GEOSTATISTICAL TOOLS FOR THE STUDY OF INSECT SPATIAL

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

DISTRIBUTION: PRACTICAL IMPLICATIONS IN THE INTEGRATED

MANAGEMENT OF ORCHARD AND VINEYARD PESTS

Pasquale TREMATERRA

Andrea SCIARRETTA

University of Molise, Department of Agricultural, Environmental and Food Sciences,

Campobasso, Italy

ABSTRACT

Spatial heterogeneity in agricultural systems is recognized as an important source of

variability to be investigated. In the evolution of IPM, patterns and processes that

influence spatio-temporal dynamics in insect populations tends to assume more

importance compared to the classical theory. Geostatistics represent a valuable tool to

investigate the spatial pattern of insect populations and to support pest control. After

an explanation of the geostatistical analysis, in the present paper we provided an

overview of practical applications in managing pests, focusing on fruit orchards and

vineyards. The utility of geostatistical tools is illustrated with examples taken from

field studies, with attention to the analysis of spatial patterns, monitoring schemes,

use of traps, scale issues, precision targeting and risk assessment maps. Potential

approaches in the context of IPM are discussed in relation to future perspectives.

Keywords: insect pests, monitoring, precision agriculture, IPM, spatial analysis

ABSTRACT

GEOSTATISTIČNA ORODJA ZA PREUČEVANJE PROSTORSKE RAZPOREDITVE

ŠKODLJIVCEV: PRAKTIČNA UPORABA OBVLADOVANJA ŠKODLJIVCEV PRI

INTEGRIRANEM VARSTVU SADOVNJAKOV IN VINOGRADOV

Prostorska heterogenost v kmetijskih sistemih je poznana kot pomemben vir

variabilnosti, ki jo je potrebno raziskati. Pri razvoju integriranega varstva rastlin (IVR),

vzorci in procesi, ki vplivajo na prostorsko - časovno dinamiko populacij škodljivcev

imajo običajno večji pomen kot pri konvencionalnem varstvu rastlin. Geostatistika

predstavlja dragoceno orodje pri proučevanju prostorske razporeditve populacij

škodljivcev in predstavlja dobro podporo pri zatiranju škodljivcev. V prispevku so

predstavljene osnove geostatističnih analiz in njihova praktična uporaba pri

obvladovanju škodljivcev, s poudarkom na škodljivcih v sadovnjakih in vinogradih kot

so: češpljev zavijač (Grapholita funebrana), jabolčni zavijač (Cydia pomonella),

breskov zavijač (Grapholita molesta), breskov molj (Anarsia lineatella), križasti grozdni

sukač (Lobesia botrana) in breskova muha (Ceratitis capitata). Uporabnost

geostatističnih orodij je ponazorjena s primeri iz terenskih raziskav, s poudarkom na

Full Prof., Via De Sanctis, I-86100 Campobasso, Italy, e-mail: tre[email protected]

Assoc. Prof., ibid.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

analizi prostorskih vzorcev škodljivcev, sistemih za njihovo spremljanje, uporabi vab,

težavah z določanjem merila, t.i. natančnem varstvu, in na izdelavi kart z oceno

tveganja. V prispevku bo tekla razprava o uporabi geostatistike v perspektivah IVR v

prihodnje.

Ključne besede: Kriging, škodljive žuželke, načrtno spremljanje, natančno kmetijstvo;

integrirano varstvo rastlin, prostorske analize

1 INTRODUCTION

Agricultural systems are intrinsically heterogeneous. In fact, they contain variable

arrangements of soils, habitats, microclimatic features, plant communities and

consequently they show an extensive variability in soil fertility, water retention, crop

productivity, and so on. Basically, this is true also when we consider single fields that

are typically composed of a central part and a border with many biotic and abiotic

parameters showing gradients and edge effects (van Helden 2010).

The same principles apply to insect populations. In this case, spatial variation is

caused by the interaction between population dynamics on the one hand and biotic or

abiotic factors on the other. Processes that influence the spatial heterogeneity include

population growth (reproduction, mortality) and dispersal (immigration, colonization,

emigration). For example, aggregations can be determined by the position of initial

immigrants influencing the behaviour of other individuals/species through the

emission of pheromones or inducing the formation of new plant volatiles. Similarly,

the colonization process is strongly influenced by birth/death rates that differ locally,

so that the total population density in the whole field will increase, while in limited

areas population will become extinct, leading to a clumped spatial pattern (Fleischer

et al. 1997).

At the landscape level, the fragmentation of farmland has resulted in a scattered

resource distribution that strongly enhances the importance of landscape structure in

determining the final spatial pattern of a pest inside and outside a crop field. In fact,

the distribution of host plants, including alternative hosts, will influence the short-

distance foraging flights of herbivores, and often also the dispersal of predators and

parasitoids (Mazzi & Dorn 2012). In the same way, the location of overwintering sites

will determine the reinvasion pattern in the following season.

In the past, many efforts have been dedicated to improving the efficiency in the

design of agricultural experiments minimizing the residual variability that in field

trials is due mainly to the spatial heterogeneity. The strong advance of the space issue

in biological sciences has arisen from the recognition that spatial variability, or

patchiness, is widespread in natural populations and this characteristic is an

interesting quantity rather than a statistical nuisance to be overcome (Schneider

1994).

In the new evolution of Integrated Pest Management (IPM) concepts, the spatial

variation in pest populations tends to assume more and more importance compared to

the classical theory. In site-specific IPM, the heterogeneity at the single field level is

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

analyzed with the aim of optimizing chemical treatments (Park et al. 2007). In area-

wide IPM, the importance of managing the whole pest population at landscape or

regional level is emphasized, for example by identifying pest shelters inside and

outside crops (Hendrichs et al. 2007). As a matter of fact, however, incorporation of

the spatial component in management plans is in practice still isolated.

In these contexts, geostatistics represent a valuable set of statistical tools to

investigate the spatial pattern of pests and to support the facilitation of practical pest

control applications.

In the present paper we provide an overview of geostatistical applications in the study

of insect spatial distribution, focusing on fruit orchards and vineyards, and their utility

in managing pests is illustrated with examples taken from field studies. Potential

approaches in the context of IPM are also discussed in relation to possible future

perspectives.

2 GEOSTATISTICS

After the advent of calculators for the capture and elaboration of experimental data,

largely accessible today thanks to new technologies such as personal computers, GPS,

remote sensing and GIS tools, statistical approaches that incorporate space in the

elaboration have found new applications in many science subjects. In this context, a

major role is played by geostatistics, first developed for mining explorations and then

adopted by many environmental disciplines such as agriculture, hydrology,

meteorology, soil sciences, fisheries, forestry, epidemiology, landscape ecology,

environmental pollution and risk assessment.

Geostatistics are a collection of statistical methods analyzing spatial dependence

among samples (autocorrelation) and obtaining estimates of the variable under study

at unsampled locations. For a detailed description of the general theory and principles,

refer, among others, to Cressie (1993), Webster and Oliver (2001), Chilès and

Delfiner (2012). Various internet sources are available for both beginners and experts

to this subject; for an overview of geostatistical methods implemented in real

applications, it is possible to consult the active list service of Ai_Geostat (1995) or the

website of GeoENVia Association (2011), that organize every two years the

International Conference on Geostatistics for Environmental Applications.

In brief, the main steps in the geostatistical analysis are:

1. Exploratory data analysis. Some elementary statistical analysis is useful to

highlight general characteristics of data. Normality of data distribution can be

evaluated using histograms and box-plots or by calculating some coefficient of

asymmetry. Skewed variables often show a proportional effect, i.e., a higher

variability in high valued areas and a lower variability in low valued areas that distort

variogram results (Manchuk et al. 2009). Although formally not required, a normal

distribution of data improves the autocorrelation analysis and can be achieved with a

logarithm transformation.

2. Estimation and modelling of spatial autocorrelation. To evaluate the spatial

variation, different tools can be used, analyzing correlation coefficient (in

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

correlograms), covariance (in covariance functions) or variance (in semivariograms).

On choosing between these methods in ecological applications, see Rossi et al.

(1992). Next, we will refer mainly to semivariograms, the most commonly used

method in geostatistics.

The experimental variogram is a graph of discrete points at particular lag intervals,

showing the semivariance of sample pairs against the distance between sampling

points. The semivariance γ for lag distance h is given by:

 







)(

)()(

)(2

)(

hxzxz



where z(x

) is a measured sample point at x

, z(x

+ h) is a measured sample at point x

+ h and N(h) is the number of pairs separated by the lag h.

Because these estimates can strongly fluctuate from point to point due to sampling

errors, a model describing the spatial variation must be fitted. Among the approaches

available for use are exponential, spherical, linear, polynomial and Gaussian functions

that can also be combined to obtain nested models (Pannatier 1996).

Figure 1: Variogram shapes of different Anarsia lineatella weekly pheromone trap catches and maps of the

corresponding spatial distribution in the investigated agro-ecosystem: A – clumped distribution without

nugget, B – clumped distribution with nugget, C – random distribution.

Figure 1 illustrates key features of a semivariogram: the nugget is the y-axis intercept;

the sill is the point at which variance no longer increases; the range corresponds to the

distance where the sill is reached. Differences in spatial variation with geographical

direction are known as anisotropy: a diverse semivariogram model can be produced

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

for each considered direction. A geostatistical rule of thumb is that each lag interval

must be represented by at least 30 pairs of points (Journel & Huijbregts 1978). This

means that a minimum number of 25-30 sample units is required to obtain variograms

with 4-5 lag classes, but often more points are necessary to accurately estimate

sample pair variances (Nansen et al. 2003). WEBSTER and Oliver (1992) pointed out

that a minimum number of 100 sampling points is needed to give reliable results, but

in a practical context it is often necessary to work with many fewer points.

Semivariogram modelling is not an easy exercise and much practice should be

devoted to this analysis. For more information on these techniques and interpretation

of results, consult Isaaks and Srivastava (1989) and Cressie (1993) for a general

overview, and Oliver (2010) for their use in an agricultural context.

3. Estimation of a surface area using interpolation procedures. In geostatistics

we can define the interpolation as a method of value estimation and/or prediction at

unsampled locations in the geographical space. The objective of interpolation is to

create continuous surfaces based on point samples. Many different methods are

available, both deterministic and probabilistic, based on the mathematical algorithms

used to compute the weights to be assigned during the interpolation; examples are

triangulation, inverse distance weighted, natural neighbour, kriging, radial basis

function, and also more sophisticated Bayesian techniques, such as the stochastic

conditional simulation (Rossi et al. 1993).

The ordinary kriging is considered the best linear unbiased predictor and is by far the

most utilized. The estimated Z at unsampled location x

is:





xZxwxZ

)()()(

where w

is the weight calculated for the sampled location x

, Z(x

) is the observed

value at x

and n is the number of locations.

The kriging weights depend on both the spatial autocorrelation measured in

variograms and the spatial configuration of the sample points around the prediction

location. Various forms of kriging have been developed to accommodate different

types of data (i.e., block kriging for mean values from local areas, universal kriging

when a spatial trend is detected, indicator kriging for binary data, cokriging for two or

more variables spatially autocorrelated, etc.).

When insect populations are sampled, it is very common to obtain count data with

many zeros. In these cases, indicator kriging represents an alternative choice. More

detailed information about this method is reported in the paragraph “Risk assessment

maps”.

It is possible to assess the quality of interpolation by computing the errors

(interpolated value minus observed value) and applying the cross-validation

procedure; various statistics can be used as a quantitative measure of quality. For

more information on geostatistical interpolation techniques, refer to Isaaks and

Srivastava (1989) and Cressie (1993).

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Different kinds of maps can be generated to visualize the results of the interpolation

process, such as contour maps, surface maps, image maps or wireframes, where the

variable densities are represented as different lines, colours, shadows or in 3

dimensions. A base map can be overlaid to show landscape features.

3 SAMPLING

Geostatistics represent a significant change in the methodology of sampling. In fact,

traditionally we need to have independent data and sampling plans are designed to

avoid correlations. On the contrary, geostatistics look for autocorrelations, and so

sampling plans became less restrictive (Sharov 1997). Moreover, the final objective of

a geostatistical survey is not to obtain the estimation of a mean, as in classic plans, but

to map the spatial variability of samples. For example, areas that are avoided because

they might be a source of bias, such as field edges, become primary areas to be

explored. Similarly, areas usually discarded because they are considered to be without

or with a low pest presence should be included: in a geostatistical survey, areas at

zero levels are as important as high density areas (Brenner et al. 1998).

Nonetheless, new aspects arise that must be considered in spatially explicit surveys. It

is known that precision, which indicates how well the mean is estimated, increases

with sampling size (Fleischer et al. 1997). Classically, sampling plans are designed to

balance such precision with the costs of sampling; in this case many sampling units

are evaluated into the field and they are used to obtain a unique mean. In geostatistical

applications, a large number of sample units are needed to perform a variogram

analysis, but sampling units are evaluated individually at each location and this results

in a poor local estimate. This effect is more accentuated when the distribution is

aggregated, a very common condition in pest populations. In such a situation, clusters

of sampling units and interpolation data with block kriging can be a solution, but the

costs of large sample sizes are often prohibitive (Fleischer et al. 1997).

In general, irregularly spaced sampling points are not a problem, especially for

kriging interpolation and this characteristic gives some freedom in setting up a

sampling design, but the orientation, scale and arrangement of sampling units can still

influence the result of geostatistical analysis. Moreover, an optimal sampling scheme

for variography can be different from that designed for kriging interpolation, so the

final purpose of our survey should be clear when the sampling plan is arranged

(Marchant & Lark 2012).

Various classical sampling schemes can be adopted, such as simple random, stratified

random, cluster, nested or systematic sampling (Wollenhaupt et al. 1997). Among

them, systematic design is generally considered more precise than simple or stratified

random (Webster & Oliver 1990), but it must be remembered that in fruit orchards

and vineyards there is usually a regular pattern composed of plants positioned at fixed

distances within and between rows, and this can strongly influence geostatistical

elaborations. Schotzko and O’Keefe (1990), evaluating the effect of sample

placement on the geostatistical analysis of Lygus hesperus Knight in lentils,

considered a staggered grid to give a better map precision than a uniform grid.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

In the case of insects, very often no prior information is available, the variation is

complex and the scale of the phenomenon is unknown. An exploratory survey of

spatial variation can help to select the appropriate size, number and location of

observations (Baldacchino et al. 2012; Marchant & Lark 2012), but in practical

situations it can rarely be done. In similar situations, it is possible to use, for example,

a cluster sampling, where clusters of individual units are selected at random and each

unit in the cluster is measured; this approach fits particularly well when populations

tend to be clustered (Gilbert 1987). Another possibility is the nested survey, where,

following a classification, clusters are subdivided, then the subdivisions are randomly

selected and further subdivided until the smallest units are identified (Wollenhaupt et

al. 1997). This approach allows the exploration of several orders of magnitude of

spatial scale in a single analysis (Kerry et al. 2012). Bacca et al. (2008) used a cluster

sampling plan to interpolate and simulate the male leaf miner Leucoptera coffeella

(Guérin-Méneville & Perrottet) distribution at different trap densities in a coffee

plantation.

Another approach can be the adaptive survey, consisting of changing sampling efforts

in the space, according to the data collected earlier (Thomson 1990). An example of

insect adaptive surveys, related to tsetse fly population, is provided by Sciarretta et al.

(2005).

4 PRACTICAL APPLICATIONS

After the first studies carried out in North America to investigate the distribution of

Pectinophoras gossypiella (Saunders) in cotton, Lygus hesperus Knight in lentil

fields, and grasshoppers in uncultivated areas (Borth & Huber 1987, Kemp et al.

1989; Schotzko & O’Keefee 1989), geostatistics have seen many applications in the

various fields of crop protection against worms or arthropods pests (for more details

see Liebhold et al. 1993; Brenner et al. 1998; Arbogast et al. 2000; Brandhorst-

Hubbard et al. 2001; Park et al. 2007; Webster 2010; Sciarretta & Trematerra 2011a)

and, in a few cases, to highlight predator and parasitoid distribution (Karimzadeh et

al. 2011; Perović & Gurr 2012).

One of the most significant examples of applications in this field was carried out in

eastern United States over the last two decades against the gypsy moth Lymatria

dispar (L.), which was introduced in North America from Europe in 1869 (Liebhold

et al. 1989). Pheromone trap catches and egg mass data were analysed using

geostatistical tools at regional scale to model the gypsy moth spatial dynamic, with

the aim of: delimiting the boundary of pest dispersion, estimating the spread rate at

the expanding population front, forecasting the spatial dynamics of moth outbreaks,

predicting the larval defoliation levels and evaluating the treatment effects (Liebhold

et al. 1991, 1998; Hohn et al. 1993; Sharov et al. 1995; Tobin et al. 2004, 2007).

4.1 Analysis of spatial patterns

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Because spatial variation is due to so many factors, generalization about the causes of

patchiness in insect populations is very problematic; often it is not possible to

understand the main factors determining the spatial pattern in a specific context or to

predict a priori any form of distribution (van Helden 2010). Often, each orchard is

unique and the features of experimental variograms are not the same also for

neighbouring fields. Further, the distribution can change according to the insects’

developmental stage, the season, the phenological status of the crop and the weather

conditions. For example, alternating periods of clumped and random patterns were

observed to be recurrent in fruit orchards and vineyards for leafhoppers, thrips and

fruit flies (Nestel & Klein 1995; Papadopoulos et al. 2003; Farias et al. 2004; Decante

& van Helden 2008; Rhodes et al. 2011). Consequently, in situ observations are

necessary to depict the spatio-temporal dynamics of a pest and descriptive maps must

be developed to have a visual representation of pest presence in the agro-ecosystem.

The study of spatial variation patterns is crucial from this point of view and the

features of semivariograms give us much information about the spatial structure of

our data.

An asymptotic function indicates an aggregated insect distribution and the range

represents approximately the extension of hot spots (areas of aggregation); on the

contrary, linear functions indicate a uniform/random distribution, with the random

component increasing with the increase of the variance variability; when the slope is

near to zero we obtain a pure nugget effect, indicating a complete lack of any

autocorrelation and a pure random distribution (Schotzko & O’Keefe 1989).

A zero nugget indicates a strong confidence in sample data, while the presence of a

nugget represents two sources of variability: the micro variance occurring at a scale

smaller than the minimum lag distance and the measurement error.

Fig. 1 shows common types of variograms underlying different insect distributions.

An index that can summarize the level of randomness is the k parameter, defined as

the ratio between the nugget and the sill, and this indicates the degree of spatial

dependence measured in the variogram (Journel & Huijbregts 1978). Values below

0.8 indicate that the distribution is aggregated; as the k parameter approaches zero, the

level of spatial dependence will become greater.

4.2 Monitoring schemes

Monitoring pest population is a key issue in IPM schemes. The objectives of

monitoring are to detect the presence or absence of pests and quantify their abundance

(and, eventually, their natural enemies) through time and space. Follow the spatio-

temporal dynamic of the population by regular, periodic sampling, monitoring allows

us to reach a decision as to whether, when and where, a pest population requires

control action.

In this context, geostatistics applied to a grid of monitoring points allows us to

obtain a map providing useful information on the pest spatial distribution, in

particular:

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

 the origin of infestations in the investigated agro-ecosystem, both inside and

outside the considered crops;

 the position and temporal dynamics of hot spots;

 the role played by cultivated and wild host plants as potential sources of

infestation;

 the effect of landscape structure on the dispersal of the pest population.

As an example, we report the studies conducted on the distribution of Grapholita

funebrana Treitschke and Cydia pomonella L. in two heterogeneous agro-ecosystems

of central Italy (Sciarretta et al. 2001; Trematerra et al. 2004).

In the case of G. funebrana, pheromone trapping was carried out inside a 12-ha plum

orchard and to the surrounding area, covering a surface of about 250 ha. The results

revealed a distribution strongly influenced by the fragmented structure of the

landscape and the presence and dissemination of host plants in the area investigated,

where adults showed a strong capacity for dispersal and movement between elements

of the landscape (Sciarretta et al. 2001). In particular, irrigation canals and the

hedgerows around the plum orchard served as corridors along which the adults passed

from one zone to another of the territory. The highest catches in the orchard were just

at the point of contact with these corridors, highlighting the movements that occur

between the plum orchard and a ravine, where there was an abundance of blackthorn,

another host plant of the insect.

For C. pomonella, the monitoring by means of pheromone traps, carried out in two

agro-ecosystems with productive apple orchards and scattered trees of apple, pear,

service and walnut, highlighted a limited dispersion of adults in the territory; catches

of male moths were clumped and the hot spots were confined to the productive apple

orchards or in small groups of wild apple, pear, service and walnut trees. The

colonized areas were isolated from each other and this suggests that strips free of host

plants around orchards may be an effective barrier against immigration from infested

zones. In this case, a strip of 200-300 m was found to be an obstacle to the movement

of the moth (Trematerra et al. 2004).

4.3 Use of traps

When attractive devices are used for a spatial monitoring scheme, some geostatistical

properties must be taken into account (Perry et al. 2002): the extent, describing the

dimension of the study area; the support, that is, the sampling unit size and

corresponding to the attractive range of the trap; the lag, i.e., the distance between

sampling units.

The grid of the traps will give different sampling results according to the

following conditions:

 when lag > support, the experimental design allows a large individual movement;

 when lag = support, the movement of individuals is more limited;

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

 when lag < support, there may be an alteration in the spatial distribution because

of the phenomena of mutual interference between traps.

Geostatistical techniques can help to establish the correct distance between

monitoring devices.

For example, Bacca et al. (2006) determined the optimal spacing of pheromone traps

for monitoring the coffee leaf miner Leucoptera coffeella in a coffee plantation, thus

allowing an efficient trap distribution in the field, finding also a significant difference

between the orthogonal directions of the plant rows.

Using experimental variograms, Epsky et al. (2010) determined the sampling range of

a female-targeted protein-based attractant for the Mediterranean fruit fly Ceratitis

capitata (Wiedemann) in various fruit crops; geostatistical results were confirmed by

combining a release-recapture experiment with the use of contour maps illustrating

the spatial distribution of recaptured flies.

4.4 Scale issues

Spatial patterns are usually strongly scale-dependent and this is true also when the

object of our investigations is a pest species. This means that the change in some

measures of pattern, i.e. extent, support and lag, will change in both the resolution and

range of measurement (Schneider 1994).

After changing the scale, prevailing processes defining that particular pattern will be

different and will consequently lead to different results. For example, if we study the

spatial structure of a pest population at the within-field level, forces such as local

population dynamics will dominate in our analysis. If we move to a landscape level,

patch composition and metapopulation processes will prevail. At a regional level,

other variables will act over the others, i.e., climatic features, altitudinal trend, genetic

drift and so on.

The choice of the appropriate scale depends on the objective of our study. If we

intend to understand the distribution of a pest inside an orchard for optimizing control

or monitoring actions, a sampling point grid will be deployed to cover every part of

the field, including peripheral sectors to verify the presence of peculiar spatial

patterns such as the border effect (van Helden 2010).

At this scale, fruit species and cultivars, in relation to their spatial location and

phenological phase, can have an important role in determining the spatio-temporal

dynamics of pests, particularly the polyphagous ones. Studies on the spatio-temporal

dynamics of C. capitata carried out to evaluate the effect of the host plants on the pest

spatial distribution, in an agricultural landscape of 500 ha located in central Italy,

showed that fruit flies were caught sequentially in orchards with host plants (i.e.,

peach, apple, pear, oriental persimmon and prickly pear) at varying times of

maturation, especially when the fruits remained on the trees (Sciarretta & Trematerra

2011b). Distributional maps provided evidence that made it possible to identify fruit

species in which the fly developed early in the season (mixed peach orchards) and

afterwards during the periodic flights.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

The experimental design will be different if we want to identify sink and sources in an

agricultural landscape. In this case, because spatial distribution can easily be affected

by landscape composition, sampling strategies should be extended to cover the whole

area and designed to adequately differentiate variable properties at each important

landscape unit, including those in which we assume the pest is not present. In these

cases, useful information will be obtained about the role played by host plants as

potential sources of infestation outside the considered crops. In the case of the

European grapevine moth Lobesia botrana (Den. & Shiff.), contour maps highlighted

that adult spatial distribution was not limited to vineyards, but its presence was high

inside olive groves, particularly during the first seasonal flight (Sciarretta et al. 2008).

The landscape structure, through the presence of elements such as hedgerows,

uncultivated fields, streams, and woodlots, which act as barriers or ecological

corridors, can have a strong effect on the dispersion of the pest population. Examples

on this topic were reported for G. funebrana, Grapholitha molesta (Busck), Anarsia

lineatella (Z.) and C. pomonella (Sciarretta et al. 2001; Sciarretta & Trematerra 2006;

Basoalto et al. 2010). The presence of overwintering sites outside deciduous orchards

was reported to influence the colonization and spread of leafhoppers into the orchards

from the surrounding vegetation (Nestel & Klein 1995).

At regional level, sampling points are often located at great distances (kilometres or

more), and this hides the population dynamic occurring at lower scales. Studies at this

level can have the objective of obtaining a general frame of the pest presence in a

large area, but investigations can also be directed to verify spatial relationships of the

pest with specific variables (Ayalew et al. 2008). For example, a study carried out on

160,000 ha in Catalonia, Spain, aimed at analyzing the current codling moth

pheromone trap spatial distribution and verifying the presence of anisotropic effects

due to predominant wind directions (Comas et al. 2012).

4.5 Precision targeting programs

The incorporation of spatial variability into an Integrated Pest Management program

is called site-specific IPM or precision targeting for IPM and relies on the use of maps

showing a pest distribution, to be used to minimize direct control tactics (Weisz et al.

1995; Brenner et al. 1998). Such an approach follows the principles of precision

agriculture, but in spite of the progress made by the latter in recent years (Oliver

2010), the practical development of site-specific IPM programs is still limited today

(Park et al. 2007; Sciarretta et al. 2011).

Among the difficulties in incorporating precision targeting into IPM are the

identification of external infestation foci, the necessity to have aggregated populations

with limited dispersal ability and the high sampling costs, which are often not

economically sustainable. Also, an evaluation of insecticide application costs, related

to the site-specific versus whole field IPM, needs to be addressed.

The development of a site-specific IPM was carried out against L. botrana in

vineyards located in a hilly landscape in Italy (Sciarretta et al. 2011). In this case, two

tactics were used: the first was directed at reducing the source of infestation from

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

outside the vineyards, and specifically from the olive groves, which were found to

host an important part of the pest population (Sciarretta et al. 2008), by establishing a

pheromone trap barrier to prevent male movements into the vineyards. The second

was to reduce the quantity of insecticides used and the treated area, focusing curative

efforts towards the sectors of vineyard with the highest level of L. botrana

oviposition, while excluding areas with low egg density. The results highlighted that

male hot spots in olive groves disappeared, and that the number of larval nests on vine

inflorescences was significantly decreased when additional traps were deployed,

compared to the period before. The site-specific control, i.e., treating only egg hot

spots with Bacillus thuringensis var. kurstaki, allowed for a decrease in the surface of

the vineyard treated and, consequently, the quantity of insecticide utilized; no

significant damage differences between whole field and site-specific IPM in

vineyards were observed when treatments were carried out against both second and

third L. botrana generations. An analysis of costs related to insecticide application in

the field highlighted that the site-specific approach was economically advantageous, if

compared to the whole field IPM, with greater damage to up to 1% of infested berries

per bunch, covered by the saving of reduced treatments (Sciarretta et al. 2011).

4.6 Risk assessment maps

One of the possible outcomes of geostatistical analysis is the creation of risk

assessment maps for pest management. Such an instrument has seen strong

development especially in epidemiological studies, and maps can be obtained merging

data from many different kinds of sources (Eisen & Eisen 2011). For example, a risk

map for L. botrana was obtained by utilizing three years’ data on larval damage, with

both the number of attacked berries per bunch and the percentage of infested bunches

(Fig. 2).

The utility of similar instruments in IPM programs was shown by Brenner et al.

(1998), who gave details on using the indicator kriging to define and quantify areas

that exceed predetermined action thresholds. In short, an indicator is a variable with

values only of 1 or 0, obtained by dividing our scale of counts into one or more

thresholds. The interpolation of the indicator variable will give the distribution of the

estimated probability that a sampling point placed in a specific location will exceed

the established threshold.

Fig. 3 illustrates the case of C. pomonella distribution in an apple orchard, where the

indicator kriging was elaborated considering an action threshold of 2 males collected

in a pheromone trap per week. In this case, the map provides support for selecting

sectors of the orchard where correct positioning of a trap will give a reliable

indication of the achievement of the threshold.

5 FUTURE PERSPECTIVES

Currently, many GIS softwares incorporate spatial analysis tools, including

geostatistics, for producing distributional maps. The widespread use of GIS-based

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

studies suggests that the utilization of geostatistical methods will become more

widespread in applied contexts and at different scales, making it easier to develop

efficient local or regional pest management plans.

Figure 2. Risk map of the Lobesia botrana larval damage sampled in a 4.5 ha vineyard. An index, obtained

multi- pling the mean number of attacked berries per bunch and the percentage of infested bunches from 3-

year data, was transformed in a scale with levels ranging from 0 (no risk) to 10 (maximum risk) and

interpolated using ordinary kriging; x and y axes are expressed as UTM coordinates.

The use of GIS technology today appears very promising in area-wide IPM programs,

where activities are conducted over large geographical areas, involving the use of

decision support systems, taking into account the pest and beneficial species

colonization and dispersal and evaluating the presence of environmental factors that,

changing across the managed area, could affect the success of an IPM program (Faust

2008). Although there are some examples of the use of geostatistics in area-wide IPM

programs (Tobin et al. 2004; Carrière et al. 2006; Smith et al. 2006; De Luigi et al.

2011), their use in fruit orchard and vineyard protection is still very limited. At this

regard, in a sterile insect release program initiated in British Columbia, Canada, since

1992 and still active nowadays, to obtain an area-wide suppression of C. pomonella

from its fruit-growing valleys (Okanagan-Kootenay SIR Program 2012), a GIS

software combined with geostatistical analysis was developed for managing moth

population and fruit damage data and to determine how key activities in the program

could be streamlined (Vernon et al. 2001, 2006).

A further improvement may arise from models that better define a pest’s spatio-

temporal dynamic. In this regard, a promising approach is space-time geostatistics,

designed for variables that vary in both time and space. They involve the use of the

variogram to characterize the variation along the time dimension as well as the spatial

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

one (Heuvelink & van Egmond 2010). The difference with respect to the classical

approach is that both these sources of variation are elaborated and their effects are

taken into account, for example, to predict the target variable at an unmeasured time

by kriging. They are not intended as temporal forecasting models, but can provide

predictions and be used to move from a series of freeze-frames to a continuous

recording of the phenomenon under study.

Figure 3: Distributional map of cydia po- monella weekly pheromone trap catches (on the left) and

corresponding risk map obtained calculating the indicator kriging for an action threshold of 2 males per trap

per week (on the right). Risk levels correspond to the estimated probability that a sampling point placed in

a speci- c location will exceed the established threshold. Black areas are the best places where to put a

monitoring trap.

The problem of the high cost of pest management in a spatial context, especially for

sampling, is currently the most serious constraint to the diffusion of geostatistical

techniques in practice. This limitation may in part be overcome if efforts are directed

to the development of intelligent Location-Aware Systems that allow automation of

trapping devices and treatment operations (Wen et al. 2009; Pontikakos et al. 2012).

Ultimately, an important shift may be achieved gradually as practices such as

sustainable agriculture, organic farming, zero-residue production and so on gain more

importance in the growing of high value crops, and as the environmental advantages

of using a reduced or zero input of chemicals are incorporated as added value in

determining the final product price.

6 REFERENCES

AI_GEOSTAT (1995): Available at http://www.ai-geostats.org (accessed 18.09.2013).

Alemany A., Miranda M.A., Alonso R., Martín Escorza C. (2006): Changes in the spatial and

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

temporal population density of the Mediterranean fruit fly (Diptera: Tephritidae) in a citrus

orchard. Spanish Journal of Agricultural Research, 4(2): 161-166.

Ayalew G., Sciarretta A., Baumgärtner J., Ogol C., Löhr B. (2008): Spatial distribution of

Diamondback moth, Plutella xylostella L. (Lepidoptera: Plutellidae), at the field and the

regional level in Ethiopia. International Journal of Pest Management, 54 (1): 31-38.

Arbogast R.T., Kendra P.E., Mankin R.W., McGovern J.E. (2000): Monitoring insect pests in retail

stores by trapping and spatial analysis. Environmental Entomology, 93 (5): 1531-1542.

Bacca T., Lima E.R., Picanço M.C., Guedes R.N.C., Viana J.H.M. (2006): Optimum spacing of

pheromone traps for monitoring the coffee leaf miner Leucoptera coffeella. Entomologia

Experimentalis et Applicata, 119: 39-45.

Bacca T., Lima E.R., Picanço M.C., Guedes R.N.C., Viana J.H.M. (2008): Sampling plan for the

coffee leaf miner Leucoptera coffeella with sex pheromone traps. Journal of Applied

Entomology, 132: 430-438.

Baldacchino F., Sciarretta A., Addante R. (2012): Evaluating the spatial distribution of

Dociostaurus maroccanus egg pods using different sampling designs. Bulletin of Insectology,

65 (2): 223-21.

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. Environmental Entomology,

39 (5): 1399-1408.

Bonsignore C.P., Manti F., Vacante V. (2007): Field and tree distribution of Capnodis tenebrionis

(Linnaeus, 1767) adults in an apricot orchard in Italy. Journal of Applied Entomology, 132:

216-224.

Borth P.W., Huber R.T. (1987): Modeling pink bollworm establishment and dispersion in cotton

with the kriging technique: 267-274. In: Proceedings of the Beltwide Cotton Production

Research Conference, Dallas, 4-8 January 2008: National Cotton Council of America,

Memphis.

Brandhorst-Hubbard J.J., Flanders K.L., Mankin R.W., Guertal E.A., Crocker R.L. (2001):

Mapping of soil insect infestations sampled by excavation and acoustic methods. Journal of

Economic Entomology, 94: 1452-1458.

Brenner R.J., Focks D.A., Arbogast R.T., Weaver D.K., Shuman D. (1998): Practical use of

spatial analysis in precision targeting for integrated pest management. American

Entomologist, 44: 79–101.

Carrière Y., Ellsworth P. C., Dutilleul P., Ellers-Kirk C., Barkley V., Antilla L. (2006): A GIS-based

approach for areawide pest management: the scales of Lygus hesperus movements to cotton

from alfalfa, weeds, and cotton. Entomologia Experimentalis et Applicata, 118: 203–210.

Chilès J.P., Delfiner P. (2012): Geostatistics: modeling spatial uncertainty, second edition. Wiley

series in probability and statistics. John Wiley and Sons, New York: 1-734.

Comas C., Avilla J., Sarasúa M.J., Ribes-Dasi M. (2012): Lack of anisotropic effects in the spatial

distribution of Cydia pomonella trap catches in Catalonia, NE Spain. Crop Protection, 34: 88-

95.

Cressie N.A.C. (1993): Statistical for spatial data, revised edition. John Wiley and Sons, New

York: 1-900.

Decante D., van Helden M. (2008): Spatial and temporal distribution of Empoasca vitis within a

vineyard. Agricultural and Forest Entomology, 10: 111-118.

De Luigi V., Furlan L., Palmieri S., Vettorazzo M., Zanini G., Edwards C.R., Burgio G. (2011):

Results of WCR monitoring plans and evaluation of an eradication programme using GIS and

Indicator Kriging. Journal of Applied Entomology, 135: 38-46.

Eisen L., Eisen R.J. (2011). Using geographic information systems and decision support systems

for the prediction, prevention, and control of vector-borne diseases. Annual Review of

Entomology, 56: 41–61.

Epsky N.D., Espinoza H.R., Kendra P.E., Abernathy R., Midgarden D., Heath R.R. (2010):

Effective sampling range of a synthetic protein-based attractant for Ceratitis capitata (Diptera:

Tephritidae). Journal of Economic Entomology, 103 (5): 1886-1895.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Farias P.R.S., Roberto S.R., Lopes J.R.S., Perecin D. (2004): Geostatistical characterization of

the spatial distribution of Xylella fastidiosa sharpshooter vectors on citrus. Neotropical

Entomology, 33 (1) (2003): 13-20.

Faust R.M. (2008): General introduction to areawide pest management. In: Koul O., Cuperus G.,

Elliott N. (eds), Areawide pest management. Theory and implementation. CAB International,

Wallingford, United Kingdom: 1-14.

Fleischer S.J., Weisz R., Smilowitz Z., Midgarden D. (1997): Spatial variation in insect

populations and site-specific integrated pest management. In: Pierce E.J. and Sadler E.J.,

(eds), The State of Site-Specific Management for Agriculture. American Society of Agronomy,

Crop Science Society of America, and Soil Science Society of America, Madison, WI: 101-

130.

GeoENVia Association (2011): Available at http://www.geoenvia.org/ (accessed 18.09.2013).

Gilbert R.O. (1987): Statistical methods for environmental pollution monitoring. Van Nostrand

Reinhold, New York: 1-321.

Karimzadeh R., Hejazi M. J., Helali H., Iranpour S., Mohammadi S. A. (2011): Assessing the

impact of site-specific spraying on control of Eurygaster integriceps (Hemiptera:

Scutelleridae) damage and natural enemies. Precision Agriculture, 12: 576-593.

Kemp W.P., Kalaris T.M., Quimby W.F. (1989): Rangeland grasshopper (Orthoptera: Acrididae)

spatial variability: macroscale population assessment. Journal of Economic Entomology, 82

(5): 1270-1276.

Kerry R., Oliver M.A., Frogbrook Z.L. (2012): Sampling in precision agriculture. In: Oliver M.A.

(ed.), Geostatistical applications for precision agriculture. Springer, New York: 35-64.

Hendrichs J., Kenmore P., Robinson A.S., Vreysen M.J.B. (2007): Area-Wide Integrated Pest

Management (AW-IPM): principles, practice and prospects. In: Vreysen M.J.B., Robinson

A.S., Hendrichs J. (eds), Area-Wide Control of Insect Pests. From Research to Field

Implementation. Springer, Dordrecht, The Netherlands: 3-33.

Heuvelink G.B.M., van Egmond F.M. (2010): Space-time geostatistics for precision agriculture. In:

Oliver M.A. (ed.), Geostatistical applications for precision agriculture. Springer, New York:

117-137.

Hohn M.E., Liebhold A.M., Gribko L.S.. (1993): A geostatistical model for forecasting the spatial

dynamics of defoliation caused by the gypsy moth, Lymantria dispar (Lepidoptera:

Lymantriidae). Environmental Entomology, 22: 1066-1077.

Ifoulis A.A., Savopoulou-Soultani M. (2006): Use of geostatistical analysis to characterize the

spatial distribution of Lobesia botrana (Lepidoptera: Tortricidae) larvae in Northern Greece.

Environmental Entomology, 35 (2): 497-506.

Isaaks E.H., Srivastava R.M. (1989): An introduction to applied geostatistics. Oxford University

Press, New York. 1-561.

Journel A.G., Huijbregts C.J. (1978): Mining geostatistics. Academic Press, London, United

Kingdom: 1-600.

Liebhold A.M., Mastro V., Schaefer P.W. (1989): Learning from the legacy of Leopold Trouvelot.

Bulletin of the Entomological Society of America, 35: 20-21.

Liebhold A.M., Zhang X., Hohn M.E., Elkinton J.S., Ticehurst M., Benzon G.L., Campbell R.W.

(1991): Geostatistical analysis of gypsy moth, Lymantria dispar, egg mass populations.

Environmental Entomology, 20: 1407-1417.

Liebhold A.M., Zhou G., Ravlin F.W., Roberts A., Reardon R. (1998): Forecasting gypsy moth

defoliation with a Geographical Information System. Journal of Economic Entomology, 91:

464-472.

Liebhold A.M., Rossi R.E., Kemp W.P. (1993): Geostatistics and geographic information systems

in applied insect ecology. Annual Review of Entomology, 38: 303-327.

Manchuk J.G., Leuangthong O., Deutsch C.V. (2009): The Proportional Effect. Mathematical

Geoscience, 41: 799–816.

Marchant B.P., Lark R.M. (2012): Sampling in precision agriculture, optimal designs from

uncertain models. In: Oliver M.A. (ed.), Geostatistical applications for precision agriculture.

Springer, New York: 65-88.

Mazzi D., Dorn S. (2012): Movement of insect pests in agricultural landscapes. Annals of Applied

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Biology, 160: 97-113.

Nansen C., Campbell J.F., Phillips T.W., Mullen M.A. (2003): The impact of spatial structure on

the accuracy of contour maps of small data sets. Journal of Economic Entomology, 96 (6):

1617-1625.

Nestel D., Klein M. (1995): Geostatistical analysis of leafhopper (Homoptera: Cicadellidae)

colonization and spread in deciduous orchards. Environmental Entomology, 24 (5): 1032-

1039.

Okanagan-Kootenay SIR Program (2012): Available at http://www.oksir.org (accessed

18.09.2013).

Oliver M.A. (2010): Geostatistical applications for precision agriculture. Springer, New York. 1-

331.

Pannatier, Y. (1996): Variowin: software for spatial data analysis in 2D. Springer, New York: 1-91.

Papadopoulos N.T., Katsoyannos B. I., Nestel D. (2003): Spatial autocorrelation analysis of a

Ceratitis capitata (Diptera: Tephritidae) adult population in a mixed deciduous fruit orchard in

Northern Greece. Environmental Entomology, 32 (2): 319-326.

Park, Y.L., Krell, R.K., Carroll, M. (2007): Theory, technology, and practice of site-specific Insect

Pest Management. Journal of Asia-Pacific Entomology, 10 (2), 89-101.

Peláez H., Marana R., Vazquez De Prada P., Puras A., Santiago Y. (2006): Comportamento local

de poblaciones de Lobesia botrana Denis & Schiffermüller (Lepidoptera: Tortricidae). Boletín

de Sanidad Vegetal Plagas, 32: 189-197.

Perović D.J., Gurr G.M. (2012): Geostatistical analysis shows species-specific habitat

preferences for parasitoids. Biocontrol Science and Technology, 22 (2): 243-247.

Perry J.N., Liebold A.M., Rosenberg M.S., Dungan J., Miriti M., Jakomulska A., Citron-Pousty S.

(2002): Illustrations and guidelines for selecting statistical methods for quantifying spatial

pattern in ecological data. Ecography, 25: 578-600.

Pontikakos C.M., Tsiligiridis T.A., Yialouris C.P., Kontodimas D.C. (2012): Pest management

control of olive fruit fly (Bactrocera oleae) based on a location-aware agro-environmental

system. Computers and Electronics in Agriculture, 87: 39–50.

Ramírez-Dávila J.F., González-Andújar J.L., Ocete R., López Martínez M.A. (2002): Descripción

geostadística de la distribución espacial de lod huevos del mosquito verde Jacobiasca lybica

(Bergevin & Zanon) (Hemiptera, Cicadellidae) en viñedo: modelización y mapeo. Boletín de

Sanidad Vegetal Plagas, 28: 87-95.

Ramírez-Dávila J.F., González-Andújar J.L., López Martínez M.A., Ocete R. (2005): Modelización

y mapeo de la distribución de las ninfas del mosquito verde Jacobiasca lybica (Bergevin &

Zanon) (Hemiptera, Cicadellidae) in viñedo. Boletín de Sanidad Vegetal Plagas, 31: 119-132.

Ramírez-Dávila J.F., Porcayo-Camargo E. (2008a): Spatial distribution and mapping of

Jacobiasca lybica (Bergevin & Zanon) (Hemiptera, Cicadellidae) egg populations in irrigated

sherry vineyards. Revista Chilena de Entomologia, 34: 37-55.

Ramírez-Dávila J.F., Porcayo-Camargo E. (2008b): Distribución espacial de las ninfas de

Jacobiasca lybica (Hemiptera, Cicadellidae) en un viñedo en Andalusia, España. Revista

Colombiana de Entomologia, 34 (2): 169-175.

Rhodes E.M., Liburd O.E., Grunwald S. (2011): Examining the spatial distribution of flower thrips

in southern highbush blueberries by utilizing geostatistical methods. Environmental

Entomology, 40 (4): 893-903.

Ribes M., Bascuñana M., Avilla J. (1998): Estudio de la distribución espacial de Cydia pomonella

(L.) y Pandemis heparana (Denis & Schiffermüller) en Torregrossa (Lleida) mediante

métodos geostadísticos. Boletín de Sanidad Vegetal Plagas, 24: 935-948.

Rossi R.E., Mulla D.J., Journel A.G., Franz E.H. (1992): Geostatistical tools for modeling and

interpreting ecological spatial dependence. Ecological Monographs, 62: 277-314.

Rossi R.E., Borth P.W., Tollefson J.J. (1993): Stochastic simulation for characterizing ecological

spatial patterns and appraising risk. Ecological Applications, 3 (4): 719-735.

Schneider D.C. (1994): Quantitative ecology. Spatial and temporal scaling. Academic Press, San

Diego, CA: 1-395.

Schotzko D.J., O'Keefe L.E. (1989): Geostatistical description of the spatial distribution of Lygus

hesperus (Heteroptera: Miridae) in lentils. Journal of Economic Entomology, 82: 1277-1288.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Schotzko D.J., O'Keefe L.E. (1990): Effect of sample placement on the geostatistical analysis of

the spatial distribution of Lygus hesperus (Heteroptera: Miridae) in lentils. Journal of

Economic Entomology, 83: 1888-1900.

Sciarretta A., Trematerra P. (2006): Geostatistical characterization of the spatial distribution of

Grapholita molesta and Anarsia lineatella males in an agricultural landscape. Journal of

Applied Entomology, 130: 73–83.

Sciarretta A., Trematerra P. (2011a): Distribuzione spaziale e lotta di precisione ad alcuni fitofagi

dannosi ai frutteti e alla vite. Notiziario sulla protezione delle piante, III Serie, 1 (2009): 7-17.

Sciarretta A., Trematerra P. (2011b): Statio-temporal distribution of Ceratitis capitata population in

a heterogeneous landscape in Central Italy. Journal of Applied Entomology, 135: 241-251.

Sciarretta A., Trematerra P., Baumgärtner J. (2001): Geostatistical analysis of Cydia funebrana

(Lepidoptera: Tortricidae) pheromone trap catches at two spatial scales. American

Entomologist, 47: 174–184.

Sciarretta A., Zinni A., Trematerra P. (2011): Development of site-specific IPM against European

grapevine moth Lobesia botrana (D. & S.) in vineyards. Crop Protection, 30: 1469-1477.

Sciarretta A., Girma Melaku, Tikubet Getachew, Belayehun Lulseged, Ballo Shifa, Baumgärtner

J. (2005): Development of an adaptive tsetse population management scheme for the Luke

community, Ethiopia. Journal of Medical Entomology, 42 (5): 1006-1019.

Sciarretta A., Zinni A., Mazzocchetti A., Trematerra P. (2008): Spatial Analysis of Lobesia botrana

(Denis and Schiffermüller) Male population in a Mediterranean agricultural landscape in

central Italy. Environmental Entomology, 37 (2): 382-390.

Sharov A.A. (1997): Quantitative Population Ecology. On-Line lectures. Available at

http://home.comcast.net/~sharov/PopEcol/ (accessed 10.4.2013).

Sharov A.A., Roberts E.A., Ravlin F.W., Liebhold A.M. (1995): Gypsy moth (Lepidoptera:

Lymantriidae) spread in the Central Appalachians: Three methods for species boundary

estimation. Environmental Entomology, 24: 1529-1538.

Smith J., Su N.-Y., Escobar R.N. (2006): An areawide population management project for the

invasive eastern subterranean termite (Isoptera: Rhinotermitidae) in a low-income community

in Santiago, Chile. American Entomologist, 52: 253-260.

Thompson, S.K. (1990): Adaptive cluster sampling. Journal of the American Statistical

Association, 85: 1050-1059.

Tobin P.C., Sharov A.A., Liebhold A.A., Leonard D.S., Roberts A.E., Learn M.R. (2004):

Management of the Gypsy Moth through a decision algorithm under the STS project.

American Entomologist, 50: 200-209.

Tobin P.C., Liebhold A.M., Roberts E.A. (2007): Comparison of mehods for estimating the spread

of a non-indigenous species. Journal of Biogeography, 34: 305-312.

Trematerra P., Sciarretta A. (2005): Activity of the kairomone ethyl (E,Z)-2,4-decadienoate in the

monitoring of Cydia pomonella (L.) during the second annual flight. Redia, LXXXVIII: 57-62.

Trematerra P., Gentile P., Sciarretta A. (2004): Spatial analysis of pheromone trap catches of

Codling moth (Cydia pomonella) in two heterogeneous agro-ecosystems, using geostatistical

techniques. Phytoparasitica, 32: 325–341.

Van Helden M. (2010): Spatial and temporal dynamics of Arthropods in arable fields. In: Oerke

E.-C., Gerhards R., Menz G., Sikora R.A. (eds), Precision crop protection – the challenge and

use of heterogeneity, Springer, New York: 51-64.

Vernon B., Thistlewood H., Kabaluk T., Smith S., Frank G., Batenburg R. (2001): Using statistical

analysis and GIS to enhance the efficiency of SIR in managing codling moth, Cydia

pomonella L., in British Columbia. In: Program and Abstracts, Joint Annual Meeting,

Entomological Society of Canada and Ontario, Niagara Falls, October, 21-24 2001.

Vernon B., Thistlewood H., Smith S., Kabaluk T. (2006): Introduction: a GIS application to

improve codling moth management in the Okanagan Valley of British Columbia. In: Esterby

S.R. (ed.), Case study in data analysis: Variables related to codling moth abundance and the

efficacy of the Okanagan Sterile Insect Release Program. Canadian Journal of Statistics, 34:

494-499.

Webster R. (2010): Weeds, worms and geostatistics. In: Oliver M.A. (ed.), Geostatistical

applications for precision agriculture. Springer, New York: 221-242.

Zbornik predavanj in referatov 13. Slovenskega posvetovanje o varstvu rastlin z mednarodno udeležbo

Rimske Toplice, 7.-8. marec 2017

Ljubljana, Društvo za varstvo rastlin Slovenije (Plant Protection Society of Slovenia), 2017

Webster R., Oliver M.A. (1990): Statistical methods in soil and land resource survey. Oxford

University Press, Oxford, UK: 1-316.

Webster R., Oliver M.A. (1992): Sample adequately to estimate variograms of soil properties.

Journal of Soil Science, 43: 177-192.

Webster R., Oliver M.A. (2001): Geostatistics for environmental scientists. John Wiley & Sons,

Chichester, UK: 1-271.

Weisz R., Fleischer S., Smilowitz Z. (1995): Site-specific Integrated Pest Management for high-

value crops: sample units for map generation using the Colorado potato beetle (Coleoptera:

Chrysomelidae) as a model system. Journal of Economic Entomology, 88 (5): 1069-1080.

Wen C., Guyer D.E., Li W. (2009): Local feature-based identification and classification for orchard

insects. Biosystems Engineering, 104: 299-307.

Wollenhaupt N.C., Mulla D.J., Gotway-Crawford C.A. (1997): Soil sampling and interpolation

techniques for mapping spatial variability of soil properties. In: Pierce E.J., Sadler E.J. (eds),

The State of Site-Specific Management for Agriculture. American Society of Agronomy, Crop

Science Society of America, and Soil Science Society of America, Madison, WI: 19-53.