Abstract

The part played by two species of pompilid wasp in population changes of P. antipodiana was investigated. A correlation of r = 0.7 between the body length of the wasp and the captured spider was found. The wasps tended to be selective, mainly capturing medium sized spiders. The two wasp species showed some niche differentiation, with the large black hunting wasp (Salius monachus) particularly favouring P. antipodiana as prey. Young spiders were generally safe from attack as their tunnels were too small for most wasps to enter. Spider mortality rates due to wasp attack varied from 12-30%. Wasp mortality rates directly attributable to spider bite was as high as 11%. Spider survival was found to be attributable to six tactics of behaviour of which the most significant was the spider failing to respond to web stimuli. The hypothesis is advanced that the main ecological impact of the wasp is to reduce intraspecific competition among spiders. The long term implication of some spiders surviving to breed for six or more years is also considered.

Recapitulation, Part I

In part I of this study (Laing, 1978) on populations of the tunnel web spider Porrhothele antipodiana, several features were described. Among them was the decline in numbers evident in most populations over the summer period. These declines coincided with the activities of pompilid wasps of the genus Salius.* A hypothesis suggesting that^** page 2 the wasps were responsible for the declines in spider numbers could be supported by the following facts:

Because the summer population declines shown by the spider populations could be as high as 30-40%. it was considered that the activities of the wasps were worthy of closer study as an aid in understanding the population dynamics of P. antipodiana.

The Wasps

Two wasp species were active around P. antipodiana webs over the summer months in Wellington. Both were members of the family Pompilidae (Psammocharidae). The large black hunting wasp, Salius monachus (plate 1), is heavily built and gives the impression

Plate 1: The black hunting wasp, Salius monachus. This individual of 18 mm body length was captured on Johnson's Hill, Karori, Wellington. The burrow shown in plate 2 was the work of this particular wasp. Even though the red hunting wasp Salius wakefieldi may also capture tunnel web spiders, the black wasp is normally responsible for most captures of these spiders.

page 3 of being well protected against piercing weapons such as the chelicerae of spiders. It also has a very smooth surface to the thorax and the abdominal plates. The significance of this feature will be discussed later in the section on spider-wasp encounters. The red wasp, Salius wakefieldi,^* is a somewhat more lightly built wasp than the black species. Measurements of the thorax of 20 black wasps and 20 red wasps gave the following figures:

The ratios of thorax length : thorax depth; and of thorax length : thorax width calculated from the above measurements were as follows:

It is apparent from these ratios that S. monachus is definitely a stouter wasp than S. wakefieldi. This fact may have some bearing on the types of spiders which are captured by each wasp, and reference to this is made in the later section on the partial niche separation of the two species. Individuals of both species may grow to a body length in excess of 20 mm, but the majority of those in the 20-25 mm range are black wasps.

In the Wellington areas studied, the wasps usually made their appearance in mid-October, though for those years when cool springs were experienced, early November was the time of appearance. Capture of tunnel-web spiders was seen between mid-November and February as shown below for 24 observations in Wellington (1974-77):

Adult pompilid wasps are nectar feeders. The spiders that they capture are sealed in burrows along with the wasp's egg in order to provide food for the developing wasp larva. It is because of this behaviour that the exact ecological status of the Pompilidae is not clear. Their relatives, the ichneumon wasps, have been called parasitoids due to their habit of laying eggs in the living tissues of host. The ichneumon larvae destroy their host, and so are not true page 4 parasites. The Pompilidae are not usually included among the parasitoids because their host /prey is paralysed by stinging before being fed to the larvae. Evans (1963, 1964), refers to the Pompilidae as predators, and to their hosts as prey, and these descriptions have been widely used. However, the fact that the adult does not feed on the prey but instead feeds as a herbivore, makes this categorisation rather different from the normal use of the term predator. The broad definition of the parasitoid as given by Solomon (1969) could accommodate the Pompilidae, whereas Andrewes (1969) leaves the question very much open. It is clear that both the terms parasitoid and predator do present some difficulties; perhaps there is some justification for using a term like ‘predatoid’ as suggested by Evans (1963).

When the spider has been captured by the wasp. one or more of its legs may be chewed off, possibly to make transport to a prepared burrow easier. It is at this stage that the wasp may be observed

Plate 2: Burrow of Salius monachus, Johnson's Hill, Karori, Wellington. Location was on a low clay bank. Burrow diameter was 12mm. The crumbled clay below the entrance is characteristic of burrows dug by this species.

page 5 drinking fluid issuing from the spider's wounds; but this is the closest the adult wasp comes to feeding on the spider. The spider is then dragged to the prepared burrow where an egg is laid on its abdomen. Finally, the burrow is carefully sealed off by the wasp. The black wasp usually excavates its burrows in a low embankment (plate 2) or a clump of grass, while the red wasp excavates in soil, or according to Miller (1971) may utilise a borer hole in a log.

Size Relationships between Wasp and Spider

The fact that hunting wasps often capture spiders far larger than themselves has been noted by most writers on the subject. Rau and Rau (1918) describe seeing a wasp dragging a spider at least five times its own weight; Petrunkevitch (1926) found that Pepsis marginata was paralysing Cyrtopholis portoricae individuals which were eight times heavier than itself. Andrewes (1969) writes of the spider being up to ten times the weight of the wasp in these associations.

In New Zealand, Quail (1903), who appears to have been the first to publish material on the Porrhothele/Salius association, described an example where the spider was grossly larger than the wasp. Miller (1971) mentions very much the same point in connection with these two species. There can be no doubt that these events do occur, and that they are quite striking visually. However, it is more usual in the New Zealand species for the wasp body length and the body length of the captured spider to be similar. To illustrate this point, a scatter diagram (Fig. 1) was drawn up from measurements of 24 wasp/captured spider pairs. These 24 sets of figures were obtained in Wellington between 1974-77, and in each case both the wasp and the spider were accurately measured. The correlation of 0.7 between wasp body length and captured spider body length is rather high for biological data, but is probably accounted for by the small sample size. For a sample of 24, the correlation of 0.7 is significant at the 1% level, and so it would be justifiable to claim a definite relationship between wasp body length and spider body length for the S. monachus/P. antipodiana association.

In terms of body weight, the New Zealand species are similar to the examples quoted earlier. An 18 mm S. monachus weighs around 0.2 g whereas an 18 mm P. antipodiana weighs around 1.0g. The weight of the spider is commonly five times that of the wasp in this association. In the case of an 18 mm S. monachus capturing a 23 mm P. antipodiana, the spider may be as much as ten times the weight of the wasp.

In this sample of 24 captures it was found that the spiders most often taken by the wasps were those in the 15-19 mm body length range:

It was felt that some confirmation of prey size favoured by the wasps was called for. The above figures were derived from observed captures only and hence represent only a small proportion of the captures that take place, most captures not being observed. Figures for the numbers of spiders missing from their webs over the summer period, and assumed to have been captured by wasps, are shown below. They were obtained from three different sites near Wellington.

From the size distribution of spiders at these sites in late October (prior to hunting wasp activity), the expected size distribution has been calculated, assuming that wasps had been unselective in their choice of spiders. Comparison of these expected values with the number of empty webs observed, shows that the wasps were clearly selective, again preferring spiders in the medium size range (Chi-square = 24.3; P < .001).

Comparisons of Behaviour of the Two Wasp Species

When both species of Salius were found active in the same area, direct comparisons of their hunting behaviour could be made. The first difference to strike the author was that the black wasp was a vigorous explorer of sheet webs and tunnel entrances, and would often disappear into the tunnel for up to ten seconds. The red wasp, on the other hand, paid more attention to gaps such as those between stones, under wood, and in crevices. It certainly walked over the sheet webs of P. antipodiana but did not spend as much time investigating the tunnel entrance or its interior as the black wasp.

The significance of these broad behavioural observations shows up when the types of spiders they capture are compared. When both wasp species were present in the one locality. and when both P. antipodiana and the brown grass spider Miturga were also present then the following pattern occurred regarding spider captures: page 7

Fig. 1: Scatter diagram showing the relationship between body length of Salius monachus and body length of P. antipodiana captured by these wasps. N = 24 and correlation coefficient = 0.7.

It is clear that at this particular site we can accept the hypothesis that there is a genuine difference between the two wasp species in their choice of spiders (Chi-square = 25.9; P < .001). Ecologically this means there is at least a partial niche separation which reduces interspecific competition to a minimum when both wasps and both spiders are found in the same locality. However, when Porrhothele is not present, S. monachus mainly captures Miturga, while S. wakefieldi will take Dolomedes and lycosid spiders in the absence of Miturga. Thus generalisations on the food selection by these two wasp species cannot be made with much confidence.

The importance of the foraging behaviour in bringing about the partial niche separation must be recognised. Behaviour which leads page 8 to investigation of the underneath of rocks and logs, with much time spent searching into crevices is more likely to result in Miturga captures. On the other hand, behaviour which leads to much investigation of webs is likely to lead to Porrhothele captures. The origin of these behavioural differences can only be postulated, but two suggestions can be put forward: (i) that each wasp species learns where to look for the type of spider it fed on during the larval stage. Thorpe (1963) thought it probable that the early olfactory cueing at the larval stage could be what directed the adult to the appropriate host in some Hymenoptera. (ii) That each wasp species inherits a slightly different combination of kinesis or taxis components and this directs their foraging in slightly different directions.

These figures on prey differences can be related to the information on the wasp characteristics presented earlier in the article. The stouter bodied S. monachus specialises in hunting P. antipodiana; it needs the stouter body to withstand this spider's long, powerful fangs which have more penetrating power than the small fangs of Miturga and similar spiders.

The Determinants of Prey Size

What determines the size of spider that any given wasp is able to capture? The author is of the view that the diameter of the spider's tunnel places the lower limit on how small a spider can be taken, due to the restriction placed on larger wasps entering the tunnel. The upper limit is likely to be due to how readily the wasp is able to subdue a spider much larger than itself. Alternatively, there is some evidence (dealt with in the section on spider survival tactics) that very large spiders may be avoided because of the problems they cause the wasp in dragging them to a burrow, and then in fitting them into the burrow.

Using information gathered mainly during the study on a crib wall population at Johnsonville, figure 2 was drawn up to show the relationship between wasp body-length and size of spider which was likely to form the prey. The red wasps at that locality had a range of from 10-17 mm body-length with a mean body-length of 13 mm. The black wasps there had a range of 13-21 mm with a mean body-length of 16 mm. The smallest red wasps were able to enter the tunnels of spiders of approximately their own body-length. However, as only a few of the wasps were this small, the young spiders (being one year old at this time) were not under much hunting pressure. The tunnels of the second and third year spiders (spiders mainly in the 15-20 mm range) were able to be entered by most of the red wasps and all but the very large black wasps.

The outcome of these size relationships was that in most cases the first year spiders were safe from attack, as the capture statistics presented earlier have shown. The second year spiders, because they page 9

Fig. 2: The likelihood of wasp attack according to the size of the spider. The very small and the very large spiders are far less likely to be subject to wasp attack; while the spiders in the 12-23 mm range are likely targets.

were susceptible to attack from both wasp species, suffered the highest capture rate as shown in the capture statistics. The large spiders were open to attack by all sizes of both wasp species; the fact that they were not taken as often as the medium spiders suggests that some selection was occurring, and this point is taken up later in the article.

Mortality Rates

2. Wasps

Entomologists who have written on the topic of hunting wasps and their spider prey have usually emphasised the ease with which the wasp subdues the spider. Rarely has any consideration been given to any mortality suffered by the wasp in these encounters. One could be forgiven for thinking that entomologists feel obliged to champion the cause of their wasps. There can be no doubting that the Pompilid wasp is a difficult insect for most spiders to subdue. Ambrose Quail, writing in 1903, put it very succinctly when he described S. monachus as a ‘regular Ned Kelly’, pointing out how impregnable the armoured body of the wasp must be to the spider's fangs. The body of the Salius wasps is not only very tough, but it is also shiny. This makes it doubly difficult for the spider's fangs to grip and pierce the wasp. During encounters between P. antipodiana and S. monachus which were observed at close quarters, the spider's fangs could be heard scratching over the wasp's body, unable to grip or penetrate. There is, however, at least one point in the wasp's body where it is vulnerable—and that is the junction between the head and the thorax. It must occasionally happen that the spider's fangs do slip into this joint, and in plate 3, an example of this is shown. The venom of P. antipodiana is certainly powerful enough to kill the Salius wasps. In experiments conducted on the German wasp V. germanica, a P. antipodiana bite lasting for two seconds immobilised the wasp in a little over five page 11

Plate 3: The wasp is not always successful in its encounters with tunnel web spiders; in this example, a female P. antipodiana, body length 22 mm, was found feeding on the headless body of an S. monachus female, body length 17 mm. It is most likely that the spider's fangs penetrated between the head and the thorax of the wasp in this encounter, for this was the area the spider was found feeding from.

seconds. In addition, as was pointed out in an earlier article by the author (Laing, 1975), the venom of P. antipodiana is often fatal to an animal as large as a mouse.

The question of how many wasps are killed by spiders during the summer months is difficult to determine with any great accuracy, for if the wasp does lose an encounter, its body will most likely remain out of sight, deep within the spider's tunnel. Analysis of prey remains in tunnels has given some pointers to the wasp mortality rate though. Over the five years 1972-76, approximately 300 wasp sightings were recorded by the author. Over the same period, 8 dead wasps were found in P. antipodiana tunnels in the same locality as the sightings were made. This gave an estimate of a 2.6% mortality rate. This must be viewed as a highly conservative estimate. The real figure, for the reason already given, is likely to be much higher. The tunnels which were opened for inspection were a small proportion of the total population, and the main reason for sampling them was to gain general information on the prey taken by P. antipodiana.

Additional evidence for wasp mortality came from detailed studies page 12 on three of the spider populations: the Paremata population studied in 1971; the Johnsonville crib wall population figures for 1975-76; and the Northland population for 1977-78. In these areas particular note was taken of the area around the sheet webs. It transpired that a number of dead wasps were found either lying on the edge of a sheet web. or on the ground adjacent to a sheet web. The indications were that these wasps had managed to make their way out of the spider's tunnel after an encounter and had died without being able to move far. The figures for these studies probably represent a more realistic account of the wasp mortality rate than the tunnel studies do.

Paremata population 1971	5 wasps active	1 killed
Johnsonville population 1975	30 wasps active	2 killed
Northland population 1977	21 wasps active	3 killed

These values suggest a possible wasp mortality of 11%. Of the wasps killed. 3 were S. monachus and 3 were S. wakefieldi, which in itself is interesting, for even though the red wasp mainly hunts Miturga, sufficient numbers of them must enter P. antipodiana tunnels for such encounters to occur.

It is worth comparing these results with the results of laboratory investigations in which the wasp is rarely troubled by the spider's defence. For example. Petrunkevitch (1926), in an extensive examination of the wasp Pepsis and its attack pattern, found that the wasp was successful in the 200 or more encounters he observed. The problem with this type of study is that taking the spider out of its natural habitat considerably reduces its ability to defend itself The spider in a dark tunnel or burrow is far better equipped to defend itself than when it is brought out into an open, well lit environment. The latter conditions favour a quick moving, highly visual animal such as a wasp. The following section on spider tactics investigates how the spider does have some protection in its natural habitat.

Spider Responses — the Tactics of Survival

The spider is faced with a formidable hunter. possesing an armoured body which largely renders the spider's weapons ineffective. In addition the sting of the wasp is powerful enough to immobilise even the largest P. antipodiana individuals. Along with these features. the wasp is an assiduous and persistent hunter; it is unlikely to miss out on investigating many of the webs in its search area. for by its very nature as a prey-capture device, the sheet web of Porrhothele must be located in the open.

Considering these facts, it seems surprising that there is not an almost 100% mortality rate among the spiders in areas where wasps are active. The question that poses itself is: what factors operate to ensure the survival of 70-90% of the spiders in a population which is facing wasp activity?

The following are some of the factors which are likely to be responsible for spider survival (see Fig. 3):