Studies of Wireworm Populations: II. Spatial Distribution

The distribution of wireworms in space is known to be not entirely haphazard. It can therefore be assumed to be dependent upon a number of environmental factors, some favourable others unfavourable, the balance of which, varying from place to place, checks the pest here, but there allows it to occur in abundance. A thorough knowledge of those environmental factors would be of great value to the economic entomologist. It would enable him to estimate, without laborious sampling, the number of wireworms likely to be found under any given set of conditions. It might enable him to hinder the favourable and assist the unfavourable factors, and so to bring about control.

One way to obtain that desirable knowledge is to choose an environmental factor arbitrarily, and study it in detail in the laboratory, with hope that the information obtained will be of use. Another way is to examine the distribution of wireworms in nature, with a view to the discovery of factors correlated with the degree of infestation.

It is the second course that has been followed in the work to be described below. The distribution of wireworms in natural pastures has been studied in large, medium and small areas; first, to discover how wireworms are in fact distributed, and secondly, in the hope that many environmental factors could be eliminated, and a few shown to be important in determining the incidence of wireworms in those areas. This study, therefore, is not an attempt to investigate exhaustively the factors that control wireworm distribution, but an exploration to discover what factors are most likely to repay detailed investigation.

The method used to extract the wireworms from soil samples has been described in the first paper of this series (Salt & Hollick, 1944). It is considered to give the complete Agriotes population of the sample other than eggs.

The distribution of wireworms within a grass field has been studied on the University Farm, 2 miles north-west of Cambridge, in a field of about 8 acres, called Spinney Pasture. This field is contiguous over part of its north-western boundary with a small spinney of larch, fir, hornbeam and other trees, and abuts on the south upon a narrow belt of deciduous trees, almost exclusively ash. Apart from those woods, the pasture is surrounded by arable land; but the fields to the south-east were ploughed from very old grass as recently as 1940 and 1941, and those to the north and west have included in their rotation a 4-year ley. From the fields to the north-east and south-east, the pasture is separated by deep ditches. Spinney Pasture has been under grass for many years but was ploughed and resown to grass in 1933 when it was taken into the University Farm. Since that time it has been continuously under grass and principally used for grazing, though it has occasionally been cut for hay.

Twenty stations were established in the main area of the field in the positions shown on the map (Fig. 1). At each station, a sample was taken in the last week of every month from January 1942 until April 1944, when the field ceased to be available to us. Sets of samples taken from the field in October and December 1941 are not included in the following account, because they were not taken at the same stations; but a special set taken on 13 March 1944 is included. We have thus a series of twenty-nine samples from each station.

The samples collected on each occasion consisted of a cylinder of soil, 4 in. in diameter and 12 in. deep. The soil was collected and examined in two parts, an upper core from the surface to 6 in. deep and a lower core from 6 to 12 in. deep. The position of the samples relative to each station is shown by the diagram on the map (Fig. 1). Had it been possible at the outset to foresee that our work would be continued, the position of the samples might have been randomized ; as it is, those of 1942 are more closely grouped about the station point than those of 1943 and 1944. The holes made by the removal of the samples were immediately filled with soil brought from the edge of the field and were capped with turf, so as to interfere as little as possible with the wireworm population and its environment. Since one 4 in. sample was taken in each square yard, just under 1 % of the soil was abstracted from the sampling area about each station.

The wireworm collections made in Spinney Pasture are composed almost exclusively of Agriotes sputator. In the whole course of our sampling of this field, we have collected 440 adult Agriotes, of which 437 were of that species. We have also identified 2430 larvae, using the character provided by Guéniat (1934), and found 2406 of them sputator. Of the 2870 identified Agriotes, then, 2843, or just over 99 %, were A. sputator. This existence of a nearly pure population of A. sputator is interesting, especially in view of the fact that in the field known as Trinity 1, only 200 yd. to the south-east, the Agriotes population is mixed and other species make up more than half of it.

The basic data for the following study of field distribution, the number of wire-worms recovered from each station at each of the twenty-nine collections made between January 1942 and April 1944, are recorded in Table 1.

Analysis of these data shows a significant variance with regard both to the months of the year and to the stations of the field. The matter of seasonal occurrence will be dealt with in another paper. The spatial distribution, our present concern, shows so great a variance (F=10·9; 1% point 1·94) that it can be concluded with certainty that the wireworm population of Spinney Pasture is not distributed at random.

This is fortunate for the purposes of the present study, because, in so far as the distribution of the wireworms in Spinney Pasture is not random, it must be due to the operation of one or more controlling influences. The field therefore provides an opportunity for further investigation of the distribution of wireworms in space, an investigation, that is, bent on the discovery of the factors controlling the distribution.

It has first to be determined how many and which stations among the twenty differ significantly from the mean infestation of the field. The mean total collection from each station is 490 wireworms, and the standard deviation of the total collections from the twenty stations is 148. The total collections from only four stations exceed that standard deviation. Stations 1, 13 and 20, from which the collections differ from 490 by 427, 244 and 286 respectively, are well outside the limit; station 5, differing by 150, is only just beyond it.

But when these four stations are left out of account, and the collections from the remaining sixteen stations are submitted to an analysis of variance, that analysis still shows a highly significant variance among stations (F = 2·35; 1 % point 2·08). Therefore, while special attention can be focused upon four stations markedly abnormal in their infestation, the distribution of the wireworms in the remainder of the field may still be expected to yield information about factors controlling wireworm distribution.

Our task now becomes the quest for environmental factors which might explain, in particular, why four stations in the field should have wireworm populations so markedly different from the other sixteen stations, two very high and two very low, and, in general, why the wireworms are distributed in the field as they are. The factors that we have been able to explore are dealt with serially below.

A glance at the map (Fig. 1) shows that there is no local association between the two stations, 1 and 13, with low populations or between the two, 5 and 20, with high populations. Stations with higher or lower populations, respectively, come between the members of the two pairs. This indicates that we have to deal with four different parts of the field, not merely with two stations in a region of low and two others in a region of high infestation. It has first to be investigated whether anything in the position of the stations will explain their infestation.

Position with reference to the surrounding fields will not do. Station 20 is nearest to the field called Charity, across a ditch to the north-east, which has been continuously arable since 1933 and probably longer, and in which the examination of twenty 4 in. samples taken to a depth of 12 in. yielded only three wireworms. Station 20 has the highest infestation. Station 1 is equally near to Bedlam, the field on the west which, although including a 4-year ley in its rotation, also has a very low wireworm content (seven wireworms in fifteen 4 in. samples taken to a depth of 12 in.). Station 1 has the lowest infestation. The other stations near Charity, 14–18 and especially 19, and those near to Bedlam, 2 and 4, are not especially heavily or lightly infested. It does not appear that either a low or a high population in different parts of Spinney Pasture can be attributed to the influence of adjacent fields.

Station 1 is sufficiently close to the narrow strip of woodland on the south for a shadow to be cast on it rather late in the morning during the winter. Frost therefore lies later in the day on this station than on any other in the field. But its small wire-worm population can scarcely be attributed to this factor, for the other station with low infestation, station 13, is completely without shade and one of the most exposed on the field.

Altitude is a topographical feature that can be more precisely investigated. Through the kindness of Dr R. D. Davies and Mr T. R. C. Fox, of the Engineering Laboratory, Cambridge, we were able to make a detailed survey of Spinney Pasture. Six-inch contours are shown on the map (Fig. 1), and the exact station levels are included for reference in Table 2. A comparison of the wireworm infestation and the altitude of all stations shows a significant correlation (r=–0·455; 5% point 0·444). Some attention is therefore attracted to the position of the stations with regard to their altitude.

Finally, the position of the stations has to be considered in relation to the use of the field. For instance, in the summer of 1942, sow pens were placed in the southerly part, in the region of stations 2–13, but not about stations 1 or 14–20. We have not a sufficiently detailed history of the field to consider such factors directly, but their influence will be reflected in the soil analyses and in the flora and fauna. The effect of carting through part of the field is considered below in the paragraphs on soil consolidation.

The soil of Spinney Pasture is a sticky, tawny-yellow clay derived from the gault (cf. Nicholson & Hanley, 1936, pp. 34–41). Below the mat of grass roots is a layer 7 or 8 in. thick where the clay is mixed with humus and appears dark grey in colour. When it dries this grey soil becomes crumbly and forms small, hard lumps very difficult to break down in the sieves. This layer and the root layer together occupy the upper 10 or 11 in. of soil and represent the region of cultivation. Beneath is the yellowish clay subsoil, in some places very sharply demarcated, elsewhere separated by a transition zone of 1 or 2 in. where the yellow clay merges gradually into the grey upper region. The yellow clay is dusted and speckled in places with calcium carbonate, and it also contains sparse, usually small, flints and pebbles. Small fibrous roots penetrate the clay at least as deep as 24 in. and leave minute pores and crevices in which small arthropods can be found. The parent gault (horizon 2 of Nicholson & Hanley, 1936, p. 35) begins to be apparent in Spinney Pasture at an average depth of 26 in. and is nine-tenths pure blue gault at an average depth of 32 in. At three stations, 14, 15 and 19, a layer of very distinctive orange-coloured sandy clay lies over the gault.

The depths of the three horizons—the beginning of the tawny-yellow clay, the first appearance of the blue gault, and the surface of the nine-tenths pure gault—are recorded for each station in Table 2. Coefficients of correlation with the wire-worm collections are given at the foot of the columns. None of these horizons appears to be significantly correlated with the wireworm infestation.

A small amount of carting is done along the north-west side of Spinney Pasture. This raises the question whether the compression of the soil due to the weight of the carts might affect the wireworm population ; and the question is apposite because the cart tracks usually pass near stations 1 and 13, both of which have a very low wireworm infestation.

Through the kindness of Mr Ronald Ede, of the School of Agriculture, Cambridge, we have been able to investigate this factor quantitatively. Mr Ede lent us an instrument designed and used by C. Culpin (1936) for measuring the resistance of soil to penetration by a steel probe. We took records with the Culpin probe in March 1944, after a period of dry weather. At most stations, four records were taken, at 18 in. north, south, east and west of the station point. At four stations, 1, 7, 13 and 20, four additional records were taken, $112$ yd. in each cardinal direction from the station point. Each record was carried to a depth of 13 in. The instrument was calibrated against a weighing machine, and the recording (cf. Culpin, 1936, Fig. 2) was then read for each inch of depth.

A summary of the measurements is given in Table 2, where the means of the readings from 1 to 4 in. at each station are shown, and also those from 5 to 8 in., and from 9 to 12 in. The entries in the table are the means of thirty-two readings at each of stations 1,7, 13 and 20; and of sixteen readings at each of the other stations.

All stations show a greater resistance to the probe in the upper 4 in., the zone of roots, than lower down; but the measurements at this level seem to bear little relation to the carting, for stations 3, 4, 15 and 18 are much consolidated though far from the carting that may possibly have affected stations 2 and 13. In any case, no correlation is indicated between consolidation at this level and wireworm infestation. Station 1 is very consolidated at the 9–12 in. level, but this can scarcely explain its low infestation, for station 6 exceeds it in respect of deep compression, and station 13, the other station with low wireworm infestation, has the lowest resistance at this depth of any station on the field. Coefficients of correlation for wireworm infestation and soil consolidation at each of the three levels are shown in Table 2. None is significant.

In March and April 1944 we thrice measured the moisture content of the soil at the twenty stations at Spinney Pasture. Immediately after the samples had been removed, a cork-borer was pushed into the wall of the hole so as to remove a small sample of earth about $34$ in. in diameter and $114$ in. long. Two such samples were taken at a depth of 3 in. and were put together into a bottle. Similar samples were taken from the hole at a depth of 9 in. The soil moisture in these small samples was weighed and calculated as a percentage of the dry weight of soil. The means of the three measurements at 3 and at 9 in. at each station are recorded in Table 2.

We cannot claim to have explored, but merely to have glanced at, the possible relation between the wireworm infestation of Spinney Pasture and soil moisture. Of this variable factor, it would have been especially desirable to have taken measurements on each of the twenty-nine occasions when wireworm collections were made; and it is not surprising that our three measurements at each station show no significant correlation with the number of wireworms found there.

On the last three occasions when Spinney Pasture was sampled, in March and April 1944, duplicate samples were removed from immediately beside those examined for wireworms. The second samples were delivered, each as an upper and a lower core, to F. Hanley, M.A., of the School of Agriculture, Cambridge, who kindly arranged for analyses to be made of the soil. We are much indebted to Mr Hanley for his willing co-operation, without which this part of our paper could not have been written.

The analyses are summarized in Table 3, where each entry is the mean of the measurements from the three upper or the three lower cores, respectively, that were analysed from each station. At the bottom of each column is given the value of the coefficient of correlation between the means in that column and the total collection of wireworms from the corresponding station, as given in Table 1.

• (a) Hydrogen-ion concentration. This was measured electrically, using a glass electrode. The average pH of the upper cores was 8·01, and of the lower cores 8·24. At no station did the pH differ markedly from these means (upper soil, σ = 0·076; lower soil, σ = 0·065). The coefficient of correlation of the wireworm infestation and the pH of the top cores, however, is 0·483, which is slightly higher than the value of r at the 5 % point and suggests that there may be some relation between the two. No such relation can be based on the analysis of the lower cores, for which the coefficient of correlation is–0·151.

• (b) Loss on ignition. The entries in Table 3 represent the percentage loss in weight after ignition of the oven-dry soil, corrected for carbon dioxide from the carbonates but not, of course, for combined water. The mean loss of the upper cores from the twenty stations (11·2±0·80%) and that of the lower cores (8·3 ±0·92%) is a common value for old pastures and calls for no special comment. This factor affords some estimate of the amount of organic material present in the soil. If it is true that young wireworms can nourish themselves on decaying organic matter in the soil, as Langenbuch (1932, p. 295) has claimed, loss of weight on ignition might be expected to bear some relation to the wireworm infestation. For the upper cores, the coefficient of correlation between the wireworm infestation and loss on ignition at the twenty stations at Spinney Pasture is–0·626, which is highly significant but, surprisingly, of a negative correlation. For the lower cores, the correlation (–0·111) is non-significant.

• (c) Organic carbon. The percentage of organic carbon in the samples was measured by Walkley’s wet oxidation method. The mean value in the upper soil was 2·97 ±0·192%, and in the lower soil 1·49 ± 0·259%. This factor, like the preceding, is related to the amount of organic matter in the soil and might, on general grounds, be considered likely to influence wireworm distribution. It is interesting, therefore, to find it related to the wireworm infestation in a manner very similar to the loss on ignition. The coefficient of correlation for the upper soil (–0·597) is highly significant of a negative correlation; that of the lower soil (–0·252) is non-significant.

• (d) Nitrogen. This was determined by the Kjeldahl method. The mean for the upper soil at Spinney Pasture was 0·321 ±0·022%, and for the lower soil 0·196 ± 0·030%. Since the nitrogen in the soil is derived largely from the organic matter present, it is not surprising to find that this factor falls into line with the two preceding in its relation to the wireworm infestation. The coefficient of correlation for the upper soil at the twenty stations (–0·606) strongly suggests a relation between the nitrogen content and the wireworm infestation. That of the lower soil (–0·283) is non-significant.

• (e) Calcium carbonate. This was measured by means of a Collins calcimeter and is recorded in Table 3 as the percentage of CaCO₃ in the soil by weight. The mean CaCO₃ content of the upper soil at the twenty stations was 7·17 ± 1·63 %, and of the lower soil 8·92 ± 3·35 %. For neither the upper nor the lower soil does the coefficient of correlation, shown at the bottom of Table 3, suggest any relation between the wireworm infestation and CaCO₃ at this order of content.

• (f) Available potash. The method used for estimating the available K₂O was a modification of that proposed by Dyer. In this method the soil is treated with a 1 % citric acid solution, using an additional quantity of citric acid equivalent to the carbonate present in the soil, and the percentage of K₂O in the extract is then determined. The measurements are recorded in Table 3 as parts per ten thousand by weight. The mean measurement for the upper soil was 1·7 ± 0·35, and for the lower soil 1·2± 0·30. The coefficients of correlation, shown at the foot of the appropriate columns of Table 3, do not support any suggestion that available potash in the amounts found at Spinney Pasture has any relation to the wireworm infestation.

• (g) Available phosphoric acid. This was measured by treating the soil with a 1 % citric acid solution, as described above for the estimation of available potash, and then determining the percentage of P₂O₅ in the extract. The mean proportion present in the upper soil from the twenty stations was 2·8 ± 0·34 parts per ten thousand by weight; and in the lower soil 1·7 ± 0·28 parts. The coefficients of correlation between the measurements of P₂O₅ and the wireworm infestation, shown at the bottom of Table 3, do not support any hypothesis of a relation between the two at Spinney Pasture.

Before taking a soil sample, we cut off the growing vegetation close to the surface of the ground with scissors, in order to avoid the needless examination of material. During that process it was a simple matter to note the different species of plants, and on twenty occasions we made a quantitative estimate of the vegetation covering each of our twenty samples. The following observations can therefore be regarded as based on 400 ‘quadrats’ each of 12·6 sq.in.

In making our estimate, we assessed the percentage of the surface occupied by each plant species. The percentage area method has been criticized by West (1938) as inadequate for the analysis of the vegetation of pastures, but it would appear to be sufficient for our purpose, especially as our observations were made in every month of the year and therefore avoid much of the error due to seasonal foliation of the different species.

A summary of the vegetation at Spinney Pasture is shown in Table 4, where for each station there is recorded (1) the number of samples out of twenty on which the plant was present, and (2) the sum of the twenty estimates of the percentage area occupied. At the foot of each principal column is also recorded the coefficient of correlation between the entries in column (2) and the total collection of wireworms made at the corresponding stations.

Before the different plants and the twenty stations are considered separately, it may be noticed that, so far as the 400 samples represent the vegetation of the field as a whole, clover occupied 12% of the area, moss 12%, Lolium perenne 40%, Agrostis stolonifera 22 %, Dactylis glomerata 8 % and Festuca ovina 2 %. Nearly 2 % of the area was recorded as bare ground, and the re maining 2 % was occupied by a miscellaneous flora.

• (1) Clover. Under this name is included both Trifolium pratense and T. repens. Clover is generally distributed throughout the field. It was present on 232 of the 400 samples, and occupied 11 ·9 % of their total area. The coefficient of correlation between the area occupied at each station and the number of wireworms collected there does not indicate any correlation between the two.

• (2) Moss. Dr P. W. Richards kindly identified specimens of moss from Spinney Pasture as Brachythecium rutabulum, but it is likely that other species not submitted to him are included in our estimates. Moss occurred at all stations but was present on only 176 of the 400 samples. It occupied a slightly larger total area than clover (12·2%), but was much less uniformly distributed. Its distribution, however, does not seem to be correlated with that of the wireworm infestation.

• (3) Lolium perenne. This is the dominant grass over the greater part of Spinney Pasture. It occurred on 325 of the 400 samples and occupied 39·9% of their total area. A marked correlation exists between our estimates of the area it covered on our samples from the twenty stations and the number of wireworms collected there, but the correlation is negative (–0·651).

• (4) Agrostis stolonifera. This grass occurred on 192 of the 400 samples and occupied 22·5% of their total area, but it was less generally distributed than Lolium. At station 2 it did not appear on any of the twenty samples ; and at stations 1, 3, 7 and 13 it was present on only a few samples and over a small area of them. The coefficient of correlation between the area it occupied on our samples and the wireworm collections from the twenty stations is highly significant (0·590) and suggests a positive relation between the two.

• (5) Dactylis glomerata. Cock’s-foot grass was recorded on 112 of the 400 samples and occupied 7·8 % of their total area. It occurred at all stations but in such variable amounts that, although it covered at one station more space than Lolium and at three stations more than Agrostis, at five other stations it was found on only one or two samples of each twenty. Its distribution within Spinney Pasture does not show any correlation with the wireworm infestation.

• (6) Festuca ovina. Only at four stations, 2, 3, 4 and 19, was sheep’s fescue at all common ; at nine other stations, small amounts were found on one or two samples ; and at seven stations it did not occur. Over the whole area of the 400 samples, it occupied 2·2%. Our wireworm collections from the twenty stations show no correlation with the distribution of fescue.

• (7) Bare ground. Small patches of bare ground, up to 6 or 7 sq.in. in extent, occurred on seventeen of the 400 samples. The area of bare ground shows no correlation with the wireworm collections.

• (8) Other flora. In this column are included one or more observations of Cnicus arvensis, Bellis, Taraxacum, Potentilla, Poterium or Ranunculus, or of the grasses Hordeum pratense, Trisetum flavescens, Poa sp. or Cynosurus cristatus. These plants occurred on forty-four of the 400 samples and occupied 1·8% of their area. The distribution of this miscellaneous flora, treated as a whole, does not appear to be correlated with that of the wireworms at Spinney Pasture.

To make a census of all the arthropod material from a standard soil sample is a laborious and time-consuming process in which we have only rarely been able to indulge. Even to collect and count the macroscopic elements of the fauna multiplies by three or four the time required for the removal of the wireworms alone. In 1942 we were unable to spare the time necessary for this extra work, and merely noted in general terms the presence of some of the larger arthropods. From February 1943, however, we collected certain groups of arthropods consistently, and noted the number of individuals present in each core. Those collections are summarized in Table 5. The entries in the table represent the total number of individuals collected at each station from the sixteen samples examined between February 1943 and April 1944.

The question will be asked, how complete are these collections of animals other than wireworms ? Of the forms similar to Agriotes in size, it is likely that we collected all the individuals present in the samples. Tests made in the course of the development of our method (Salt & Hollick, 1944, p. 55) failed to recover additional specimens from the residues. Of the smaller forms listed in Table 5, the Staphylinidae and the Sciarinae, it is possible that our apparatus failed to recover some of the very young larvae. But on the occasions when we tried to collect the complete arthropod fauna, by adding to our apparatus sieves fine enough to retain the smallest Collembola and mites, we recovered no additional specimens of Staphylinidae and only two additional sciarine larvae. This gives us confidence that the kinds of animals listed in Table 5 were almost completely collected from the soil samples. It must be borne in mind, however, that the different habits of some of those animals may lead to a smaller proportion of their true population being present in the samples than is the case with wireworms. Many spiders living on the surface of the ground, for instance, probably made good their escape when the samples were being taken ; while a much higher proportion of the diplopod population may habitually live below 12 in. deep and so have escaped collection.

• (1) Chilopoda. These predaceous animals are said to feed principally on earth-worms and soft-bodied arthropods. There seem to be few observations of their feeding habits in nature, and it is at least possible that wireworms are included in their diet. We collected a total of 1146 centipedes from the 320 samples. This number, giving an average of 3·6 per sample, is surprisingly high for such large predators. Their distribution at the twenty stations shows a highly significant correlation with that of the wireworms (r = 0·610).

• (2) Diplopoda. Although millipedes are not predaceous, we collected them as being common and conspicuous members of the soil fauna. A total of 515 were found in the 320 samples. The coefficient of correlation (0·445) of millipedes with wireworms at the twenty stations is just significant at the 5 % point.

• (3) Spiders. These are surface-dwellers rather than a part of the soil fauna, but they must be considered as agents possibly affecting wireworm distribution because they have been recorded as preying upon adult elaterids (Subklew, 1938, p. 539). The distribution of the 284 individuals we collected in Spinney Pasture was probably not correlated with the wireworm population, the coefficient of correlation, 0·425, being a little less than r at the 5 % point.

• (4) Coleoptera. Carabidae and Staphylinidae are listed separately in Table 5 ; the other families of Coleoptera are lumped together. Neither the predaceous carabids nor the Coleoptera of various habits seem to be connected with the wireworm distribution. Among the staphylinids, the adults show no correlation with the occurrence of wireworms ; but the larvae, so far as our collections may be valid, are negatively correlated to a highly significant degree (r=–0·597).

• (5) Hymenoptera. Parasitic Hymenoptera, especially Proctotrypoidea, were collected, and also ants. The 129 parasites do not appear to be correlated with the incidence of the wireworms, but are perhaps too few for satisfactory demonstration. In the case of ants, some samples contained very many individuals, while others, though still within the territory of an ant colony, contained very few. We have therefore not used the number of individuals but rather the number of samples containing ants, as the measure of their distribution at the twenty stations. On this basis ants and wireworms occur together to a highly significant degree (r = 0·579).

• (6) Díptera. Only of Nematocera was a sufficient number of individuals collected to make-their listing worth while. Of that group, the Sciarinae were most common and are recorded separately in Table 5. The Sciarinae gave an example of their curious gregarious habit in the occurrence of 2595 larvae in a single sample. Whether that sample is included or not, the Sciarinae do not seem to be concerned with wireworm distribution. The other Nematocera occurred more frequently where wireworms were few, the coefficient of correlation (–0·642) being highly significant.

When we come to summarize the foregoing investigations, we find that of the forty-three correlation coefficients that have been calculated, twelve are significant. Nine of them are highly significant; those, namely, between wireworms and loss on ignition, organic carbon, nitrogen, Lolium, Agrostis, chilopods, staphylinid larvae, ants and Nematocera. The three that are significant only at the 5 % point are the correlations between wireworms and altitude, pH, and diplopods. This list is long, but it can be reduced by further analysis.

The three highly significant correlations found between wireworms and nitrogen, organic carbon, and loss on ignition probably represent three observations of the same phenomenon. The nitrogen in the soil is derived largely from organic matter there present, and the amount of organic carbon and the loss on ignition are in large part measurements of the same material. Calculation of the partial correlation coefficients shows that, after eliminating the effects of the other two factors, the correlation between wireworms and loss on ignition is–0·348, between wireworms and nitrogen–0·261, and between wireworms and organic carbon–0·033. It would appear, then, that in this complex of factors the greatest part of the correlation is due to that between wireworms and loss on ignition.

The next-significant correlations to be considered are those with the grasses Lolium and Agrostis. The former is negatively, the latter positively, correlated with the wireworm infestation. Since the vegetation of our samples was measured as the percentage area occupied, so that where one floral element was abundant others were necessarily scanty, it may be that one of these two correlations is spurious and merely represents the reciprocal of the other. We do not know of any observations indicating that wireworms are attracted by Agrostis or repelled by Lolium and have no biological grounds for a choice between the two correlations. On statistical grounds, however, since the partial correlation coefficient between wireworms and Agrostis after adjustment for inequalities in Lolium is 0·179, while that between wire-worms and Lolium after similar adjustment for Agrostis is–0·380, we are led to suppose that the negative correlation between wireworms and Lolium is the more important.

Of the five factors so far considered, the two that claim most attention, loss on ignition and Lolium, are both negatively correlated with the incidence of wireworms. The two are correlated between themselves to a highly significant degree (r = 0·615, with 18 d.f.). We should like to know whether the root system of Lolium is so abundant or so combustible as to control, in Spinney Pasture, the distribution of the loss on ignition, but we have not that information. What can be said is that the partial correlation coefficient of wireworms with Lolium in the absence of loss on ignition is–0·433, that between wireworms and loss on ignition in the absence of Lolium is–0·377. Also, that wireworms are known to occur in large numbers in black fen soil where the loss on ignition is much greater than the average of 11·2% found at Spinney Pasture. From these considerations it would appear more likely that the effective correlation is that between wireworms and Lolium, and that the correlations between wireworms and the three measurements of organic matter result from it.

Even if that suggestion (it has no higher status) is correct, the correlation between wireworms and Lolium may still be secondary. It may be due, for instance, to the preference for a Lolium habitat of some predator which in that particular habitat reduces the wireworm population to a lower level than elsewhere.

Among the faunal elements two, chilopods and ants, are positively correlated with wireworms, and two, staphylinid larvae and Nematocera (other than Sciarinae) are negatively correlated. When we try to evaluate the significance of these correlations we are at once confronted with a question we cannot answer. Shall we expect the numbers of an effective predator to be positively or negatively correlated with wire-worms in our collections or, indeed, will the numbers show any correlation at all? What little is known about the numerical interaction of predators and their prey (cf., for example, Lotka, 1925, and Gause, 1934) indicates that the numbers of a predator fluctuate after a lapse of time with those of the prey. But the information available has to do almost exclusively with the distribution of the two in time, especially over successive generations. Little of a quantitative nature is known about the relative distribution of the two in space; and nothing at all about their distribution in such a medium as the soil, where movement of the predator is impeded and where its perception of the prey must be limited to a relatively short distance. In looking for a possibly effective predator of wireworms in our collections, therefore, we cannot restrict ourselves to those members of the fauna for which we have found a significant correlation.

It is likely, however, that any predator that may control the wireworm distribution at Spinney Pasture is one that feeds on small wireworms. This is indicated by the characteristic size-composition of the wireworm population of the field. At Sheep’s Green, the wireworm population can be represented by a roughly triangular histogram (Salt & Hollick, 1944, Figs. 4, 5). At Spinney Pasture, the population is composed of an excessive number of very small wireworms and a small and only gradually decreasing number of medium-sized and large wireworms (Salt &Hollick, 1944, Fig. 6). The great reduction in wireworm numbers at Spinney Pasture occurs just before the larvae reach a length of 6 mm. If a predator is concerned with this, it need not be a large animal and it may be expected to be a small one.

For this reason, we do not put much weight on the correlation between wireworms and chilopods. Their comparatively large size and their elaborate apparatus of poison fangs would be wasted on prey so much smaller than themselves. It may be of interest to record that at Spinney Pasture the distribution of chilopods is correlated to a highly significant degree with that of the diplopods (r = 0·637, with 18 d.f.).

The positive correlation between the occurrence of ants and the incidence of the wireworm infestation at Spinney Pasture is supported by observations made at another field. There, a station situated in a region conspicuously populated by ants consistently gave high wireworm counts. The association, if it is real, must be an indirect one. Perhaps the presence of ants deters a predator.

The nematocerous larvae collected are almost entirely non-predaceous forms and cannot be supposed to affect the number of wireworms directly. We are rather surprised to find them negatively correlated with the wireworms. They are not significantly correlated with the occurrence of Lolium (r = 0·387) or with loss on ignition (r = 0·367, with 18 d.f.).

The final group that shows a highly significant correlation, the staphylinid larvae, includes many carnivorous and some parasitic forms. It might be supposed that they would be associated with dipterous larvae, but the coefficient of correlation between staphylinid larvae and Nematocera is 0·317, which is non-significant, whereas the partial correlation coefficient between wireworms and staphylinid larvae in the absence of Nematocera remains significant at–0·541. Moreover, these staphylinid larvae are about the size that would be postulated for an arthropod predator attacking wireworms about 4–6 mm. long. These facts, statistical and biological, give some basis for an hypothesis that the correlation between staphylinid larvae and wireworms at Spinney Pasture is no coincidence, but an effective association.

One group, not included in Table 5, remains in our minds as possibly concerned in the destruction of small wireworms. That is the Acarina. Mites are extremely numerous in the soil of Spinney Pasture. We have already recorded finding the population to be at least 200 million acarines per acre (Salt & Hollick, 1944, p. 63), and more recent work, still to be published, shows that estimate to be far too low. We have repeatedly found mites on medium-sized and large wireworms. The role of mites in the wireworm environment deserves serious investigation.

The correlation coefficients between wireworms and altitude, pH, and diplopods are significant only at the 5% level of probability. That between wireworms and diplopods can be dismissed as of little immediate interest, since these animals are not predaceous. Wireworms occur in soils of such very different acidity and alkalinity that we cannot put much weight on the observed correlation within the comparatively narrow range of pH at our stations. The correlation with altitude suggests a study of the relation between altitude, soil moisture and wireworms, possibly with a floral link interpolated; but our measurements of soil,moisture, as already explained, are regrettably inadequate.

It appears, then, that among the factors we have studied, those most clearly bound up with the wireworm distribution within Spinney Pasture are three: the amount of organic matter in the soil, the density of the grass Lolium, and the number of staphylinid larvae. Calculation of the partial correlation coefficients among the four variables—wireworms, loss on ignition, Lolium and staphylinid larvae—shows that, after eliminating the effects of the other two factors in each case, the coefficient of correlation between wireworms and loss on ignition is–0·335, between wire-worms and Lolium–0–388, and between wireworms and staphylinid larvae–0·473. Statistical analysis therefore points to the staphylinid larvae as the most important single factor we have studied ; but we are not satisfied that our data are sufficient on biological grounds for that conclusion, and prefer to consider all three of these factors as being worthy of further study.

For the present, that is as far as we can carry our search for the factors that control wireworm distribution within a field. Later in this paper we shall give reasons to suggest that further progress in the analysis of the problem will be slow until other considerations have been added to the discussion, namely, those concerned with the composition of the wireworm population, rather than merely its size.

The distribution of wireworms within a plot has been studied on part of a field known as Sheep’s Green, near the Fen Causeway, Cambridge. This is a low-lying common of about 7 acres, intersected by small streams and ditches, and sparsely planted with willows and Lombardy poplars. Some of the field is marshy, but most of it is under rough grass which is only lightly grazed and rarely, if ever, cut for hay. It is unlikely that Sheep’s Green was ever under cultivation.

Our work there has been concentrated in an area of about $14$ acre near the centre of the field, bordered on one side by a brook and on others by a tarred path and a shallow ditch. On this area nine stations were arranged, 5 yd. apart, in a row running north and south; and seven stations, similarly spaced, in a nearly parallel row 15 yd. to the west (Fig. 2). At each station a sample was taken at the middle of every month from July 1941 until September 1942. Extra samples were taken in June, July and August 1942. After an interval of nearly 2 years, further samples were removed in May, June and July 1944. From each of the sixteen stations, then, we have examined twenty-one samples.

The samples taken on each occasion consisted of cylinders of soil, 4 in. in diameter and 12 in. deep. Each sample was removed and examined in two parts, an upper core from the surface to 6 in. deep and a lower core from 6 to 12 in. deep. The position of the samples relative to each station is shown by the diagram on the map (Fig. 2). Holes made by the removal of the samples were immediately filled with soil brought from an adjacent part of the field, and were capped with turf, so as to interfere as little as possible with the wireworm population and its environment. The primary samples were spaced a yard apart, but the extra samples and those taken in 1944 came between them so that, over the whole period of 3 years, 1·3% of the soil was removed from the immediate sampling area of 15 sq.yd. about each station.

The elaterid beetles collected from these samples include species of Agriotes, Athous and Adrastus, but in this paper, except for a paragraph on p. 29, we are concerned only with the Agriotes. Of that genus we have collected from Sheep’s Green a total of 6921 individuals, among which there were fifty-one adults and 2542 larvae over 6-5 mm. long. All the adults and 2475 of these larger larvae have been identified; the other sixty-seven large larvae cannot be readily named because the critical parts are injured or missing. We have not identified larvae smaller than 6·5 mm. because it is difficult and takes longer to see the distinguishing characters in small larvae, but the wireworms above 6·5 mm. in length can be taken to be sufficiently representative of the whole population. Of those identified, fifty adults and 2333 larvae were A. sputator, one adult was A. obscurus, and 142 larvae were obscurus or lineatus, which we have not attempted to distinguish in the larval stage. It appears from these figures that about 94% of the Agrietes population of Sheep’s Green is A. sputator. But the whole field is not uniformly infested with that species, and we shall have occasion later (p. 29) to show that obscurus-lineatus forms a high proportion of the Agriotes population at certain stations.

More than half of our wireworm collections from Sheep’s Green were obtained from square-yard and square-foot samples with which we are not here concerned. From the twenty-one 4 in. samples taken at each station, we collected a total of 2317 Agriotes, as set out in Table 6. These are the collections on which the present discussion of plot distribution is based.

It requires no statistical treatment of the data of Table 6 to show that there are consistent differences between the sixteen stations at Sheep’s Green in respect of their wireworm infestation. An analysis of variance of the data confirms that the variation between the means for the sixteen stations is far greater than that between different samplings at the same station (F=9·21 ; 1 % point 2·11), and thus forbids any hypothesis of random distribution

Inspection of the table also shows that throughout the period covered by these samples the stations forming the easterly row were twice as heavily infested as those forming the’ westerly row. But this cannot be taken to indicate that the area is simply divided into two parts, one of high infestation traversed by the easterly row, and one of low infestation traversed by the westerly row. Closer inspection shows that one station of the westerly row, 16, has a high infestation comparable with that of most stations in the easterly row ; while the stations at the two ends of the easterly row, 1 and 9, each have a low infestation similar to most stations in the westerly row.

The sixteen stations therefore fall into two groups of eight. One group, comprising stations 2–8 and 16, includes most of the easterly row. This group provided 1916 wireworms; nearly 83% of the whole collection. These stations averaged 11·4 wireworms per sample, the standard deviation being 3·9. An analysis of variance on the 168 collections shows the difference between stations to be slightly greater than can be attributed to sampling variation, using a 5 % level of significance. The other group, comprising stations 9–15 and 1, includes most of the westerly row. This group provided 401 wireworms, only 17% of the total. These eight stations averaged 2-4 wireworms per sample, with a standard deviation of 0-91. Analysis of the 168 collections from this group of stations gives no indication of a significant variation between stations as compared with the variation between separate collections from a station.

Not only is the difference between these two groups of stations very marked, it is also very persistent. It appears, as can be seen in Table 6, at the beginning of the sampling, in July 1941, and it is a regular feature of each collection until September 1942. Then, after a period of 19 months when no samples were taken, the difference reappears in the same order of magnitude in the collections of May, June and July 1944.

We can conclude that there is a marked and persistent non-random distribution of the wireworms within this experimental plot of about J acre. It now lies before us to consider what factors might have brought about that distribution.

The immediate environs of our plot at Sheep’s Green seem unlikely to affect the distribution of wireworms within the experimental area. The brook that runs near the westerly row (Fig. 2) has a steep eastern bank, and the surface of the water in it is usually about 2 ft. below the general level of the ground. The brook is balanced on the other side of the plot by a shallow ditch which lies just off the edge of our map. It is usually waterlogged, and carries running water in wet seasons. The position of the tarred path and of the trees is indiscriminate with respect to the rows of stations.

The plot itself has no marked features. The distance between the two rows is 15 yd., and between adjacent stations only 5 yd., so that larger wireworm larvae presumably, and the adults certainly, are able to move from one station to another. From general considerations it would seem unlikely that topographical features vary enough within a level plot of less than $14$ acre to have much influence on the distribution of wireworms.

One topographical feature at Sheep’s Green, however, appears to belie that view. Six-inch contours, for the survey of which we are indebted to Mr T. R. C. Fox, are shown on the map. It is apparent that, on the whole, the westerly row passes across a shallow depression, while the easterly row runs over a plateau. Further than this, stations 1 and 9 lie in depressions, and station 16 on higher ground; so that the exact station altitudes, recorded in Table 7, show a highly significant correlation (r = 0·091) with the total wireworm collection from each station. No one would suppose that wireworms are influenced by 6 in. of altitude as such, and it will be shown later that this correlation is one of several bound up with the soil at Sheep’s Green.

The soil at Sheep’s Green is river alluvium and can be described as a shallow clay loam with heavy clay beneath. The loam is light brown in colour, very calcareous, and varies in depth from about 6 in. to at least 30 in. A mechanical analysis, kindly made for us by Mr Hanley, showed 12·8% coarse sand, 7·9% fine sand, 8·4% silt, 32·7% clay, 14·8% loss on ignition and 21·4% CaCO₃. The clay beneath is grey, mottled with reddish brown. Both layers contain many recent shells of which the most numerous are Limnaea pereger Müller and Bithynia tentaculata Linn.

The depth of the loam at each of the sixteen stations is given in Table 7. The considerable variation, from $712$ to 25 in., is noteworthy. When the depth of the loam is considered with reference to the measurements of altitude, it appears that the clay underneath lies relatively level, and that the surface contours are principally due to the varying depth of loam lying on the clay table. Calculation shows that there is a highly significant correlation (r = 0·683) between the depth of the loam and the number of wireworms at the sixteen stations.

There is no carting across Sheep’s Green and, although it is walked upon a good deal, there is no trodden path across our experimental plot. Measurements of the soil consolidation were made with the Culpin probe (see p. 7, above) 1 ft. north, south, east and west of each station point, and the records were read for each inch of depth to 12 in. There were therefore forty-eight readings at each station. The means of the sixteen readings from 1 to 4 in. at each station are given in Table 7, as also are those from 5 to 8 in., and from 9 to 12 in.

The mean resistance at all depths (24·6 lb.) was almost exactly the same as at Spinney Pasture, but, unlike that field, where the surface layer was most consolidated, the soil at most stations at Sheep’s Green provided a gradually increasing resistance as the probe went down. Coefficients of correlation between the resistance at the three depths and the wireworm population at each station are recorded at the bottom of the table. None is significant.

The amount of moisture in the soil at Sheep’s Green was measured in connexion with the samples taken in May, June and July 1944. The measurements were made by the same method as that used at Spinney Pasture (p. 8), and the weight of soil moisture was calculated as a percentage of the dry weight of soil. Means of the three measurements at 3 in. depth at each station are recorded in Table 7, as also are those at 9 in. depth.

At both depths the mean measurements show a highly significant correlation with the number of wireworms collected. The coefficient of correlation between wire-worms and soil moisture at 3 in. depth is–0·755; at 9 in. depth–0·628.

On the last three occasions when Sheep’s Green was sampled, in May, June and July 1944, duplicate samples were removed from immediately beside those examined for wireworms. They were delivered to Mr Hanley at the School of Agriculture, Cambridge, who very kindly arranged for analyses to be made of the soil. The methods used in this case were the same as those used for the soil of Spinney Pasture, and need not be detailed again. The results are summarized in Table 8, where each entry is the mean of the measurements of the three upper or the three lower cores, respectively, from each station.

• (a) Hydrogen-ion concentration. The average pH of the upper cores was 7·98 ± 0·04, and of the lower cores 8·31 ±0·04. The coefficients of correlation between the total number of wireworms collected at each station and the pH of the upper and of the lower cores are 0·029 and 0·040 respectively. Wireworm distribution does not appear to be correlated with pH within this range.

• (b) Loss on ignition. The mean loss on ignition of the upper cores from the sixteen stations was 14·8±·4%, which is 3·6% more than at Spinney Pasture. Of the lower cores it was 8·2 ±0·6%, practically the same as at the other field. The coefficient of correlation with wireworms is, for the upper cores,–0·234, and, for the lower cores, 0·007. Neither is significant.

• (c) Organic carbon. The soil at Sheep’s Green contained about twice as much organic carbon as that at Spinney Pasture. The mean amount in the upper soil was 6·27 ± 0·47%, and in the lower soil 2·56 ± 0·23%. The coefficients of correlation with wireworm infestation,–0·279 and 0·334 for the upper and lower soils respectively, are non-significant.

• (d) Nitrogen. The amount of nitrogen in the soil at Sheep’s Green, like the organic carbon, was about double the amount present at Spinney Pasture at both levels. For the upper soil, the mean measurement was 0·658 ± 0·042%; for the lower soil, 0·329 ± 0·079%. The coefficient of correlation for the upper soil is–0·335 and for the lower soil 0·238. Neither of these values is significant. It is interesting that these three measures of the organic material present in the soil appear to be uncorrelated with the incidence of wireworms at Sheep’s Green, although they are significantly correlated at Spinney Pasture. The difference will be discussed below.

• (e) Calcium carbonate. The soil at Sheep’s Green contains a great deal of CaCO₃; the mean value for the upper soil being 21·39 ±3·50% and for the lower soil 27·23±4·05%. It would appear that the wireworms tend to avoid the higher concentrations, for the coefficient of correlation between the number of wireworms at the sixteen stations and the CaCO₃ in the upper soil is–0·533, and in the lower soil–0·577. Both values exceed the 5% point of significance.

• (f) Available potash. The mean measurement for the upper soil was 1·1±0·3 parts and for the lower soil 0·7± 0·2 parts per ten thousand by weight. The co-efficients of correlation with wireworm infestation, 0·086 for the upper and 0·262 for the lower soil, do not suggest any correlation between the two.

• (g) Available phosphoric acid. The mean proportion present in the upper soil at the sixteen stations was 7·8±1·8 parts per ten thousand, and in the lower soil 6·6 ± 1·8 parts. The coefficient of correlation between the number of wireworms and the measurements of P₂O₅ are–0·316 for the upper soil and–0·425 for the lower soil, neither of which is significant.

Our study of the vegetation at Sheep’s Green was conducted on quite different lines from that at Spinney Pasture. Instead of making a quantitative estimate of the flora of each station, we set ourselves to find out whether there was a major qualitative difference in the plant cover of the easterly and westerly rows which might explain their very different wireworm populations. Having found such a difference, we extended our observations to see whether it held good for samples taken else-where in Sheep’s Green and in other fields.

The principal plant species concerned were the grasses Hordeum pratense, Dactylis glomerata and Agrostis stolonifera; thistles, Cnicus arvensis, and buttercups, Ranunculus acris, repens and bulbosus. Other species, including Phleum pratense, Lolium perenne, Festuca ovina, Poa sp., Trifolium repens and Crepis virens, occurred so much less commonly that they can be left out of the following summary account.

To compare the flora of the two rows in some detail, we examined the vegetation of twelve quadrats, each 1 yd. square, six along the line of the easterly row and six along the westerly row. The number of thistle plants (many of them small) was counted, and the percentage area occupied by buttercups and by the different species of grasses was estimated. The result, summarized in Table 9, shows that while the principal grasses differ little between the two rows, both the thistles and the buttercups were more numerous in the westerly row. This is especially noticeable in the case of Ranunculus, which occupied 23 % of the area of the quadrats in the westerly row but only 1 % in the easterly row.

This set of observations is supported by. another. As each of our soil samples was removed, the presence or absence of thistles and buttercups in its near vicinity was noted. Thistles were marked present if they occurred within about a yard, butter-cups if they were within a foot of the sample. This was done at each of our samplings except that of September 1941, so that we have twenty records for each station spread over a period of 3 years. The information is summarized in the first two columns of Table 10, where each entry represents the number of samples out of twenty on or near which the weed occurred. It is evident at a glance that both thistles and buttercups were much more plentiful in the westerly than in the easterly row. On the average, there were 5-8 observations of Cnicus per station in the easterly row and 11 in the westerly. For Ranunculus, the difference was even more marked; the means being 2·9 observations per station in the easterly row and 10 per station in the westerly row.

This clear difference between the two rows led us to investigate a little further the apparent avoidance of these two weeds by wireworms with a view to distinguishing between them. Pairs of 4 in. samples were taken in such a manner that one sample was immediately adjacent to a weed, and included part of its rootstock, the other was 1 yd. away to the west, or, if that site was occupied by a weed, 1 yd. to the east. The wireworms found in the two samples were then compared.

For Cnicus arvensis, twenty-one pairs of samples were examined, five pairs at Sheep’s Green, eight pairs at Spinney Pasture, and eight pairs at a field near Swavesey, Cambs. The twenty-one samples near the plants contained a total of 247 wireworms; those away from the plants contained 214. The differences between the twenty-one pairs are non-significant (t=1·09; P>5%). There appears to be no general correlation between the occurrence of thistle roots in a soil sample and the number of wireworms it contains.

For Ranunculus only eleven pairs of samples were examined, five concerning R. acris and six R. repens, all collected at Sheep’s Green. The eleven samples near the plants contained a total of thirty-eight wireworms, those away from the plants contained seventy-three. Although these totals seem to indicate avoidance of Ranunculus, the /-test shows that the differences between the eleven pairs are non-significant (t=1·77; P>5%). We wish that we had done more pairs of samples involving Ranunculus. So far as they go, these samples do not support any hypothesis that the low wireworm infestation of the westerly row was due to the number of buttercups there, and the superficial correlation may be due to a positive connexion between soil moisture and Ranunculus similar to the negative correlation between soil moisture and wireworms.

When we began regular sampling at Sheep’s Green we were without assistance and were unable to spare time for the collection of the fauna other than wireworms. Only for the samples taken in 1944 have we faunal records comparable with those of Spinney Pasture, and they are too few to provide a safe basis for detailed correlations. It is worth recording, however, that the group of eight stations having a high wireworm infestation gave an average of 8·7 chilopods, 2·1 Coleoptera (mostly Staphylinidae), 38 ants and 0·9 Diptera (mostly Nematocera) per sample per station. The other eight stations, having a low wireworm infestation, gave 5·2 chilopods, 3·8 Coleoptera, 3·6 ants and 3·1 Diptera per sample per station in the 1944 samples. For what they are worth these group averages show positive correlations between wireworms and chilopods and ants, and negative correlations between wireworms and the other two, just as we found at Spinney Pasture. The positive correlation between wireworms and ants is especially marked, not only in the actual figures available but also in the frequent mention of ants in the notes we made when collecting samples in which many wireworms were subsequently found.

The wireworm fauna of Sheep’s Green includes species of Adrastus and Athous as well as Agriotes. Of Adrastus we found only sixteen individuals; but of Athous sp., probably haemorrhoidalis, we collected 547. Most of the Athous were obtained from square-foot and square-yard samples, but exactly 200 were found in samples taken at the sixteen stations, among which they were distributed as shown in Table 10. At most of the stations a correspondence can be seen between the number of Athous collected and the infestation by Agriotes, but the correlation is spoiled by station 1, which provided more Athous than any other. The coefficient of correlation for all stations is 0·344, which is non-significant; but when station 1 is left out of account, the remaining fifteen have a coefficient of correlation of 0·687, which is highly significant.

One of the most interesting things about the distribution of wireworms at Sheep’s Green is the manner in which the different species of Agriotes occurred. From the samples taken at the sixteen stations we collected fifteen adults and 1008 larvae over 6·5 mm. long. One adult was an obscurus, the other fourteen were all sputator. Of the larvae, 843 were sputator and 123 were obscurus or lineatus, which we have not attempted to distinguish in the larval stage. The other forty-two larvae could not be identified with certainty because the critical parts were injured or missing. The distribution of the 857 identified sputator and the 124 obscurus-lineatus at the sixteen stations is shown in Table 10, together with the percentage of obscurus-lineatus among the identified individuals.

It is immediately apparent that the two groups of stations distinguished by their total wireworm infestation are markedly different in the proportion of obscuruslineatus in their populations. Stations 2-8 and 16, having a high wireworm infestation, average only 2% of obscurus-lineatus-, stations 9-15 and 1, having a low wire-worm infestation, average 52%. The negative correlation between the numbers of obscurus-lineatus and of sputator collected at the sixteen stations is highly significant (r =–0·727, with 14 d.f.). This unexpected result raises a number of problems which will be discussed below.

In the foregoing pages, it has been shown that the number of wireworms at the different stations at Sheep’s Green was correlated to a highly significant degree with the altitude, with the depth of loam overlying the clay, and with the amount of soil moisture present at 3 and at 9 in. depth. We have first to try to discover which of these factors, if any, really influences the wireworm distribution.

On the scale of our plot there can be no direct effect of altitude on wireworms, and to explain this correlation we must look for associated factors. They are easily found. As would be expected, the altitude of the stations was negatively correlated with the amount of soil moisture at 3 and at 9 in. depth (r =–0·839 and r=–0·800, respectively, with 14 d.f.). Altitude was also closely correlated with the depth of loam (r = 0·929), for the simple reason that the varying depth of loam lying on a clay table largely determined the surface contours of the plot. These facts allow us to neglect altitude, and to look for an effective factor in the depth of loam or the soil moisture.

The coefficient of correlation between the incidence of wireworms and depth of loam was ±0·683, that between wireworms and soil moisture at 3 in. depth was–0·755, and that between wireworms and soil moisture at 9 in. depth was–0·628. All of these are highly significant; but the three factors are so closely interrelated that each coefficient derives support from the others. The calculation of partial correlation coefficients allows a clear discrimination among them. After eliminating the effects, in each case, of the other two factors, the partial correlation coefficient between wireworms and depth of loam becomes only +0·022, that between wire-worms and soil moisture at 3 in. depth–0·476, and that between wireworms and soil moisture at 9 in. depth +0·195, the sign having changed. There is therefore no doubt that the most important factor of these three is the amount of soil moisture at 3 in. depth. This conclusion is supported by other calculations, including some in which the number of wireworms in the top cores only were correlated with the soil-moisture measurements at 3 in. depth, and those from the bottom cores only with the measurements at 9 in. depth. These latter calculations add the information that, in the bottom cores alone, depth of loam was more important than the soil-moisture measurements at 9 in.

Interest in the correlation between wireworms and soil moisture at Sheep’s Green centres about the fact that the correlation is negative. So much laboratory work has shown the dependence of wireworms on high humidity (Subklew, 1934; Evans, 1944) and the tendency for wireworms to migrate from dry to moist soils (Langenbuch, 1932 ; Lees, 1943 a) that one is accustomed to think of them as preferring very damp soils. These field data from Sheep’s Green (supported by the data from Spinney Pasture, although there the correlation does not reach the conventional level of significance) show wireworms to be much less numerous in the damp parts of the field than in the drier parts. This may have been due, of course, to the intervention of some other quite different factor—perhaps the Ranunculus we shall mention below—but we should like to suggest that the aeration of damp soils should be brought into consideration. Dr Lees tells us that in his experiments on the reactions of Agriotes to soil moisture (Lees, 1943b), he used sand 65% saturated, because, when the moisture content was higher, wireworms were asphyxiated. The sand used in his experiments was saturated when water weighing 23 % of the dry weight of sand was added. At Sheep’s Green, the average soil-moisture content at 3 in. depth at stations 1 and 9–15 was 34·3% of the dry weight of the soil. The saturation points of the sand used by Dr Lees and the soil at Sheep’s Green were doubtless very different. But the question remains whether wireworms living in soil of high moisture content, where, owing to a shallow soil overlying clay, their burrows are likely to be flooded, are liable to asphyxiation. The part played by aeration of the soil in the apparent effects of soil moisture on wireworms seems to us worthy of closer attention than it has yet received.

The correlation between calcium carbonate and wireworms is significant at the 5 % level for both the upper and the lower cores. This factor was not correlated with the wireworm distribution at Spinney Pasture, but there the average CaCO₃ content of the soil was only a third as high as at Sheep’s Green, This suggests that while wireworms are indifferent to chalk at lower concentrations, they may tend to avoid it when its concentration reaches between 20 and 30%.

In the plot at Sheep’s Green no correlation was found between wireworms and organic carbon, nitrogen, or loss on ignition, although at Spinney Pasture there was in each case a highly significant negative correlation. At Sheep’s Green each of these factors occurred at a much higher level than at Spinney Pasture—the percentage of organic carbon and of nitrogen was about twice as high. This affords evidence that wireworms do not avoid the organic matter in the soil as such. It bears out the suggestion made above (p. 16) that the correlation at Spinney Pasture was due to avoidance of the grass Lolium rather than of the organic content of the soil.

There was a large amount of Ranunculus on the westerly row of stations where there were few wireworms. Comparative samples taken elsewhere in Sheep’s Green, however, failed to show a significant correlation in general between the occurrence of buttercups and of wireworms. Unfortunately, our method of surveying the vegetation at Sheep’s Green, from row to row rather than from station to station, does not allow statistical comparison with the soil factors discussed above. We can only point out that the amount of Ranunculus, possibly in relation to the amount of soil moisture, is a factor worth further study.

The presence of Agriotes obscurus and/or lineatus at some of the stations at Sheep’s Green, and the highly significant negative correlation between the occurrence of obscurus-lineatus and sputator, raises several important questions. Does obscuruslineatus usually inhabit wetter soil with a higher proportion of weeds among the grass? Some such qualitative difference in the habitat of the species is clearly indicated by our data (Table 10). The absence or scarcity of obscurus-lineatus at some stations cannot be explained as an accident of colonization, for our samples were collected over a period of three years, ample time for migration or colonization across 5 or 10 yd. distance. When they occupy the same site, is obscurus-lineatus actually antagonistic to sputatorl Has a mixed population, including obscuruslineatus, a lower maximum density than a nearly pure population of sputator’, that is, would there be more Agriotes in the westerly row if there were not so many obscuruslineatusl If the answer to this last question is in the affirmative, it is possible that the distribution of gross numbers of Agriotes at Sheep’s Green, all the data with which we have been occupied, may be partly dependent on the difference in the distribution of the two species.

So far, then, as we have been able to carry our observations on the small experimental plot at Sheep’s Green, the factor most closely correlated with the infestation by Agriotes is the amount of soil moisture present at 3 in. depth. The prevalence of buttercup plants among the grass may also be important, but our evidence about this factor is less satisfactory. Much interest attaches to the occurrence of Agriotes ‘obscurus-lineatus’ at several stations, and its numerical preponderance at some stations where the total Agriotes population is very small ; but the general significance of the observation cannot be ascertained until more is known of the interactions of populations of obscurus-lineatus and sputator.

In order to obtain information about the micro-distribution of wireworms, we have examined four square-yard samples. Each of them included all the soil from an area i yd. square, removed to a depth of 6 in., in eighty-one separate parts, each 4 in. square. The careful removal of the eighty-one separate quadrats occupied several hours, and it is possible that there was some migration of wireworms across the boundaries of adjacent quadrats in the course of the work. But the effect of such movements on our results is probably negligible; in the first place because wire-worms tend to move up and down rather than sideways, in the second place because any lateral movements that did occur would be so ill-directed as to be practically random, and in the third place because it is doubtful whether even the large larvae could move through the soil quickly enough to cross more than one boundary before the adjacent sample was taken.

The first square-yard sample to be described, Square Yard A, was removed from Spinney Pasture half-way along its north-eastern edge. From that sample we obtained 1100 wireworm larvae and sixteen adults. The larvae were distributed in the different quadrats as shown in Fig. 3 a.

The question is, were the wireworms distributed at random, independently of one another, among the eighty-one quadrats? If they were, the numbers of wire-worms in the different quadrats should follow a Poisson series. That hypothesis is readily contradicted (χ² = 29·47 with 8 d.f.). The larvae are not distributed independently and at random within the square yard.

This result can be tested against the other three square-yard samples we have examined.

• (1) Square Yard B was taken in the eastern corner of Spinney Pasture near station 19. From it were obtained 1334 larvae, one pupa and forty-five adults of Agriotes. The distribution of the larvae among the eighty-one quadrats is recorded in Fig. 4a.

• (2) Square Yard C was taken at Sheep’s Green about 25 yd. south-east of station 9, and therefore well away from the experimental plot. It provided only 342 larvae and two adult Agriotes together with five Athous. The distribution of the larval Agriotes is shown in Fig. 4c.

• (3) Square Yard D was taken at Sheep’s Green, about midway between stations 5 and 15. We collected from it a total of 1139 larvae, sixteen pupae and two adult Agriotes, together with eighty-two individuals of Athous. The distribution of the Agriotes larvae is shown in Fig. 4c.

The coefficients of dispersion of the larvae in these three square yards are 3·33, 2·90 and 6·24 respectively. All are significantly different from unity and indicate that the larvae were not distributed independently and at random, and that in each case they were aggregated.

We can therefore draw the general conclusion that wireworm larvae in a grass field are not distributed at random, even within areas so small as 1 sq.yd., but tend to be aggregated into small groups.

Following the procedure adopted with the field and the plot, we should now go on to explore the environmental factors correlated with the distribution of wire-worms within a square yard. When we attempt to do so, we are faced with an entirely different set of considerations. Within a square yard, many of the factors that we have hitherto investigated cannot be supposed to vary enough to influence the distribution of wireworms. Differences in topography are absurd, and differences in many properties of the soil are negligible, in the matter of inches that separates the eighty-one quadrats. Those factors can certainly be left out of account. Vegetation and the fauna are left.

Before taking each of the four square-yard samples, we carefully mapped the vegetation covering it. The map of Square Yard A is shown in Fig. 3 c. But the vegetation observed on the surface of such a small area at any particular date bears little relation (1) to the root system beneath the surface, on which the wireworms feed; (2) to the vegetation of the area at other times of the year; and (3) to the vegetative cover that was present 1, 2, 3 or 4 years previously, when the different year groups of larvae were deposited and began to feed. We are therefore of the opinion that, although vegetative factors may be important in determining wire-worm distribution, their importance cannot be revealed, though it may be illustrated, by the comparison of quadrats in square-yard samples.

Similar considerations apply to the fauna. Predators capable of destroying wire-worms must be active enough to move easily through the soil of so small an area. Shall we expect them, at the particular moment of time when the sample was taken, to be numerous in a quadrat where wireworms are numerous, because they have moved there in search of prey ; or shall we expect wireworms to be few where predators are numerous because they have been eaten ? And surely the predators that ate the young wireworms in a quadrat 2 or 3 years ago have passed on, in more than one sense ! At this stage of the analysis, we consider it useless to attempt a formal correlation either between wireworms and vegetation or between wireworms and predatory animals, on the basis of these square-yard samples.

But the very considerations that seem to stifle the investigation in one direction, stimulate it in another. The difficulty of dealing with the vegetation and the fauna of small areas brings out in strong relief the dependence of present distribution on past history. It reminds us that to understand completely the present distribution of wireworms in a plot, we should have to be able to trace the movements of each larva and to discover the influences that brought it where it is. That we cannot do. But we can take a step towards it, and perhaps all that is needed for practical purposes, by considering the distribution, not simply of the total population of wire-worms, but of the several age groups included in it ; by having regard, that is, to the composition as well as the size of the wireworm population.

To this point in the account of our investigation we have been concerned exclusively with the observed distribution of wireworms. It is time now to turn our attention from the distribution to the processes that brought it about. Scrutiny of the problem suggests that within areas of the size of a field or smaller, three processes are involved. (1) Oviposition. Adult females, in four or five successive years, have selected the particular places in which they have laid their eggs. (2) Movement. The larvae hatching from those eggs were free to redistribute themselves, and some have moved about during 4 years, others during 2 or 3 years, but the youngest ones during only part of 1 year. (3) Survival. Many of the larvae originally hatched have died, and it cannot be assumed that the agents of death were regularly distributed.

These processes are obscured in any investigation that deals only with the total number of wireworms. But if data are available on the composition of the population, an attempt can be made to distinguish the separate effects of the three processes and to estimate their relative importance. For instance, the distribution of the recently hatched larvae, in August and September, must closely resemble the distribution of egg-laying, since these very small larvae cannot have moved far from where they hatched. Again, the distribution of the 4-year-old larvae must principally represent the effects of survival and movement, for those larvae have survived through 4 years and, during that time, have redistributed themselves in response to influences quite different from those that induced their parents to lay. For analysis of the processes of distribution, therefore, information is needed on the age as well as on the distribution of the wireworms.

Some, but not all, of the means for that analysis are now available. Miss J. F. Blacklocks (1944, and in unpublished work) has shown how larvae of Agriotes sputator can be separated into eight growth stages which represent, not absolute, but relative age groups. If all of our thousands of larvae had been assigned to those eight growth stages, we could proceed now to consider the distribution of each of the age groups they so nearly represent. Unfortunately, that laborious task is scarcely begun and is not likely to be completed in the near future. Nevertheless, a preliminary attack on the problem can be made. In an earlier paper (1944, p. 58) we have remarked that the size groups of larval wireworms show peaks which may indicate age groups. On the average, the smaller larvae will be the younger, and the larger larvae the older. All of our larvae are measured. We can therefore use the measurements that are available in place of the age determinations that are not, and, in a rough and preliminary fashion, push our analysis of the spatial distribution of wireworms a little further.

The collections of Square Yard A provide the first example. We have divided the 1100 larvae into three size groups, arbitrarily, but with guidance from the shape of the population-size histogram. This gives a group of 463 small larvae nearly a year old, for the sample was collected in June, before the new generation of eggs had hatched. They were distributed as shown in Fig. 3 d. The 392 medium-sized larvae, 2 and 3 years old, and the 245 large larvae, 3 and 4 years old, were distributed as shown in Fig. 3 e and f respectively.

The coefficient of dispersion of the small larvae is 2·52, which is highly significant of aggregation. That of the medium-sized larvae is 2·01, which is also significant of aggregation. But the large larvae have a coefficient of dispersion of only 1·21, which is not significantly different from a random distribution. The larvae, gathered together in groups when they are young, seem gradually to disperse and to reach a random distribution when they are old.

The other three square-yard samples give similar results. In each case, groups of small, medium-sized, and large larvae were separated arbitrarily, after reference to the size-frequency histogram. The distribution of the larvae comprising those groups is shown in Fig. 4 d–l, and the coefficients of dispersion for the twelve size groups of all four square-yard samples are set out for comparison in Table 11A.

In every case, the small larvae show a coefficient of dispersion very much greater than 1·318 and therefore highly significant of aggregation. The medium-sized larvae were also significantly aggregated, although the coefficients are lower than those of the small larvae. The large larvae, on the other hand, were distributed in three samples in a manner that did not significantly differ from random. In the fourth sample, Square Yard D, the coefficient of dispersion of the large larvae slightly exceeds the limit of probability of a random distribution, but it is likely that in this case we have set the lower limit of size too low, for the group includes 369 larvae out of a total collection of 1139. When only the 119 largest larvae are included in it, the coefficient of dispersion of this group falls to 1·06, which is indicative of a random distribution.

The coefficient of dispersion is not entirely satisfactory for the analysis of our square-yard samples. It serves for quadrats taken at random, and wastes the additional information to be obtained from the arrangement of our quadrats within the square yard. To meet this case, Prof. Fisher has suggested that we make an analysis of variance on the sums of squares between the eighty-one quadrats and remove the component for the 9 sq.ft. This discloses what share the larger grouping contributes to the total distribution. The coefficient of dispersion of the remainder can then be tested as having 72 degrees of freedom. The results of these calculations are set out in Table 11B.

The new calculations do not materially alter the situation. In each square yard there is a decrease in the value of the coefficient from the small larvae, through the medium-sized to the large. The principal result of the further analysis lies in the fact that, in almost every case, the new coefficient is smaller than the old. This indicates that the square-foot blocks contribute a disproportionate share of the aggregation: that the clumps of larvae tend to occupy a greater space than a single 4 × 4 in. quadrat. In the case of the medium-sized larvae of Square Yard C, indeed, the aggregation within the 9 square feet is so great as to reduce the coefficient of the remainder to non-significance.

These analyses of four square-yard samples provide a view of two of the processes of wireworm distribution. It appears that, as a result of the habit of the adult females of laying a number of eggs more or less together in a selected place, the young wireworms are markedly aggregated. In the course of time the growing wireworms disperse and, in a pasture, gradually assume a random distribution.

Two further comments are to be made. First, the date of collection of the square-yard samples affects the degree of aggregation of the group of small larvae. Square Yards A and C were taken in June, before the current season’s eggs had hatched. The small larvae of these collections had had nearly a year to disperse from the oviposition site. Square Yards B and D were collected in August. The smallest larvae in these samples had therefore had less than a month in which to disperse. To some extent these differences appear in Table 11 ; but our crude separation of the larvae into three size groups obscures what we expect will be much more evident when separation into age groups has been accomplished.

Secondly, the dispersion from oviposition clumps seems to end, under grass, in a random distribution of the older larvae. Where the feeding conditions are less uniform than they are in the upper soil of a pasture, the old larvae may regroup themselves into aggregations which are not relics of the oviposition clumps but new aggregations formed in favourable feeding sites. This is marked in arable land. In June 1943, forty samples were taken in a field sown with peas; twenty on the rows and twenty in the interrows. The samples from the rows contained 151 wire-worms, of which eighty-four were longer than 8 mm.; the interrow samples contained forty-seven wireworms of which twenty-one were longer than 8 mm. The row samples contained over three times as many wireworms as the interrow samples, but four times as many large wireworms.

A similar aggregation of large larvae at feeding sites may take place in pastures if especially palatable food occurs among otherwise uniform grass. That does not seem to be represented in our surface samples, but it does seem to have occurred in the one square-yard sample we carried to a lower level. Underneath Square Yard A, a second tier of quadrats was removed, from 6 to 12 in. deep. From these, eighty-one quadrats, 661 larvae and one adult Agriotes were obtained, additional to those in the upper 6 in. These larvae were separated into three groups with the same limits of size as those of the upper tier, and their distribution is shown in Fig. 3 g–i. The coefficient of dispersion of the small larvae is 2·88, which is highly significant of aggregation; that of the medium-sized larvae is 1·17, which indicates a random distribution; that of the large larvae is 138, which is again significant of aggregation. This suggests that, in the lower soil, beneath the general tangle of roots near the surface, the larvae disperse earlier in search of food, and then tend to concentrate about certain deeper root systems that are more attractive than others.

Applying our analysis of the processes of distribution to conditions at Sheep’s Green, we must first emphasize the marked and persistent differences in the wire-worm populations found at stations only a few yards apart. Stations 1 and 2 were only 5 yd. apart; some samples, indeed, only 1 yd. apart. Yet the total collection from station 1 was sixty-three wireworms; from station 2, 206 (Table 6). The only sample from station 1 that contained more than six wireworms was the sample of April 1942, the one nearest to station 2. A similar difference is found between the adjacent stations 8 and 9. Between stations 15 and 16 the difference is even greater, fifty-three wireworms as against 337; and two of the three samples from station 15 that contained more than five wireworms were samples taken 2 yd. south of the station point and nearest to station 16. There can be no doubt that both adult beetles and large wireworms are capable of moving 5 yd., the distance that separates adjacent stations. Either, therefore, adults avoid laying eggs at stations 1, 9 and 15, and larvae do not move to them, or most of the eggs and larvae fail to survive at those stations. The latter explanation seems to us untenable because, if eggs and larvae were at any season numerous at those stations, and subsequently died, our monthly (sometimes half-monthly) samples should have revealed that fact; which they did not. We must therefore conclude that adults and larvae both tended to avoid certain parts of Sheep’s Green as a matter of behaviour; the adults did not lay there, the wireworms did not move there to feed.

Since a crude separation of the wireworm larvae into three size groups has served to show something of the processes of distribution on the scale of a square-yard sample, it would seem logical to make use of similar divisions in our collections of larvae from the plot and the field. For two reasons, however, that is not possible. First, for statistical correlation with measurements of various factors, accurate numbers of the wireworms concerned are essential. The arbitrary separation into size groups that we have found satisfactory for a qualitative analysis would not be sufficiently precise for quantitative work leading to the calculation of a correlation coefficient. Secondly, with a group of samples taken all at one time, such as the eighty-one quadrats of a square-yard sample, a single arbitrary selection of size limits suffices to separate the larvae into the three groups. But for a collection of samples taken in every month of the year, such as those from the field and the plot, different limits would have to be fixed for each monthly collection to allow for the growth of the larvae. The accumulated error from the fixing of twelve arbitrary limits would be so great as to give no confidence in the size of the groups so fixed.

For these two reasons we cannot now reopen the problems of field and plot distribution and consider them again, as we should like to do, with regard to population composition. To overcome this difficulty by separating the larvae into their true age groups, and then to evaluate the three processes of distribution on a field and plot scale, would be an important step forward in the problems of wireworm distribution. Correlations between the youngest larvae and the various environmental factors would lead directly to a knowledge of the field conditions that attract the adult beetles to lay. Those between the old larvae and other factors would reveal something of the movements and much more of the survival and natural destruction of wireworms. There is little hope that our thousands of larvae can be examined and put into their age groups in the near future. But when that can be done, we shall be able to return to these problems of spatial distribution in another paper.

The distribution of wireworms on a scale larger than that of a field lies outside the limits of this investigation. However, in the course of our work we have made two sets of observations bearing on geographical distribution, and it will be more convenient to mention them here than to put them under another title.

It was shown by the National Wireworm Survey of 1939-40 that the wireworm populations of grass fields decrease from the south and east of England to the north and west (Adv; Entom. Conf. 1944). In the northernmost counties of England, the species of Agriotes cease to be important agricultural pests ; and in the southern half of Scotland they become increasingly uncommon. We can find no records of sputator or lineatus north of the Grampians, and only four of obscurus, three from Aviemore and one from Nethy Bridge, kindly communicated to us byMrD. K. Kevan.

Consideration of the life history of wireworms and of what is known of their temperature relationships, together with the climate of the British Isles, led us to think it likely that the northern limit of Agriotes might be set by the minimum temperature required for pupation.

Briefly, the argument is this. Temperature could only be efficient at the northern limit if it were lethal in winter to the larvae or adults, or if it were too low in summer for the egg or the pupal stage. In view of the low temperatures that wire-worms are able to withstand (cf. Falconer, 1945), and considering that the average temperature of Cambridge and Aberdeen is the same in December and January and that the daily minimum temperature is actually lower at Cambridge than at Aberdeen from November to March (Bilham, 1938, pp. 316, 327), the first alternative can be put aside. As for the second, eggs are laid in the region of Manchester from 1 May (Miles & Cohen, 1941), long before the warmest time of the year. Egg-laying and hatching do not seem to require high temperatures. Pupation, however, takes place only when the soil is at its maximum temperature, and the pupal period occurs in late July and during August. In those two months, the daily maximum temperature of the air at Aberdeen is 9° F. lower than that of Cambridge ; and the soil is warmed by sunshine at Cambridge for 6·1 hr. a day on the average, but at Aberdeen for only 4·7 hr. (Bilham, 1938).

With these considerations in mind, we set up a simple experiment, intended to be preliminary to further work. Early in June, 176 large Agriotes larvae, apparently full grown and ready to pupate, were isolated in rearing tubes; and forty-four (thirty-four sputator, ten obscurus-lineatus) were put into each of four constant-temperature rooms, at 10, 15, 20 and 25° C. It is unnecessary for our present purpose to describe the methods and results of the experiment in detail, and it will suffice to record that within the next 5 months, thirty larvae pupated at 25° C., twenty-one at 20° C., six at 15° C. and none at 10° C. Of the six that pupated at 15° C., three were obscurus and three were sputator.

More critical work remains to be done. So far as it goes, this preliminary experiment sfiows that a soil temperature somewhere between 10 and 15° C., and probably near 15° C., is required for pupation. It may well be that on account of the cooler summers and the shorter periods of sunshine in the northern half of Scotland, the soil there does not reach and maintain the necessary temperature.

The second matter has to do with the geographical distribution of particular species of wireworms. In the report of the National Wireworm Survey (Adv. Entom. Conf. 1944, p. 26) information is given of the relative abundance of obscurus, lineatus and sputator in the several Advisory Provinces. These data, based on the collection of adult beetles, show that obscurus is the dominant species to the north and west, and lineatus to the south and west, and that sputator reaches its highest proportion in the east and south-east of England.

This last is the region in which wireworm populations as a whole reach their highest numbers. So far as the data go, therefore, they indicate a correlation between a high proportion of sputator and high wireworm populations.

It will be noticed that this is the same correlation as that found at Sheep’s Green. On a geographical scale, as well as on a $14$-acre plot, high populations of Agriotes occur where there are most sputator, lower populations where obscurus (or obscuruslineatus) is dominant. This supports the suggestion made above (p. 31) that sputator reaches a higher maximum population than the other species, or than mixed populations.

The coincidence can be carried a step further. The parts of England where sputator reaches its highest proportion are the driest parts of the country. The stations-at Sheep’s Green where sputator formed the high wireworm populations were the driest stations. There appears, therefore, to be a general agreement between high wireworm populations, the prevalence of sputator, and low moisture, both on a geographical scale and within a $14$-acre plot. A comparative study of populations of sputator and of obscurus-lineatus in relation to soil moisture seems to be needed.

A paper on the spatial distribution of wireworms might be expected to deal with three principal questions. The first, ‘What is the distribution?’ requires a descriptive answer. The second, ‘What caused that distribution?’ calls for analysis. The third, ‘How did the observed distribution come about?’ requires again a descriptive answer, but this time the description not of a state but of a process.

The spatial distribution of wireworms has been described above as it was found in a field, a plot, and in four areas of 1 sq.yd. The descriptions provided are, of course, peculiar to those particular areas, and to the particular period or time of sampling, but on two matters they provide general information.

First, in each of our experimental areas, the wireworms were distributed in a non-random fashion. Within a grass field of 8 acres, certain parts were consistently more heavily infested than others; and during 28 months, one part had a wireworm population only one-eighth as large as that of the remainder of the field. Within a grass plot of $14$ acre, the distribution was strikingly non-random over a period of 3 years; some places being consistently three or four times as heavily infested as other places only 5 yd. away. Within each of four square-yard samples examined at a particular time, the wireworms were aggregated into groups to an extent far greater than can be attributed to chance. All this is not to say that wireworms in general are not distributed at random, although that may seem likely. What it does show is that upon none of these scales, neither within a field, nor within a square yard of it, can wireworms be assumed to be distributed at random.

Secondly, in our square-yard samples, the small larvae were markedly aggregated, the medium-sized larvae more dispersed, the large larvae distributed at random. Since fields are but mosaics of such small areas, the distribution of wireworms under grass can be described in general as forming a sort of pattern—a group of small larvae here, a less compact group of medium-sized larvae there, with the large larvae more generally distributed among them—a pattern which is repeated again and again over the field, differing in the number and spacing of its elements from place to place so as to be never quite the same although retaining an essential similarity. This is a zoological example of a phenomenon that botanists have already met and that A. S. Watt (1943) has recently described as a ‘pattern’.

The distribution of wireworms in an area having been described and shown to be non-random, the second question arises, ‘What caused that distribution?’ An answer to this question involves analysis of the wireworm environment, and study of the many factors composing it. That is so large a subject that no single investigation could be expected to lead to a complete answer. In the present study we have measured a number of edaphic factors in each of two larger areas and have subjected the measurements to statistical analysis so as to show up the factors most likely to be important (summaries, pp. 16 and 29). But not until those factors, and others, have been studied on many other fields in different parts of the country will it be possible to define with any certainty the conditions that govern the distribution of wireworms.

Since some factors of the wireworm environment have been the subjects of detailed laboratory studies, we might be expected at this point to relate our findings to those studies. For several reasons that would be unprofitable. In the first place, few of the published data refer specifically to Agriotes sputator, the species with which we have been mainly concerned. Observations reported above (p. 31) show that it may be quite incorrect to assume sputator to be physiologically similar to other Agriotes, let alone to other unspecified ‘wireworms’. Secondly, so far as the published data concern distribution, they deal principally with the limits of tolerance of wireworm larvae for certain factors. But some of those limits are scarcely ever met in nature in England and, in any case, the problem of wireworm distribution within fields or from field to field requires information, not of the limits between which wireworms can exist, but of the conditions between those limits that encourage or discourage the building up and maintenance of high populations. Thirdly, and most important, practically all the published data have to do with large wireworms. Little or nothing is known even about the limits of tolerance or the behaviour of the small larvae, or of their ability to move away from adverse conditions. And yet these small larvae comprise a half or three-quarters of a natural population. Finally, the conditions controlling the distribution of wireworm populations in nature are unlikely to depend upon any single factor, but rather upon a much more subtle combination of many factors. In view of these considerations, it is not surprising that the experimental work available, excellent as some of it is, does not go far towards answering the question under discussion.

The third and final question that we might be expected to deal with in discussing wireworm distribution is ‘How was the observed distribution brought about?’ To that question we have devoted a separate part of this paper (p. 36), where we have suggested that three distinct processes are involved—oviposition, larval movements and survival.

We have been able to demonstrate the first two of those processes in a preliminary fashion, as they affect the micro-distribution of wireworms. Further than that we cannot go at present. Throughout this investigation of the distribution of wireworms we have naturally been involved with some of the major problems and difficulties of ecology. This final question, about the processes of distribution, brings us squarely up against the greatest of them—the impossibility of understanding distribution without having regard to succession. This is a difficulty that pervades the whole subject, but our investigation of the micro-distribution of wireworms affords an example of classroom clarity. A group of eggs is laid and the larvae gradually disperse, like the ripples from a stone dropped into water. But when the dispersion is only one-quarter completed, another group of eggs is laid, not usually in exactly the same place, and a second wave of dispersion begins. At any one time, at least four such dispersions of larvae are taking place, and the relic of a fifth can be found in the distribution of the pupae or of the adults in their pupal chambers. These waves of dispersion, naturally, are not independent; they interact, and further complicate the situation that the observer finds when, at a particular moment of time, he comes to describe the distribution. What he finds, expressed as a number of individuals distributed thus and thus in space, can have little meaning unless time is introduced into the description and the evolution of the observed distribution is traced.

In short, spatial distribution and seasonal distribution are inseparable, and any study of the one without a study of the other is bound to be incomplete. For that reason this paper on spatial distribution cannot be brought to its real conclusion until another paper, on the seasonal distribution and population composition of our collections, has been written.

Several acknowledgements of assistance in particular matters are made in the text; here we must thank a number of benefactors without whose consent, co-operation, and assistance this work would have been impossible. Mr W. S. Mansfield, Director of the University Farm, and Mr G. W. Teasdale, Cambridge Borough Surveyor, kindly gave us permission to carry on this investigation in fields under their charge. We are very grateful to Prof. J. Gray for allowing us to continue the messy work of examining soil samples in his Laboratory for over four years; and to Dr F. Kidd for providing us during that time with refrigerator space in the Low Temperature Research Station. Mr F. Hanley and Mr D. J. Finney have given us invaluable help. Mr Hanley not only arranged for the chemical analysis of our soil samples, but also advised us in matters to do with the soil, and he has read the relevant parts of this manuscript. Mr Finney has been our mentor in statistics, and we owe a great deal to his patience. He has very kindly read and criticized the whole manuscript. This long-continued investigation could not have been carried on without a grant from the Agricultural Research Council which paid for materials, transport and assistance.