SUPPLEMANTARY DATA
Table S1. Categories used to assign parasitoid species to functional groups. Parasitoids were reared
from herbivores on 10 organic and 10 conventional farms in the south-west of England. The rearing
information and previous literature was used to assign them to one of these functional groups. The
number of species and abundance for each functional group is the total for all 20 farms sampled in
the study.
Functional
group
Number
Host feeding niche
Host stage killed
1
External feeders
Larva
9
19
2
External feeders
Pupa
4
5
3
External feeders
Pre-pupa
11
13
4
Webspinners
Larva
10
11
5
Webspinners
Pupa
12
8
6
Webspinners
Pre-pupa
25
152
7
Leafminers
Larva
6
470
8
Leafminers
Pupa
54
6755
9
Leafminers
Pre-pupa
9
15
10
Attacks >1 host group
Larva
1
3
11
Attacks >1 host group
Pupa
0
0
12
Attacks >1 host group
Pre-pupa
4
15
13
Parasitoids
2
3
14
Leafminers
21
99
number
Hyperparasitoid of any
stage
larva or pupa (applies to
some eulophids)
of species
Abundance
Robustness to parasitoid species loss: Testing the effect of network species richness
Despite presenting the results using proportions (i.e. the proportion of herbivore species
released from biological control), it remains a possibility that any observed differences between
webs derived from organic and conventional farms are a consequence of differences in numbers of
species, rather than structural properties of the webs. We used a rarefaction procedure to test if this
was the case. We generated 100 random webs by selecting herbivores and parasitoids from the more
species-rich web in each pair of organic and conventional farms, until the number of herbivores
equalled that in the corresponding target web (the less species rich of the pair). In most cases it was
also possible to ensure that the number of parasitoids also equalled the number in the target web.
We did this by selecting individual interactions with replacement and with the probability of
selecting a particular intereation determined by its frequency of observation in the source web. This
continued until the target species richness of herbivores and parasitoids was obtained. The
frequency of a given interaction was then taken to be the number of times that interaction occurred
in the simulated web.
There was little difference in the extinction simulation results obtained between the random
webs and the source web, suggesting that our rarefaction procedure was effective. In the most-toleast cases, the lines representing the source webs fell entirely within the 95% confidence limits for
the random webs or on the rare occasions where they differed it was by a very small amount. With
least-to-most extinction in five out of the ten pairs the random webs begin losing herbivore control
with less removal of parasitoids than either of the actual webs. In the remaining five pairs there is a
non-significant tendency for this to occur. This corresponds to proportionately fewer parasitoids
being removed before some of the herbivores experience a reduction of more than 50% control.
This may imply a small tendency for herbivores attacked by small numbers of parasitoids to be
associated with specialist parasitoids slightly more often in random webs compared to real webs.
Taking the pairs shown in Figure 1 in turn, pair 1 shows significant differences between the
organic and conventional members of the pair at a number of points in both the most-to-least and
least-to-most cases. The conventional web loses less control initially and then departs from the
95% confidence interval at three other points. Pairs 2 and 3 show a very similar pattern with the
conventional web departing from the 95% confidence interval between approximately 0.2 and 0.6
extinction of parasitoids. With most-to-least extinction the conventional web loses more control in
this section whereas with least-to-most extinction it is the organic web that loses more control. For
pairs 5, 6 and 8 there is only a significant difference in the least-to-most case with the less specious
web (organic for pair 6, conventional for pairs 5 and 8) beginning to lose control after slightly more
extinction of parasitoids. Pair 7 appears to be somewhat unusual with the organic and conventional
members of the pair behaving in quite different ways over much of the range. For most-to-least
extinction the organic web loses control rapidly at the start and significantly faster than webs of the
same size derived from the larger, conventional web. The members of the pair again diverge
significantly at around 0.3 parasitoid extinction (the organic web losing more control) and 0.6
parasitoid extinction (the organic web losing less control). For least-to-most extinction the organic
web loses significantly more control initially but appears significantly more robust between 0.4 and
0.7 parasitoid extinction and again after 0.9 parasitoid extinction. Pair 9 has the conventional web
retaining significantly more control in the most-to-least case between 0.4 and 0.5 parasitoid
extinction and significantly less control between the same two points in the least-to-most case. Both
pair 9 and pair 10 have both webs losing initial control less quickly than random webs in the least to
most case.
Figure S1. The simulated parasitoid removal output for each pair of farms. Each graph shows the
proportion of herbivores released from control as parasitoid species are removed the farm networks.
Solid black line represents the organic farm, the black dashed line the conventional farm, the grey
solid line the median of random webs drawn from the more specious of the farm pairs, and the grey
dotted lines illustrate the maximum and minimum range of the random web. A. shows most-toleast scenarios and B. shows least-to-most scenarios.
Fig. S1.
A. Pair 1. Most-to-least
B. Pair 1. Least-to-most
A. Pair 2. Most-to-least
B. Pair 2. Least-to-most
A. Pair 3. Most-to-least
B. Pair 3. Least-to-most
A. Pair 4. Most-to-least
B. Pair 4. Least-to-most
A. Pair 5. Most-to-least
B. Pair 5. Least-to-most
A. Pair 6. Most-to-least
B. Pair 6. Least-to-most
A. Pair 7. Most-to-least
B. Pair 7. Least-to-most
A. Pair 8. Most-to-least
B. Pair 8. Least-to-most
A. Pair 9. Most-to-least
B. Pair 9. Least-to-most
A. Pair 10. Most-to-least
B. Pair 10. Least-to-most