(**a**) **Moran`s Global I test** (**b**) **Getis-Ord general G(d) test**

**Figure 2.** Results of Moran's gGlobal I Test and the Getis–Ord general G (d) test on beds available in Extremadura. Source: Own material from calculations made with ArcGIS 10.3.

From the results obtained for each of these tests the lack of coincidence can be appreciated; this may be due to the weak relationship of the variable which means that this relationship is not maintained in all the tests carried out. In turn, this weak relationship may be a sign of an equal bid on the part of the public administration of Extremadura for the development of accommodation infrastructures apart from the resources and the tourist potential of each tourist territory in the region.

On the other hand, the study carried out up to now only takes into account the analysis of the beds at a global level, i.e., considering the whole of the target territory. The fact that a global distribution pattern is not identified, or that if it is identified it is weak, does not imply that it is impossible for there to be groups of lodgings at a local level with values of the variable which are very different from the average values of their neighbours. For this reason, Moran's local I test has been calculated; this will allow us to obtain further knowledge on the distribution of beds for tourists at a local level. The results obtained can be observed in Figure 3.

In view of Figure 3, it can be concluded that certain areas exist in which groups of accommodation establishments offer a higher number of beds than those located in neighbouring areas. In the case of the two main towns of the region, both of which are provincial capitals, the existence of several accommodation establishments with high values of the variables in comparison with the value of neighbouring locations can be observed. In these two locations, therefore, one can speak of the existence of two clusters of high values of the variables, i.e., of two clusters in which the accommodation establishments offer a higher number of beds than those located in neighbouring areas.

In the same way, this technique will allow the identification of atypical spatial entities, i.e., those with a high (or low) value of the variable which are surrounded by low (or high) values. In this case, one can observe the existence of three accommodation establishments with a high number of beds compared with the average offered by neighbouring establishments. In turn, there are also two accommodation establishments which offer a small number of beds and are surrounded by neighbouring establishments offering on average a higher number. Finally, two isolated lodgings were also identified with a high number of beds in comparison with the average. In all cases, the results obtained will not be taken into consideration in the analysis performed as we are here concerned with findings related to isolated accommodation establishments, which although they are significant in

terms of the administration of the same, do not allow the forming of accommodation groups which sugges<sup>t</sup> that the results can be extended to the entire territory.

**Figure 3.** Moran's local I for tourist places in Extramadura. Source: Own material from calculations made with ArcGIS 10.3.

Once the analysis of the distribution of the variable of the beds available in the territory of the region has been completed, it can be concluded in the first place that there is a slight tendency towards the concentration of the values of the variable in the total space of the region. Secondly, the existence is confirmed of two groups of accommodation clusters offering a high number of beds in comparison with the average for the territory and which are located in the two main centres of population: Badajoz and Cáceres.

Subsequently, we proceed to repeat the analysis performed for the second variable which is the subject of this study, the occupation level, which will allow the identification of the level of adjustment existing between supply and demand for the region of Extremadura.

Figure 4 shows the results obtained for Moran's global I test and the Getis–Ord general G (d) test. As can be seen in Figure 4, statistic I has a value of 0.1594, which means a score *z* of 8.7278; it can thereforebeconfirmedthatatasignificancelevel of1%thereisapatternoftheconcentrationofthe

variable in the space. In other words, at a confidence level of 99% it is confirmed that the similar values of the variable have a tendency towards concentration in the space.

**Figure 4.** Moran's global I test and the Getis–Ord general G (d) test on the occupation level in Extremadura. Source: Own material from calculations carried out with ArcGIS 10.3.

For its part, the Getis–Ord general G (d) test gives a statistic with a value of 0.0863, with the score *z* being 3.9279, and a p-value of practically zero. This implies that a significance level of 1% confirms that the variable tends to concentrate in the space and moreover is concentrated at high values.

As we can see, both results coincide and confirm the existence of a global pattern of the concentration of the occupation level variable in high values of the same. However, the contrast of the two results obtained with respect to the global distribution of the variables allows us to confirm the initial suspicions regarding the results of the expansionary growth policies used as development strategies in the region. These have allowed the tourist places in the region to have a practically homogeneous distribution pattern, thus only obtaining confirmation of a weak tendency to concentrate in the high value space that is not confirmed in the totality of the tests performed. On the other hand, the results obtained for the occupation level do confirm the existence of a strong tendency to group high values in the region, which suggests the existence of a serious imbalance in terms of the efficiency of these places that should be studied in an exhaustive way with the objective of being able to create strategic lines that fit the reality of each of the territories in which the region is divided, with the ultimate goal of achieving sustainable tourism development throughout the territory.

The local analysis of the variable by means of Moran's local *I* test allows determining whether groups with similar behaviour are created and their locations. The results of this analysis can be seen in detail in Figure 5.

In view of the results obtained, the presence of different groupings of spatial clusters with similar values as to the occupation level is confirmed, i.e., they show a high or low level of adjustment between supply and demand in comparison with the average value to be expected in the space analysed.

Firstly, we can observe the presence of three clusters of high values of the occupation level variable located in the towns of Badajoz, Mérida, and Cáceres. These three territories are therefore characterised by the presence of a high number of lodgings with higher occupation levels than would be expected from the average value of their neighbouring territories. If we take the occupation level as a proxy indicator, we find that these are the three territories in the region with the best adjustment between supply and demand.

**Figure 5.** Moran's Local I for the occupation level. Source: Own material from calculations carried out with ArcGIS 10.3.

For their part, the presence of two clusters of low values is observed in the territories of Trujillo-Miajadas-Montánchez and the Villuercas-Ibores-Jara Geopark. In these territories, therefore, we find a set of accommodation establishments with the common characteristic of a low occupation level with regard to the average value to be expected in the month and territory analysed. In accordance with the above reasoning, these two territories are those with the worst adjustment between supply and demand, owing to which they must concentrate on developing strategies allowing the increasing of said occupation level so that the creation of the existing beds for tourists fulfils the mission of creating wealth and employment in the territories analysed.

Finally, the presence of different spatial outliers and outlier types is observed, such as high outliers and low outliers. As can be seen, these are not grouped together in any of the territories in question but are rather isolated cases of accommodation establishments with a casuistry causing this situation owing to specific management, or to obtaining better or worst results than those to be expected in the territory in which they are to be found. Although these results could represent a valuable output for each specific accommodation establishment, they go beyond the scope of the global objective of this research, which consists of the location of groupings, which as their situation is similar, can help the administration to characterise them, and as from this starting point to create common lines of strategy which allow the optimum tourist development of each territory analysed.

As a general conclusion of the distribution pattern of the occupation level variable, it should be stressed that in contrast to what occurred with the beds available, on this occasion a strong tendency to the concentration of high values in the space can be observed. This conclusion is confirmed with the existence of various clusters of high and low values, which requires an analysis of the same before carrying out any task of tourist planning in the interests of achieving more e fficient management.

#### **6. Discussion and Conclusions**

The analytical possibilities of the inclusion of GIS for the application of the techniques of spatial statistics have opened up many options for achieving highly exhaustive analyses in the territories and their spatial dimension. By means of the application of said techniques it is possible to identify spatial distribution patterns of the variables under study in a given territory and also to identify groupings of entities with a similar or di fferent behaviour to that which could be expected in a homogeneous distribution of the variable. The main advantages deriving from the identification of these clusters is that they will allow the joint planning of more extensive territories, in which owing to their characteristics it is possible to analyse the initial situation and at the same time the existing needs to establish a common goal, making use of the advantages of this joint planning regarding management.

It is for this reason that more and more studies are concentrating on analysing the distribution patterns of the variables related to tourism in the space, with the aim of improving the planning and managemen<sup>t</sup> of the territory. These studies are characterised by being essentially focussed on demand, concentrating their e fforts on studying the distribution in the space of the accommodation capacity of the territory.

This study expands this perspective, completing the analysis of the supply with a detailed analysis of the distribution of the occupation level, taking this last variable as a proxy indicator of the satisfactory level of adjustment between supply and demand in the territory. By means of this analysis therefore, the objective of getting to know the e fficiency of each territory with regard to the hotel beds o ffered is achieved, identifying accommodation groupings with similar characteristics in the territory and which therefore can be examined for correct planning and managemen<sup>t</sup> in a joint manner.

Moreover, this analysis also presents the novelty, regardingthose tudied on the region previously, of using a geostatistical approach which allows the using as an analytical unit the maximum level of territorial disaggregation possible, the location of the establishment itself. As a result, the contributions of this research allow on the one hand exhaustive knowledge of the distribution pattern of the two variables which are the subject of study in the month of reference used, and on the other hand the laying down of the methodological foundations allowing their repetition at di fferent moments in time, allowing in this way the considerable enriching of the results obtained. In this sense, and after performing various tests in accordance with the characteristics of the sample used, the decision was taken to consider as the most suitable neighbourhood criterion a fixed band distance of 15.79 miles.

With regard to the main results obtained from this study, it should be emphasised that the beds available in the total territory analysed have a weak tendency towards the concentration of values in the space. However, given that this relationship is not significant in the Getis–Ord general G (d) test, it cannot be confirmed whether this tendency is the result of the concentration of high or low values of the variable.

From a local perspective, the presence has been detected of two clusters of high values in the main population centres of the region, i.e., in the towns of Cáceres and Badajoz.

The analysis of the global distribution of the occupation level variable identifies a strong tendency towards concentration in the space, and moreover this spatial concentration occurs with high values. It is therefore a variable with the characteristic of a contagion e ffect which allows the increase of a variable in a given space, favouring also the value taken in its proximal locations, which is a clear indication that joint managemen<sup>t</sup> has the possibility of benefitting a wider extension than that of the territory itself.

In relation to the study of the distribution of the occupation level at a local scale, the presence is identified of three spatial clusters of high values located in Badajoz, Cáceres, and Mérida. These territories are therefore characterised by concentrating a high number of accommodation establishments with high occupation levels in comparison with what would be expected in a homogeneous distribution of the variable in the territory.

With regard to the administration of this result, it must be taken into account that although these three territories stand out as being the most e fficient in relation to the beds they provide, the result only implies that they have a good initial situation as the average occupation level of each of these territories in the month analysed varies from 38.2% in Badajoz, 40.2% in Cáceres, and 60.7% in Mérida as has been calculated from the data available from the Extremadura Tourism Observatory. In all cases it can be observed that there is clearly room for improvement, and that there is no evidence to assume that the same are exceeding the maximum load capacity to ensure sustainable management, owing to which said territories should in the first place find out whether this situation remains constant over time. Secondly, the regional administration should analyse the characteristics of the o ffer of tourist products, which at the moment places them in this position, and finally analyse the possibilities of the territory in accordance with their allocation of resources so as to improve the initial situation.

On the other hand, the analysis has also allowed the identification of two accommodation clusters with low occupation values in relation to the figures for accommodation in their neighbouring locations; these are the territories of Trujillo-Miajadas-Montánchez and the Villuercas-Ibores-Jara Geopark. The average occupation levels of these territories are 20.7% and 16.7%, respectively. These figures are below the regional average, which is 25.8% of the beds available for the month being studied.

We are therefore concerned with two territories with a demand deficit which they must resolve if the beds they o ffer are to contribute to the creation of wealth and employment. It would be recommendable to perform an exhaustive analysis on the supply of the type of tourist product these territories currently o ffer with the aim of determining whether there is a strong dependence on any specific tourist type with a marked seasonality which is causing this situation. If necessary, the existing offer should be diversified with the aim of palliating this deficit and improving the results. In short, once the similarity of the initial situation has been detected, the analysis of the reasons will give rise to a series of common needs which will be those allowing the articulation of the lines of action to bring about the more e fficient managemen<sup>t</sup> of their resources.

To conclude this analysis of the results, it should be pointed out that both in the local analysis of the distribution of the beds available and in the respective occupation level a series of accommodation establishments appear which represent both clusters of similar values (HH or LL) and an outlier of HL and LH values. These results have not been detailed among the main conclusions of this study, as all of them are found in the space in isolation, owing to which said findings, although they may constitute a valuable output for the managers of these establishments, are beyond the scope of this study, the purpose of which is the comprehensive (or global) managemen<sup>t</sup> of the territory.

One of the main limitations of this research is that the results obtained are restricted to the timeframe used, July 2015. In order to resolve this shortcoming, we propose as a future line of research the extending, by means of the repetition of the methodology proposed, of the results to a wider timeframe, preferably annual to limit the possible e ffects of seasonality in each of the territories, so as to find out whether the clusters identified are maintained throughout the whole of the tourist season, and also the possibility that new associations may appear so as to expand the knowledge on the region of Extremadura. Likewise, it would be enriching to complete the analysis of the tourist reality of the region and be able to design properly adapted development strategies, disaggregate the results obtained according to the type of accommodation, taken into account variables such size, quality, the profile or ownership, among others, to eliminate the possible e ffects that the peculiar characteristics of the results might be introducing into each one of them.

**Author Contributions:** Conceptualization, M.C.R.R. and M.S.R.; methodology, M.C.R.R. and M.S.R.; software, M.C.R.R. and M.S.R.; validation, M.C.R.R. and M.S.R.; formal analysis, M.C.R.R. and M.S.R.; investigation, M.C.R.R. and M.S.R.; resources, M.C.R.R. and M.S.R.; data curation, M.C.R.R. and M.S.R.; writing—original draft preparation, M.C.R.R. and M.S.R.; writing—review and editing, M.C.R.R. and M.S.R.; visualization, M.C.R.R. and M.S.R.; supervision, M.C.R.R. and M.S.R.; project administration, M.S.R.; funding acquisition, M.S.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This publication is part of the research carried out within the research project "Análisis de factores críticos para el desarrollo turístico de Extremadura (IB-18015)". This project is funded by the Ministry of Economy and Infrastructure of the Junta de Extremadura and by the European Regional Development Fund (ERDF). The article processing charge (APC) was financed by Junta de Extremadura with ERDF funds (GR18124)

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
