soil loss. Wilderness values may inhibit relocations of trails so this might be an option that could be used only in select locations (excessively steep or aligned closely to the fall line). A third option is to shorten the time between regular maintenance visits for each trail (Birchard and Proudman, 2000). This option would likely be the least economic and have impacts on visitation and wilderness due to the more frequent presence of workers or closures of trails for work. Some authors commonly recommend preventing soil loss by keeping grades of less than 10–12 % (Hooper, 1988; Hesselbarth et al., 2007), trail slope alignment higher than 22∘ (Olive and Marion, 2009) and trail slope ratio less than 0.5 (IMBA, 2004). Our survey found that 24 % of the evaluated trails exceed a grade of 15 %. Additionally, only 6 % of the evaluated trails are aligned within 22∘ of the fall line, which makes the dispersion of water more challenging and thus increases soil erosion. Soil erosion would likely have been much higher than assessed if not for the substantial amount of granitic rock in the soils and the extensive use of rock steps – see also Fig. 2. Moore et al. (2012) stated that unmaintained trail impacts are negatively perceived by visitors and have an overall adverse effect on visitor experiences, providing support for proper trail design and maintenance. Ballantyne et al. (2014) recommended that management should seek to minimize the creation of informal trails by hardening popular routes and centralizing visitor flow. Different walking track types can have an effect on different vegetation characteristics (Hill and Pickering, 2006). In some cases closure of recreational sites and trails can be a solution for trail degradation. However, to improve soil properties, long periods of closure are necessary due to slow soil-forming processes (Özcan et al., 2013; Sharratt et al., 1998; Brevik and Fenton, 2012). Another option is the importance role of trampling tolerant vegetation communities which can be both resistant and resilient to increased use (Pickering and Growcock, 2009). Restoring damage to natural vegetation and soils by human use in alpine environments can be extremely challenging due to the severity of the environment which restricts plant growth and increases the potential for soil erosion (Scherrer and Pickering, 2006).

Regarding methods, a distance-based technique, in which measurements are made at regular spatial intervals, is quite time-consuming. A technique of sampling at 20 m intervals can be used to assess typically 5–7 km of track per day in remote areas (Hawes et al., 2006). Our experiences confirmed time consumption so there will be fair discussion about practicality to repeat these measurements as a part of a potential monitoring program. A combination with GIS-based methodologies could be a more effective tool (Hawes et al., 2013; Ballantyne et al., 2014; Ólafsdóttir and Runnström, 2013) to examine the relationship between trail condition assessment and local physical properties, such as elevation, gradient, soil type and vegetation cover. For further trail monitoring, a recommendation to consider is the possibility of increasing precision of measurements (submeter accuracy GPS units, smaller intervals for measurements between sampling points of 30 m; this will increase time capacity). Lidar-derived terrain models could greatly speed up collection of measurements (Nadal-Romero et al., 2015). Maximum incision and trail width are the most significant predictors of CSA which can be used for simplifying during measurements. Measurements of CSA could be influenced as well by boundary determinations (historic vs. recent erosion). Trail work prior to this study could have also impacted the precision of measurements taken. For example, previous side-hill work along the trail may have also altered the final estimation of soil loss since these practices would have helped keep soil in place. It is also important to add the presence of a trail border into the point sampling from what can be used for analysis, especially with the trail width indicator. Contrary to the original methodology for simplification, we slightly modified categories of trail surface.

This study has set out to understand and assess impacts on the trail network of the RMNP study area to gain information on how abiotic factors such as grade, elevation, surface type and trail slope alignment can influence trail conditions. Additionally, we looked at how visitation type (e.g., people vs. horses) and level of use impacted trails and finally, we wanted to determine which factors were most important and what connection between factors existed. This information could then be used to help managers reduce the effects of increased visitor use on trails and other resources of the park.

After assessing the trail conditions within the study area we found that grade, elevation, surface type and trail slope alignment were all important factors to determine trail degradation. It does appear that certain factors, such as trail slope alignment, are more important than others when considering soil loss or trial widening. Furthermore, factors such as grade and elevation are important factors when considering loss of vegetation. We would recommend that any robust trail monitoring program would include monitoring all the aforementioned trail indicators and factors.

When we looked at the different types and volume of trail usage, only factors such as maximum incision, trail width and soil loss were considerably affected. Horse trails tended to be more incised, wider and had more soil loss than trails closed to horses, meaning use type was important to consider when maintaining or constructing trails. When looking at how use levels impact trail conditions, only one indicator was impacted significantly – tread width – meaning that more users only seem to make trails wider. This make sense since there will be more users passing by each other along the trail.

We believe that any solid trail monitoring program would use all trail indicators and factors mentioned. Each factor and indicator is either directly or indirectly connected to each of the others, so omitting any may increase the likelihood of misunderstanding what is causing trail degradation. For example, trail slope alignment and grade together were the strongest indicators for predicting soil loss, though indicators such as usage type also impacted soil loss. As visitation of protected areas is going to increase as populations grow and face new challenges from changing climates, it is important to continue to monitor, learn and adapt if these areas are to remain accessible while protecting the valuable resources found within them.