2. Materials and methods

Our study site was located in the south of the Czech Republic (Figure 1). This area of South Bohemia with about 10,057 km<sup>2</sup> extends between 400 and 800 m above sea level and is known for many localities of different orchid species, including even critically endangered species in the Czech Republic, such as Liparis loeselii or Malaxis monophyllos. The advantage of this area is in quite a small human population density, which allows preserving a natural environment suitable for endangered species.

The orchid family is regarded as one of the largest and most diverse taxa of this rank in the flowering plant kingdom, with estimates of 880 genera and about 20,000–35,000 species [6–8]. Orchids are found in many different habitats but not in areas that are extremely cold or dry throughout the year [9]. Many characteristics, such as great species richness, its specific role in ecosystem, or endangered situation make it crucial to explore the distribution and conservation status of Orchidaceae [10]. It is an important group with respect to conservation biology

Species distribution models (SDMs) are a useful tool, which is now often applied in many branches of biogeography, conservation biology, and ecology [13], especially when threatened species are concerned [14]. These numerical tools combine species occurrence records with environmental data [13]. In combination with GIS techniques, these models are especially important and useful for predicting occurrence of rare species [15] especially in areas where certain parts are not fully explored. The species distribution models are then the only means

In our study, we used the maximum entropy algorithm in the MaxEnt application [16–19]. This algorithm uses maximum entropy and Bayesian methods to estimate the probability distribution of each species based on their presence or absence. Since becoming available in 2004, MaxEnt has been utilized extensively for modeling species distributions. This approach was used by conservation practitioners for predicting the distribution of a species from a set of occurrence records and environmental variables [19, 20] as well as in numerous other fields of biology and ecology that cover diverse aims across ecological, evolutionary, conservation, and biosecurity applications [19]. Despite long history of studies on orchids, only a minute part of previous papers concerning distribution, phytogeography, or conservation strategies of this taxonomic group included application of species distribution models, for example, see [21–24]. Presence-only modeling methods require exclusively a set of known species occurrences together with predictor variables such as topographic, climatic, edaphic, biogeographic, and/ or remotely sensed data [17, 18]. As an output from the MaxEnt program, we get extensive information, for example, maps of distribution of suitable niches and contribution of input

Here, we show an example of using the species distribution models for analyses of orchid species occurrence in the Czech Republic. We estimated which climatic, environmental, and other associated factors influence the distribution of two selected species and tried to find a new, yet unknown, localities in area selected. A similar approach was previously used in the study

Our study site was located in the south of the Czech Republic (Figure 1). This area of South Bohemia with about 10,057 km<sup>2</sup> extends between 400 and 800 m above sea level and is known for many localities of different orchid species, including even critically endangered species in the Czech Republic, such as Liparis loeselii or Malaxis monophyllos. The advantage of this area is

concerning conservation of orchid species in the Greek island of Crete [3].

enabling prediction of biodiversity for the group in question in such areas.

[11] being at the frontline of extinction [12].

134 Selected Studies in Biodiversity

variables to the model.

2. Materials and methods

As a source of data, we used information from 5 databases—the database of the Nature Conservation Agency of the Czech Republic [25], the Czech National Phytosociological Database, and the Floristic Documentation, both deposited at the Department of Botany and Zoology, Faculty of Science of the Masaryk University in Brno [26], the database of the South Bohemian Branch of the Czech Botanical Society [27] and the database of the inheritance of the late František Procházka (10,000 items, digitized from original cards). All data from these databases are deposited in one comprehensive database at the Global Change Research Institute (CAS), Department of Biodiversity Research in České Budějovice, but in order to protect the orchids in the localities, there is no public access to either of these databases.

During 2014–2016, we visited as many localities as possible to check, whether a selected orchid species is still present there or not. If the species was found, the number of flowering plants was counted and all important information, such as accurate GPS coordinates, how the locality looked like, or if it was mown or not was registered. The total of 428 localities was checked.

Because of special demands of methods in MaxEnt we used, only the two most numerous species were incorporated in all analyses. The first one was Dactylorhiza majalis (Rchb.) P.F. Hunt & Summerh., which lives in wet meadows, and the second species was Platanthera bifolia Rich., which flourishes in light deciduous forests.

Figure 1. Map of the study site in the Czech Republic.

A set of environmental and habitat variables was prepared using available datasets for the Czech Republic. They were divided into two groups according to its spatial scale and ecological meaning (Table 1).

KVES is a list of 40 types of habitat type, named as KVES\_1, KVES\_2, …, KVES\_40. For example, KVES\_4 means alluvial and wet meadows, KVES\_5 means dry grasslands, and so on (see Table 2 for further examples). According to our knowledge, encompassing many years of orchid research, and to the information published in literature on ecological requirements of individual

Determinants of Orchid Occurrence: A Czech Example http://dx.doi.org/10.5772/intechopen.74851 137

consolidated layer of ecosystems • Alluvial and wet meadows

• Vegetation of standing waters • Wetlands and coastal vegetation • Peat bogs and water springs • Rocks and brushes • Swamps and marshes • Mixed forests

• Sports and recreational areas • Agricultural meadows • Dominant habitat type

solar\_rad Solar radiation—total amount of incoming solar insolation (WH/m<sup>2</sup>

summer\_days Number of summer days (with temperature exceeding 25C) per year

trop\_days Number of tropical days (with temperature exceeding 30C) per year

vert\_het Vertical heterogeneity (standard deviation of altitude)

• Urban green areas, gardens, parks, or cemeteries

op\_buff Amount of arable land in the buffer zone of 250 m from particular orchid species (%)

• Habitat heterogeneity (amount of different types of habitats)

)

• Dry grasslands • Mesophilic meadows • Oak and oak-hornbeam forests

• Beech forests • Dry pine groves • Natural shrubs

orna\_p\_buff Amount of arable land in the square of 500 to 500 m (%)

precipitation Total precipitation per year (mm) reactivity Reactivity of rocks in a bedrock

temp\_1 Mean yearly temperature (C)

TPI Topographic position index

veg\_season; veg\_sez Duration of vegetation season

Table 2. Description of all important factors used in all analyses.

temp\_2 Temperature variability during year (C)

zapl\_pl Periodically flooded areas (binary variable)

slope Slope of terrain ()

Code Description

dem Altitude

KVES • 4 • 5 • 6 • 10 • 12 • 13 • 17 • 18 • 19 • 20 • 21 • 23 • 30 • 33 • 34 • 39 • maj • var

Alkali Alkalinity of rocks in a bedrock

frost\_days Number of freezing days per year

All analyses were conducted by the MaxEnt program version 3.3.2 [17–19]. In this program, we first performed the jackknife procedure, which told us how the species reacts to different environmental factors. Two different blue bars are always displayed in the resulting figure. The length of the dark-blue bar tells us, how large the impact of the selected factor is. The length of the light-blue bar tells us, how much information would be lost, if the corresponding factor were excluded from the analysis. Thus, deletion of a factor associated with the long light-blue bar would cause a large loss of explanatory power of the model. Then we performed three analyses for each species.

Before describing these, we have to elucidate the meaning of one factor used in the analyses that consists of 40 subfactors: the meaning of the "consolidated layer of ecosystems" (KVES) [28].


Table 1. Description of variables used in the analyses.

KVES is a list of 40 types of habitat type, named as KVES\_1, KVES\_2, …, KVES\_40. For example, KVES\_4 means alluvial and wet meadows, KVES\_5 means dry grasslands, and so on (see Table 2 for further examples). According to our knowledge, encompassing many years of orchid research, and to the information published in literature on ecological requirements of individual

A set of environmental and habitat variables was prepared using available datasets for the Czech Republic. They were divided into two groups according to its spatial scale and ecolog-

All analyses were conducted by the MaxEnt program version 3.3.2 [17–19]. In this program, we first performed the jackknife procedure, which told us how the species reacts to different environmental factors. Two different blue bars are always displayed in the resulting figure. The length of the dark-blue bar tells us, how large the impact of the selected factor is. The length of the light-blue bar tells us, how much information would be lost, if the corresponding factor were excluded from the analysis. Thus, deletion of a factor associated with the long light-blue bar would cause a large loss of explanatory power of the model. Then we performed

Before describing these, we have to elucidate the meaning of one factor used in the analyses that consists of 40 subfactors: the meaning of the "consolidated layer of ecosystems" (KVES) [28].

> frost\_days precipitation solar\_rad summer\_days trop\_days veg\_season temp\_1 temp\_2 KVES slope

KVES\_5 KVES\_6 KVES\_20 KVES\_21 KVES\_39 KVES\_maj KVES\_var orna\_p\_buff TPI veg\_sez vert\_het zapl\_pl

KVES KVES\_4 KVES\_6 KVES\_var op\_buff reactivity

Dactylorhiza majalis Platanthera bifolia

dem frost\_days precipitation solar\_rad summer\_days trop\_days veg\_season temp\_1 temp\_2 KVES slope

KVES\_4 KVES\_5 KVES\_6 KVES\_20 KVES\_21 KVES\_39 KVES\_maj KVES\_var orna\_p\_buff TPI veg\_sez vert\_het zapl\_pl

alkali KVES KVES\_5 KVES\_var op\_buff reactivity solar\_rad vert\_het

ical meaning (Table 1).

136 Selected Studies in Biodiversity

three analyses for each species.

Analysis 1 dem

Analysis 2 KVES\_4

Analysis 3 alkali

Table 1. Description of variables used in the analyses.


Table 2. Description of all important factors used in all analyses.

orchid species [29–32], we suspected that these factors might be important for determination of the occurrence of these species and therefore we included them into the analyses. KVES without a number means the presence of the certain habitat class, therefore it is a categorical variable. If this proves to be statistically significant, it means that the occurrence of the corresponding orchid species depends on some habitat type. Sometimes also the environmental heterogeneity (here called KVES\_var—see Table 2), expressed as the number of different KVES types per unit area (sometimes also called "grain size" in the literature, especially in the landscape ecology jargon) may be important—large KVES\_var means that the landscape consists of a mosaic of many small units like fields, pastures, meadows, forests, and so on, which usually indicates low-intensity agriculture and subsequently a likely good habitat for protected species. Therefore, the KVES\_var is sometimes included in our analyses. Similarly, variable KVES\_maj provides information about dominant habitat type within the assessed zone.

tells us at which position the locality is in the terrain—for example, whether it is on the top of a hill, in a valley, or near a depression. The information about periodical floods (zapl\_pl) help us to determine whether the studied species prefer dry or wet areas or whether the probability of occurrence is higher in wet or dry localities. Another important factor influencing the occurrence of orchid species is the amount of arable land near the selected locality (orna\_p\_buff and op\_buff). These two similar factors have a great impact on the distribution of orchids because with the increasing amount of arable field in the vicinity of localities, the probability of occurrence of studied species decreases rapidly, almost to zero. Arable lands are highly influenced by humans and full of artificial nutrients that are not suitable for the occurrence of orchids in general. The duration of vegetation season (veg\_sez) was also added into this analysis but it has no important influence on the distribution of the studied species because the length of the vegetation season does not differ across the whole country. The last important environmental variable is vertical heterogeneity (vert\_het). This factor explains how much rolling is the landscape near the selected locality, so how many of different altitudes comprises

Determinants of Orchid Occurrence: A Czech Example http://dx.doi.org/10.5772/intechopen.74851 139

The final analysis, Analysis 3, then uses only those factors, which proved to contribute to the determination of the presence of the orchid species studied, which followed from the previous two analyses. These factors were selected as the most significant ones from the first and second analyses and their linking into one analysis should determine which of them has the highest impact on the occurrence of studied species in the selected area (see Table 1). It could be just one, as well as a combination of more of them. The influence of alkalinity and reactivity of rocks in bedrock of a particular locality was added into this analysis [34] because according to literature, particular orchid species prefer only one or two rock types [29, 30, 32]. The final

The detailed description of all factors used in each of these analyses is shown in Table 1 and

The jackknife procedure in Figure 2 indicates that many of the variables included in this analysis have a certain impact on this species. However, in Central Europe, because of the rather flat terrain, the mesoclimatic variables reflect the position in a particular region (such as South Bohemia, or Northern Moravia or so) rather than exact position of the point considered. In other words, the same set of mesoclimatic conditions characterizes a relatively large area, rather than a particular point. Therefore, the set of mesoclimatic variables found in the localities was characteristic for South Bohemia rather than for occurrence of orchids. For example, in Figure 3, there is not a clear trend, as the values are only precipitation values in the particular localities. Therefore, neither precipitation, nor other mesoclimatic variables were used for the

the description of each important factor used in all analyses is shown in Table 2.

the area. All of these factors are also explained in Table 2.

potential distribution map was then created based on this analysis.

3.1. Dactylorhiza majalis (Rchb.) P.F.Hunt & Summerh

3. Results and discussion

3.1.1. Analysis 1: climatic factors

The KVES variable was used in Analysis 1, as described later. For any orchid species, particular vegetation types might be characteristic—for example, KVES\_4 (alluvial and wet meadows) may—according to our knowledge—characterize a typical habitat for Dactylorhiza majalis. Thus, in subsequent analyses, only those vegetation types, which we suspected as candidates for description of the presence of the corresponding orchid species, were selected, as described in Table 1. Detailed description of the particular KVES values is given in Table 2 only for those KVES factors used in the analyses. So, the three analyses were as follows.

In Analysis 1, the influence of climatic variables and other basic abiotic gradients on orchid distribution was studied. The list of these factors is shown in Table 1 and their description in Table 2. The climatic data were obtained from the Global Change Research Institute of the CAS and a climate character from a timeline of 1981–2011 was created. Besides of the climatic factors, we also added KVES and slope of the terrain [33] as additional factors that could influence the distribution of Dactylorhiza majalis and Platanthera bifolia. This analysis was aimed to test, to which extent climate may affect the occurrence of the studied orchid species. However, at least some of other most important environmental nonclimatic factors had to be included, too, in order not to indulge into a purely climatic model, which does not seem to be appropriate in our case—our knowledge and literature information tells us that climate itself is not able to fully explain presence of orchid species in these temperate and rather flat regions [29–32]. There was no risk in including these additional factors—if our expectation did not come true, then these factors would just prove to be not significant.

As the results of Analysis 1 were not describing the presence of the studied species sufficiently in either of the studied species, we performed Analysis 2, which was more specific to selected environmental variables—particular KVES values. We selected these according to our experience and to the indications given in orchid literature—description of ecological requirements of the studied orchid species [29–32]. We also added the topographic position index (TPI), information about periodical floods (zapl\_pl), and the amount of arable land in the square of 500 to 500 m (orna\_p\_buff) and similarly the amount of arable land in the buffer zone of 250 m from particular orchid species (op\_buff), duration of vegetation season (veg\_sez), and vertical heterogeneity (vert\_het; see Table 1) as they might be important for the occurrence of particular orchid species. TPI classifies the landscape into slope position and landform category and tells us at which position the locality is in the terrain—for example, whether it is on the top of a hill, in a valley, or near a depression. The information about periodical floods (zapl\_pl) help us to determine whether the studied species prefer dry or wet areas or whether the probability of occurrence is higher in wet or dry localities. Another important factor influencing the occurrence of orchid species is the amount of arable land near the selected locality (orna\_p\_buff and op\_buff). These two similar factors have a great impact on the distribution of orchids because with the increasing amount of arable field in the vicinity of localities, the probability of occurrence of studied species decreases rapidly, almost to zero. Arable lands are highly influenced by humans and full of artificial nutrients that are not suitable for the occurrence of orchids in general. The duration of vegetation season (veg\_sez) was also added into this analysis but it has no important influence on the distribution of the studied species because the length of the vegetation season does not differ across the whole country. The last important environmental variable is vertical heterogeneity (vert\_het). This factor explains how much rolling is the landscape near the selected locality, so how many of different altitudes comprises the area. All of these factors are also explained in Table 2.

The final analysis, Analysis 3, then uses only those factors, which proved to contribute to the determination of the presence of the orchid species studied, which followed from the previous two analyses. These factors were selected as the most significant ones from the first and second analyses and their linking into one analysis should determine which of them has the highest impact on the occurrence of studied species in the selected area (see Table 1). It could be just one, as well as a combination of more of them. The influence of alkalinity and reactivity of rocks in bedrock of a particular locality was added into this analysis [34] because according to literature, particular orchid species prefer only one or two rock types [29, 30, 32]. The final potential distribution map was then created based on this analysis.

The detailed description of all factors used in each of these analyses is shown in Table 1 and the description of each important factor used in all analyses is shown in Table 2.
