Insectivorous bats in semi-arid agroecosystems − effects on foraging activity and implications for insect pest control

Insectivorous bats in semi-arid agroecosystems − effects on foraging activity and implications for insect pest control

Agriculture, Ecosystems and Environment 261 (2018) 80–92 Contents lists available at ScienceDirect Agriculture, Ecosystems and Environment journal h...

3MB Sizes 0 Downloads 225 Views

Agriculture, Ecosystems and Environment 261 (2018) 80–92

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage:

Research Paper

Insectivorous bats in semi-arid agroecosystems − effects on foraging activity and implications for insect pest control Idan Kahnonitch, Yael Lubin, Carmi Korine


Mitrani Department of Desert Ecology, Swiss Institute for Dryland Environmental and Energy Research, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev,Sede Boqer Campus, 8499000, Midreshet Ben-Gurion, Israel



Keywords: Bats Species richness Activity Agroecosystem Agrochemicals Landscape heterogeneity

Insectivorous bats populate habitats that adjoin or overlap with agricultural lands, and tend to use cultivated land for foraging and commuting. The goal of the study was to assess the principal factors influencing the activity and species richness of insectivorous bats in a semi-arid Mediterranean agroecosystem. We hypothesized that bat activity and species richness are influenced by the anthropogenic factors that are typical of agroecosystems, such as fragmentation of the landscape and loss of natural habitat, agrochemical use, presence of powerlines and roads, and proximity to urban areas. We recorded bats in a diversified semi-arid Mediterranean agroecosystem in 2012 and 2013 and estimated the effect of various anthropogenic and environmental factors on their activity. The proportion of natural and semi-natural habitats at the landscape and at the plot scale were the most important predictors of total bat activity, and of the activity the two most common species recorded, Pipistrellus kuhlii and Tadarida teniotis, both known to be synanthropic. Indeed, P. kuhlii had a positive association with the proximity to bodies of water and to settlements. Total bat activity was negatively associated with the use of agrochemicals. Thus, in line with our predictions, both the proportion of natural land cover in the environment and the use of agrochemicals play an important role in determining bat distribution in agricultural environments. Ecological inferences based on our results can be used to develop management schemes, such as restoring patches of natural vegetation near and within farmlands, to increase the suitability of agroecosystems as habitats for insectivorous bats. These could contribute both to the protection of endangered bat species and to bio-control of insect pests.

1. Introduction Insectivorous bats that use cultivated land for foraging may play an important role in regulating nocturnal insect populations in natural and agricultural ecosystems (Boyles et al., 2011; Williams-Guillén et al., 2016). Over the recent decades, as environment-friendly pest control became more prevalent, research has focused on methods to enhance the activity of arthropod pest control agents (Barbosa, 1998; Rusch et al., 2017). However, methods designed to enhance the activity of insectivorous bats for pest control purposes have scarcely been developed (but see Brown et al., 2015). Lack of ecological information regarding many species of bats is probably the main factor that limits practical research aimed to integrate insectivorous bats as conservation bio-control agents (Kunz et al., 2011). Specifically, information is needed on bat foraging behavior in agricultural lands, since many ecological processes in these habitats differ from those of natural habitats (Gliessman, 1998; Tscharntke et al., 2012). Several factors can influence insectivorous bat activity in ⁎

agricultural lands. Insectivorous bats are affected by agrochemicals directly, through poisoning (Bayat et al., 2014) and possibly indirectly through the negative effect that pesticides exert on insect populations (e.g. Pisa et al., 2015). A few studies showed that bats prefer organic farms over non-organic farms (e.g. Wickramasinghe et al., 2004), while others showed no difference in bat activity between organic and nonorganic farms (Pocock and Jennings, 2008). To our best knowledge, a quantitative analysis is lacking on the effect of agrochemical use on foraging patterns of bats in agroecosystems. Human settlements in proximity to agroecosystems can alter bat activity both positively and negatively (Williams-Guillén et al., 2016). Bat species are generally categorized into three groups, according to their response to urban environment (reviewed by Russo and Ancillotto, 2015): Avoiders (e.g. most Rhinolophids) are affected by the scarcity of natural roosts and food resources in urban environments (Russo et al., 2002) or by disturbances such as artificial illumination (Boldogh et al., 2007); adapters (e.g. Tadarida teniotis) are able to exploit some resources in the urban environment (e.g. roosting sites; Marques et al.,

Corresponding author. E-mail address: [email protected] (C. Korine). Received 24 April 2017; Received in revised form 23 October 2017; Accepted 1 November 2017 Available online 24 April 2018 0167-8809/ © 2017 Elsevier B.V. All rights reserved.

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Fig. 1. Distribution of sampling points of the research area in the southern Judea plains, Israel. Points were produced using ArcGIS 10.2 software (ESRI).

2004); and exploiters (e.g. Pipistrellus kuhlii) are able to exploit novel resources in urban environments, for example foraging at street lamps (Rydell and Racey, 1995), outcompeting other bats. Other factors, such

as roads and powerlines may affect bat activity. Roads and their associated traffic have various distance-dependent negative effects on bats such as a barrier to movement (Zurcher et al., 2010) and increased 81

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

in 2012 and 63 polygons in 2013 (total of 78 polygons in both years combined; Fig. 1). Due to seasonal changes in crop types, only 43 polygons out of the 2012 sample were included in the 2013 sample. The samples included six vegetation types (Table A in Supplementary file): (1) deciduous orchards (almond, peach and apple), (2) citrus orchards (lemon and orange), (3) olive groves, (4) grapevines, (5) natural woodland and shrubland and (6) vegetables crops (cabbage, cauliflower, kohlrabi, watermelon, beets and fennel). In the data analysis, we divided the vegetables crops into three separate categories: (1) watermelon, (2) vegetables of Brassica oleracea and (3) beets and fennel.

mortality (Russell et al., 2009). Radiation of low frequency electromagnetic fields (ELE-EMFs) from electric powerlines may cause aversive responses in foraging bats, as was found in some other mammals (Burda et al., 2009). Habitat heterogeneity is generally lower in agroecosystems than in natural ecosystems (Gliessman, 1998) and occurs at a different scale and pattern (Cabell and Oelofse, 2012). While in natural habitats, the variety of land cover types (compositional heterogeneity) and the grain size of each land cover type (configurational heterogeneity) are rather fixed to topography and water availability, in agroecosystems, these two components of landscape heterogeneity are variably shaped by changing agricultural practices (Fahrig et al., 2011). Accumulating research shows that farmland heterogeneity is a key predictor of biodiversity at multiple scales (Benton et al., 2003, Turner et al., 2001) and through various mechanisms (Tscharntke et al., 2012). Studies of bats in agroecosystems found positive relationships between bat foraging activity and species richness and some features of spatial heterogeneity in farmland, such as the number of vegetation strata (Estrada et al., 2006) and crop boundaries with natural vegetation (Kelly et al., 2016). Natural or semi-natural land cover within agricultural landscapes has a positive effect on species diversity and abundance at multiple spatial scales (Duelli and Obrist, 2003), as seen in studies of various taxonomic groups (e.g. Pluess et al., 2010; Gabriel et al., 2006). Nevertheless, for some mammals, including bats, the nature of this relationship seems species specific (Silva et al., 2005; Numa et al., 2005). Previous studies of bats in agroecosystems were conducted in mesic regions. Agriculture is progressively advancing into semi-arid regions, where environmental conditions and anthropogenic effects on potential natural enemies of crop pests may be different. For example, water may be a critical resource, while urbanization may be less extensive than in mesic regions. Our aim was to determine how and to what extent a range of natural and anthropogenic factors affect bat activity in a semiarid agroecosystem. We hypothesized that bat activity patterns and community structure in agroecosystems are related to environmental factors which originate in human activity typical of agroecosystems. We predicted that foraging activity and species richness of bats will increase with increasing landscape heterogeneity, proximity to high quality bodies of water, and the proportion of natural land cover, and will be negatively correlated with higher levels of agrochemical inputs and other anthropogenic disturbances. We also predicted that anthropophilic species typical of more mesic regions will be found in association with urban areas in the semi-arid agroecosystem.

Bat foraging activity was sampled using the AnaBat (Titley Electronics, Ballina, Australia). Each detector was placed at the sampling point, on the ground, in a 45° angle facing east, and recorded bat calls throughout the night, from sunset till sunrise. Since we used 2–5 detectors on each night, we could not survey all types of habitat simultaneously. We varied the sampling sites on a given night such that multiple habitat types were sampled each night. Furthermore, to prevent sampling bias in the data analysis of bat activity, we took into account temporal variation in environmental conditions (minimum night temperature and wind velocity; see Environmental factors). Each sampling point was surveyed one night in a year. Samples were not taken during the period of eight days before and after the full moon. The sampling point in each polygon was located 28.5 m westward of the polygon centroid in order to correct for the eastward-biased detecting range (considering the positioning angle of 45°). This correction was done on the basis of the approximate maximum detection range of 60 m of the AnaBat detectors (Parsons, 1996). When we analyzed bat activity we accounted for the proximity of the sampling point to the polygon's border (d (j, i) in the heterogeneity index; see next). This was done since the area of the polygons was variable and bat activity measured in the sampling locations could be influenced by edge effects. We monitored foraging activity by identifying and counting bat passes (Fenton, 1970). Prior to recording sessions, all detectors were calibrated to a sensitivity level of 6 in a new SD2 AnaBat detector, using a constant and monotonic ultrasonic signal. Calls were analyzed manually using the software AnalookW version 3.8 v. Identification of bat calls and species was done on the basis of known species-specific acoustic characters (Benda et al., 2008; Russo and Jones, 2002).

2. Materials and methods

2.3. Anthropogenic factors

2.1. Study area and sampling points

2.3.1. Agrochemical inputs We used three measures to estimate the effects of agrochemical inputs: one that was based on the actual detailed pesticide application records from the farmers and plant-protection inspectors, and two additional estimates based on published calculations of cost and returns for pesticides certified by the Agricultural Extension Service of the Israeli Ministry of Agriculture (MoA). Actual application records were available for only 27 and 35 plots during 2012 and 2013, respectively. The analyses based on MoA protocols were conducted on all plots. Estimate based on actual pesticide applications: 1. Averaged agrochemical load for the growing season – Based on the hypothesis that intensive spraying regime has long term effects (Freemark, 1995), we computed an index that estimates the average agrochemical load in each sampling location, limited to the relevant growing season, as follows:

2.2. Monitoring bat activity

The study area is a diversified ecosystem comprising natural, seminatural and agricultural patches, stretching over 140 km2 in the southern Judean plains of Israel in a transition zone between the Mediterranean and the semi-arid climatic zones (Ben-Arieh, 2001). Agricultural land (orchards and crop fields) comprise 50.7% of the study area, natural and semi-natural land 34.2%, settled area (residential, industrial and roads) 14.7% and bodies of water 0.4%. We collected data on bat activity in summer of 2012 and of 2013, from June to September (except one record in May 2013). Initially, during spring-summer of 2012, we conducted a field survey of the study area to record the landscape inventory, using a GPS device (GPSMAP 76, Garmin International, Kansas, USA) and orthophoto images of the area. The data collected in the survey were transferred into a computerized GIS system using ArcGIS 10.2 software (ESRI, California, USA). The study area was divided into digitized GIS polygons representing distinct vegetation patches, out of which we selected 78 polygons ranging from 2243 m2 to 493,459 m2 (mean 52,005 m2), with mean nearest-neighbor distance of 685 m (SD ± 362 m), measured from the sampling point in each polygon. We used 58 polygons for bat sampling


AVRSL i = 1/Ts i · ∑ Tar·m j j·PHIj j= 1


AVRSLi − Averaged spraying load for sampling point i Tsi − Time period of the growing season for the crop of sampling 82

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

point i (in months). mj − The amount of the substance applied in spraying treatment j (g/dunam). Tarj − A categorical factor for the type of spraying treatment j (five categories representing insect pests, pathogens, weeds, plant growth regulators or other interventions). PHI(j) − The time interval between the last spray treatment and the safe harvesting time for immediate consumption (in days) as published by the Agricultural Extension Service of the Israeli Ministry of Agriculture (MoA, Indices for the entire sample based on MoA protocols: 2. Ranked averaged agrochemical load for the growing season – The values of this index were computed by the same formula of AVRSL index (Eq. (1)) but using data obtained from MoA protocols instead of the detailed records. The product values were then clustered into five categories and scaled by order with the Jenks natural breaks classification method (Jenks and Caspall, 1971). 3. Spatially-proportioned averaged agrochemical load for the growing season – Since the use of agrochemicals impact a larger area than the specific location of application (Gabriel et al., 2010; FuentesMontemayor et al., 2011), we computed this index to account for the spatially weighted effect of agrochemical application in the area around the sampling point. The values of this index were computed in the same way as the ranked AVRSL index and were assigned to all of the polygons in the research area. To generate the weighed product we used the same method we used to compute the index of proportion of natural land cover, and multiplied each polygon area by the value of its ranked AVRSL index.

2.4. Anthropogenically-induced environmental factors 2.4.1. Effective distance to bodies of water Because bats drink on the fly, the size and shape of the surface area of the pool of water are important factors in the ability of the flying bat to drink (Razgour et al., 2010). We computed the following index to account for the combined effect of the proximity of the sampling point to the nearest bodies of water, the quality of water and its accessibility to bats:




∑ j= 1

qw / di,w

w= 0


2.4.2. Landscape heterogeneity Our landscape heterogeneity index evaluates the complexity of arrangement of the different vegetation patches in space with respect to their proximity to the sampling point. The index formula was based on landscape indices and concepts that appear in the program “FRAGSTATS” (McGarigal and Marks, 1995; Schindler et al., 2013) and was constructed following Dufour et al. (2006) definition of heterogeneity. The heterogeneity value for each sampling point is the distanceweighted sum of the heterogeneity values of all the polygons within a range of 600 m from the sampling point. This buffer size corresponds to the average nearest-neighbor distance (685 ± 362 m) between sampling points and is adequate to detect significant spatial variation. In addition, 600 m is an appropriate buffer size considering the spatial pattern of foraging activity of bats and their prey, conforming to the average area of foraging for local species and closely-related bat species (e.g. Nicholls and Racey 2006; Marques et al., 2004), the distance to roosts (Boughey et al., 2011), and the movement distances of a number of local agricultural insects (e.g. Sciarretta et al., 2008; Basoalto et al., 2010; Mata et al., 2016). This method ensures that as the number of polygons in the range increases, indicating higher spatial diversity, the sum of them is likely to be higher as well. The heterogeneity of a sampling point is:

NTL − mean Nightly Traffic Load, width − the number of lanes in the road. j − Road crossing the research area (N]6), max − the maximum value among the roads jn. The road disturbance index was calculated as follows:



αw,i − The azimuth of the water source w to the sampling point i. q − Quality factor (1–5), i − sampling point (N]78), w − body of water (N]10) di,w − Distance from sampling point i to all bodies of water (w) within a range of 6000 m from the sampling point i. We ranked the quality of each body of water in an ordinal scale, i.e. quality factor (qw), according to two parameters: (1) the quality of the water in terms of pollution and content of organic matter and (2) the estimated accessibility of the water surface to flying bats. The sum of the quality-distance ratios for the bodies of water in range with respect to the referenced sampling point was weighted by the ratio of the azimuth variation and the quality-distance ratios variation (see Eq. (4)). Azimuth variation was calculated using circular statistics (Jammalamadaka and Sengupta, 2001) and was integrated in the index since greater azimuth variation indicates wider distribution of available water sources.

2.3.2. Disturbance of roads We assume that the negative effect of roads on bats is a function of traffic volume and the number of lanes in the road (“road class”, as defined by Bennett et al., 2013). Our road index is based on three parameters: distance to the road, the road's width and its nightly traffic load ( The width and the nightly traffic load of each road were combined to create a disturbance factor (DF) for each road, as follows:

DFj = NTL j/NTL max ·widthj/widthmax

1 + var(α w,i) · 1 + var(qw / di,w)

DFJ/d(j, i) (3)

n (j , i) − The number of roads within a radius of 2000 m from the sampling point i. d(j, i) − The distance in meters from the sampling point i to the road j



∑ j= 1

2.3.3. Radiation of electromagnetic fields (EMFs) Four high-voltage power transmission lines known to produce EMF cross the research area. We computed the distance between each sampling point and the nearest power line within a range of 85 m, since within a distance of approximately 85 m from high-voltage power transmission lines (161 KV) the magnetic field flux diminishes to approximately 0.8 mG (Kaune, 1993), a level found to cause disorientation in birds (Ritz et al., 2004) and an avoidance reaction in bats, though at a higher frequency range (Nicholls and Racey, 2009). As the area beyond a range of 100 m from a power line of 161Kv exerts a negligible flux of EMF (Kaune, 1993), we assigned all sampling points outside of the 85 m range with the value of 100m.

perimeterj areaj

⋅SFj⋅Stj/d(j, i) (5)

LHETXi − Heterogeneity level of a sampling point i with units of meter−2, j − Near polygon in the range of 600 m from i, Stj = Number of strata levels in j, SFj − Averaged similarity factor of j with respect to its intersecting polygons, n (j , i) − Number of polygons j within a range of 600 m from i, perimeterj/areaj– Geometrics of j measured in meters, d(j, i) − Distance between j polygon and the sampling point i. The averaged similarity factor (SFj) expresses the degree of similarity in terms of land cover type between the focal polygon (j) and the 83

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

variable can take various distributions, including “zero inflated models” for a distribution with excess zero counts (Stasinopoulos and Rigby, 2007). Four dependent variables were included in the data analysis: (1) Total bat activity (passes per night), (2) P. kuhlii activity (passes per night), (3) T. teniotis activity (passes per night), (4) and species richness (number of species per night). The data of each year (2012 and 2013) were analyzed separately in a similar procedure, with minor changes in the set of models due to exclusion of different associated pairs of variables. To select the most probable models we used the AIC index. Since we constructed the models based on sample size of up to 63 cases (n) and with up to 10 parameters (k), we used the corrected AICc (Burnham and Anderson, 2002) which incorporates a second-order bias correction. All models with an AICc score ≤2 in comparison with the best model were considered as probable and “equivalent” to the best model. For each dependent variable, we constructed a set of constraint models with the possible combinations of explanatory variables excluding combinations that exhibited multicollinearity. Fitting the constraint models was done using the RS algorithm (Rigby and Stasinopoulos, 1996) with maximum 100 iterations. In order to determine whether the relationship to bodies of water is linear or quadratic we ran both quadratic and linear models and used the set whose best model yielded the lowest AICc score. Fitted models whose algorithms failed to converged and models whose estimated overdispersion was much higher than the unconstrained model (Δ > 50) were excluded from the analysis. We evaluated the goodness of fit for the leading models (AICc ≤2), computing: (1) the generalized R-squared value for each model and (2) the ΔAICc value of the intercept-only model and the best model. To evaluate the importance of each parameter in explaining the modeled variance we computed (3) the evidence ratio for nested models and (4) the maximum likelihood ratio test for nested models. To evaluate the power of effect of each parameter we used (5) Wald z-test for models where the overdispersion parameter was not modeled (i.e. Poisson and YULE distribution) and (6) the t-test for all other models. In addition, to evaluate the partial contribution of each of the model's parameters in explaining the variance in the recorded bat activity, we calculated (7) the partial correlation for each parameter, taking into account all other parameters in the model in which the focal parameter is nested (Johnson and Wichern, 2002). In cases where categorical nominal variables were included in the analysis (e.g. vegetation type) or when the raw count data were spatially autocorrelated, we used the partial Mantel test (Smouse et al., 1986). The relationship between bat activity and agrochemical use or the proportion of natural vegetation were also presented as follow; we calculated the randomized quantile residuals (Dunn and Smyth, 1996) of (1) the best model of those explanatory variables and (2) its nested model (explanatory variable excluded). We then subtracted the residuals of the best model from the residuals of its nested model and plotted them against the values of the explanatory variable. Pearson correlation coefficient was used to assess the strength of relationship between the two data sets. For the dependent variables, bat species richness and activity, we constructed different distribution models (Poisson, negative binomial, etc.). We ranked the models using log likelihood, AICc values and graphical plots for visualization and compared zero-inflated models over their non-zero-inflated analogs (e.g. zero-inflated Poisson versus ordinary Poisson). When more than one distribution model was chosen by the above measures, we constructed the constraint models with the explanatory variables (e.g. heterogeneity level, spraying load, etc.) for all the nominated distribution models and chose the distribution whose best model yielded the lowest AICc score and the highest R-squared value. To test for spatial autocorrelation in the dependent variables we conducted a Moran’s I test using the function Spatial Autocorrelation (Global Moran's I) in ArcGIS (ESRI). If spatial autocorrelation was found in the raw data, we further tested for spatial autocorrelation in the residuals of the selected models. Spatial autocorrelation in the residuals

polygons surrounding and intersecting it. In order to assess the degree of similarity between the different land-cover types, we constructed two dissimilarity matrices: one accounting for the general similarity and one for the taxonomic similarity. 2.4.3. Proportion of natural and semi-natural land cover To evaluate the proportion of natural and semi-natural land in the area surrounding each sampling point we followed Pluess et al. (2010) with two modifications: (1) we used a gradient of four distance levels: 100 m, 300 m, 600 m and 1000 m. On that basis, we created four annuli around the focal sampling point whose areas were formed by the intervals between each ring. (2) We used the inverse mean distance from each annulus to the sampling point to weight the partial contribution of each radius category to a final weighted average of the proportion of natural and semi-natural land cover. 2.5. Environmental factors We measured ambient temperature during each sampling night in each sampling location using a calibrated iButton data logger (Dallas Semiconductor, Maxim Integrated Circuits, USA). The sensors were placed horizontally, 2 m above ground and programmed to record temperature every 30 min. Wind measurements were taken from six meteorological stations in and around the research area. In order to increase the reliability and accuracy of the input, for each sampling point we took the mean value of two measurements from different meteorological stations closest to the sampling point. Instead of using the average wind velocity per night we used an index combining the mean wind speed and the inflated proportion of high speed wind gusts. The inflation was created as wind gusts that exceeded 2.5 m/s were duplicated in the calculation, as follows:

PSW = (N(Wind speed > 1.5) + N(Wind speed > 2.5) )/N


PSW- Inflated proportion of strong wind gusts, N- Number of sampling intervals per night

WATXi = Vwi·PSWi


Vw i−combined mean wind speed measurement for sampling point i. Wind regime is influenced by topography, where highly exposed sites like topographical peaks and ridges are more prone to high wind velocities than more sheltered valleys. In order to determine the extent to which each sampling point is topographically exposed and prone to strong wind we combined two measures into one index: (1) the mean slope of the circular area with a radius of 10 m surrounding the sampling point, and (2) the elevation differences between the sampling point and the lowest point in the 60 m radius area surrounding it. The exposure index was calculated as follows: TEXPOXi = slopei ·(Alt i − Alt min)


Alti − The altitude in sampling point i. Altmin − The minimum altitude in a range of 60 m from the point i. 2.6. Data analysis We tested for pair-wise associations between the explanatory variables, since strong multicollinearity among those variables may bias regression analyses and may indicate a spurious relationship (Graham, 2003). To estimate the severity of multicollinearity we used the variance inflation factor (VIF) with the conservative threshold of 5 (Studenmund, 2001). Models that included a pair of variables that exceeded a VIF of 5 were excluded from the analysis. To test the relationship between the explanatory variables and the dependent variables we used Generalized Additive Models for Location Scale and Shape (GAMLSS: Stasinopoulos and Rigby, 2007). GAMLSS fits regression type models where the distribution of the response 84

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

plot scale (AVRSL), but only in 2013 (Table 2B and Fig. 4). The variable of landscape heterogeneity (LHETX) was included in the leading models of 2012 and 2013. However, a significant positive effect was found only in 2013 (Table 2B). Distance from settlements (SET) was also included in the leading models of both years, but had a significant negative effect only in 2012 (Table 2A), i.e., bat activity was higher near the settlements. Effective proximity to bodies of water (WATX.Q) had a significant positive effect on total bat activity only in 2012 (Table 2A) with a quadratic link of the distance parameter prevailing over the linear link. Specific environmental factors influenced the two dominant bat species, often in different ways. For P. kuhlii, the raw data of 2013 showed spatial autocorrelation but not in the residuals of the leading models. P. kuhlii accounted for the majority of the bat passes and thus, results of the analysis of P. kuhlii are for the most part similar to those of total bat activity, with a few differences: (1) P. kuhlii activity was positively associated with Brassica crops (C.bras; P = 0.033, Fig. 2b), but, unlike total bat activity, it was not associated with fennel and beets (C.beet). (2) In 2013, P. kuhlii activity was positively associated with natural land cover only at the plot scale (V.nat; P = 0.018, Fig. 2b), but not at the landscape scale (SNATX). (3) Unlike total bat activity, P. kuhlii activity was not affected by the use of agrochemicals in the landscape scale (SPAGKSL), nor in the plot scale (AVRSL). We found spatial autocorrelation in the raw data of T. teniotis in 2012 and in the residuals of the leading models. Therefore, we included only GLS-fitted models in the analysis below. T. teniotis activity was mostly influenced by natural and semi-natural land cover. In 2012, vegetation type was the most important predictor for T. teniotis activity (VEGT; P < 0.0001). All the vegetation type variables had a significant negative association with T. teniotis except the natural vegetation variable (V.nat), which had no significant effect. When we ran the model of T. teniotis activity with the vegetation type variable alone, V.nat had a significant positive effect on T. teniotis activity with an estimate of 15.30 and P < 0.0001, while all the agricultural crop types had significant negative effects. Kruskal–Wallis test, however, showed only a marginal effect of vegetation type on T. teniotis activity with P values of 0.060 and 0.067 in 2012 and 2013, respectively. The difference between these results and the strong significant effect in the GLS results is likely related to the spatial autocorrelation found in the raw data of T. teniotis activity in 2012, as the Kruskal–Wallis test did not incorporate the spatial context of the data and the GLS modeling did. This explanation is supported by the partial Mantel test results, showing an increase of 20.2% in the correlation between T. teniotis activity and vegetation type when the spatial context was included in the test. The positive effect of semi-natural land cover type on T. teniotis appeared again in 2013 at the landscape scale (SNATX; P < 0.0001), which was by far the most important predictor of T. teniotis activity. Distance to power lines (PWRL) was included in the leading models of T. teniotis in 2012 and 2013, with a significant positive effect in 2012 (P < 0.0001), thus T. teniotis’s activity increased as the distance from powerlines increased.

Table 1 Explanatory variables used to fit the models for the response variables. Variable name



Averaged spraying load Heterogeneity Min. temperature Natural


Power lines Ranked spraying load Roads Semi-natural


Settlements Spray load (spatial average) Topographic Exposure Water Water (Q)


Wind velocity Vegetation type Crop citrus Crop deciduous Crop olives Crop grapevines Natural vegetation

WNDX VEGT: C.cit C.dec C.oli V.nat

Monthly averaged reported spraying load Landscape heterogeneity Index Nightly minimum temperature Proportion of natural land cover (gradientbased) Effective distance to near powerline Averaged general ranked spraying load Road disturbance index Proportion of semi-natural land cover (Gradient-based) Distance to the nearest settlement Spatially proportioned averaged general ranked spraying load (gradient-based) Topographic exposure index Effective proximity to water Effective proximity to water with quadratic decay of distance Wind velocity and strong wind gusts index

Crop veg: Crop brassica

C.veg: C.bras

Crop watermelon Crop beet and fennel

C.wat C.beet


Citrus orchards Deciduous orchards Olives groves Grape vineyards Natural land cover (maquis and garigue form) Vegetables: Cabbage, cauliflower, kohlrabi (Brassica oleracea cultivars) Watermelon Winter vegetables (fennel/beets)

of a model indicates a spatial pattern that the model does not sufficiently explain (Dormann et al., 2007). To avoid biased conclusions, such models were excluded from the analysis. We conducted a partial Mantel test (Smouse et al., 1986) to account for the impact of the spatial pattern on coefficients of explanatory variables in the models that were not excluded. For the reported spray load (AVRSL), the sample size was not large enough to detect a spatial pattern and for analysis in which all models were excluded due to spatial autocorrelation, we used generalized least squares (GLS), which fits the models with correction for the spatial pattern (Dormann et al., 2007). We used pair-wise comparisons on all the vegetation types in the surveyed plots (Table 1). We used a Kruskal-Wallis test for analysis of variance and the Brunner-Munzel test for pair-wise comparisons for data that were not normally distributed (Shapiro-Wilk test). Both of those methods are robust to the presence of unequal variances in the data (Fagerland and Sandvik, 2009). 3. Results We identified 10 species of insectivorous bats: Pipistrellus kuhlii, P. pipistrellus, Tadarida teniotis, Rhinolophus hipposideros, R. ferrumequinum, Otonycteris hemprichii, Taphozous nudiventris, Myotis nattereri, Eptesicus serotinus and Plecotus christii. In the two sessions of 2012 and 2013 combined, we identified 11,204 bat passes during 50 nights of recording (an average of 224.1 passes/night), of which 80.9% belonged to P. kuhlii, 14.1% belonged to T. teniotis and 1.8% belonged to the other remaining eight species (3.2% of the calls could not be identified). Total bat activity was positively related to natural and semi-natural land cover in the landscape scale (NATX) in 2012 (Table 2A, Fig. 3A and supplement Table B), and (SNATX) in 2013 (Table 2B and Fig. 3B), and also at the plot scale (V.nat) in 2013 (Table 2B and Fig. 2A ). There were significant differences in bat activity by vegetation type in 2013 (Kruskal–Wallis, P = 0.012, Fig. 2.A), differences that were largely due to P. kuhlii activity (Fig. 2.B). Total bat activity was higher at sites with less agrochemical use both at the landscape scale (SPAGKSL) and at the

4. Discussion In the present study, we identified factors affecting diversity and patterns of activity of bats in agricultural land located between the Mediterranean and the semi-arid climatic zones. Overall, bat activity was primarily associated with patches of natural land cover and was affected negatively by agrochemical application. Hence, in line with our predictions, both the proportion of natural land cover in the environment and the use of agrochemicals play an important role in determining bat distribution in agricultural environments. Since a large portion (80.9%) of bat activity as recorded in both years was of P. kuhlii, inferences from the results regarding general patterns of bat activity in agroecosystems with different species composition should be made with caution. 85

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Table 2 Environmental factors affecting total bat activity for all sites in 2012 (A) and 2013 (B). Shown are variables that were included in at least one of the leading models [see Table B in Supplementary file]. Averaged reported spraying load (AVRSL), separated with bold line, was tested with a subset of n = 35 sites in 2013 [see Table 5. A in Supplementary file]. SE − Standard error of the estimate, P − Significance level of estimates (model coefficients) for the relevant test, DF − Residual degrees of freedom for the relevant test, E. Ratio − Evidence ratio for variable (j), the Wi ratio of the model with j to the nested model without j, P. Mantel test and P. correlation − the partial association of the variable (j) to bat activity (see text). Δ2 mod is the% of models containing variable (j) among the leading models with ΔAICc < 2. A. 2012 Variable


T test

LR test

E. Ratio







1.75 −5.E-04 0.08 0.01 1.71 0.01 −0.01 −0.16 −0.01

0.359 1.E-04 0.024 0.005 0.828 0.006 0.004 0.130 0.006

1.E-05 4.E-05 0.001 0.030 0.045 0.089 0.119 0.230 0.337

48 49 51 48 48 49 48 48 48

9.E-06 7.E-04 8.E-04 0.022 0.025 0.077 0.114 0.223 0.337

1 1 1 1 1 1 1 1 1

B. 2013 Variable

T test

LR test





VEGT V.nat (In-pt) C.wat C.cit C.oli C.beet C.bras C.dec LHETX TEXPOX SNATX WNDX SPAGKSL SET MTEMP PWRL

4.82 −1.39 −1.34 −0.86 −0.93 −1.15 0.40 −0.29 0.03 −0.01 1.46 −1.60 −0.26 −1.E-04 0.04 0.01

0.251 0.306 0.444 0.321 0.351 0.571 0.396 0.310 0.008 0.004 0.500 0.565 0.130 9.E-05 0.037 0.008

2E-25 3E-05 0.004 0.009 0.011 0.048 0.319 0.358 9.E-04 0.003 0.005 0.007 0.050 0.110 0.236 0.276

52 52 52 52 52 52 52 52 52 58 58 52 58 51 51 51






4372.4 81.4 78.8 3.3 3.0 1.2 0.8 0.4 0.4




0.58 −0.42 0.26 0.19 0.23 0.12 −0.15 0.14 −0.25

7.E-07 0.001 0.050 0.185 0.101 0.399 0.289 0.324 0.073

100 100 100 67 73 53 27 20 7

E. Ratio





8.E-04 0.002 0.005 0.006 0.046 0.273 0.227 0.300


Δ2 mod

P. correlation

Δ2 mod

P. Mantel test r







1 1 1 1 1 1 1 1

66.5 42.8 14.6 9.98 2.23 0.39 0.45 0.37

0.13 −0.19 0.24 −0.10 −0.03 −0.05 −0.02 −0.06

0.107 0.005 0.006 0.053 0.401 0.312 0.541 0.324

80 20 20 80 20 20 20 20



P. correlation −0.31 0.085


areas, which in turn may lead to higher activity of bats in agricultural patches that are close to non-crop patches. Proximity to settlements had a strong positive effect on P. kuhlii activity in 2012. This seems contradictory to the positive effect of natural habitats on its activity, as discussed above, since anthropogenic disturbance is expected to be greater with decreasing distance to settlements. Other studies, however, also showed that when comparing several types of habitats, activity of vespertilionid bats, e.g. P. kuhlii, is significantly higher in both natural habitats and human settlements (Walsh and Harris, 1996; Rainho, 2007). As suggested by Walsh and Harris (1996), this exceptional affinity of vespertilionid bats to settled areas may reflect an adaptation to utilize certain elements of these areas. P. kuhlii is known to roost in buildings (Yom-tov and Kadmon, 1998) and forages within human settlements (Russo and Jones, 2003; Rainho, 2007), and was positively associated with anthropogenic activity (Lisón and Calvo, 2013). T. teniotis is also known to roost in manmade structures (Yom-tov and Kadmon, 1998), but there was no effect of the proximity to settlements on its activity in either 2012 or 2013. Vegetation type had a significant effect on total bat activity in 2013. Since P. kuhlii in 2013 accounted for 77% of bat passes and was the only species whose activity was significantly correlated with vegetation type, selective foraging behavior of this species may provide an explanation. Goiti et al. (2003) found that the diet selectivity of P. kuhlii changes with season and location, leading them to define P. kuhlii as a

The positive effect of natural and semi-natural land cover on bat activity was the strongest environmental factor in term of consistency and significance of effect and was represented by three variables: (1) Natural land cover (V.nat), (2) proportion of natural land cover (NATX), and (3) proportion of natural and semi-natural land cover (SNATX). Results of other studies show a similar trend. FuentesMontemayor et al. (2011) found that the percentage of non-crop land cover in an agroecosystem had a positive effect on Pipistrellus pipistrellus at various scales and similarly, the occurrence of forest fragments within agricultural land had a positive effect on Neotropical bats (Numa et al., 2005). The mechanism underlying the relationship between natural land cover in agroecosystems and bat activity may be related to the availability of more suitable habitats or habitat fragments such as non-crop patches within or between cultivated lands. Patches of woody and natural vegetation may provide a less disturbed and more sheltered environment from which bats can leave for foraging bouts in the cultivated lands (Verboom and Spoelstra, 1999). The positive association of bat activity with natural land cover may be related also to availability of insect prey. Abundance and diversity of arthropods in cultivated land is often positively correlated with the percentage cover of natural and semi-natural land cover in the landscape scale (Pluess et al., 2010; Fuentes-Montemayor et al., 2011). Non-crop patches may sustain source populations of prey insects that can move into the cultivated


Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Fig. 2. Bat activity in 2013 according to the different land cover types in the southern Judea plains, Israel. Letters mark significant differences between vegetation types by Brunner-Munzel test (P < 0.05). (A) Total bat activity. (B). Activity of Pipistrellus kuhlii. Results are presented as mean ± SE.

“selective opportunist”. Other studies have found P. kuhlii to exhibit a dynamic and selective use of foraging sites in response to spatial and temporal changes in composition and availability of different prey types (Haffner and Stutz, 1985; Barak and Yom-Tov, 1989). The variability of P. kuhlii activity with regard to vegetation type in the present study may reflect changes in prey composition and availability of prey insects across different crop types. The exceptionally high activity level of P. kuhlii in plots of Brassica oleracea crops (C.bras) in the present study

may also be related to the composition and abundance of the insect fauna in this crop. Two pest moths, Hellula undalis and Plutella xylostella, occur only in B. oleracea (Tzafrir, 2012) and thus may explain the higher activity of P. kuhlii in B. oleracea plots compared to plots of the two other types of vegetable crops. Landscape heterogeneity, as was measured in our study, had a significant positive effect on both total bat activity and on P. kuhlii activity in 2013. Three main mechanisms might be responsible for the positive 87

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Fig. 3. Difference (Δ) of the residuals of the best model of total bat activity that includes (A) natural land proportion index (Natx) in 2012 or (B) natural and seminatural land proportion index (Sntax) in 2013, and the residuals of their closest nested model without Natx or Sntax, respectively. The difference (Δ) is plotted against the index of Natx and Snatx, respectively. Correlations were tested with Pearson coefficients, for (A) Natx R = 0.934 (P < 0.0001) and (B) Santx R = 0.607 (P < 0.0001).


Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Fig. 4. Difference (Δ) of the residuals of the best model of total bat activity that includes (A) Spatial average of ranked spraying load index (SPAGKSL) in 2013 or (B) Averaged reported spray load index (AVRSL) in 2013, and the residuals of their closest nested model without SPAGKSL or AVRSL, respectively. The difference (Δ) is plotted against the index of SPAGKSL or AVRSL, respectively. Correlations were tested with Pearson coefficients, for (A) SPAGKSL R = −0.806 (P < 0.0001) and (B) AVRSL R = −0.887 (P < 0.0001).


Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

or fields. Activity of P. kuhlii was positively associated with proximity to bodies of water in 2012. In arid and semi-arid environments, bodies of water are substantially more abundant in agricultural than in natural habitats (Casas et al., 2012). Since the research area is situated at the border of a semi-arid climatic zone, availability of bodies of water in such an agroecosystem can be regarded as an anthropogenic feature. The affinity of P. kuhlii to bodies of water was showed in other studies in various climatic zones, including Mediterranean (Russo and Jones, 2003), and arid (Korine and Pinshow, 2004). Unlike P. kuhlii, the association of T. teniotis to water sources seems climate dependent. While in European habitats, T. teniotis shows no particular association to water sites (Marques et al., 2004; Rainho, 2007), and in arid areas it has to drink frequently and relies on the availability of ponds (Razgour et al., 2010). We did not find an effect of water sources on T. teniotis, which may be due to the presence of sufficient sources, to the ability of this species to range widely and find distant sources, or to an artefact of the measure we used (the index of WATX.Q), which included water-troughs that are probably too small to be accessed by less maneuverable bats such as T. teniotis (Altringham, 1996). Tadarida teniotis is the only species that was negatively affected by the proximity to power lines among bat species occurring in the research area. This bat forages higher above ground (10–50m, Aulagnier et al., 2008) than P. kuhlii, which flies close to ground (Korine and Pinshow, 2004). The height of power lines is approximately 10 m (Amirshahi and Kavehrad, 2006) and they are sometimes higher in hilly topography. Consequently, given the same horizontal distance from the foraging bat to a powerline, T. teniotis will forage at an actual distance that is shorter than P. kuhlii. In addition, T. teniotis prefers foraging in uncluttered space (Aulagnier et al., 2008) and thus may avoid powerlines physically, rather than due to EMF radiation. We suggest that a case-focused investigation is needed to reach a better understanding on this topic. Disturbance of roads (ROADX) was included in the leading models explaining activity of the focal bat species, but had non-significant effects. With regard to P. kuhlii, this could be because we did not discriminate in the analysis between lit and unlit roads. Since Pipistrelle bats are known to regularly use lit roads for foraging, at least in central Western Europe (Rydell and Racey, 1995), lit and unlit roads may have opposite effects and if analyzed without discrimination (as the variable ROADX in our study) may lead to an overall non-significant effect on bat activity. The effects of environmental factors on bat activity in agroecosystems shown in this study and in others should help to develop batfriendly management schemes, which will encourage bat activity for agricultural purposes of pest control. For example, landscape heterogeneity can be enhanced with landscape engineering to increase crop diversification in the landscape scale (Gurr et al., 2004). Patches of natural woodland (or maquis, in the case of the focal research area) near or within the cultivated land can be preserved or rehabilitated, with adequate connectivity between them, to support populations of insectivorous bats and enhance their activity in the nearby agricultural plots. Reduction in the use of agrochemicals may be important as well, especially for species that show higher sensitivity to agrochemicals. Water-troughs can be placed in strategic locations to enhance activity of specific group of bat species that can utilize them. The extent to which these management schemes and modifications can be achieved economically needs further study, and especially, long-term studies are needed to take into account year-to-year variation in bat activity. Does foraging by bats provide economic benefits that exceed the reduction of yield due to implementation of such bat-friendly schemes? In cases where the answer is not certain, adding the value of those bat-friendly schemes in aspects of conservation may turn the weight in their favor. In conclusion, foraging activity of insectivorous bats in the focal agroecosystem was significantly associated with several important anthropogenic and environmental factors. This knowledge of elements

effect: (1) a direct effect of landscape heterogeneity on spatial niche partitioning of bats, (2) an indirect effect by spatial and temporal partitioning of insect niches, and (3) a specific effect of linear elements in the landscape. Landscape heterogeneity increases spatial niche partitioning and thereby reduces effects of interspecific competition (Tilman and Kareiva, 1997). This relationship may explain the occurrence of greater overall biodiversity in heterogeneous agricultural landscapes (Fischer et al., 2006) compared to homogenous agricultural landscapes (Altieri and Nicholls, 2004). For bats, greater landscape heterogeneity could enable overlap of some rarer species, either due to the presence of habitat refuges or to availability of a wider selection of potential prey. Agroecosystems often include linear elements, such as bushy hedgerows separating arable fields, tree lines in an open landscape and the linear borders between grassland or crop patches and woodland patches (Burel, 1996). Linear elements contribute to the spatial heterogeneity in agroecosystems and as their cover increases, the level of spatial heterogeneity increases too (Bennett et al., 2006). Consequently, where bat activity is found to be positively affected by spatial heterogeneity, it is often difficult to separate the relative contribution of linear versus other elements of the landscape heterogeneity. Linear elements provide a rich foraging ground due to their function as windbreaks. There is a well-documented phenomenon of aggregation of insects on windbreaks due to the special climatic conditions that occur when wind velocity is high (Pasek, 1988). Linear elements, such hedgerows, may serve as a shelter from predators by reducing visibility to predators (Kunz and Fenton, 2006) and may be used as acoustic landmarks thus playing an important role in the ability of bats to navigate (Jenkins et al., 1998). However, it appears that linear elements do not differ in their contribution to landscape connectivity compared to other structural element of different shapes, at least for most temperate bat species (Frey-Ehrenbold et al., 2013). The explanation for the lack of an effect of landscape heterogeneity on T. teniotis activity may be that the methods we used were inadequate considering its biology and focal habitat characteristics. We suggest that the scale of measurement we used was too small to evaluate an effect of heterogeneity on T. teniotis. This species has a larger home range size than that of P. kuhlii (Marques et al., 2004), and a larger portion of its diet consists of large moths (e.g. Noctuidae) that undertake long-distance nocturnal flights (Mata et al., 2016). Total bat activity and activity of both species were negatively associated with agrochemical inputs. The results accord with many studies that found poisoning effects of pesticides on bats (reviewed by Bayat et al., 2014) and higher bat activity in organic versus non-organic farms (e.g. Wickramasinghe et al., 2004). In addition, the results presented in Table 2B and Fig. 4 show specific effects of agrochemicals on total bat activity. Averaged spraying load (AVRSL) had a significant negative effect on total bat activity and non-significant effect when related only to agrochemical applications over the period before sampling time. We therefore conclude that the use of agrochemicals had a long-term effect on total bat activity, which may be related to a decrease in abundance and diversity of insects (Somerville and Walker, 1990). The significant negative effect of the spatially averaged spray load variable (SPAGKSL) on total bat activity supports the long-term effect hypothesis as in addition to the seasonal averages of agrochemical application, it also accounts for the effect at the landscape scale that by definition acts in long-term processes. Long-term spraying, as measured by the seasonal averaged spraying load variable (AVRSL), also suggests an effect on the behavior of foraging bats. Insectivorous bats often show fidelity to specific foraging sites (Perry, 2011) and foraging sites such as conventional orchards that are frequently sprayed may have unpredictable shifts and sudden reductions in insect abundance (Cohen et al., 1994). Since fidelity to foraging site is linked to predictability of resources in the site (Kapfer et al., 2008), bats are less likely to show fidelity to foraging sites with unpredictable resource availability such as frequently sprayed orchards 90

Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

that influence bat foraging activity can be used to design the landscape structure and adapt agricultural management to enhance bat activity and thus, to improve control of nocturnal insect pest populations. To reach an optimal balance of benefits and risks, special attention should be given to the variation among bat species in their response to different environmental and anthropogenic factors.

Fenoy, E., Pérez-Martínez, C., Sánchez, P., Bonachela, S., Elorrieta, M.A., 2012. Farm ponds as potential complementary habitats to natural wetlands in a Mediterranean region. Wetlands 32, 161–174. Cohen, J.E., Schoenly, K., Heong, K.L., Justo, H., Arida, G., Barrion, A.T., Litsinger, J.A., 1994. A food web approach to evaluating the effect of insecticide spraying on insect pest population dynamics in a Philippine irrigated rice ecosystem. J. Appl. Ecol. 31, 747–763. Dormann, C.F., McPherson, J.M., Araújo, B.M., Bivand, R., Bolliger, J., Carl, G., Davies, R.G., Hirzel, A., Jetz, W., Kissling, D., Kühn, I., Ohlemüller, R., Peres-Neto, P.R., Reineking, B., Schröder, B., Schurr, F.M., Wilson, R., 2007. Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30, 609–628. Duelli, P., Obrist, M.K., 2003. Regional biodiversity in an agricultural landscape: the contribution of seminatural habitat islands. Basic Appl. Ecol. 4, 129–138. Dufour, A., Gadallah, F., Wagner, H.H., Guisan, A., Buttler, A., 2006. Plant species richness and environmental heterogeneity in a mountain landscape: effects of variability and spatial configuration. Ecography 29, 573–584. Dunn, P.K., Smyth, G.K., 1996. Randomized quantile residuals. J. Comput. Graph. Stat. 5, 236–244. Estrada, C.G., Damona, A., Hernándezb, C.S., Soto Pintoc, L., Núñeza, G.I., 2006. Bat diversity in montane rainforest and shaded coffee under different management regimes in southeastern Chiapas. Mexico.Biol. Conserv. 132, 351–361. Fagerland, M.W., Sandvik, L., 2009. Performance of five two-sample location tests for skewed distributions with unequal variances. Contemp. Clin. Trials. 30, 490–496. Fahrig, L., Baudry, J., Brotons, L., Burel, F.G., Crist, T.O., Fuller, R.J., Sirami, C., Siriwardena, G.M., Martin, J.L., 2011. Functional landscape heterogeneity and animal biodiversity in agricultural landscapes. Ecol. Let. 14, 101–112 (). Fenton, M.B., 1970. A technique for monitoring bat activity with results obtained from different environments in southern Ontario. Can. J. Zool. 48, 847–851. Fischer, J., Lindenmayer, D.B., Manning, A.D., 2006. Biodiversity, ecosystem function, and resilience: ten guiding principles for commodity production landscapes. Front. Ecol. Environ. 4, 80–86. Freemark, K., 1995. Assessing effects of agriculture on terrestrial wildlife: developing a hierarchical approach for the US EPA. Landsc. Urban Plan. 31, 99–115. Frey-Ehrenbold, A., Bontadina, F., Arlettaz, R., Obrist, M.K., 2013. Landscape connectivity, habitat structure and activity of bat guilds in farmland-dominated matrices. J. App. Ecol. 50, 252–261. Fuentes-Montemayor, E., Goulson, D., Park, K.J., 2011. Pipistrelle bats and their prey do not benefit from four widely applied agri-environment management prescriptions. Biol. Conserv. 144, 2233–2246. Gabriel, D., Roschewitz, I., Tscharntke, T., Thies, C., 2006. Beta diversity at different spatial scales: plant communities in organic and conventional agriculture. Ecol. App. 16, 2011–2021. Gabriel, D., Sait, S.M., Hodgson, J.A., Schmutz, U., Kunin, W.E., Benton, T.G., 2010. Scale matters: the impact of organic farming on biodiversity at different spatial scales. Ecol. Lett. 13, 858–869. Gliessman, S.R., 1998. Agroecology: Ecological Processes in Sustainable Agriculture. CRC Press, Boca Raton. Goiti, U., Vecin, P., Garin, I., Saloña, M., Aihartza, J.R., 2003. Diet and prey selection in Kuhl’s pipistrelle Pipistrellus kuhlii (Chiroptera: vespertilionidae) in south-western Europe. Acta Theriol. 48, 457–468. Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. Ecology 84, 2809–2815. Gurr, G., Wratten, S.D., Altieri, M.A., 2004. Ecological Engineering for Pest Management: Advances in Habitat Manipulation for Arthropods. Csiro Publishing, Collingwood. Haffner, M., Stutz, H.P., 1985. Abundance of Pipistrellus pipistrellus and Pipistrellus kuhlii foraging at street-lamps. Myotis 23, 167–172 (86). Jammalamadaka, S.R., Sengupta, A., 2001. Topics in Circular Statistics. World Scientific Press, Singapore. Jenkins, E.V., Laine, T., Morgan, S.E., Cole, K.R., Speakman, J.R., 1998. Roost selection in the pipistrelle bat, Pipistrellus pipistrellus, (Chiroptera: vespertilionidae), in northeast Scotland. Anim. Behav. 56, 909–917. Jenks, G.F., Caspall, F.C., 1971. Error on choroplethic maps: definition, measurement, reduction. Ann. Assoc. Am. Geogr. 61, 217–244. Johnson, R.A., Wichern, D.W., 2002. Applied Multivariate Statistical Analysis. Prentice hall, Englewood Cliffs, New Jersey. Kapfer, G., Rigot, T., Holsbeek, L., Aron, S., 2008. Roost and hunting site fidelity of female and juvenile Daubenton's bat Myotis daubentonii (Kuhl, 1817) (Chiroptera: vespertilionidae). Mammal. Biol. 73, 267–275. Kaune, W.T., 1993. Assessing human exposure to power-frequency electric and magnetic fields. Environ. Health Perspec. 101, 121. Kelly, R.M., Kitzes, J., Wilson, H., Merenlender, A., 2016. Habitat diversity promotes bat activity in a vineyard landscape. Agric. Ecosyst. Environ. 223, 175–181. Korine, C., Pinshow, B., 2004. Guild structure, foraging space use, and distribution in a community of insectivorous bats in the Negev Desert. J. Zool. 262, 187–196. Kunz, T.H., Fenton, M.B., 2006. Bat Ecology. University of Chicago Press, Chicago. Kunz, T.H., Braun de Torrez, E., Bauer, D., Lobova, T., Fleming, T.H., 2011. Ecosystem services provided by bats. Ann. N.Y . Acad. Sci. 1223, 1–38. Lisón, F., Calvo, J.F., 2013. Ecological niche modelling of three pipistrelle bat species in semiarid Mediterranean landscapes. Acta Oecol. 47, 68–73. Marques, J.T., Rainho, A., Carapuço, M., Oliveira, P., Palmeirim, J.M., 2004. Foraging behaviour and habitat use by the European free-tailed bat Tadarida teniotis. Acta Chiropterol. 6, 99–110. Mata, V.A., Amorim, F., Corley, M.F., McCracken, G.F., Rebelo, H., Beja, P., 2016. Female dietary bias towards large migratory moths in the European free-tailed bat (Tadarida teniotis). Biol. Let. 12, 20150988.

Acknowledgements We thank A. Peters, E. Zaady and A. Harari for the advice and fruitful discussions, to I. Giladi and E. Gavish-Regev for valuable advice, two anonymous reviews and to A. Tsoar for the assistance in field work. We also thank G. Rozenfeld, E. Harcavi, M. Porat and I. Shahar for assistance in becoming familiarized with the local farming practices. The study was supported by Nekudat Hen (Yad HaNadiv Foundation) and the Ministry of Agriculture and Rural Development (grant 8570713-14). This is publication number 944 of the Mitrani Department of Desert Ecology. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at References Altieri, M.A., Nicholls, C., 2004. Biodiversity and Pest Management in Agroecosystems (No. 2). Food Products Press, New York. Altringham, J.D., 1996. Bats: Biology and Behaviour. Oxford University Press, New-York. Amirshahi, P., Kavehrad, M., 2006. High-frequency characteristics of overhead multiconductor power lines for broadband communications. IEEE J. Sel. Areas Commun. 24, 1292–1303. Aulagnier, S., Paunovic, M., Karataş, A., Palmeirim, J., Hutson, A.M., Spitzenberger, F., Juste, J., Benda, P., 2008. Tadarida Teniotis. The IUCN Red List of Threatened Species. Accessed April 2016). Barak, Y., Yom-Tov, Y., 1989. The advantage of group hunting in Kuhl's bat Pipistrellus kuhlii (Microchiroptera). J. Zool. 219, 670–675. Barbosa, P.A., 1998. Conservation Biological Control. Academic Press, CA. Basoalto, E., Miranda, M., Knight, A.L., Fuentes-Contreras, E., 2010. Landscape analysis of adult codling moth (Lepidoptera: tortricidae) distribution and dispersal within typical agroecosystems dominated by apple production in central Chile. Environ. Entomol. 39, 1399–1408. Bayat, S., Geiser, F., Kristiansen, P., Wilson, S.C., 2014. Organic contaminants in bats: trends and new issues. Environ. Int. 63, 40–52. Ben-Arieh, Y., 2001. The agricultural land use in Lachish region, semi-arid zone in Israel. In: Aaronsohn, R., Lavsky, H. (Eds.), A Land Reflected in Its Past: Studies in Historical Geography of Israel, Collection of Essays by Yehoshua Ben-Arieh. The Magnes Press and Yad Izhak Ben-Zvi, Jerusalem, pp. 77–96 (Hebrew). Benda, P., Dietz, C., Andreas, M., Hotovy, J., Lucan, R.K., Maltby, A., Meakin, K., Truscott, J., Vallo, P., 2008. Bats (Mammalia Chiroptera) of the Eastern Mediterranean and Middle East. Part 6. Bats of Sinai (Egypt) with some taxonomic, ecologic and echolocation data on this fauna. Acta. Soc. Zool. Bohem. 72, 1–103. Bennett, A.F., Radford, J.Q., Haslem, A., 2006. Properties of land mosaics: implications for nature conservation in agricultural environments. Biol. Conserv. 133, 250–264. Bennett, V.J., Sparks, D.W., Zollner, P.A., 2013. Modeling the indirect effects of road networks on the foraging activities of bats. Landsc. Ecol. 28, 979–991. Benton, T.G., Vickery, J.A., Wilson, J.D., 2003. Farmland biodiversity: is habitat heterogeneity the key? Trends Ecol. Evol. 18, 182–188. Boldogh, S., Dobrosi, D., Samu, P., 2007. The effects of the illumination of buildings on house-dwelling bats and its conservation consequences. Acta Chiropterol. 9, 527–534. Boughey, K.L., Lake, I.R., Haysom, K.A., Dolman, P.M., 2011. Effects of landscape-scale broadleaved woodland configuration and extent on roost location for six bat species across the UK. Biol. Conserv. 144, 2300–2310. Boyles, J.G., Cryan, P.M., McCracken, G.F., Kunz, T.H., 2011. Economic importance of bats in agriculture. Science 332, 41–42. Brown, V.A., de Torrez, E.B., McCracken, G.F., 2015. Crop pests eaten by bats in organic pecan orchards. Crop Prot. 67, 66–71. Burda, H., Begall, S., Červený, J., Neef, J., Němec, P., 2009. Extremely low-frequency electromagnetic fields disrupt magnetic alignment of ruminants. Proc. Natl Acad. Sci. USA 106, 5708–5713. Burel, F., 1996. Hedgerows and their role in agricultural landscapes. Crit. Rev. Plant Sci. 15, 169–190. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference: a Practical Information-theoretic Approach, second ed. Springer, New York. Cabell, J.F., Oelofse, M., 2012. An indicator framework for assessing agroecosystem resilience. Ecol. Soc. 17, 18. Casas, J.J., Toja, J., Peñalver, P., Juan, M., León, D., Fuentes-Rodríguez, F., Gallego, I.,


Agriculture, Ecosystems and Environment 261 (2018) 80–92

I. Kahnonitch et al.

Rydell, J., Racey, P.A., 1995. Street lamps and the feeding ecology of insectivorous bats. In: Racey, P.A., Swift, S.M. (Eds.), Symposia of the Zoological Society of London, 67, The Society, 1960–1999, pp. 291–307 (London). Schindler, S., von Wehrden, H., Poirazidis, K., Wrbka, T., Kati, V., 2013. Multiscale performance of landscape metrics as indicators of species richness of plants, insects and vertebrates. Ecol. Indic. 31, 41–48. Sciarretta, A., Zinni, A., Mazzocchetti, A., Trematerra, P., 2008. Spatial analysis of Lobesia botrana (Lepidoptera: tortricidae) male population in a Mediterranean agricultural landscape in central Italy. Environ. Entomol. 37, 382–390. Silva, M., Hartling, L., Opps, S.B., 2005. Small mammals in agricultural landscapes of Prince Edward Island (Canada): effects of habitat characteristics at three different spatial scales. Biol. Conserv. 126, 556–568. Smouse, P.E., Long, J.C., Sokal, R.R., 1986. Multiple regression and correlation extensions of the Mantel test of matrix correspondence. Syst. Zool. 35, 627–632. Somerville, L., Walker, C., 1990. Pesticide Effects on Terrestrial Wildlife. Taylor and Francis, London. Stasinopoulos, D.M., Rigby, R.A., 2007. Generalized additive models for location scale and shape (GAMLSS) in R. J. Stat. Softw. 23, 1–46. Studenmund, A.H., 2001. Using Econometrics: A Practical Guide. Addison Wesley Longman, New York. Tilman, D., Kareiva, P., 1997. Spatial Ecology: the Role of Space in Population Dynamics and Interspecific Interactions. Princeton University Press, New Jersey. Tscharntke, T., Tylianakis, J.M., Rand, T.A., Didham, R.K., Fahrig, L., Batáry, P., Bengtsson, J., Clough, Y., Crist, T.O., Dormann, C.F., Ewers, R.M., Fründ, J., Holt, R.D., Holzschuh, A., Klein, A.M., Kleijn, D., Kremen, C., Landis, D.A., Laurance, W., Lindenmayer, D., Scherber, C., Sodhi, N., Steffan-Dewenter, I., Thies, C., Van der Putten, W.H., Westphal, C., 2012. Landscape moderation of biodiversity patterns and processes − eight hypotheses. Biol. Rev. 87, 661–685. Turner, M.G., Gardner, R.H., O’Neill, R.V., 2001. Landscape Ecology in Theory and Practice: Pattern and Process. Springer-Verlag, New York. Tzafrir, G., 2012. Growing Kohlrabi. Ministry of Agriculture, Extension Service. Dept. of Vegetables, Israel(in Hebrew). Documents/gidul_colraby_2012.pdf (last Accessed April 2016). Verboom, B., Spoelstra, K., 1999. Effects of food abundance and wind on the use of tree lines by an insectivorous bat, Pipistrellus pipistrellus. Can. J. Zool. 77, 1393–1401. Walsh, A.L., Harris, S., 1996. Foraging habitat preferences of vespertilionid bats in Britain. J. Appl. Ecol. 508–518. Wickramasinghe, L.P., Harris, S., Jones, G., Vaughan-Jennings, N., 2004. Abundance and species richness of nocturnal insects on organic and conventional farms: effects of agricultural intensification on bat foraging. Conserv. Biol. 18, 1283–1292. Williams-Guillén, K., Olimpi, E., Maas, B., Taylor, P.J., Arlettaz, R., 2016. Bats in the anthropogenic matrix: challenges and opportunities for the conservation of Chiroptera and their ecosystem services in agricultural landscapes. In: Voigt, C.C., Kingston, T. (Eds.), Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer International Publishing, pp. 151–186. Yom-tov, Y., Kadmon, R., 1998. Analysis of the distribution of insectivorous bats in Israel. Div. Distrib. 4, 63–70. Zurcher, A.A., Sparks, D.W., Bennett, V.J., 2010. Why the bat did not cross the road? Acta Chiropterol. 12, 337–340.

McGarigal, K., Marks, B.J., 1995. FRAGSTATS: Spatial Pattern Analysis Program for Quantifying Landscape Structure. General Technical Report PNW-GTR-351. US Department of Agriculture, Forest Service, Pacific Northwest Research Station. Nicholls, B., Racey, P.A., 2006. Contrasting home-range size and spatial partitioning in cryptic and sympatric pipistrelle bats. Behav. Ecol. Sociobiol. 61, 131–142. Nicholls, B., Racey, P.A., 2009. The aversive effect of electromagnetic radiation on foraging bats—a possible means of discouraging bats from approaching wind turbines. PLoS One 4, e6246. Numa, C., Verdú, J.R., Sánchez-Palomino, P., 2005. Phyllostomid bat diversity in a variegated coffee landscape. Biol. Conserv. 122, 151–158. Parsons, S., 1996. A comparison of the performance of a brand of broad-band and several brands of narrow-band bat detectors in two different habitat types. Bioacoustics 7, 33–43. Pasek, J.E., 1988. Influence of wind and windbreaks on local dispersal of insects. Agric. Ecosyst. Environ. 2, 539–554. Perry, R.W., 2011. Fidelity of bats to forest sites revealed from mist-netting recaptures. J. Fish Wildlife Manage. 2, 112–116. Pisa, L.W., Amaral-Rogers, V., Belzunces, L.P., Bonmatin, J.M., Downs, C.A., Goulson, D., Kreutzweiser, D.P., Krupke, C., Liess, M., McField, M., Morrissey, C.A., Noome, D.A., Settele, J., Simon-Delso, N., Stark, J.D., Van der Sluijs, J.P., VanDyckand, H., Wiemers, M., 2015. Effects of neonicotinoids and fipronil on non-target invertebrates. Environ. Sci. Pollut. Res. 22, 68–102. Pluess, T., Opatovsky, I., Gavish-Regev, E., Lubin, Y., Schmidt-Entling, M.H., 2010. Noncrop habitats in the landscape enhance spider diversity in wheat fields of a desert agroecosystem. Agric. Ecosyst. Environ. 137, 68–74. Pocock, M.J., Jennings, N., 2008. Testing biotic indicator taxa: the sensitivity of insectivorous mammals and their prey to the intensification of lowland agriculture. J. App. Ecol. 45, 151–160. Rainho, A., 2007. Summer foraging habitats of bats in a Mediterranean region of the Iberian Peninsula. Acta Chiropterol. 9, 171–181. Razgour, O., Korine, C., Saltz, D., 2010. Pond characteristics as determinants of species diversity and community composition in desert bats. Anim. Conserv. 13, 505–513. Rigby, R.A., Stasinopoulos, D.M., 1996. A Semi-parametric additive model for variance heterogeneity. Statist. Comput. 6, 57–65. Ritz, T., Thalau, P., Phillips, J.B., Wiltschko, R., Wiltschko, W., 2004. Resonance effects indicate a radical-pair mechanism for avian magnetic compass. Nature 429, 177–180. Rusch, A., Bommarco, R., Ekbom, B., 2017. Conservation biological control in agricultural landscapes. Adv. Bot. Res. 81, 333–360. Russell, A.L., Butchkoski, C.M., Saidak, L., McCracken, G.F., 2009. Road-killed bats, highway design, and the commuting ecology of bats. Endang. Species Res. 8, 49–60. Russo, D., Ancillotto, L., 2015. Sensitivity of bats to urbanization: a review. Mammal. Biol. Zeitschrift für Säugetierkunde 80, 205–212. Russo, D., Jones, G., 2002. Identification of twenty-two bat species (Mammalia: chiroptera) from Italy by analysis of time-expanded recordings of echolocation calls. J. Zool. 258, 91–103. Russo, D., Jones, G., 2003. Use of foraging habitats by bats in a Mediterranean area determined by acoustic surveys: conservation implications. Ecography 26, 197–209. Russo, D., Jones, G., Migliozzi, A., 2002. Habitat selection by the Mediterraneanhorseshoe bat, Rhinolophus euryale (Chiroptera: rhinolophidae) in a rural areaof southern Italy and implications for conservation. Biol. Conserv. 107, 71–81.