Visitor demographics, site-type and activities determine the occurrence and severity of environmental impacts at nature-based tourist destinations

Abstract

1. Increasing demand for access to nature has the potential to increase environmental impacts. Identifying links between increased visitor intensity and habitat damage in context-specific studies is an oversimplification which does not account for visitor demographics or activity characteristics.
2. We compared the prevalence of tourism-related threats/pressures across protected areas in Europe, Ireland, and in sites close to the Wild Atlantic Way (WAW) tourism route along the west coast of Ireland. To better understand the drivers of tourism-related impacts, we analysed impact occurrence (IO) and impact severity (IS) at 43 sites along the WAW using a database of over 6000 visitor movement, demographic and behaviour records gathered by a national tourism agency (Fáilte Ireland).
3. Threats/pressures related to tourism and recreation are prevalent and widespread across protected sites in Europe (49% of sites) with significantly higher (p < 0.001) prevalence of these threats in protected sites in Ireland (58% of sites). Over 33% of protected sites in Ireland are located within 2 km of a WAW Discovery Point and 64% of these WAW adjacent sites have known threats/pressures related to recreation and/or tourism, which is significantly higher than Irish and European averages (p < 0.001).
4. Visitors to WAW sites ranged mostly within 1 km of tourist destinations (99%) and over 75% of all visitors had no observable impact on the receiving environment. Site-type, tourist demographics and tourist activities were informative predictors of IO. IS was influenced by total number of impacts, activity intensity and site-type. Site-type together with visitor activity occurrence and activity intensity were significant for IO and IS, highlighting the importance of understanding local conditions as well as visitor characteristics for impact avoidance/mitigation.
5. Tourism-related pressures and threats are widespread among protected sites in Europe and Ireland. We demonstrate the need to shift the focus from analysing discrete ecological responses in response to generalized tourism pressure to understanding the causes of impacts to mitigate them at source. This will help align tourism and conservation goals.

1 INTRODUCTION

Nature-based tourism (NBT) is increasing in popularity as public perception shifts towards valuing the ancillary benefits of nature, such as wellness (Jiang, 2020; Leung et al., 2008; Winter et al., 2020). As the popularity of NBT increases ‘green’ branding and credentials influence the competitiveness of a tourism destination (Chin et al., 2018; Huybers & Bennett, 2003; Wong et al., 2021) which influences tourism performance (Bazargani & Kiliç, 2021). However, the fundamental drivers of tourism are primarily economic and social factors with environmental attention only being considered through legal compliance measures (Buckley, 2012; León-Gómez et al., 2021; McKercher, 1993). This is problematic as tourism developments and related services often have negative environmental impacts (León-Gómez et al., 2021; Liu et al., 2018; Monz et al., 2013; Ren et al., 2019; Zhong et al., 2011). Moreover, with increasing prosperity and global transport, access to nature-based destinations is increasing, but awareness of impacts is weak (Davenport & Davenport, 2006). Tourism can therefore have negative impacts on the resources that support the tourism industry, threatening the sustainability of NBT (Font & McCabe, 2017), and with a clear need for investment into the natural assets which underpin the tourism industry (Rodríguez et al., 2020; Sørensen & Bærenholdt, 2020).

Sites protected for conservation are popular and nature is a key motivation for tourism, which often conflicts with conservation status (Eagles et al., 2002; Jones et al., 2021; Kc, 2021; King et al., 2012; Reinius & Fredman, 2007; Spenceley et al., 2021). Participation in outdoor activities and NBT has long been increasing in countries such as the United States (Outdoor-Foundation, 2022) and Australia (Tourism & Transport Forum, 2017) and has been particularly evident since the COVID-19 pandemic (Askew & Bowker, 2018; Beery et al., 2021). The eco-tourism market was estimated at $181.1 billion in 2019 with a compound annual growth rate of 14.3% from 2021 to 2027 (Allied-Market-Research, 2021), showing a clear trend for increased demand globally. Within Europe, a substantial volume of tourists identify nature as a primary motivation for their holiday (Table 1).

TABLE 1. Six European countries with the highest tourism numbers which are primarily motivated by nature (CBI, 2020)

	Outbound overnight trips 2018, in millions	Percentage of holidaymakers who see nature as a primary motive for choosing a holiday destination	Indication of number of outbound overnight trips with nature as a primary motive, in millions
Germany	109.0	15%	16.35
France	53.3	14%	7.46
Netherlands	20.8	25%	5.20
Belgium	13.1	19%	2.49
Poland	12.0	19%	2.28
Czech Republic	7.4	26%	1.92

Tourism impacts contribute to loss of biodiversity and ecosystem services, potentially undermining the attractiveness of tourism destinations (Chen, 2010; Huybers & Bennett, 2003; Wong et al., 2021). Recreation- and tourism-related activities are commonly cited as issues for protected sites (Eagles et al., 2002; European Environmental Agency, 2018; Jones et al., 2021; Kc, 2021; Spenceley et al., 2021). The environmental effects of tourism can be evident at a local-scale but broader/landscape-scale interactions are more difficult to identify (Macdonald & Wester, 2021). Local-scale impacts are often considered to be context specific and enterprises that operate on larger scales often fail to recognize their role in the protection/maintenance of the locations which they operate from (Weaver, 2001). Arif et al. (2022) showed that recreational pressures in the drawdown areas of freshwater systems can have both negative and positive effects on ecosystem functioning depending on the context and/or scale. This disparity of scale limits the application or influence of identified local impacts on forward planning or policy in a more general context (Dertien et al., 2021; Ruban, 2021).

The primary benefits of NBT are employment, business opportunities and income (Thapa et al., 2022) and a study by Ozcan et al. (2022) showed tourism and globalization decreased environmental damage in countries such as the UK and France. Despite known impacts, time spent in nature positively influences environmental behaviours and decision-making (DeVille et al., 2021; Fenitra et al., 2021; Pasek & Ratkowski, 2021); which is beneficial for broader conservation goals. Currently, the literature expresses the need for multidisciplinary approaches and a clear lack of consideration for the experiences, management, design and planning of NBT (Mandić & McCool, 2022) which could be used as conservation resources if framed correctly.

The COVID-19 pandemic caused the closure of many sites and trails resulting in natural recovery of these sites which could lead to a global rethinking from mass tourism towards a more sustainable-tourism model (Jones et al., 2021; Spenceley et al., 2021). Eagles (2003) suggests that the negative impacts of tourism on park resources are less influenced by absolute numbers of visitors, and more influenced by weak tourism policy, management and staffing, often caused by very low levels of finance. Furthermore, Chen et al. (2022) demonstrated that strong tourism policies can result in unintended environmental effects due to the complexities involved. This is particularly relevant where nature is used as a branding recourse or attraction; specifically, in and around national parks or nature reserves (King et al., 2012; Rhama et al., 2020; Vecco & Caust, 2019; Weaver, 2001). This can be positive for biodiversity, if the footfall presents an opportunity to educate and facilitate engagement with nature and influence environmental decision-making (Massingham et al., 2019). A soft approach, using education and guidance of both the tourist and the tourism operator, can be taken to foster environmental stewardship as a means to minimize potential impacts (Della Lucia et al., 2021; King et al., 2012; Zhang et al., 2020); however, this approach requires understanding of visitor impacts at a macro level and requires a national policy and implementation structure.

Increasing visitor intensity has been used generally as a proxy for environmental impacts (Monz et al., 2013); considering visitor behaviours as a single factor and assuming all visitors have consistent levels of impacts. Moreover, if increasing visitor numbers as a proxy for environmental damage is constantly being reaffirmed it can create misalignments between tourism and conservation goals.

Winter et al. (2020) and Gundersen et al. (2022) identified; the need for transdisciplinary collaboration at a variety of scales to achieve sustainable tourism. Political, social, cultural and economic factors are important to consider with respect to NBT in protected areas (Rhama et al., 2020). Trends in NBT research show a high demand for increased access to nature reserves; however, the research has a bias towards low- to moderate-income counties (Thapa et al., 2022) and income availability is known to influence environmental impacts (Ozcan et al., 2022). There is a need to prioritize responsible low-key tourism—particularly in protected areas—to align tourism and conservation goals; this is reliant on increased understanding of visitor interactions (Jurkus et al., 2022). Recent research has shown a conflict of interest between tourism and nature protection which varies between different demographics (Tolvanen et al., 2020). Arif et al. (2022) identified links between recreational activities and ecosystem functionality which varies in different contexts; these results indicate a clear conflict between conservation goals for increased ecosystem functionality and tourism goals for increased access/engagement.

In Europe, there are a network of protected sites known as Natura 2000 sites (Natura sites) designated under the European Habitats Directive (1992) [Council Directive 92/43/EEC on the Conservation of Natural Habitats and of Wild Fauna and Flora] and Birds Directive (2009) [Council Directive, 2009/147/EC on the conservation of wild birds] which provide a legal framework for Special Areas of Conservation (SACs) and Special Protection Areas (SPAs). Within the legal framework in Europe, there is a list of 413 threats/pressures for protected sites upon designation; 53 (12%) of these threats/pressures have attributes that relate to recreation and/or tourism (European Environmental Agency, 2018). Here, we harness monitoring data from the Natura 2000 sites across Europe and the Wild Atlantic Way (WAW) Visitor monitoring programme in the Republic of Ireland. Ireland was selected as a study country as it is a high-income country in Europe which specifically brands its natural assets with campaigns such as ‘Green Ireland’ and the ‘WAW’. The NBT literature shows a deficiency of research across North America, Europe and high-income countries (Thapa et al., 2022). In addition, Ireland has high volumes of self-directed tourism.

We use a multi-scale study of tourism threats/pressures to the Natura 2000 network across Europe, the spatial co-incidence of tourist branded and Natura sites at the national scale in Ireland as well as visitor demographic and activity drivers of environmental impact richness and severity at a regional scale across 43 tourist destinations in Ireland. The regional-scale analysis used a multi-site, multi-year visitor monitoring dataset from a regional tourism initiative in Ireland, the WAW.

To identify the prevalence of tourism threats/pressures across Europe, we assessed recreation- and tourism-associated threats/pressures recorded as part of the monitoring programmes for Europe's Natura 2000 network of protected areas. At a regional scale in Ireland, we identified the proximity of sites promoted for tourism with Natura sites and the prevalence of recreation and/or tourism threats/pressures across all protected sites within 2 km of promoted tourism sites. Finally, visitor demographics and activity engagement were investigated as factors hypothesized to drive impact richness and intensity at a regional scale across 43 sites in Ireland. By including site type, we evaluated whether the impacts observed are independent of local-scale or context-specific factors.

2 MATERIALS AND METHODS

Data were gathered from the European Environmental Agency (2018) for known threats/pressures recorded for Natura sites and prevalence of the 53 threats/pressures that are related to recreation and/or tourism activities (Appendix I) was determined.

The prevalence of tourism threats/pressures was assessed at a national scale, for the Republic of Ireland (Ireland), to determine consistency between the European and national scales. Within Ireland, a further targeted approach was taken focusing on the areas surrounding the WAW, which has 191 ‘discovery point’ destinations across the west coast (Figure 1).

FIGURE 1

Distribution of 604 Natura 2000 sites, (SACs and SPAs), in Ireland relative to the 191 wild Atlantic way discovery points, sites surveyed and the Wild Atlantic Way (WAW) travel route: Image demonstrating erosion caused by tourists at Strand hill.

R-Studio (4.0.3) was used throughout the analysis (R Development Core Team, 2020). The gDistance function in the package gdistance (Van Etten, 2017) was used to identify the nearest SAC and SPA to each of the discovery points; those within 2 km were selected. The number of SACs and SPAs with known threats/pressures related to tourism or recreation were counted as well as the number of sites for each threat/pressure, at both scales. These data were recorded as presence absence data only which limits their interpretability. A chi-squared test of association was used to determine if the prevalence of tourism pressures was consistent across Europe, Ireland and the WAW-associated sites.

We used visitor observation data collected as part of the environmental monitoring of the WAW over 43 sites for 3 years (Failte Ireland, 2017, 2018, 2019). Direct observation has been identified as the optimal process for collecting human distribution data (Marion et al., 2020). Analysis was constrained by this pre-existing data; for example, the activities and impacts data had no connecting metadata and therefore could only be considered independently. Monitoring sites were chosen to represent a variety of tourist destinations, management structures and site-types, spread throughout the WAW (Figure 1); the monitoring process was undertaken for compliance with national and international legislation.

Data were collected between 2017 and 2019; each year 14 new sites were selected and one site, Dursey Sound, was surveyed in all years. An environmental steering committee guided the site selection process annually ensuring geographical spread and variety of site-types were selected; the approaches for each year were:

• 2017 Estuaries and areas with high suitability for wading birds;
• 2018 Signature Discovery Points; and
• 2019 Sites within both SACs and SPAs.

At least two surveyors surveyed each site, three for large sites where the full extent of the site cannot be observed from two vantage points. Visitors on site were monitored using ArcPro (2.7.1) (ERSI, 2020) along with ESRI ArcGIS enabled apps Survey123 (ERSI, 2020b) and Collector (ERSI, 2020c). All surveys were conducted during the peak tourism season in Ireland (June–August) and surveys were conducted twice at each site for 8 h. Efforts were made to record every visitor on site, but where visitor numbers were high, recording yields were not 100%. Each group arriving to site was recorded as a single observation, visitor demographic data were collected through direct observation and included perceived age category (child, teenager, adult or elderly). The time of arrival, mode of transport and time of departure were also recorded (where possible); it is important to note that not all records represent the entire activity profile for the group due to view obstruction or non-returning groups, etc. Visitor movement patterns were recorded using the Collector app, producing a polyline map of the group's movements. All activities undertaken were recorded using a coding system which ranked activities by the intensity of the activity (low, moderate and high). For example, walking, jogging or sprinting would be three equivalent activities in each of the activity intensity index (AII) categories (Appendix II). Similarly, all impacts observed were recorded using a coded referencing system with low, medium and severe impact classes (Appendix II).

Data from Fáilte Irelands WAW visitor monitoring programme were used to evaluate the drivers of impact occurrence (IO) and impact severity (IS) using a set of explanatory variables (Table 2).

TABLE 2. Definitions of response and explanatory variables

	Definition
Response variables
Impact occurrence (binary)	Any impacts observed yes/no
Impact severity	Proportion of the total impacts that were high- or moderate-level impacts. Impact severity was a binomial proportion variable where successes were high and moderate impacts (combined), and failures were the number of low impacts
Explanatory variables
Distance travelled^a	Length of the polyline used to track and map visitor movements
Group-size^b	Total number of people
Group-type	Category of group composition resulting in 14 group-types; further defined in Table 3
Site-type	Category of site general features—built infrastructure, beach, Clifftop and highly managed
Site	Site identity (random effect)
Number of activities undertaken^b	Total number of activities which the group participated in
Activity Intensity Index^b	The exertion level of the activity: for example, Walking—Low; Jogging—Moderate, Running—High This variable was calculated as a proportion using the total number of moderate and high-level activities (combined) as a proportion of the total number of activities
Total Impacts^a	Total number of impacts observed
Year	Year of observation

• ^a Not used in IO or IS models; therefore, it was not standardized.
• ^b Numeric or count data with were standardized.

The distances of all visitor movements were used to determine the distribution of visitor distances travelled upon arrival at the site. The demographic information was used to create a ‘group identity’ resulting in 14 different group-types (Table 3).

TABLE 3. Group-type category definitions

Group-type	Definition	Count
Couple	Two adults only	2215
Individual	One adult only	742
Family	One or two adults and under 18s > 0	957
Family with elderly	One or two elderly individuals with one or two adults and under 18s > 0	44
Individual elderly	One elderly individual	161
Elderly Couple	Two elderly individuals	389
Elderly Group	Elderly individuals >2 and no others	101
Large Adult Group	Adults only >5 individuals	133
Small Adult Group	Adults only >2 but <5 individuals	592
Large Mixed Group	Adults, children, and elderly (inclusively) >5 individuals	155
Large Mixed Adult Group	Adults and elderly only >5 individuals	48
Small Mixed Adult Group	Adults and elderly only <5 individuals	196
Mixed Small Group	Adults, children, and elderly (inclusively) <5 individuals	104
Under 18s	No adults or elderly (0) and number of under 18s is >0	73

The total number of activities (activity richness) and total number of impacts (impact richness) for each group observation were calculated. IS was modelled as a binomial proportion of the number of medium and high-level impacts relative to the number of low-level impacts. Higher values for IS indicate a high proportion of high-/medium-level impacts and lower values of IS are dominated by low-level impacts. The AII was calculated as the proportion of all activities that were moderate or high level. A ‘site-type’ factor categorized the site according to dominant features: beach, clifftop, hard infrastructure and heavily managed.

Five of the variables (indicated in Table 2 by a *) were standardized using the scale function. A two-step modelling approach was undertaken which focused on specific hypotheses: (1) What variables explain the occurrence of impacts (binary response, binomial errors); (2) if impacts occurred what factors explained the IS (proportion of impacts which were moderate or high, binomial errors). For all models, site was used as a random effect.

Models were fit using generalized linear mixed effects regression model with binomial error distribution (GLMER; Bates et al., 2014). The maximal model was constructed for both response variables (IO and IS). The reference-level group selected for the model was a couple at a built infrastructure site, as a couple is viewed to be a standard tourism unit and built infrastructure sites were anticipated to have the lowest potential for impacts. For the IO model, only second-order interactions could be fit. For the IS model, third-order interactions were fit. The MuMin Package (Barton & Barton, 2015) was used to run all possible nested models and rank them according to their AICc values. The AICc values were used as a model selection criterion, models with the lowest AICc value were the models of best fit (Akaike, 1974). Equivalent models were any models with AICc values within 2 units of the best model; where this occurred, the simpler model (lowest number of explanatory variables) was selected. Models were assessed to ensure the assumptions were met. The variables were assessed for collinearity using a correlation matrix and the models were inspected for overdispersion. Further post-hoc tests were undertaken using the emmeans package in R (Lenth et al., 2018) to compare the estimated marginal means—using the ‘Tukey’ method to adjust for multiple comparison—for effects of each level of the categorical variable (full details are provided in Appendix III).

The data were observational; therefore, there was no meaningful way to discern between tourism and local recreation—therefore, all observations are labelled as ‘visitor’. Direct engagement studies would be needed to discern any potential difference. The activity and impact data were recorded with no connecting reference metadata. Future studies should connect these data to identify which activities have the highest rates of impacts associated with them. The data were non-exhaustive, meaning that data capture was not 100% at any of the sites; at high volume sites not all visitors' behaviours were recorded. Moreover, not all information could be gathered for all observations due to operational logistics during the survey. The study was limited in its scope in that it did not cover the full litany of possible NBT site-types. We interpret the results with these limitations in mind.

3 RESULTS

The data show that 49% of all Natura sites have recreation- and/or tourism-related activities among the known threats/pressures for the site (Table 4). 58% of the 604 Natura sites in Ireland have known threats related to tourism or recreation (Table 4), which is significantly higher than the European average (χ² = 16.08, df = 1, p < 0.001). Over a third of all of Ireland's Natura sites are located within 2 km of a WAW discovery point (Table 4). Furthermore, there are discovery points within or directly adjacent (within 100 m) to 23% of Natura sites along the WAW (Appendix IV). Of the Natura sites within 2 km of discovery points 64% of them have known threats/pressures related to recreation and/or tourism which is significantly higher than the averages for Ireland and Europe (χ² = 31.93, df = 2, p < 0.001).

TABLE 4. Counts of sites within the Natura 2000 Network with known threats/pressures related to tourism or recreational pressures at various scales

Scale/region	Sites	SACs	SPAs	Total^a
Across Europe	Number of Sites	3328	789	3444 (338)
	Sites with known tourism pressures	1498 45%	471 60%	1682 (144) 49%
Ireland	Number of Sites	439	165	604
	Sites with known tourism pressures	239 54%	110 67%	349 58%
Within 2 km of WAW discovery points	Number of Sites	129	73	202
	Sites with known tourism pressures	75 58%	55 75%	130 64%

• ^a () number of sites that are designated as both an SAC and SPA.

The prevalence (number of sites) where the 53 activities from the known threats/pressures associated with recreation, tourism or both were identified within the Natura 2000 network across Europe, Ireland and the WAW are shown in Table 4. The prevalence of the top five threats/pressures at each scale are shown in Figure 2 and further detail is provided in Appendix V.

FIGURE 2

(a) Top five threats/pressures at Natura sites in Europe (outermost circle), Ireland (middle) and along the wild Atlantic way (central donut) * hunting and outdoor sports and leisure activities, recreational activities were joint 5th for the Wild Atlantic Way (WAW). (b) The proportions of other pressures were also graphed for comparison; Appendix V gives the full breakdown of all pressures.

The most common threat/pressure for European Natura sites is hunting (14% of sites), whereas in Ireland it is Walking, horse-riding and non-motorized, occurring at 37% of sites nationally and 23% of sites along the WAW (Figure 2).

The median distance travelled by groups was from the discovery point was 202 m; the furthest distance observed was 5.66 km. 94% of all observations were less than 1 km and 99% of all observations were less than 2 km from the discovery point. Less than 1% (0.88%) of observed visitors travelled beyond 2 km from the discovery points.

Two equivalent models were found to best explain IO (ΔAIC_c = 0.5). The model with the lower AICc contained an interaction between Group-size and AII, which did not significantly contribute to model performance (β = −0.08, SE = 0.050, z = −01.64, p = 0.1012; full Model in Appendix VI). Therefore, the simpler model was selected as the model of best fit. This model contained the following six fixed effects: Site-type, Group-type, Activity Richness; AII; Group-size; and the interaction between Activity Richness and AII. The model explained almost 70% of the variance in IO (Marginal R² 0.6, Conditional R² 0.698).

Activity richness and AII had positive effects on IO (β (richness) = 1.8, SE = 0.13, z = 10.93, p < 0.001; β (intensity) = 0.43, SE = 0.45, z = 23.89, p < 0.001). Similarly, the interaction between AII and activity richness had a positive effect (β = 0.04, SE = 0.11, z = −5.79, p < 0.001). The number of individuals in a group positively affected observable impact of the group (β = 0.12, SE = 1.13, z = 1.25, p < 0.05).

Most group-types were not significantly different from the IO of a couple at a built infrastructure site with the following exceptions: Families and Large Mixed Adult Groups were significantly less likely to have impacts than the reference level group-type (β (families) = −0.33, SE = 0.12, z = −2.06, p < 0.04; β (large mixed adults) = −1.86, SE = 0.16, z = −1.37, p < 0.03). In all, 33 of the 48 records for Large Mixed Adult Groups were recorded as large coach tours and a further 5 were minibus tours.

Under 18s groups and individual adults were significantly more likely to have impacts than the reference group (β (under 18s) = 0.54, SE = 0.07, z = 2.02, p < 0.002, β (individual adults) = 0.49, SE = 0.33, z = −0.65, p < 0.03; respectively).

Clifftop sites were the only site-type that had a significant effect on IO compared to the reference site-type (β = 0.87, SE = 0.99, z = 2.28, p < 0.008).

The graph in Figure 3 shows the strength of association of each of the fixed effects detailed in Table 5 for the IO model. All variables above one has a positive effect and all those below one has a negative effect.

FIGURE 3

Odds ratios for each of the fixed effects in the final observable impact model; showing the 95% confidence intervals (* indicates significant terms in the model; red indicates a negative effect and blue indicates a positive effect compared to the reference value).

TABLE 5. Impact occurrence model outputs (GLMER, family = binomial)

Predictors		Estimates (β)	Standard errors	z-value (n = 5910)	p
(Intercept) [Couple, Built Infrastructure]		−2.55	0.03	−5.788	<0.001
Group-type [Family]		−0.33	0.12	−2.062	0.039
Group-type [Mixed Over 18 Large Group]		−1.86	0.16	−1.373	0.022
Group-type [Individual Adult]		0.54	0.07	2.018	0.001
Group-type [Under 18]		0.48	0.33	−0.650	0.022
Group-type [Small Adult Group]		0.91	0.58	−0.087	0.086
Group-type [Elderly Couple]		−0.30	0.27	3.421	0.170
Group-type [Elderly Group]		−0.28	0.36	0.238	0.516
Group-type [Large Adult Group]		−0.11	0.32	−0.314	0.754
Group-type [Individual Elderly]		0.07	0.31	−0.132	0.812
Group-type [Mixed Over 18 Small Group]		0.04	0.13	−2.289	0.816
Group-type [Mixed Large Group]		−0.04	0.29	0.233	0.895
Group-type [Mixed Small Group]		0.28	0.38	0.115	0.908
Group-type [Family with Elderly]		−0.05	0.22	1.716	0.931
Site-type [Cliff Site]		0.86	0.99	2.283	0.007
Site-type [Highly Managed]		0.70	0.85	0.933	0.212
Site-type [Beach Site]		1.41	2.15	2.721	0.351
Group-size		0.12	1.66	1.247	0.044
Activity Richness		1.79	0.13	10.929	<0.001
Activity Intensity Index		0.43	0.45	23.885	<0.001
Activity Richness*Activity Intensity Index		0.04	0.11	−5.788	<0.001
Random effects
σ ^2		3.29
τ 00 Site		1.06
Intraclass correlation coefficient		0.24
NSite		42
Observations		5910
Marginal R^2/conditional R^2		0.600/0.698

• Bold p values are signicant factors.

For IO, the interaction between AII and total number of activities shows that the positive relationship between the number of activities is strengthened if a higher proportion of the activities being undertaken are at a high or moderate intensity level (Figure 3).

The model of best fit for IS contained the following fixed effects: total impacts, total activities, proportion of activities high or moderate, site-type, year and the interaction terms total activities*proportion of activities high or moderate, total impacts*proportion of activities high or moderate, total impacts*total activities and total impacts*total activities*proportion of activities high or moderate (Table 6). Total number of impacts, total number of activities and proportion of activities high or moderate all had significant positive effects on IS (β (number impacts) = 0.54, SE = 0.17, z = 5.47, p < 0.001, β (number activities) = 0.54, SE = 0.28, z = 3.38, p < 0.001, and β (AII) = 0.37, SE = 0.13, z = 4.22, p = 0.001; respectively).

TABLE 6. Impact severity model outputs (GLMER, family = binomial)

Predictors	Estimates (β)	Standard error	z-value (n = 1354)	p
(Intercept) [Built Infrastructure, 2017]	−0.79	0.23	−1.57	0.116
Total Impacts	0.54	0.17	5.47	<0.001
Total Activities	0.54	0.28	3.38	0.001
Proportion of Activities High or Moderate	0.37	0.13	4.22	<0.001
Site-type [Beach Site]	−0.38	0.33	−0.78	0.431
Site-type [Cliff Site]	−1.96	0.08	−3.67	<0.001
Site-type [Highly Managed]	−1.69	0.12	−2.70	0.007
Year [2018]	1.42	1.76	3.36	0.001
Year [2019]	0.48	0.58	1.36	0.173
Total Activities* Proportion of Activities High or Moderate	−0.19	0.06	−2.84	0.004
Total Impacts* Proportion of Activities High or Moderate	−0.09	0.04	−2.55	0.011
Total Impacts* Total Activities	−0.19	0.06	−2.58	0.009
Total Impacts* Total Activities* Proportion of Activities High or Moderate	0.06	0.03	2.35	0.019
Random effects
σ ^2	3.29
τ 00 Site	0.59
ICC	0.10
NSite	40
Observations	2167
Marginal R^2/conditional R^2	0.17/0.26

• Bold p values are signicant factors.

Figure 4 shows the strength of association of each of the fixed effects in the final IS model graphically.

FIGURE 4

Odds ratios for each of the fixed effects in the final impact severity model, showing the confidence intervals (* indicates significant terms in the model; red indicates a negative effect and blue indicates a positive effect compared to the reference value).

There were a number of second- and third-order interactions which significantly contributed to model performance. The interaction between total number of impacts and AII shows that AII strengthens the positive effect of impact richness when numbers are low; however, as number of impacts increase the negative interaction weakens the effect (Appendix VI Figure 1). AII strengthened the effect of activity richness when numbers are low; however, at high number of activities, activity richness has the stronger influence (Appendix VI Figure 2). For IS, the total number of impacts strengthens the effects of the total number of activities at low levels; however, as activity numbers increase this relationship weakened (Appendix VI Figure 3).

The number of observations at each of the different site-types varied across each of the years (Table 7). IS also varied with respect to year with 2018 having lower IS identified than other years (Figure 5).

TABLE 7. Breakdown of the number of observations at each site-type for each of the years

Site-type	2017	2018	2019
Beach site	250	239	1208
Built infrastructure	139	106	320
Cliff site	65	2368	446
Highly managed	130	247	402

FIGURE 5

Box plot of impact severity levels within each year of the study. The median and interquartile range are shown within the box (* indicate significant difference).

Site-types had varying levels of moderate/high IS. Beach sites and built infrastructure were identified to be statistically similar, whereas cliff sites and highly managed sites had lower levels of impacts when compared to the others (Figure 6). All data and code for the work presented above is available online (Torsney & Buckley, 2022).

FIGURE 6

Box plot showing impact severity levels at each of the site-types observed with median and interquartile range shown within the box (* indicate significant difference).

4 DISCUSSION

Our results show that visitor pressures are widespread throughout the protected sites across Europe (49%). This presents potential conflicts in tourism goals and conservation goals and impedes sustainability (Buckley, 2012; McKercher, 1993; Thapa et al., 2022; Tolvanen et al., 2020). We also show that activity dynamics are key drivers of both IO and IS; which supports the concept that visitor behaviours are complex (Tolvanen et al., 2020) and using general visitor numbers is an over simplification of impact (Monz et al., 2013). Therefore, management practices should identify and appropriately manage the activities available at nature-based destinations to reduce overall impacts. In addition, the study showed 75% of all visitor observations resulted in no identifiable effects on the environment. While not all impacts can be directly observed and ascribed to individuals or groups this result highlights the feasibility of ‘no impact tourism’ being an achievable target and that the prevalence of impacts are not attributable to all tourism.

We show that threats/pressures related to tourism occur at nearly half of all protected sites across Europe, with higher proportions in Ireland and at almost two-thirds of protected sites close to a major tourism discovery route. We highlight key damaging activities that have high occurrence rates at protected sites, which vary across the continental and regional scales shown. The complexities of tourism activities, behaviours and attitudes are known to influence environmental action (Tolvanen et al., 2020). The threats/pressures data are recorded in the EEA Natura 2000 database in a form that gives little understanding with regard to the intricacies of the threats; for example, ‘Outdoor sports and leisure activities, recreational activities’ was recorded at 10% of sites across Europe with no further detail provided. We therefore used a detailed study of tourism movements and behaviours to identify the key drivers of impacts. IOs were low—less than 25% of visitor observations had impact records. These results help to dispel the myth that increasing visitor numbers alone is inherently damaging to a nature-based destination. To achieve no impact from NBT, policy measures are needed to incorporate environmental considerations into the consent and decision-making frameworks for tourism at national and regional levels; specifically with regard to protected sites to incentivize the alleviation of threats and pressures related to tourism. Given the widespread nature of the problem identified in protected sites in Europe action is needed; policy measures could include the need for NBT developments which facilitate access or activities within protected areas to be required to submit environmental visitor management plans or introduce licencing processes for activities at or adjacent to protected areas to provide regulated and controlled access in a systematic and structured manner. Existing protocols which explore these concepts include:

• Limits of Acceptable Change (Jurkus et al., 2022; McArthur, 2000);
• Visitor Impact Management Model (Jurkus et al., 2022; McArthur, 2000);
• Visitor Activity Management Programme (Jurkus et al., 2022, McArthur, 2000);
• Visitor Experience and Resource Protection Model (Jurkus et al., 2022, McArthur, 2000);
• Tourism Optimization Management Model (Jurkus et al., 2022, McArthur, 2000); and
• Integrated Monitoring and Adaptive Management System (IMAMS; Jurkus et al., 2022, McArthur, 2000).

The number of visitors per group was not a strong predictor of IO in isolation and does not contribute to the IS for the group. This is an important finding as the literature focuses on identifying links between visitor numbers and impacts more broadly (Monz et al., 2013) or oversimplifying visitor behaviours/interactions. Transdisciplinary approaches to understanding the complexities of visitor behaviours and complex planning/management dynamics are well supported in the literature (Gundersen et al., 2022; Jurkus et al., 2022; Thapa et al., 2022; Winter et al., 2020). This research shows a need to focus on site-type, tourist demographics and activity interactions when trying to understand IO. Increased education, awareness and guidance material are needed to support the tourism industry in understanding site-type from an ecological perspective to understand the known sensitivities threats or pressures associated with broadscale classes of NBT destinations (beach sites, cliff top sites, etc.). The IS model was more complex with second- and third-order interactions moderating the main effects. However, site-type, activity richness and intensity and number of impacts were all significant as main effects. Therefore, site management should focus on understanding their site and the availability of activities with a view to ensuring the activities can be facilitated in a manner aligned with the sensitivities of the site to ensure impacts are mitigated. The advancement of this understanding may be reliant on ‘citizen science’ or increased remote sensing technologies (Jurkus et al., 2022) as studies of this scale (5910 groups observed) are rare. Resources and training are needed at a national level to provide a platform to facilitate standardized replicate datasets which can be analysed effectively. This approach could facilitate on the ground understanding and learnings being harnessed by national decision-makers.

There is a need to complement the focus on discrete ecological responses to tourism—such as vegetation cover with respect to trampling and subsequent erosion (Bateman & Fleming, 2017; Monz et al., 2013)—to understanding the sources of effects. Looking at visitor behaviours, attitudes and activities in a broader context is where NBT research needs to focus in a transdisciplinary approach (Arif et al., 2022; Gundersen et al., 2022; Jurkus et al., 2022; Tolvanen et al., 2020; Winter et al., 2020). This shift could facilitate the alignment of conservation and tourism goals and enable informed conversations/knowledge exchanges at planning level where interventions are needed (Mandić & McCool, 2022). Tourism-typologies are widely recognized in the tourism literature with many studies showing the variant behaviours or motives of different types of tourists (Mehmetoglu & Normann, 2013) from various demographics (Ozcan et al., 2022; Thapa et al., 2022; Tolvanen et al., 2020). This research shows that group-types also have different environmental behaviours and therefore can connect to discussions of Tourism-typology, although further research is needed to interrogate this concept further with direct engagement surveys. Integrating these concepts into tourism management practices can help to reduce overall impacts which degrade a site, thus promoting a more sustainable model for tourism. It is important to build demographic understanding into the policy and decision-making framework; particularly with respect to promotional material to target low impact demographics and build targeted impact alleviation campaigns.

Site-type was a significant contributor of both IO and IS. Therefore, tourism planning/management processes need to understand the type of site being promoted to ensure it can sustainably host the activities being offered/marketed (Bazargani & Kiliç, 2021; Chin et al., 2018; Font & McCabe, 2017; Liburd & Becken, 2017; Zhang et al., 2020). Clifftop sites had higher rates of observed impacts which is probably due to the topography and gradient of the sites; steep embankments and irregular ground at these sites leave them susceptible to soil slippage when walkers use trails at these sites. Conversely, clifftop sites were seen to have lower IS than built infrastructure sites, which may be due to the limited range of activities which are typically supported at these sites. Site managers can use these findings to promote path fidelity on site, particularly at steep areas. The impacts at managed sites were highly variable which is expected as management practices are site specific with minimal continuity between sites. There is a need to have overall management practices and guidelines set out at a high level to ensure NBT destinations are managed in a way that is cognisant of the reliance of tourism on nature (Elbe et al., 2009; Huybers & Bennett, 2003; Steele, 1995; Vecco & Caust, 2019; Zaremba et al., 2018).

The results show that visitor demographics influence IO; therefore, marketing specifically towards demographics with lower impact could lead to better outcomes. Marketing campaigns are known to be successful in eliciting sustainable visitor behaviours if implemented correctly (Font & McCabe, 2017; Liburd & Becken, 2017; Zhang et al., 2020). Most group-types behaved consistently with the reference group (adult couple); however, families and mixed large over 18s groups had significantly lower rates of observed impacts. This is likely to be due to the managed/controlled nature of such groups. Families are usually constrained in their ranging behaviours and activity patterns. Mixed large over 18s groups are mostly large coach tours which are often guided with potential for context-specific issues to be communicated to the visitor at arrival to a site. Individual adults are more free roaming and likely to engage in activities which are off the beaten track, or at more sensitive/vulnerable locations. Similarly, groups of unsupervised under 18-year-olds had significantly higher impact rates, which may be due to antisocial behaviours associated with teenagers (Dishion & Patterson, 2006). Marketing campaigns and on-site management measures should be targeted towards specific demographics to avoid impacts; both approaches have been shown to have positive effects (Elbe et al., 2009; Font & McCabe, 2017; Zhang et al., 2020). Direct engagement surveys are needed to further investigate the motives and demographics of visitors; this can help disentangle local recreational pressures and tourism pressures.

In general, visitor intensity has been used as a proxy for environmental impacts (Monz et al., 2013). Interestingly, although group-size significantly contributed to the overall model performance the effect size was relatively low (β = 0.12). The number of activities and intensity of activities undertaken had greater influence on observable impact. It is important to note that increasing visitor numbers is often a fundamental target embedded in tourism strategies and although group-size is a factor, it needs to be considered together with site dynamics, tourism demographics and activity interactions. Understanding environmental attitudes and behaviours are important considerations so as not to oversimplify interactions (Fenitra et al., 2021). There needs to be a shift away from focusing on visitor numbers for performance and/or environmental damage. Moreover, tourism targets should be expanded to focus on increasing engagement, awareness and interaction to activate the learning opportunities presented by tourism; this could increase environmental stewardship and responsible behaviours (Fenitra et al., 2021; Jurkus et al., 2022; Tolvanen et al., 2020) to alleviate existing conflicts.

Activity intensity (the proportion of moderate or high-level activities) had significant contributions to both the observable impact and IS models. For IO, the positive interaction between activity intensity and total number of activities shows that the relationship with number of activities is strengthened if the activities being undertaken are at a high or moderate intensity level. Positive main effects, negative two-way interactions and a positive three-way interaction between total impacts, total activities and activity intensity in the IS model however show that there is a threshold value for the combination of these factors in determining IS which needs to be understood from a management perspective (Steele, 1995). The IMAMS could be used with direct community involvement focused on activity types, intensities and frequencies; this approach could provide valuable learnings to assess the relative merits or successes of various management interventions.

Understanding the predictors of IO and IS particularly relevant when the site managers are somewhat removed from the tourism hierarchy such as council owned beaches or clifftop walks or when the tourism destination is within 1 km of a protected site as the 99% of visitor observations were recorded within 1 km of sites. Distance-based impact density issues needs to be extended into the tourism hierarchy of planning to ensure adequate resources and communication chains are in place to facilitate sustainable management (Liburd & Becken, 2017). Understanding the activities that have the highest IO rates across protected sites could help to focus management action to control these impacts and possibly reduce their effects. This is particularly important in an Irish context where there are widely recognized threats to protected areas from walking, horse-riding and non-motorized vehicles. Guidelines should be produced for regulating activities which have a known environmental impact and funds made available for conflict resolution in this regard.

The study was limited in its scope in that it did not cover all possible NBT site-types. Similarly, the study had a short time period; year was not significant with respect to IO however, it was significant for IS. 2018 had lower rates of moderate/high impacts, this is thought to be due to the 2018 sites being the ‘signature discovery points’; the 15 sites along the WAW which are key mass tourism destinations in Ireland. At mass tourism destinations movement patterns and activities may be constrained by high volumes of tourists doing coordinated activities and therefore there are fewer free roaming incidences for impacts to occur. Further studies focusing on direct engagement studies at a wider variety of sites could help to further understand the role of demography and site type in IO and IS and disaggregate tourism effects from local amenity use. However, there are no significant limitations identified with regard to the reported findings of the study.

Further research is needed to identify the distribution of impacts around tourism destinations. It is expected that impacts are concentrated around the receiving locations (such as carparks) and that there is a distance threshold where impacts are negligible. Understanding the spatial distribution of impact would help to focus interventions. A clearer understanding is needed with respect to the relationships between activities and impacts, that is, do specific activities cause specific impacts or impact types. This could be achieved through a more refined data collection system which links impacts to activities where they are apparent. To further expand the scope of our understanding, applying these methods to a wider array of site-types would help to understand broader site interactions. Additional nature-based site-types which were not included in this research would be riparian destinations, woodland parks and/or mountain trails. Using average daily visitor numbers could provide insight into capacity thresholds within site-type. Further research to garner a deeper understanding into tourism-typologies are needed with respect to environmental impacts. This could be achieved by intercepting visitors upon departure, having captured their movement and behaviours on site and collecting information relating to environmental literacy, cultural background, motive for attending the site, etc. This could deepen our understanding of the intricacies within group-type which are currently not known with respect to environmental impacts.

AUTHOR CONTRIBUTIONS

Andrew Torsney and Yvonne M. Buckley had substantial contributions to conception and design, analysis, and interpretation of data, and revising the paper. Anrew Torsney oversaw the data collection, performed the data analysis and wrote the first draft.

ACKNOWLEDGEMENT

Open access funding provided by IReL.

FUNDING INFORMATION

Research funded by CAAS Ltd. and supported by Fáilte Ireland, National Tourism Department Authority of Ireland. Yvonne M. Buckley is funded by the Irish Research Council Laureate Awards 2017/2018 IRCLA/2017/60.

CONFLICT OF INTEREST

The authors confirm that there are no conflicts of interest with regard to the submitted article with regard to the Conflict-of-Interest policy statement of the Ecological Solutions and Evidence journal.

Open Research

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1002/2688-8319.12207.

DATA AVAILABILITY STATEMENT

The data and code that support the findings of this study are openly available in GitHub: https://github.com/TorsneyStat/Predictors-of-impact-occurrence-and-severity-of-impacts/releases/tag/tourismmanagement and archived in Zenodo https://doi.org/10.5281/zenodo.7457424 (Torsney & Buckley, 2022).