Abstract
Potatoes are among the most profitable agricultural crops in arable farming in the Netherlands and consequently are grown as frequently as conditions allow. As a result Dutch farmers experienced huge problems with potato cyst nematodes during the last 50 years. In Chapter 1 an outline is presented of the situation of potato cyst nematode control in the 1980's, the problems, possible solutions and the research initiated, of which a part is presented in this thesis.
In 1984 emphasis was put into research concerning the efficiency of soil fumigation on heavy marine clay soils where the majority of Dutch seed and consumption potatoes are grown. Both laboratory and field experiments were carried out to investigate the efficiency of 1,3 dichloropropene and metamnatrium. This required the processing of thousands of larvae suspensions. One of the most labourious and tedious occupations in nematological research is the counting of individuals of the pathogen.
In Chapter 2 A GOP302 image analysis system  Context Vision, Sweden  was used to automate the counting of large numbers of larvaesuspensions of Globodera rostochiensis and G. pallida . These suspensions originated from hatching tests, which were conducted to estimate percentage mortality in field and lab experiments of nematodes exposed to nematicides. The result is called ANECS ( A utomatic NE matode C ounting S ystem), a software program that can count up to 64 compartments with larvae suspensions successively without the aid of an operator. A special object carrier was developed. Images of up to eight object carriers (512 larvae suspensions) can be stored and image analysis can be suspended to offoffice hours. The time needed to count one compartment was reduced by 80% to one minute compared to 'manual' labour while the time for probe preparation remained the same. The percentile error is highest at very low larvae densities (<20 per suspension) and is caused by pollution with small fibres carried by air during the handling of the larvae suspensions. This problem can be minimised by setting up cleanlaboratory procedures. At least 95% of the larvae originating from hatching tests were recognized and counted. The program can and has been be adapted to count other nematode species or to suit more complicated problems like counting both larvae and eggs in one suspension.
In Chapter 3 errors due to subsampling and laboratory procedures affecting the expected value and variance of cyst counts were investigated. Several fields, infested with potato cyst nematodes ( Globodera rostochiensis and G. pallida ) were sampled by collecting bulk samples of approximately 70 cores amounting to 1.8 or 2.5 kg soil from a number of square metre plots located in a regular grid pattern over a 0.33 ha area. Bulk samples from five fields, IV, were thoroughly mixed and from one field, VI, lightly mixed, and subsequently divided into three subsamples of approximately equal weights. Two, sometimes three, subsamples were elutriated separately. Cysts were elutriated by two commercial laboratories, 1 and 2, and separated from the debris and counted at two research laboratories, 0 and 3.
Random bulk samples from five fields, IV, were divided into three portion after thoroughly mixing and taken to Laboratory 0, to compare elutriation precision and accuracy of commercial and scientific laboratories and to check the quality of mixing. To this purpose, pairs or triples were divided into classes. The expected value of the variance within pairs was estimated per class and could be described by a distribution function analogue to a negative binomial distribution, but with three in stead of two parameters. Cysts appeared to be randomly distributed in the well mixed samples, resulting in a binomial or trinomial distribution between pairs or triples. The expected values of the coefficient of variation associated with elutriation were 3.6, 9.6 and 5.5% in the Laboratories 0, 1 and 2, respectively. The upper 95% confidence limit,δ _{0.95} , of coefficients of variation associated with elutriation in Laboratories 1 and 2, were estimated by the differences in 95% upper limits of coefficients of variation between the Laboratories 1 and 2 on the one hand and Laboratory 0 conversely. This difference,δ _{0.95} , ranged from 73% to 42% for Laboratory 1 and from 43% to 19% for Laboratory 2 if 10 to 100 cysts were counted in samples. The consequences of these laboratory errors for the accuracy of sampling methods for both research and extension purposes are discussed.
Nematicide trials require reliable results concerning the effect caused by the fumigant. The percentage mortality of potato cyst nematodes can be estimated by comparing the hatchability of untreated larvae with that of nematicide treated larvae. In Chapter 4 , research is described to improve the quality of hatching tests. Hatching tests using potato root diffusate are labourious and yield quite variable results. Sources of variability were identified and analysed, and solutions were presented. A method was developed to conduct hatching tests using inert materials so that the total variation at the end of the test is minimized.
A number of hatching tests was carried out to increase reliability, optimize the method and limit the amount of work. Thus, it was possible to obtain a coefficient of variation ( cv ) of the hatching process which is in accordance with the combined errors expected when a certain number of cysts is treated and eggs are used in a hatching test. An Appendix is provided listing the different errors and ways to calculate and cope with them.
The results indicate that the hatching process is no longer an important source of variation for the end result. All variation higher than expected could be explained by variation between replications of batches with the same treatment, indicating that small differences in nematicide application cause major differences in the end result. The treatment effect was more important in field experiments than in laboratory experiments.
The hatching curve could be described adequately by a loglogistic curve with 3 parameters (λfinal number of hatched larvae,αtime,βslope parameter). Addition of a fourth parameter (γ, incubation time) improved the fit of the hatching curve significantly. Using the loglogistic model, final hatch can be predicted with a certain error before the actual hatching test ended, but in general final hatch is underestimated. When an error of 5% is accepted, the length of time required to perform a hatching test of a laboratory experiment can be reduced by 80% for untreated batches and by 40 to 80% for batches treated with nematicides. Acceptable reduction is negatively correlated with the concentration of the fumigant used.
Hatching tests with cysts originating from field experiments are unsuitable for prediction using a time limited data set. In cyst batches from the field compound hatching curves could be distinguished in 4 out of 6 fields, indicating that the soil samples contained at least two fractions of cysts with different hatching responses. Prediction would cause a significant underestimation of final hatch and consequently an overestimation of mortality.
Because of its high vapour pressure, 1,3dichloropropene is primarily used on marine clay soils. In Chapter 5 a laboratory experiment is described investigating the two stereoisomeres of 1,3dichloropropene for their efficiency in killing nematodes. Batches of increasing numbers of Globodera rostochiensis cysts were exposed to a range of concentrations of the (E) and (Z)isomers of 1,3dichloropropene. The cysts were of identical origin. Temperature during treatment was 10 ^{o}C, humidity 100% and time of exposure 8 days. The integrals of concentration time products ( CT ) created were 0, 3, 7, 14, 31, 60, 125, 242, and 437μg/ml·day for the (E)isomer and 0, 3, 16, 59, 240, and 419μg/ml·day for the (Z)isomer. Survival was estimated with hatching tests 1.5, 3, and 7 months after treatment.
The relationship between dosage of (E)isomer and numbers of hatchable nematodes followed a loglogistic equation at all hatching dates. Hatchability, and therefore lethal dosages, increased as hatching tests were more delayed. Seven months after treatment, practically all treated nematodes had recovered and hatchability of treated and untreated nematodes was the same. A loglogistic relationship was also found for dosage (Z)isomer and numbers of hatchable nematodes 1.5 month after treatment. When hatching tests of nematodes treated with the (Z)isomer were delayed till 3 and 7 months after treatment, the results were better explained by a compound model, assuming two independent loglogistic effects, one stimulating hatch at low dosages and one reducing hatch at all dosages. Only the (Z)isomer of 1,3dichloropropene was effective as a nematicide.
Chapter 6 presents research concerning the efficiency of standard doses of 1,3 dichloropropene in fields with a high silt content. Three fields of marine clay soil were fumigated with 150 l/ha 1,3dichloropropene (DD) (Teleone II ^{TM}, Shell 95 ^{TM}). On three dates after application, concentrations of Z and E 1,3dichloropropene were measured per 5 cm layer of soil to a depth of 40 cm and integrals of concentration time products were calculated. When the fumigant was no longer detectable, a top soil treatment with either 150 l/ha metamsodium or 180 kg/ha dazomet (active compound methyl isothiocyanate) was applied, followed immediately by autumn ploughing. Soil samples were taken before and after fumigation and after the top soil treatment to extract potato cyst nematodes (PCN). Survival was determined by means of hatching tests. Mortalities after the DD treatment, defined as 100  % survival, were estimated per 5 cm layer of soil to a depth of 30 cm to construct dosage response curves. Fumigation with DD killed 48, 48 and 72% of the PCN per field, respectively. Accelerated breakdown of DD by microorganisms accounted for the two lower mortality rates.
The additional top soil treatment with metamsodium increased mortality to 90% or more. Dazomet, however, was less effective (53 and 80%) considering that twice as much of the active compound was applied as in the metamsodium treatment. Multiplication of hatched larvae originating from the injection layer after the DD treatment was 25% less than that from untreated plots. This was caused by a lower fraction of larvae developing into cysts. PCN could be retrieved from soil layers as deep as 80 cm below the surface. Fumigation reached only a fraction of the infested soil, down to 2530 cm. The infestation foci were so small compared to the standard minimum area fumigated (1 ha) that 90% of the active compound would be wasted on noninfested soil. Soil fumigation, whether or not combined with an additional top soil treatment, will seldom be profitable. Monitoring for infestation foci is recommended.
As soil fumigation was not a viable option to keep potato cyst nematodes in check on heavy marine clay soils, another way of control had to be found. Research was focussed at the development of sampling methods for the detection of small infestation foci with high reliability (≥90% probability of detection). Precautionary soil fumigation can be avoided, the area where a control measure has to be applied can be minimized to the actual infestation, and detection occurs so early that (partially) resistant potato cultivars can be grown without significant yield reduction as population densities are still low.
Research in the Flevopolders yielded promising results. Therefore, in 1990, a research program was initiated to develop new sampling methods for the detection of patchy infestations of potato cyst nematodes ( Globodera rostochiensis and G. pallida ) with known accuracy in all potato cropping areas of The Netherlands. Patchy infestations in cropping areas of the provinces of Zeeland, Friesland, Groningen and Drente were sampled to validate a model based on data from cropping areas in Flevoland and to determine whether one detection method could meet the requirements of all cropping areas in The Netherlands. The results are presented in Chapter 7 . Eighty two fields were presampled to locate patchy infestations using a coarse sampling grid (8 · 3 m). Parts of thirty seven fields, containing one or more foci, were sampled intensively by extracting at least 1.5 kg of soil per square metre (1.33 · 0.75 m). Forty foci were analysed for spatial distribution characteristics of cysts using Generalized Linear Models (GLM's) and classical Multiple Linear Regression Analysis, differing in assumptions about the distribution of the input variable (number of cysts per kg of soil).
The results showed that the data from all investigated cropping areas fit well to an exponential model with two parameters, the length and width gradient parameters. Significant differences in these parameter values between cropping areas could not be demonstrated. As both parameters follow a normal distribution, the probability of any combination of these parameters can be described by a bivariate normal distribution. Gradient parameters were correlated but significant correlations between these parameters and certain variables, such as the nematode species involved ( G. pallida or G. rostochiensis ), the time interval between sampling and the last potato crop, soil type, cropping frequency and cyst density in the focus centre could not be demonstrated. It can be concluded that one detection method for small infestation foci suffices for all investigated cropping areas. Its expected accuracy is independent of soil type, potato cyst nematode species, cropping frequency or time interval between sampling and last potato crop.
In Chapter 8 the model for infestation foci developed in the previous chapter was applied for practical usage. A computer program called SAMPLE was developed to evaluate existing and create new sampling methods for the detection of patchy infestations or 'foci' of the potato cyst nematode ( Globodera spp.). By combining a model for the medium scale distribution of cysts, which provides the expected population densities at each position within the focus, and a model for the small scale distribution within square metres (negative binomial distribution) SAMPLE allows to simulate sampling procedures.
The importance of the parameters of the two distribution models  the length and width gradient parameters for the medium scale distribution and the aggregation factor k of the negative binomial distribution for the small scale distribution  was investigated by sensitivity analyses. The aggregation factor k proved to be less important when calculating the average detection probability of a focus than the length and width gradient parameters. Several existing versions of the statutory sampling method used in The Netherlands were tested for their performance on a standard infestation focus with a central population density of 50 cysts/kg soil.
The standard focus is small enough to use resistant potato varieties as a control measure without noticeable yield reductions in a 1:3 potato crop rotation. As the statutory soil sampling methods did not perform with the desired average detection probability, set at 90%, the program was used to develop several new sampling methods for focus detection and to investigate their performance. SAMPLE is a tool to develop sampling methods on demand for every possible combination of characteristics required for use by seed and ware potato growers (recommendations for optimum control measures leading to maximum returns, Integrated Pest Management) and by governments (legislation, quarantine and export protection).
For advisory purposes a model is required describing the relation between the number of potato cyst nematodes and tuber yield. A stochastic model with biologically relevant parameters was available. In Chapter 9 the direct relation between the number of potato cyst nematodes and plant growth is described and used to deduce the relation between nematode density and yield reduction of total plant weight and tuber yield. The relation between small and medium initial population densities and the relative total plant weight was derived as cross sections at right angles to the time axis of a growth model with three dimensions: time after planting t , relative total plant weight Y and relative growth rate r _{p} /r _{0} . The relative growth rate is the (constant) ratio between the growth rate r _{p} of plants of a certain weight at a nematode density P and the growth rate r _{0} of (younger) plants of the same weight without nematodes. Therefore, the ratio between the time after planting that plants need to reach a certain weight in the absence of nematodes and at nematode density P, t _{0} /t _{p} equals the ratio r _{p} /r _{o} (2).
The relative growth rate r _{p} /r _{o} = k + (1 k )0.95 ^{P/T} ^{1}for P > T and = 1 for P ≤T (3). Formally, k is the minimum relative growth rate as P →∞. As a result the arbitrary equation y = m + (1 m )0.95 ^{P/T} ^{1}for P > T and = 1 for PT (6) also applies to the relation between small and medium initial population densities and relative total plant weight. T is the tolerance limit, below which growth and yield are not reduced by nematodes; m is the relative minimum yield.
The relations between small and medium initial population densities of potato cyst nematodes and relative tuber weight of potatoes can be derived from the growth model in an analogous way. However, there is one complication: tuber initiation does not start at the same haulm weight in plants with and without nematodes, but at the smaller haulm weight the larger the nematode density. As a consequence, tuber weights of plants with a certain total weight at nematode density P are not equal to those of plants with the same total weight without nematodes, but r _{p} Δ t units of weight larger,Δ t being the difference between the actual time of tuber initiation and the time total plant weight becomes the same as that of plants without nematodes at the initiation of tuber formation.
Relative total and tuber weights of plants with 'early senescence' and at large nematode densities are smaller than estimated by the model and equation (2). This indicates that at large initial population densities growth reducing mechanism(s) become active that were not operating at smaller densities.
In Chapter 10 an advisory system is presented for the management of potato cyst nematodes ( Globodera pallida) . It emphasizes the use of partially resistant potato cultivars, which provide the possibility of keeping population densities of potato cyst nematodes at a low level in short fixed rotations. Using stochastic models based on the population dynamics of potato cyst nematodes and the relation between preplant nematode densities and relative yield it is possible to calculate the probabilities of population development and the reductions in yield caused by these population densities. A simulation model is developed which integrates both models, using the frequency distributions of some of the most variable parameters relevant to a particular combination of potato cultivar and nematode population. Also, the natural decline in population density when nonhosts are grown is incorporated in the model.
The model makes it possible to calculate the probability of a certain yield reduction, given a certain potato cultivar, nematode population and rotation. Therefore, it becomes feasible for a farmer to evaluate risks and the costs of different control measures in fixed rotations. The application of this model in the starch potato growing areas could lead to significant improvements in financial returns and a major reduction of the use of nematicides.
In Chapter 11 we describe the 'Seinhorst research program' initiated by Dr J.W. Seinhorst, former head of the Nematology Department of the IPODLO. It consists of an empiric philosophy, the scientific methods applied, and the models developed at the IPODLO during the last 45 years of nematological research, including the 13 years in which the research described in this thesis was carried out. All theories of the Seinhorst research program are developed by searching for recurring regularities (patterns) in a collection of observations, named 'the empirical base'. To prevent " ghost theories from sloppy data " all assumptions underlying the empirical base are carefully described in theories with respect to methodology and technology, including statistics.
The patterns to be recognized are summarized by mathematical equations, which must be connected with biological processes to bridge the gap between 'normal' language and mathematical language for the description of biological theories. Often, the patterns result from more than one biological process. If so, the basic patterns are disentangled from one another using a method of pattern analysis. The procedure is best carried out when only a limited number of more or less congruent patterns are involved. Therefore, attention must be given to the choice of the hierarchic level and the complexity of the investigated system. Investigations proceed from simple experimental systems to complex natural systems at a hierarchic level that is neither so high that manifesting processes are very dissimilar nor so low that one runs the risk of describing processes irrelevant for the purpose of the investigation.
In the 'Seinhorst Research Program' this purpose is finding methods for improvement of financial returns of host crops attacked by plantparasitic nematodes through calculating risks of nematode population development and subsequent yield reduction. Pattern analysis yields theories about causes of phenomena observed at the investigated hierarchic level and about properties of processes at the nearest lower hierarchic level. Predictions at the next higher hierarchic level are made by synthesizing several patterns in (stochastic) simulation models. Synthesis is also applied to compound patterns of processes in simple experimental systems, with the objective to explain complicated patterns in complex systems.
In the Conclusions ( Chapter 12 ) an overview is presented of the practical results and aspects of the research effort described in this thesis. Some comments are made on the present state of affairs concerning potato cyst nematode control in the Dutch seed and ware potato growing areas.
Original language  English 

Qualification  Doctor of Philosophy 
Awarding Institution  
Supervisors/Advisors 

Award date  9 Dec 1998 
Place of Publication  S.l. 
Publisher  
Print ISBNs  9789054859604 
Publication status  Published  1998 
Keywords
 potatoes
 globodera
 plant parasitic nematodes
 control
 nematicides
 techniques
 sampling
 detection
 models