Spatial distribution of Ants Nests in SE Spain

1

Abstract

Spatial distribution of Ant Nests in SE

Spain

Bachelor thesis: Bart ter Mull

14 August, 2018

Student number: 10530568

2

Abstract

The Guadalentín basin is one of the driest places of Europe and is prone to desertification and land degradation. In these Semi-arid climates, the lack of a full vegetation cover and soil properties play an important role in soil erosion processes. Soil erosion is influenced by soil properties, such as crusting, porosity and water content. Aggregate stability is considered less important than vegetation cover, but is still significant. Soil properties are being altered by organisms such as harvester ants (e.g Messor Bouvieri), which collect seeds from Stipa tenuissima and bring them to their nests. The accumulated nutrients and increased soil aggregate stability and water content in the ant nest result in different soil conditions compared to the surrounding area. These harvester ants are ecosystem engineers, which alter soil properties and have influence on the hillslope and catchment scale. There is less scientific information available on microorganisms influencing semi-arid geo-ecosystems. In the research area harvester ants were located. The geosystem characteristics of the observed ant nests were compared to random generated points. The results show that there is no significant difference between the geosystem variables (slope, elevation, geomorphology, soil and lithology). The total variance explained by the variables is 74%. The distribution of ant nests can be described as dispersed. The harvester ants have preference for seeds especially for Stipa marcochloa tenacissima. It is important to understand these processes in order to understand the impact of harvester ants on geosystems in semi-arid climate.

3

Content

Introduction ` 4

Background information 5

Details fieldwork area 5

Methods 6

2.1 Data collection 6

2.1.1 Mapping ant nests 6

2.1.2 Vegetation 6

2.1.3 Geomorphology 6

2.2 Data processing 7

2.2.1 T-test and Chi-square 7

2.2.2 CATPCA 7 2.2.3 Spatial distribution 7 Results 9 3.1 Maps 9 3.1.1 Lithology 9 3.1.2 Soil 10 3.1.3 Geomorphology 11 3.1.4 Slope 12 3.1.5 DEM 13 3.2 Vegetation 14 3.3 Statistical 16 3.3.1 Spatial distribution 16 3.3.2 K-mean clustering 16 3.3.3 CATPCA 17

3.3.4 T-test and Chi-square 18

Discussion 19 4.1 Maps 19 4.1.1 Lithology 19 4.1.2 Soil 19 4.1.3 Geomorphology 19 4.1.4 Slope 19 4.1.5 DEM 19 4.1.6 Vegetation 19 4.2 Statistical 20 4.2.1 Nearest Neighbour 20

4.2.2 T-test and Chi-square 20

4.2.3 CATPCA 20 4.3 General 20 5. Conclusion 22 6. References 23 7. Acknowledgements 25 8. Appendix 26

4

Introduction

In a semi-arid climate, soil properties play an important role in desertification, land degradation and surface erosion processes. Vegetation cover also has an important role in soil erosion. An increase in vegetation cover results in a decrease of soil erosion (Lavee et al., 1998; Quinton et al., 1997; Castillo et al., 1997). The soil properties are often influenced by organisms, such as earthworms, termites and ants. However, in the semiarid study area in South East Spain no earthworms and termites are present. The harvester ants Messor Bouvieri & Messor Brabus are the most dominant soil ecosystem engineers (Cammeraat et al., 2002). The harvester ants collect seeds individually or in groups. The nutrients in the seeds accumulate in the ant nest and therefore affect the soil properties. In addition, the water infiltration rate and porosity increase as well (Bruyn and Conacher, 1990). Harvester ants do not only alter soil properties, but also have an effect at hillslope and catchment scale, ecosystem functioning and hydrological processes (Cammeraat et. al., 2002).

It is important to understand these relations in order to understand the impact of harvester ants on geosystem processes. More knowledge about location and geosytem preference for harvester ant nests can be used to better understand land degradation, agricultural practices and to improve soil quality (Cammeraat et al., 2002)

The aim of this research is to gain more insight on the geosystem and vegetation preferences for harvester ants. This results in the following main research question: Is the location of ant nests influenced by the slope, altitude, soil, lithology, geomorphology or vegetation type? In order to answer the main research question the following sub research question are obtained. (1) What is the spatial distribution of the harvester ant nest locations?

(2): What is the relation between the variables, slope, altitude, soil,

lithology, geomorphology and vegetation type?

(3) What is the harvester ant’s foraging diet? For statistical analysis the following hypothesis are tested.

Hypothesis 1:

H0: there is no significant difference between the variables for the observed ant nests and random generated ant nests.

HA: there is a significant difference between the variables for the observed ant nests and random generated ant nests.

Hypothesis 2

H0: the ant nests are randomly distributed HA1: the ant nests are clustered

HA2: the ant nests are dispersed

Figure 1, Messor bourvieri (Bourvieri, 2017)

5

1.1

The harvesting ant messor family inhabit semi-arid and semi-arid climates and can be found in the western Mediterranean and North-Africa (Bernard, 1968). They harvest a selective diet of seeds. M. Bouvieri mainly forage seeds from dry scrubs and the M. Barbarus forage seeds from dry shrubs, but also sub humid and humid patches (Azcárate and Peco, 2002). The harvester ants have a preference for large seeds, as there is a significant relation between body size of ants and the size of the seeds. No relation between the size and the weight of the seeds exists, but the larger ants are able to crack the seeds open more easily (Willott and Compton, 2000). The harvester ants mainly collect seeds from Stipa marcochloa tenacissima and Anthyliss. The ants also sometimes collect plant residuals and rarely collect animal remains (arthropod corpses, snail or vertebrate faeces (Cerdá and Retana, 1994)).

Microclimatic conditions influence the surface activity of M. Barbarus. For instance, the threshold temperature for M. Barbarus activity is 9 °C. Humidity is also a controlling factor for ant activity. The ant activity peaks with a temperature between 20-25°C, with a moderate humidity (Azcarate et al., 2007). Therefore, in the winter there is less foraging and they rely more on their food reserves (Cerdá et al., (1998). Study of Cerd́a and Retana (1994) also reveal that temperature is a controlling factor for ant activity for M. Bouvieri. Other abiotic factors, such as: rainfall sunlight and wind, are also controlling factors for ant activity (Briese and Macauley, 1980). According to previous research the ant nest densities will be approximately between 0.1- 10% of the soil surface (Eldrige, 1993; Cammeraat et al. 2002; Jones and Wagner 2004; Lane and BassiriRad 2005).

According to Whitford (1978), Briese and Macauley (1980) Crist and MacMahon (1992) and Hobbs (1985), the M. Bouvieri show their preference by the density of foraging. They form a highway of ants and in the end they spread out to search for food (Cerdá et al., (1998). The direction of these foraging

columns is only temporary and might change during the day, depending on the seed availability. The length of their columns is usually between 4 to 5 meters and sometimes larger than 10 meter (Cerdá and Retana, 1994). The workers work individually and only carry one seed at the time.

1.2 Introduction - Details Fieldwork area

This research is focused on a specific area called ‘la Alquería’ in the Guadalentin basin, which is located east of Lorca in SE Spain (Figure 3). The research site has a size of 800 by 1500 meters and has patchy vegetated hill slopes. The area is interesting because it has been monitored for the past several years. In 2013 and 2016 a drone flight collected a sub-meter resolution DEM and multispectral images which provides very detailed information on (micro)topography of the study area. The climate is semi-arid with mild winters and warm/hot summers (syllabus, 2017) and an annual average rainfall of 270 mm per year (Cammeraat, 2004). Most of the rain falls in the spring and autumn which results in a water deficit during the summer. The average temperature is around 16-18 degrees Celsius with highest temperatures occurring during the months July and August and the lowest temperatures occuring in January (Navarra Hervas, 1991). The altitude varies between 580 and 680 meter. In the area, the main vegetation is dominated by Stipa marcochloa tenacissima grass and Pinus halepensis.

6

2. Methods

At first primary research was done, mainly literature research. Followed by data collecting with a fieldwork week in SE Spain. At last the collected data is processed using, statistical analysis in IBM, Geographic information system and Matlap. The Method diagram can be found in figure 4.

Figure 4, Methods diagram

2.1 Methods - Data Collection

For the data collection, the research site ‘la Alquería’ in the Guadalentín basin in SE is Spain was visited. The fieldwork lasted for 5 days and started on 23-04-2017.

2.1.1 Methods - Data Collection - Mappings ants nest

The research area can be seen in the transparent rectangle of figure 5. Marle de Jong and I searched for ant nests. Once an ant nest was located, the location was determined using the collector app or gps location on our phones. For these points location, the slope and altitude was also noted down. Afterwards each ant nest was labelled to prevent duplication and to give structure while exploring the fieldwork area.

2.1.2 Methods - Data Collection - Vegetation

For 15 ants nest locations the surrounding vegetation was determined. A rope of 10 meter was used to mark out an area around the ant nest in a 10-meter radius. In this 314m2

area all the vegetation species were determined using the syllabus of the course Desertification and Land Degradation (2016) or by taking pictures and samples and compare these with literature. Next to the species, the amount of each species was noted.

2.1.3 Methods - Data Collection - Geomorphology

During the week of fieldwork the geomorphological map was sketched using ortho photos and field observations. The map was further elaborated in the ArcGis Studio in Amsterdam. The geomorphological map was computed for the whole area in figure 1. The landscape units that are identified are: Pediment landscape (1) and river processes (2): 1: Residual mountains, face slope, dip slope, pediment intact, Eroded pediment, agricultural land and 2: gully/channels, fluvial deposits. In addition, in same area a slope and DEM map was made using the data of Jan Timmerman (2017) in ArcGis.

The Soil and Lithology were identified in the field by my fellow student Marle de Jong. Together we elaborated these maps in ArcMap in the ArcGis studio in Amsterdam.

For each map in ArcGis the following geographic coordinate system used was: GCS_WGS_1984

7

2.2 Methods - Data processing

A frequency table for further analysis was made for geomorphology, slope and DEM. In addition, a density table was computed for the Geomorphology.

2.2.1 Methods - Data processing -

Independent t-test and chi-square test

To answer the main research question the observed ant nests location need to be compared with points were no ants nest are located. In order to do so, in ArcGis a buffer zone of 10 meters was created around all the ant nest locations. Within the fieldwork area and outside the buffer zones, 34 random points were generated by ArcGis (figure 6). 34 random points were generated equalling the amount of ants nest we found during the fieldwork week. For all of these random points and for the real ants nests, the unit of slope, geomorphology, soil, lithology , altitude and was noted. With this set of data loaded in IBM SPSS 23 a t-test and chi-square test was used to determine if there is a significant difference between the observed ant nests and the random generated points. The variables are soil, lithology, geomorphology, slope and elevation of which geomorphology, soil and lithology are categorical. The response variable is binary, value zero means no nest and value one means that a nest is present.

2.2.2 Methods - Data processing - Categorical Principal component analysis (CATPCA)

In order to determine the interrelations among a set of variables with regard the underlying structure of these variables, principal component analysis was used in IBM SPSS 23. According to Kass and Tinsley (1979) the rule of thumb is having between 5 and 10 observations per variable. With the PCA only the observed ant nests locations were used, which were 34. Therefore, the variables for this analysis are slope, elevation, geomorphology, soil and lithology.

Before analysis the slope and elevation data were standardized and geomorphology, soil, lithology data were used as categorical data. The variables were standardized by dividing each element with the relevant standard deviation. They were standardized because the variables have a different measurement scale and now is it possible to compare the effect size (Field, 2009).

2.2.3 Methods - Data processing - Spatial distribution

To check what kind of spatial distribution there is from the observed ant nest locations the nearest neighbour tool in ArcGis was used. The nearest neighbour index is based on the average distance from each feature to its nearest neighbouring feature and therefore calculates the z-score and p-value (ArcGis, 2017). The average nearest neighbour tool in

8 ArcGis uses the Euclidean method and the output could be dispersed, clustered or random.

Further analysis with k-mean clustering was done, to determine if there are hidden patterns or grouping in the data set. The elbow method was used to determine the amount of k-clusters based on the distance to the centroid of a cluster (Matlab, 2017). The x and y coordinates of the observed ant nests point locations are loaded in Matlap. The elbow method was used to investigate the amount of clusters, which is 3.

9

3. Results

In this section the statistical analysis and the composed maps (scale: 1:3500m) of Geomorphology, slope, elevation, soil and lithology are presented. The lithology and soil maps are only shown to give an impression of the spatial distribution of the units to clarify the analysis. These maps are more thoroughly discussed in the Bachelor thesis of Marle de Jong. Maps obtained without random generated points can be found in appendix. Frequency tables for the maps can also be found in the appendix.

3.1 Results – Maps

The maps are plotted with the observed ant nests locations (Ants_Fieldwork) shown as

green dots. The red stars are the random generated points within the fieldwork area. For the figures with histograms the blue bar represents the number of observed ants nests (e.g. obs. Lithology) and the orange bar is the number of random generated points (e.g. RLithology).

3.1.1 Results – Maps - Lithology

The lithology map is presented in figure 7 with a corresponding legend. The most of the ant nests locations (70,5%) were found in the white Limestone area. Figure 8 shows a frequency distribution in a histogram of observed and random generated ant nests. It is clear that most ant nests observed and generated were found in the white limestone.

Legend

^ Random_generated_points Ants_Fieldwork

Lithology

<all other values> Gis_Code Marls white Marls Pink Marls yellow/white/green/pink Marls white/yellow Limestone white Limestone yellow Limestone eroded calcrete Calcrete 0 5 10 15 20 25 30

obs. Lithology Vs. RLithology

Lithology Rlithology [n]

(counts)

Figure 7, Lithology map (ArcGIS)

10

3.1.2 Results - Maps - Soil

Figure 9 represent the soil map of the research site. The most dominant soil type is calcaric leptosol and accounts 47% of the observed ant nests locations. The frequency distribution of the observed ant nests per soil unit is presented in figure 10. The y-axis is the number of counts and the x-axis is the soil unit.

Legend ^ Random_generated_points Ants_Fieldwork Soil Gis_code 0 Leptosol Calcaric Calcisol Leptic Calcisol Skeletic Calcisol Petric Calcaric Regosol Leptic Calcaric Regosol Calcaric 0 5 10 15 20 25 30

Calisol Petric Calcisol Skeletic Leptosol Calcaric Regosol Calcaric Regosol Leptic Calcaric

Calsisol Leptic

obs. Soil Vs. RSoil

Soil Rsoil [n]

(Counts)

Figure 8, Lithology histogram (Excel)

Figure 9, Soil Map (ArcGIS)

11

3.1.3 Results – Maps – Geomorphology

Figure 11 shows the spatial distribution of the observed and random generated ant nests for the geomorphology map. It is clear that in figure 11 the most observed ant nests are in the geomorphological unit Pediment intact (79,4%). Likewise the Pediment intact is the largest area in the research site (71%). Figure 12 shows the frequency for each geomorphological unit.

In table 1 gives an indication of the ant nest density in m2_{/ant nest. The density is only given} for the observed ant nest for each geomorphological unit. Table 1, Ant nest density

Table1, Density ant nests Geomorphology (Excel) Geomorphological unit Area (m2) Ants nests (n) Density (m2_/per ant nest) Pediment eroded 8620 3 2873 Pediment intact 43090 28 1539 Faceslope 3808 1 3808 River incision 381 0 - Gullies/channels 4431 2 2216 Fluvial deposits 1054 0 - Total 61384 34 1805 Legend ^ Random_generated_points Ants_Fieldwork Geomorphology Gis_Code Dipslope pediment Faceslope pediment <Null> River incision Residual mountain Faceslope mountain Dipslope mountain Pediment intact Pediment eroded Gullies / Channels Fluvial deposits Argicultural land Roads 0 5 10 15 20 25 30

Pediment eroded Faceslope mountain

Fluvial deposits Gullies/ Channels Pediment Intact Dipslope mountain

obs. Geomorphology Vs. RGeomorphology

Geomoprhology Rgeomorphology [n]

(counts)

Figure 12, Geomorphology histogram (Excel)

12

3.1.4 Results – Maps - Slope

The figure 13 of the slope is uncorrected for the vegetation. Therefore, unrealistic high gradient value can be found. The lighter spots in figure 13 show steeper slope values than the darker areas. The slope values that are represented are manually corrected for the Vegetation. For the observed ant nests the steepest slope found is 22 degrees and the lowest slope is 5 degrees.

3.1.5 Results – Maps – DEM

0 5 10 15 20 25 30 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

Slope Vs. RSlope

Figure 13, Slope (ArcGIS)

13

3.1.5 Results – Maps - DEM

The digital elevation model in figure 15 is uncorrected for vegetation. The brighter the area the higher the area is. The values in figure 16 are corrected for vegetation manually. The largest altitude difference between the observed ant nests is 30 meters.

Legend ^ Random_generated_points Ants_Fieldwork alqueria2016_dsm_high Value High : 708,323 Low : 561,261 580 590 600 610 620 630 640 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

Altitude Vs. RAltitude

Figure 16, Observed altitude vs. random generated altitude (Excel)

14

3.2.1 Results - Vegetation

Figure 17 gives an indication the vegetation that was encountered during the fieldwork week. The amount of the vegetation encountered within in a 10 meter radius is shown in figure 18 on the y-axis and the corresponding location is given on the x -axis. The colour of the bar represent a vegetation specie. Figure 19 gives an overview where the locations are in the fieldwork area.

The most dominant specie is Stipa tenuissima With a total count of 1454. Second, is Rosemarinus officinalis with a count of 915 and Helianthenum syriensis is the third most dominant specie with 688 observations.

0 20 40 60 80 100 120 140

Vegetation Around Ant Nests

stipa Pinus halepensis Helianthenum syriensis

Rosmarinus officinalis Cenista Spec. Teucrium pseudochamaepytis cistus clusi Collomia grandiflora bloem Scorzonera angustilifola Pistacia Lentiscus

Figure 17, Locations of vegetations samples (Timmerman, 2017

15 Figure 19, Locations of vegetations samples

16

3.3 Results – Statistical

The statistical results are showed in this section, additional information about the results can be found in appendix.

3.3.1 Results - Statistical – Spatial

Distribution – Average Nearest Neighbour

Figure 20 shows the outcome of the average nearest neighbour analysis in ArcGIS. The average nearest neighbour shows that spatial distribution of spatial distribution of ant nests is significantly dispersed. For additional results see appendix.

3.3.2 Results – statistical – spatial distribution – k-mean clustering

In figure 21 the result of spatial k-mean clustering are shown. There are 3 clusters (yellow, red and blue). Yellow has 4 ant nests, Red has 22 ant nests and blue has 8 ant nests.

Figure 21, Cluster analysis (Timmerman, 2017)

17

3.3.3 Results – Statistical - CATPCA

Figure 22 and 23 show the results of the principal component analysis for the observed ant nests points. The variables used are: soil, geomorphology and lithology (categorical) and elevation and slope (numerical). Figure 24 shows the total variance accounted for both dimensions. The variance for dimension 1 and 2 can be found in the appendix. The total

variance explained is 74%. _{Figure 24, Variance per dimension}

Figure 23, Biplot PCA observed ant nests (SPSS) Figure, 22 Variance Accounted for Total

18

3.3.4 Results – Statistical – T-test &chi-square test

Table 2 and 3 show the results of t-test figures 25, 26 and 27 show the results of crosstab and chi-square test. Elevation T-test Mean Std p-value Obs. Ant Nests 614,24 6,334 0,216 Rand. Ant Nests 616,21 6,682

Table 3, Results t-test elevation

Slope T-test Mean Std p-value Obs. Ant Nests 14.65 7.471 0,317 Rand. Ant Nests 13.06 5.337

Figure 27, Total variance of observed ant nests (SPSS) Figure 26, Chi-Square test obs. Ant nests * random generated ant nests

Figure 25, Chi-square test obs. Ant nests * random generated ant nests

19

4. Discussion

At first the results of the maps, vegetation and statistical analysis are discussed. Followed by the discussion of the main research question, sub research questions and hypothesis. Secondly, the discussion about what could approve the research and as last recommendations for further research.

4.1 Discussion – Maps

4.1.1 Discussion – Maps - Lithology

Figure 6 shows the distributions between observed and random points seems to be equal except for the Marls yellow/green/pink. There are no observations of ant nests in this lithological unit. It is remarkable since Marls white/yellow is quite a large area and that there are only 2 observations of ants and 8 random generated points.

4.1.2 Discussion – Maps - Soil

In figure 8, the distribution of soil is depicted. It is notable that both ant points random and initial have the highest preference for a Calcaric Leptosol, which is maybe due to the fact that this is the largest area of the fieldwork site (figure 7).

4.1.3 Discussion – Maps - Geomorphology

For the geomorphology map it is notable that most ant locations initial and random are located in the pediment intact landform. This seems sensible since this is the largest landform unit with 43090 m2_{. In addition, this} pediment is the most dense with 1539 m2 per ant nest. Other landscape forms have more surface per ant nest. Further, it is remarkable that only 1 observed ant nests is located in the river landscape form. Ants nests located river landscape form might wash away during a rainfall event. There is only 1 random point generated in this river landscape type. These river landscape form types are relative small in size compared to the land form types and that might explain this finding. This difference seems to be small but it might be significant, because it is very unlikely that ants nests are located in river landscape units because they will wash or drawn in these areas.

4.1.4 Discussion – Maps- Slope

This map is not corrected for vegetation. The random intense white dots are larger vegetation Pinus halepensis. The map has a pixel value of 2,3 by 2,3 cm and the slope is calculated over these vegetation. That causes unrealistically high slopes. Still on average the random generated points are a little bit higher than the observed points. In this altitude range there seems not be a preference for a certain altitude for ant nests.

4.1.5 Discussion – Maps- DEM

The average of the DEM is a little bit higher for the random generated points than the observed points ant nests. However, the difference is not significant. The DEM is also like the slope not corrected for vegetation. Again the larger white dots are Pinus halepensis , These dots are higher than the surrounding vegetation. None of these observed points or random points are located in these dots. Although, there might be an error of 0.4 – 0.6 meter for the dots who are located in the Stipa tussocks.

In order to correct the values of the slope and DEM there should be written an algorithm that automatically corrects the slope and DEM. The algorithm could be made from the results of Jeroen Kooijman bachelor thesis.

4.1.6 Discussion – Maps - Vegetation

Figure 18 shows clearly that the main vegetation type of the research area is Stipa. Only for research location 6 there are more Rosmarinus. It is notable that location 6 is by far the closest to an incision. The soil could be wetter and steeper here and that might influence the nearby vegetation. At location 6 there are a lot of Pinus halepensis compared to the other sites. This might also indicate that there is a higher water content. Further, it is remarkable that location 60 has very different surrounding vegetation than the other locations. Location 60 is located outside the fieldwork area of ant nests (figure 18). However, Stipa was the most common vegetation, but, there were also species

20 present that were not encountered in the other 15 locations. For instance Anthylis (35 times) and Lithodora fruticose (17 times). Location 60 is located on the edge of a residual slope and Anthylis is known as a species that grows on a northern slope or degrading sites or recolonizing sites. Harvester ants also forage on Anthylis. To answer the sub research question 3, harvester ants diet consist dominantly on seeds from stipa or anthylis, and in lesser amount plant residuals and rarely on animal faeces .

4.2 Statistical

4.2.1 Discussion – Statistical – Nearest Neighbour

The spatial distribution analysis average nearest neighbour, results that the observed ant locations are significantly dispersed with a p-value of 0.01 and a z-score of 2,6. Therefore the data is not clustered. This answers the sub research question 1 and rejects the null hypothesis of hypothesis 2. The spatial distribution of harvester ant nests is dispersed. However, there seemed to be a spatial cluster in the middle of the fieldwork area. Therefore, an additional k-means cluster analysis was made (figure 19). The k-means cluster analysis resulted in 3 spatial clusters. The ant locations seem to varied a lot along the x-axis and less around the y-axis. It is clear that the middle cluster has the most point locations, while the yellow cluster in figure 19 has only 4 point locations.

The left yellow area in figure 19 might be less studied but also the vegetation on this side was more sparse and also smaller than in the middle. The left site is partly on a eroded pediment see figure 5. In this area the calcrete layer is gone so the initial limestone bedrock is left. Limestone has a lower water content. The lower content might explain that vegetation is more sparse and that there is less abundance of seeds for harvester ants in this particular area. In the middle of figure 19 in red, shows the largest cluster in the fieldwork area, this might be due to the fact that the area has been studied most because, 3 other fellow students

were working in this particular area and they sometimes discovered an ant nest.

4.2.2 Discussion – Statistical - T-test &chi square test

To answer second sub research question, there is no relationship between the variables slope, altitude, soil, lithology or geomorphology. Therefore hypothesis 1 is not rejected. None of the observed variables is significantly different from the random generated points. This might be due the fact there are not many observations. It is remarkable that the p-value for the soil is very high. This indicates that there is not much difference between the soil units between observed and random generated points. For the other variables there is more difference between observed and random generated points but still far from significantly different

4.2.3 Discussion – Statistical – CATPCA

The total explained variance is 74,4% by the 5 variables used. Meaning there is a loss of 25.6% of information. It is notable that the total variance explained by each variable is over 0.8 and for the slope 0.94 which is good. Figure 21 shows that the geomorphology and elevation are positively correlated. The lithology and soil are uncorrelated to the geomorphology and elevation because, the angle is around 90 degrees. The soil and lithology and slope are not correlated.

4.3 Discussion – General discussion

We found a limited amount of ants during the fieldwork week 23-04-2017 till 28-04-2017. According to (azcárate, 2007) the optimal temperatures are between 20 -25 ºC degrees however, during the fieldwork week the temperature were lower as you can see figure 26.

21 Figure 26, Temperature Lorca April (Accuweather, 2017). This might declare why there was not much foraging in the beginning and morning of the fieldwork week. Foraging depends also on food availability (azcárate, 2007). More food is available later in the season and the temperatures are higher. Flowering of the plants is also related to soil moisture content. For instance, in 2015 the plants did not flower and therefore did not have seeds because of drought. A year later there was abundant flowering because of the rainfall in March (seeds need time to grow and ripe) (Erick Cammeraat, personal communication, 21-06-2017). Our results might therefore be different when comparing to a research which is done later in the season.

The method of measuring and determining plants within a 10 meter rope in radius around an ant nests might be simple and fast, but is not an exact measurement strategy. The rope might get stuck behind a Stipa or other shrub or you might hold the rope to loose or too tight

resulting in smaller of larger distance from the ant nests. To be more accurate a pole and measuring laser could be used.

At the beginning of the fieldwork week I assumed harvest ants forage within 10 meters of their nests. Although, I encountered a highway larger than 10 meters. Therefore it would be better to select next time a radius of 20 meters.

The field work area was relatively small and might not be representative for the whole habitat of Messor bourvieri and Messor barbarus in the Mediterranean. Not all soil, geomorphology, slope, altitude and lithology types that are present in the Mediterranean are found in the research site. The fieldwork area was located on a southern slope, where no anthylis plants grow while harvester ants can also forage on the seeds of anthylis. For further research it might be interesting to investigate a whole catchment area and indicate all the geosystem variables and also the vegetation in a radius of 20 meters for each ant nest. Also vegetation should be noted down at locations where there are no ant nests present so a statistical analysis can be performed. More ants nest and more different landscape units should give a more detailed description about the geosystem and vegetation preference for harvester ant nests.

22

5. Conclusion

Taking all the above into account we can conclude that the location of the harvester ant nests are not significantly influenced by the variables: slope, elevation, lithology, geomorphology and soil. Although, maybe later in the season when there are more harvester ant nests the results might be different than the currently observed results. Because, more observation of ants nests will improve the power for statistical tests and might therefore lead to more significant results. Meanwhile, the distribution of the harvester ants Messor bourvieri and Messor barbarus nests is significantly dispersed. Although, there seems to be a cluster in the middle of the field work area. The harvester ant have preference for large seeds from the Stipa marcochloa tenacissima and anthylis, but also forage on plant residuals and rarely on animal faeces .

23

6. References

Accuweather, (2017). Temperature April, Lorca, Spain. Retrieved on 24-06-2017 from:

https://www.accuweather.com/nl/es/lorca/308558/month/308558?monyr=4/01/2017

Agresti, A., (2007). An Introduction to Categorical Data Analysis. University of Florida

Azcárate and Peco (2002), Spatial patterns of seed predation by harvester ants (Messor Forel) in Mediterranean grassland and scrubland

.

ArcGis, (2017). Average Nearest Neighbour. Retrieved on 26-06-2017 from:

http://pro.arcgis.com/en/pro-app/tool-reference/spatial-statistics/average-nearest-neighbor.htm

Bernard, F., 1968. Les Fourmis d’Europe Occidentale et Septentrionale. Masson et Cie, Paris Barbarus, (2017). image of Messor Barbarus. Retrieved on 23-06-2017 from:

http://www.ants-kalytta.com

Bouvrieri, (2017). image of Messor bouvieri. Retrieved on 23 – 06 -2017 from:

https://encrypted-tbn0.gstatic.com

Briese,D.T.,andMacauley,B.J.(1980).Temporal structure of an ant community in semi-arid Australia. Aust. J. Ecol. 5: 121–134.

Cammeraat, E. L. (2004). Scale dependent thresholds in hydrological and erosion response of a semi arid catchment in southeast Spain. Agriculture, ecosystems & environment, 104(2), 317-332

Cammeraat, L. H., Willot, S.J., Compton, S.G., Incoll, L.D., (2002). The effects of ants’ nests on the physical, chemical and hydrological properties of a rangeland soil in semi-arid Spain. Physical Geography, University of Amsterdam.

Castillo, V. M., Martinez-Mena, M., Albaladejo, J., (1997). Runoff and soil loss response to vegetation removal in a semiarid environment. Soil Sci. Soc. Am. J. 61, 1116–1121.

Cerdá, X., and Retana, J. (1994). Food exploitation patterns of two sympatric seed-harvesting ants Messor bouvieri (Bond.) and Messor capitatus (Latr.) (Hym.: Formicidae) from Spain. J. Appl. Ent. 117: 268–277.

Cerdá, X., Retana, J., and Manzaneda, A. (1998). The role of competition by dominants and temperature in the foraging of subordinate species in Mediterranean ant community. Oecologia. 117:404-412. Crist, T. O., and MacMahon, J. A. (1992). Harvester ant foraging and shrubs-steppe seeds: Interactions of seed resources and seed use. Ecology 73: 1678–1779.

Eldridge D. J., (1993). Effect of ants on sandy soils in semiarid eastern Australia: local distribution of nest entrances and their effect on infiltration of water. Aust. J. Soil. Res. 31, 509–518.

Google Maps, (2017). Fieldwork location ‘Lorca, Spain’. Retrieved on 12-08-2017 from:

https://www.google.nl/maps/place/30800+Lorca,+Murcia,+Spanje/@39.0052699,-9.1430819,2217110m/data=!3m1!1e3!4m5!3m4!1s0xd64e6382d7cb5e1:0xc8867e729b04ca6a!8m2! 3d37.6735925!4d-1.6968357

24

Field, A., (2009). Discovering statistics using spss: (and sex drugs and rock ‘n’ roll). Third edition. London. ISBN 978-1-84787-906-6

Jones J. B., Wagner D, (2006). Microhabitat-specific controls on soil respiration and denitrification in the Majave desert: the role of harvester ants and vegetation. W. N. Am. Naturalist 66, 426–433.

Haagse, Pugnaire and Incoll (1994). Seed production and dispersal in the semi-aris tussock Stipa marcochloa tenacissima during masting. Journal of Arid Environment 31, 55-65

Hobbs, R. J. (1985). Harvester ant foraging and plant species distribution in annual grassland. Oecologia 67: 519–523.

Lambregts, C., (1996). The impact of Harvesting Ants on Soil Surface Properties of a Degraded Mediterranean Ecosystem. Master thesis, Department of Geography University of Amsterdam Lane Lavee, H., Imeson, A.C., Pariente, S., 1998. The impact of climate change on geomorphology and desertification along a mediterranean–arid transect. Land Degrad. Dev. 9, 407–422

BassiriRad H., (2005). Diminishing effects of ant mounds on soil heterogeneity cross a chronosequence of prairie restoration sites. Pedobiologia 49, 359–366. Lavee, H., Imeson, A.C.,

Pariente, S., (1998). The impact of climate change on geomorphology and desertification along a editerranean–arid transect. Land Degrad. Dev. 9, 407–422.

Bruyn, L.A., Conacher, A.J., (1990). The role of termites and ants in soil modification: a review. Aust. J. Soil Res. 28, 55–93.

Hervas, F., (1991). Factors determining the spatial variability of seed densities in shrub steppe ecosystem the role of harvesting ants. Journal of arid environments 32, 181192

Matlab, (2017) Cluster Analysis. Retrieved on 26-06-2017 from:

https://nl.mathworks.com/discovery/cluster-analysis.html

Quinton, J.N., Edwards, G.M., Morgan, R.P.C., (1997). The influence of vegetation species and plant properties on runoff and soil erosion: results from a rainfall simulation study in southeast Spain. Soil Use Manage. 13, 143–148.

Syllabus (2017) Fieldwork manual southeast spain, Course on desertification and land degradation. IIS, University of Amsterdam

Timmerman, J., (2017). Comparing soil losses in heterogeneous vegetated semi-arid hillslopes by remote sensing of topography and vegetation patterns. Master Thesis.

Whitford,W.G.(1978).Foraging in seed harvester ants Pogonomyrmexspp. Ecology59:185– 189. Willott, S., Compton, S., and Incoll, L., (2000). Foraging, Food Selection and Worker Size in the Seeding Harvesting Ant Messor bouvieri. Oecologia, Vol. 125, No.1 (2000), pp. 35-44

25

7. Acknowledgements

I would like to thank the following people for their support: At first, Erik Cammeraat., thank you for being my supervisor. I had a amazing time in Spain also thanks for the useful notes and insights. My fellow students Marle de Jong, Cynthia van Leeuwen, Jeroen Kooijman and Olaf de Haan for helping me finding ant nests. Especially, Jeroen and Olaf who also helped me with my grammar and spelling. Furthermore, Rutger van de Leur for the insights he gave me in statistics and as last my family for their support.

26

8. Appendix - Maps

27

28

29

30

31

32

Appendix Frequency table

Altitude RAlititude

33

Lithology RLithology

Geomorphology RGeomorphology

34

Appendix - T-Test

Elevation

35

36

38

Appendix – Matlab – K-mean Cluster analysis

%% Bart ter Mull 10530568 %% k-mean cluster analysis clear all close all clc x = [1.830034818 1.830087254 1.830422787 1.830242696 1.830608403 1.830311641 1.830227297 1.829705741 1.829764673 1.830575783 1.829992761 1.831773746 1.832221324 1.832294906 1.831671984 1.830981086 1.829853571 1.830030364 1.830103045 1.8298648 1.829681339 1.82965494 1.82967071 1.829307153 1.828819807 1.828394092 1.828045462 1.828291919 1.830007282 -1.829876602 -1.828115596 -1.827910126 -1.828158592 -1.830575823]'; y = [37.7860836 37.78604377 37.78605163 37.7860684 37.78621324 37.78644621 37.7863848 37.78624818 37.78608284 37.78621545 37.78596726 37.78602015 37.7858572 37.78533654 37.78549683 37.78551691 37.78571799 37.78556264 37.78537151 37.78546439 37.78533234 37.78546176 37.78569626 37.78614741 37.78599184 37.78596035 37.78566747 37.78565291 37.786297 37.78633654 37.78498811 37.78554005 37.78558797 37.78545687]'; hp = convhull(x,y) plot(x(hp),y(hp),'r-',x,y,'b+') %% k- means clustering for i = 1:1:10

[IDX, C, Sumd] = kmeans([x,y],i); ssd(i) = sum(Sumd);

end

plot(ssd,'bo-', 'MarkerFaceColor','g')

xlabel ('clusters'), ylabel('Total Sum square Distances') % elbow = 3 clusters = 3 [IDX, C] = kmeans([x,y],clusters) scatter(x,y,20,IDX, 'filled') xlabel 'x' ylabel 'y'

title 'k-mean cluster analysis'