ABSTRACT
Title of Document:THE INFLUENCE OF PREDATOR SPECIES
RICHNESS ON PREY MORTALITY:
IMPLICATIONS TO CONSERVATION
BIOLOGICAL CONTROL.
Scott Asher Lewins, Master of Science, 2006
Directed By:Professor, Pedro Barbosa, Department of
Entomology
Understanding how changes in biodiversity affect the function of
agroecosystems is paramount to conservation biological control. The Species
Assemblage Control Hypothesis predicts increasing species richness of predator
assemblages can increase the assemblages? ability to suppress pests. I hypothesized
that an increase in species richness of a predator assemblage leads to an increase in
prey mortality and predator species identity can alter the relationship. An assemblage
of predators identified from an assessment of a collard agroecosystem was evaluated
to find that only some predators fed on larval Pieris rapae, they did not differ in their
per capita consumption, and some intraguild predation occurred. In testing the
hypotheses I found a significant relationship between predator species richness and
prey mortality; however, predator species identity altered the relationship. These
findings highlight the importance in understanding predator assemblages before
conservation decisions that effectively suppress pests can be made.
THE INFLUENCE OF PREDATOR SPECIES RICHNESS ON PREY
MORTALITY: IMPLICATIONS TO CONSERVATION BIOLOGICAL
CONTROL.
By
Scott Asher Lewins.
Thesis submitted to the Faculty of the Graduate School of the
University of Maryland, College Park, in partial fulfillment
of the requirements for the degree of
Master of Science
2006
Advisory Committee:
Dr. Pedro Barbosa, Chair
Dr. Galen Dively
Dr. Professor Paula Shrewsbury
? Copyright by
Scott Asher Lewins
2006
ii
Acknowledgements
I have a great many people to thank. My research committee members Drs.
Paula Shrewsbury, Galen Dively and my major advisor, Dr. Pedro Barbosa for their
feedback and support throughout this project. Carlo Moreno, my collaborator, with
whom I shared much of the pain and enjoyment of this project. Members of the
Barbosa lab, Dr. John Kemper, Dr. Astrid Caldas, Dr. Raul Medina, Eric Lind,
Megan Paustian, and Gwen Shlicta, research assistants Kelsey Phelps, Rebeka
Friedrich and Nat Lichten, and UMD farm managers Kevin Conover, Mike Newell
and Mark Spicknall were all integral in the completion of this project. Identifications
were made with the help of Dr. Michael Gates, Dr. Thomas Henry, and Dr. Charles
Staines. A special thanks to Grace Pazdan for her endless support and undying love
during this undertaking.
iii
Table of Contents
List of Tables......................................................iv
List of Figures......................................................v
Chapter 1: The Influence of Predator Biodiversity on Pest Suppression: Historical
Perspectives........................................................1
Chapter 2: Evaluation of Predator Community Affecting Pieris rapae on Collards in
Maryland..........................................................9
Introduction......................................................9
Methods.......................................................10
Study Sites....................................................10
Community Assessment.........................................11
Constructing Species Abundance Distributions........................13
Results/Discussion................................................13
Chapter 3: Testing Predator-Prey and Predator-Predator Relationships...........22
Introduction.....................................................22
Methods.......................................................25
Study System..................................................25
Mesocosm Design..............................................29
Experimental Protocol...........................................30
Results.........................................................33
Discussion......................................................34
Chapter 4: The Relationship Between Predator Species Richness and Prey Mortality.
................................................................43
Introduction.....................................................43
Methods.......................................................46
Study System..................................................46
Mesocosm Design..............................................49
Experimental Protocol...........................................49
Results.........................................................52
Discussion......................................................54
Appendix.........................................................64
References........................................................72
iv
List of Tables
Table 3.1. Determination of predators of 1
st
 instar P. rapae. In 2005, comparisons
were made between mean P. rapae larval mortality imposed by each predator
and that in a no-predator treatment..................................38
Table 3.2. Per capita consumption of 1
st
 instar P. rapae. In 2006 the per capita
consumption of the three predators species collected in sufficient numbers, and
that were found to consume 1st instar P. rapae, was determined. Per capita
consumption represents the mean mortality in the presence of each predator
species minus background mortality. Therefore, per capita consumptions reflect
adjusted mortalities. The design was balanced, in that sample size was 15 for all
treatments, unlike in 2005.........................................39
v
List of Figures
Figure 2.1. Sampling design..........................................17
Figure 2.2. Foliar abundance distribution. Abundance distribution of all foliar
predator morpho-species collected during sampling period from June through
August 2004. Darkened bars represent web-building, social arthropods......18
Figure 2.3. Aerial abundance distribution. Abundance distribution of aerial predator
morpho-species species collected during sampling period from June through
August 2004...................................................19
Figure 2.4. Epigeal abundance distribution. Abundance distribution of epigeal
predator morpho-species species collected during sampling period from June
through August 2004............................................20
Figure 2.5. Assemblage species abundance distribution. Species abundance
distribution of non-web-building and non-social predators collected in the foliar
microhabitat. This represents the assemblage of potential predators of 1
st
 instar
P. rapae that were evaluated in chapter 3, with the addition of Pardosa sp  and
Pterostichus lucublanduds, two numerically dominant predators found in the
epigeal microhabitat.............................................21
Figure 3.1. Feeding trials to determine which predators feed on larval P. rapae. Each
predator treatment was compared to the no predator control. Significant
differences were found when C. maculata, C. septempunctata, P. maculiventris
and N. roseipennis were compared to the no predator control.* denotes p-values
less than 0.05, and error bars represent standard error of the mean..........40
Figure 3.2. Per capita consumption of 1
st
 instar P. rapae by each predator species.
There was no significant difference in consumption between any of the
predators. Error bars represent standard error of the mean.................41
Figure 3.3. Test for Intraguild Predation. The first set of bars represents C. maculata
in the presence of C. septempunctata (left) control (center) and in the presence of
P. maculiventris (right). The second set of bars represents C. septempunctata in
the presence of C. maculata (left), control (center) and in the presence of P.
maculiventris (right). The third set of bars represents P. maculiventris in the
presence of C. maculata (left), control (center) and in the presence of C.
septempunctata (right).* Denotes a p-value less than 0.05................42
Figure 4.1. Generalized results graphs. All possible results for an experimental
design with three levels of species richness............................59
vi
Figure 4.2. Relationship between predator richness and prey mortality. Adjusted P.
rapae mortality was pooled across all treatments with the same richness and then
compared between treatments with 1 species, 2 species and 3 species........60
Figure 4.3. C. maculata (Cmac) in one, two and three species assemblages. In figure
4.3a, treatments with C. maculata alone, treatments with C. maculata and C.
septempunctata (C7), and treatments with all three predators were compared.
Figure 4.3b differs in that the two-predator treatments contained C. maculata
and P. maculiventris (Pmac), however the same comparisons were made.....61
Figure 4.4. C. septempunctata (C7) in one, two and three species assemblages. In
figure 4.4a, treatments with C. septempunctata alone, treatments with C.
septempunctata and C. maculata (Cmac), and treatments with all three predators
were compared. Figure 4.4b differs in that the two-predator treatments contained
C. septempunctata and P. maculiventris (Pmac), but the same comparisons were
made.........................................................62
Figure 4.5. P. maculiventris (Pmac) in one, two and three species assemblages. In
figure 4.5a, treatments with P. maculiventris alone, treatments with P.
maculiventris and C. maculata (Cmac), and treatments with all three predators
were compared. Figure 4.5b differs in that the two-predator treatments contained
P. maculiventris and C. septempunctata, however the same comparisons were
made.........................................................63
1
Chapter 1: The Influence of Predator Biodiversity on Pest
Suppression: Historical Perspectives.
For hundreds of years citrus growers in ancient China would place predaceous
ant colonies, Oecophylla smaragdina subnitida Emery, between trees to protect their
harvest (DeBach 1964). The use of predators as biological control agents in
conventional U.S. agriculture began with the introduction of the Vedalia Beetle,
Rodolia cardinalis Mulsant, in 1889 to combat cottony cushion scale, Icerya
purchasi Maskell, an introduced species that was threatening the citrus industry in
California at that time (Doutt 1967, Caltagirone 1981). The latter is an example of
classical biological control. In this approach to biological control, an exotic natural
enemy, i.e. a parasitoid, pathogen or predator, is intentionally introduced from the
region of origin of the pest to the new region of the pest for the purpose of
establishing long-term suppression usually of a coevolved pest (Eilenberg et al.
2001). Augmentative biological control is the repeated release of native or non-native
natural enemies by way of mass rearing programs in order to increase their abundance
in habitats where their populations are low or non-existent. Both classical and
augmentative biological control have historically been where much of the success in
biological control has been achieved (Caltagirone 1981, Denoth et al. 2002).
However, in addition to the successes there have been numerous examples in which
the release of an introduced species for the purpose of biological control of a pest has
resulted in negative consequences to non-target species (Howarth 1991, Pearson and
2
Callaway 2003, Stiling 2004). It is perhaps for this reason that a third approach,
conservation biological control, has recently received increasing interest among
researchers.  Conservation biological control employs tactics that enhance the
survival and/or performance of native natural enemies in order to enhance their
effectiveness (Barbosa 1998). This can be achieved through the manipulation of the
habitat in ways that benefit natural enemies such as providing alternative food like
pollen or nectar, providing a microclimate for natural enemies to seek refuge from
environmental extremes or pesticides, and providing habitat for alternative prey
(Landis et al. 2000, Barbosa et al. 2005).
In contemporary agriculture, generalist predators have typically been viewed
as ineffective at natural suppression of pests and have been neglected as biological
control agents in comparison to their specialists counterparts (predators and
parasitoids) (Chang and Kareiva 1999). Despite this disparity, there is little evidence
that suggests indigenous generalist predators cannot effectively suppress pests. A
2002 review of generalist predators found that in about 75% of manipulative field
studies, generalist predators significantly reduced pest numbers (Symondson et al.
2002). Further, they found that assemblages (sensu Claridge 1987) of generalist
predators were equally as effective at lowering pest numbers as single predators, and
in most cases, their predation led to an increase in yield or decrease in plant damage.
More recently, the increased interest in the effect of biodiversity on ecosystem
function (Chapin et al. 2000, Loreau 2000, Tilman 2000) has resulted in a
corresponding increase in research on how the biodiversity of predator assemblages
affect their ability to control pests (Ives et al. 2005, Snyder et al. 2005, Wilby et al.
3
2005, Straub and Snyder 2006). The evidence suggests that increased biodiversity can
enhance biological control. But is all of the biodiversity of predator assemblages
necessary for effective biological control? If the entire assemblage is not necessary
for effective pest suppression, then conservation efforts could focus on tactics that
benefit the subset of the assemblage actually responsible for pest suppression. Current
conservation biological control tactics may not benefit all predators similarly
(Barbosa et al. 2005). Therefore, it may be more appropriate to evaluate the effect of
potential tactics on the assemblage subset responsible for suppression. In order to
identify the subset of a predator assemblage most likely to impose mortality on the
target pest, it is necessary to determine (a) what predators, potentially interacting with
the target pest actually feed on the target pest, (b) the relative effectiveness and
differences in effectiveness among the predators that feed on the target pest, and (c)
whether the predators in the group that do consume the target pest engage in any
antagonistic interactions like intraguild predation, which can potentially reduce the
impact of the assemblage.
The increased interest in the influence of biodiversity on ecosystem function
has highlighted the importance of understanding how the species in a predator
assemblage interact and ultimately influence the assemblage?s ability to suppress
pests. Research has shown that increasing natural enemy species richness (a measure
of biodiversity) can lead to decreased biocontrol because of negative interactions like
interference, cannibalism and intraguild predation (Rosenheim et al. 1993, Hochberg
1996, Denoth et al. 2002). Debach (1974) went as far as suggesting that biological
control would be most effective if relatively few specialized natural enemies were
4
employed as biological control agents. That is, with agents that are able to survive
within the same range of the pest, capable of high reproductive capacity relative to
the pest and highly efficient at searching for the pest. The Species Assemblage
Control Hypothesis (SACH) epitomizes the opposing view. Central to SACH is the
belief that an assemblage of generalist predators, if conserved properly, can
suppression associated prey populations more effectively than a single predator
species (Riechert and Lockley 1984, Provencher and Riechert 1994, Riechert and
Lawrence 1997). Riechert and Lockley (1984) argue that certain characteristics of
generalist predator assemblages (spiders in this case) make the assemblage as a whole
more effective then any single agent of natural suppression.  They contend that the
assemblage is self-dampening because of factors like cannibalism and intraguild
predation, lending stability in periods of low prey availability. In times of high prey
availability, the assemblage can exhibit a strong numerical response through
aggregation and reproduction. They go on to suggest that as predator richness in an
assemblage increases, a more diverse set of foraging behaviors and sizes should be
present, which should enhance the probability that prey of various sizes and species
will be killed. This anticipated advantage of greater species diversity is what has been
term sampling effect (Loreau et al. 2001). On the other hand, biological control
theory predicts that increasing predator richness should lead to an increase in
potentially negative predator-predator interactions, and limit effective suppression of
pests (DeBach 1974, Rosenheim et al. 1995). However, there have been examples of
predator assemblages suppressing pests despite the occurrence of negative predator-
predator interactions (Lang 2003, Snyder and Ives 2003, Snyder et al. 2004).
5
The relationship between biodiversity of predator assemblages and pest
suppression is of great importance due to recent trends in agriculture that have led to
a loss of predator biodiversity (Snyder et al. 2005). Increased biodiversity has been
shown to increase pest suppression (Cardinale et al. 2003, Snyder et al. 2006). The
goal of this study was to elucidate the relationship between biodiversity and pest
suppression, specifically the relationship between species richness of a predator
assemblage and pest suppression.
Larvae of the cabbage white butterfly, Pieris rapae Linnaeus, are significant
economic pests of crucifers worldwide (Schmaedick and Shelton 2000) and in
Maryland (G. Dively personal communication). Larvae are preyed on by a myriad of
generalist predators (Dempster 1967, Ashby 1974, Schmaedick and Shelton 2000).
Life tables constructed by Dempster (1967) indicated that 1
st
 instars are the most
vulnerable life stage.  Assemblages of generalist predators, which could potentially
affect 1
st
 instar P. rapae, have been identified in crucifers across a wide range of
geographic regions such as Japan (Suenaga and Hamamura 2001), Hawaii (Hooks et
al. 2003), and the Northwestern (Snyder et al. 2006) and Midwestern continental
United States (Schellhorn and Sork 1997). Therefore this study system, consisting of
collards, P. rapae and the associated assemblage of generalist predators found on
collards in Maryland, is an appropriate focus for research on the relationship between
predator species richness and pest suppression.
Mesocosm studies, using protocols that are scaled-down versions of field
experiments have been used by researchers as an intermediate approach between field
studies and laboratory Petri dish studies (Dinter 2002, Finke and Denno 2004,
6
Madsen et al. 2004). Laboratory studies done in the sterile environment of a small
arena (such as a Petri dish) may be over simplified. The circumstances created in
these simplified arenas may foster interactions that are unlikely to occur in the field.
Conversely field studies, although most relevant in an applied context, can be costly,
time intensive, and logistically impractical. Mesocosms provide a useful alternative.
By conducting this study in mesocosms I was able to simulate, to a certain extent, the
environment present in a collard field, while at the same time maintaining certain
experimental parameters constant. The results can then be used to infer what might
take place under field conditions.
Given that some assemblages of predators have been shown to be more
effective at controlling pests than a single predator species, the number and identity of
predator species in an assemblage influences the assemblages effectiveness, and that
interactions between predator species may impact the effectiveness of the
assemblage, I chose to explore two questions. Does the level of pest suppression
provided by the assemblage increase as the number of predator species in an
assemblage increases? And, how does species identity of predators in the assemblage
affect the relationship between predator species richness and pest suppression? Thus,
I hypothesized that an increase in the species richness of an assemblage of predators
leads to an increase in prey mortality imposed by the assemblage. Further, I
hypothesized that predator species identity can alter the relationship between species
richness of a predator assemblage and prey mortality.
7
In order to test these hypotheses it was necessary to accomplish the following
objectives: (1) I evaluated the assemblage of generalist arthropod predators found in
collard agroecosystems in Maryland to assess the relative abundance of species in the
assemblage and their identity, (2) I then further evaluated the predators identified
through this assessment to confirm that they fed on larval P. rapae, their relative
effectiveness and differences in consumption of P. rapae larvae, and determine if
they engaged in intraguild predation in the absence of P. rapae, (3) I then tested the
hypotheses that increasing predator species richness increases the mortality imposed
on larval P. rapae and that predator species identity can alter the relationship between
species richness of an predator assemblage and prey mortality using the cohort of
predator species selected as a result of the evaluations.
Objective (1) provided information on the assemblage of generalist arthropod
predators found in collard agroecosystems in Maryland. The assessment of the
predator assemblages within each microhabitat (epigeal, aerial, and foliar) was
conducted with three common sampling methods: pitfall trapping, sweep, and visual
sampling. From this assessment, I generated species abundance distributions (SADs)
for each of the microhabitats in order to identify the numerically dominant and
subdominant predators. It is from the SADs that I identified a subset of predators in
their presumed order of importance (assuming the most abundant species present is
the most important in the suppression of the focal pest species P. rapae). Predators
most likely to interact with P. rapae larvae were identified to the species level. For
objective (2) I performed a series of feeding trials in small mesocosms wherein I first
determined which of the predators present in the collard agroecosystem actually fed
8
on P. rapae larvae by comparing P. rapae mortality in the presence of each predator
species to that of the no predator control. Then, I estimated the per capita
consumption for the predators that were determined to feed on P. rapae using a
second more rigorous series of feeding trials where a no predator control was used to
adjust for background mortality. In order to evaluate the potential for intraguild
predation to occur between those predators, I undertook a third set of trials in which
combinations of two individuals of a different species were combined in the absence
of prey. The results of the feeding trials (objective 2) enabled me to choose the
appropriate predators for the experimental assemblage needed to test my hypotheses.
In objective (3), I tested my hypotheses by manipulating the number of predator
species in assemblages within larger mesocosms to determine if there were significant
changes in larval P. rapae mortality as species richness increased, and explored the
affect species identity had on the hypothesized relationship.
9
Chapter 2: Evaluation of Predator Community Affecting Pieris
rapae on Collards in Maryland.
Introduction
In contemporary agriculture, generalist predators have typically been viewed
as ineffective at natural suppression of pests and have been neglected as biological
control agents in comparison to their specialists counterparts (predators and
parasitoids) (Chang and Kareiva 1999). Despite this disparity, there is little evidence
that suggests indigenous predators cannot effectively suppress pests. A 2002 review
of generalist predators in biological control found that in about 75% of manipulative
field studies, generalist predators significantly reduced pest numbers (Symondson et
al.). Further, they found that assemblages (sensu Claridge 1987) of generalist
predators were equally as effective at lowering pest numbers as single predators, and
in most cases, their predation led to an increase in yield or decrease in plant damage.
More recently, the increased interest in the effect of biodiversity on ecosystem
function (Chapin et al. 2000, Loreau 2000, Tilman 2000) has resulted in a
corresponding increase in research on how the biodiversity of predator assemblages
affect their ability to suppress pests (Ives et al. 2005, Snyder et al. 2005).
Larvae of the cabbage white butterfly, Pieris rapae Linnaeus, are significant
economic pests of crucifers worldwide (Schmaedick and Shelton 2000) and in
Maryland (G. Dively personal communication). Larvae are preyed on by a myriad of
10
generalist predators (Dempster 1967, Ashby 1974, Schmaedick and Shelton 2000).
Life tables constructed by Dempster indicate that the most vulnerable life stage is the
1
st
 instar.  Assemblages of generalist predators, which could potentially affect 1
st
instar P. rapae, have been identified in crucifers across a wide range of geographic
regions such as Japan (Suenaga and Hamamura 2001), Hawaii (Hooks et al. 2003),
and the Northwestern (Snyder et al. 2006) and Midwestern continental United States
(Schellhorn and Sork 1997).  In this study, I evaluated the assemblage of generalist
arthropod predators found in collard agroecosystems in Maryland in order to assess
the relative abundance of predator species in the assemblage and their identity. The
predators identified through this assessment were further evaluated to confirm
whether they fed on larval P. rapae, and determine if they engaged in intraguild
predation, in the absence of P. rapae (see chapter 3). The cohort of predator species
selected as a result of these evaluations were then used to test the hypotheses that
increasing species richness increases the mortality imposed on larval P. rapae and
that predator species identity can alter the relationship between species richness of an
predator assemblage and prey mortality (see chapter 4).
Methods
Study Sites
Study sites included the Wye Research and Education Center (WREC) on the eastern
shore of Maryland and the Central Maryland Research and Education Center
(CMREC) - Upper Marlboro (UP) facility. Two plots of collard greens (Vates
variety) were planted on May 2 and June 1, 2004 at WREC and on May 8, 2004 at
11
CMREC. The collards were grown using standard agricultural practices. The two
plots at WREC were conventionally tilled, and the two at UP were not tilled. The
plots were all 23 x 33m and the rows in a plot were approximately 1 meter apart from
each other.  There were 25 rows per plot at WREC and 26 per plot at UP.
Community Assessment
The community of arthropods on collards, including P. rapae and the assemblages of
generalist predators in three microhabitats (epigeal, foliar and aerial), were assessed
once a week for six weeks starting five weeks after the collards were planted (from
early June to late July 2004). The assessment of the predator assemblages within each
microhabitat was conducted with the following common sampling methods: pitfall
traps (for epigeal), sweeping (aerial), and collections by hand (foliar).
The pitfall trap sampling within the epigeal microhabitat was conducted using
two 473ml clear plastic cups (with a 9.7 cm diameter opening; Solo Cup Co.
?
,
Urbana, Illinois), one inside the other. A plastic plate (Solo Cup Co.
 ?
, Urbana,
Illinois) serving as a roof and fastened 2 cm above the trap with three 7.62cm bolts
was used to prevent the cups filling with rain water. Approximately 60ml of
automobile antifreeze was added to each trap, as a killing agent and preservative.
Once weekly, pitfalls were left in the field for 24 hours, after which their contents
emptied. Nine traps were placed in 1m X 1m grids within each plot. This was
accomplished by replacing a plant, every eight meters of every eighth row, thus each
pitfall was approximately 8m apart from adjacent traps (Figure 2.1). The contents of
each trap were emptied into a clean plastic cup, labeled according to site, date, time,
plot No., row No., location within row, and identified as a pitfall collection. Samples
12
were brought back to the lab, washed, and stored in 95% EtOH until they were
identified.
The community inhabiting the aerial microhabitat was sampled with sweep
nets. A standard 30cm diameter sweep-net composed of linen net was used for sweep
samples. The samples consisted of ten replicate sweep transects; each comprised of
ten double swings taken while walking ten paces. Each sweep was initiated at a
randomly selected row and starting point using a random numbers table and sweeps
sampled the air immediately above the plants. The first and last five rows, as well as
the first and last eight meters inward from plot edges (Figure 2.1) were excluded from
sampling to minimize edge effect. The sweeping took place once a week at the
approximate same time of day, for six weeks. The contents of each replicate sweep
were individually bagged, labeled according to site, date, time, plot No., row No.,
location within row, identified as a sweep sample and kept on ice until brought back
to the lab where they were placed into 95% EtOH for later identification.
The foliar assemblage of arthropods actually represented the assemblage of
arthropods found on foliage as well as stems but was described as ?foliar
assemblage.? It was sampled by visually inspecting individual collard plants and hand
collecting all arthropods. The visual inspection of all the above ground parts of ten
randomly selected collard plants was conducted by carefully searching the individual
plant for five minutes. All of the arthropods found on the observed plant were
collected into glass vials containing 95% EtOH with the aid of feather forceps
(BioQuip
?
, Rancho Dominguez, CA). The first eight meters from the edge along the
entire perimeter of the plots (Figure 2.1) were excluded in the visual sampling to
13
minimize edge effect. The samples were all taken at the approximate same time every
week. The vials were labeled according to site, date, time, plot No., row No., location
within row, identified as a visual sample, and brought back to the lab for
identification.
Constructing Species Abundance Distributions
Species abundance distributions (SADs) for each of the microhabitats were generated
in order to compare the structures of the predator communities in each microhabitat,
with the aim of identifying which species were numerically dominant and
subdominant predators. The abundances of all predator morphospecies collected from
the four plots were pooled to construct SADs for each microhabitat separately. All
life stages were included in the SADs. It is from the SADs that I generated a list of
predators in rank order to provide insight into their potential importance as predators
of P. rapae (assuming the most abundant species present is the most important).
Results/Discussion
Due to the large number of individuals collected, and the difficulty in
identifying to the species level, specimens were identified to family and sorted by
species and morphospecies. In the foliar microhabitat, 239 individuals representing 8
species and 44 morphospecies, were collected (Figure 2.2). Sweeping of the aerial
microhabitat yielded 567 individuals representing 14 species and 31 morphospecies
(Figure 2.3). From within the epigeal microhabitat, 3060 individuals representing 19
species and 151 morphospecies were collected (Figure 2.4). Abundance distributions
14
for the microhabitats were informative because they show the relative abundances for
each morphospecies. Barbosa et al. (2005) developed a method of comparing
assemblages using ?Robin?s Curves,? however statistical comparisons of SADs
between assemblages found in the different microhabitats would not be appropriate
because not all of the individuals collected have been identified to species.
Although there were numerous predator species in all three microhabitats, not
all are predators of P. rapae. Certain foraging behaviors and life-history traits of
predators, particularly in relation to the traits and behavior of P. rapae make it
unlikely that they would find, attack and consume larval P. rapae.  For example,
given that larval P. rapae rarely, if ever, leave their natal plant except to pupate
(Harcourt 1961, Jones 1977) the predator species most likely to be important
mortality factors would be in the foliar microhabitat. Within the foliar microhabitat,
only the adults of each species were considered, because testing both adults and
larvae/nymphs of each predator species would have created far too many
combinations to evaluate. Further, identification of immatures was not always
possible (especially spiders). Web-building spiders were excluded from consideration
as P. rapae predators because they would be unlikely to interact with relatively
sessile larval P. rapae. Social predators, such as ants, also were excluded because
their social behavior makes testing individual predators difficult and unlikely to
produce accurate data since the use of nests in mesocosm trials would be unfeasible.
The remaining predators in the foliar predator assemblage were Nabis roseipennis,
Coleomegilla maculata, Lygus lineolaris, Coccinella septempunctata, Podisus
maculiventris, Geocoris punctipes, and Chauliognathus marginatus (Figure 2.5).
15
Lycosids and carabids feed on P. rapae larvae (Dempster 1967, Ashby 1974,
Schmaedick and Shelton 2000), so Pardosa sp (Lycosidae) and Pterostichus
lucublanduds (Carabidae), two numerically dominant predators found in the epigeal
microhabitat (216 and 116 individuals collected, respectively), also were evaluated.
After confirming the status of these predators as P. rapae predators, I was then able to
use them to test the relationship between increases in predator species richness and
increases in larval P. rapae mortality and how the relationship is affected by predator
species identity.
The subsets of generalist predator assemblages responsible for suppression of
P. rapae in collard agroecosystems in other regions are very similar to those in
collards in Maryland. The common predators of 1
st
 instar P. rapae on collards in New
York state include C. maculata, Nabis americoferus, Lygus lineolaris, Orius
insidiosus, and Pardosa milvina (Schmaedick and Shelton (2000). Predators of aphids
on collards, and potentially P. rapae, in Washington state include C. maculata, C.
septempunctata, Geocoris pallens, and a combination of N. americoferus and Nabis
alternatus (Snyder et al. (2006). The only two predator species selected to be apart of
the assemblage tested in mesocosms that haven?t been reported in crucifers was  P.
maculiventris, which is commonly known to feed on lepidopteran larvae, and C.
marginatus, which was found along the collard field margins in great numbers during
the 2005 season.
From this evaluation of the assemblage of generalist arthropod predators
found in collard agroecosystems in Maryland, I was able to ascertain those predators
most likely to interact with, and inevitably influence P. rapae populations.  The
16
predators were then tested in mesocosm trials to confirm that they fed on larval P.
rapae, and determine if they would feed on each other in the absence of P. rapae (see
chapter 3). The assemblage was ultimately tested in mesocosms to examine the
relationship between species richness of a predator assemblage and mortality of larval
P. rapae imposed by the assemblage and the influence of predator identity on that
relationship (see chapter 4).
17
Figure 2.1. Sampling design
18
Figure 2.2. Foliar abundance distribution. Abundance distribution of all foliar
predator morpho-species collected during sampling period from June through
August 2004. Darkened bars represent web-building, social arthropods.
0
20
40
60
80
100
120
1 6 11162126313641
Rank
Number of Individuals Collected
19
Figure 2.3. Aerial abundance distribution. Abundance distribution of aerial predator
morpho-species species collected during sampling period from June through
August 2004.
0
50
100
150
200
250
300
350
1 6 11 16 21 26 31
Rank
Number of Individuals Collected
20
Figure 2.4. Epigeal abundance distribution. Abundance distribution of epigeal
predator morpho-species species collected during sampling period from June
through August 2004.
0
100
200
300
400
500
600
700
1112131415161718191101111121131141151
Rank
Number of Individuals Collected 
21
Figure 2.5. Assemblage species abundance distribution. Species abundance
distribution of non-web-building and non-social predators collected in the foliar
microhabitat. This represents the assemblage of potential predators of 1
st
 instar
P. rapae that were evaluated in chapter 3, with the addition of Pardosa sp  and
Pterostichus lucublanduds, two numerically dominant predators found in the
epigeal microhabitat.
Nabis
roseipennis
Coleomegilla
maculata
      Lygus
    lineolaris
Coccinella
septempunctata
Podisus
maculiventris
Geocoris
punctipes
Chauliognathus
marginatus
0
20
40
60
80
100
120
Number of Individuals Collected
22
Chapter 3: Testing Predator-Prey and Predator-Predator
Relationships.
Introduction
The use of predators as biological control agents in conventional U.S.
agriculture began with the introduction of the Vedalia Beetle, Rodolia cardinalis
Mulsant, in 1889 to combat cottony cushion scale, Icerya purchasi Maskell, an
introduced species that was threatening the citrus industry in California at that time
(Doutt 1967, Caltagirone 1981). The latter is an example of classical biological
control. In this approach to biological control, an exotic natural enemy, i.e. a
parasitoid, pathogen or predator, is intentionally introduced from the region of origin
of the pest to the new region of the pest for the purpose of establishing long-term
suppression usually of a coevolved pest (Eilenberg et al. 2001). Augmentative
biological control is the repeated release of native or non-native natural enemies by
way of mass rearing programs in order to increase their abundance in habitats where
their populations are low or non-existent. Both classical and augmentative biological
control have historically been where much of the success in biological control has
been achieved (Caltagirone 1981, Denoth et al. 2002). However, in addition to the
successes there have been numerous examples in which the release of an introduced
species for the purpose of biological control of a pest has resulted in negative
consequences to non-target species (Howarth 1991, Pearson and Callaway 2003,
23
Stiling 2004). It is perhaps for this reason that a third approach, conservation
biological control, has recently received increasing interest among researchers.
Conservation biological control employs tactics that enhance the survival and/or
performance of native natural enemies in order to enhance their effectiveness
(Barbosa 1998). This can be achieved through the manipulation of the habitat in ways
that benefit natural enemies such as providing alternative food like pollen or nectar,
providing a microclimate for natural enemies to seek refuge from environmental
extremes or pesticides, and providing habitat for alternative prey (Landis et al. 2000,
Barbosa et al. 2005).
In contemporary agriculture, generalist predators have typically been viewed
as ineffective at natural suppression of pests and have been neglected as biological
control agents in comparison to their specialists counterparts (predators and
parasitoids) (Chang and Kareiva 1999). Despite this disparity, there is little evidence
that suggests indigenous generalist predators cannot effectively suppress pests. A
2002 review of generalist predators found that in about 75% of manipulative field
studies, generalist predators significantly reduced pest numbers (Symondson et al.
2002). Further, they found that assemblages (sensu Claridge 1987) of generalist
predators were equally as effective at lowering pest numbers as single predators, and
in most cases, their predation led to an increase in yield or decrease in plant damage.
More recently, the increased interest in the effect of biodiversity on ecosystem
function (Chapin et al. 2000, Loreau 2000, Tilman 2000) has resulted in a
corresponding increase in research on how the biodiversity of predator assemblages
affect their ability to suppress pests (Ives et al. 2005, Snyder et al. 2005, Wilby et al.
24
2005, Straub and Snyder 2006). The evidence suggests that increased biodiversity can
enhance biological control. But is all of the biodiversity of predator assemblages
necessary for effective biological control? If the entire assemblage is not necessary
for effective pest suppression, then conservation efforts could focus on tactics that
benefit the subset of the assemblage actually responsible for pest suppression. Current
conservation biological control tactics may not benefit all predators similarly
(Barbosa et al. 2005). Therefore, it may be more appropriate to evaluate the effect of
potential tactics on the assemblage subset responsible for suppression. In order to
identify the subset of a predator assemblage most likely to impose mortality on the
target pest, it is necessary to determine (a) what predators, potentially interacting with
the target pest actually feed on the target pest, (b) the relative effectiveness and
differences in effectiveness among the predators that feed on the target pest, and (c)
whether the predators in the group that do consume the target pest engage in any
antagonistic interactions like intraguild predation (IGP), which can potentially reduce
the impact of the assemblage.
Larvae of the cabbage white butterfly, Pieris rapae Linnaeus, are significant
economic pests of crucifers worldwide (Schmaedick and Shelton 2000) and in
Maryland (G. Dively personal communication). Larvae are preyed on by a myriad of
generalist predators (Dempster 1967, Ashby 1974, Schmaedick and Shelton 2000).
Life tables constructed by Dempster (1967) indicated that 1
st
 instars are the most
vulnerable life stage.  Assemblages of generalist predators, which could potentially
affect 1
st
 instar P. rapae, have been identified in crucifers across a wide range of
geographic regions such as Japan (Suenaga and Hamamura 2001), Hawaii (Hooks et
25
al. 2003), and the Northwestern (Snyder et al. 2006) and Midwestern continental
United States (Schellhorn and Sork 1997). In this study, the predators identified
through the field assessment of the predator community (chapter 2) were further
evaluated to confirm that they feed on larval P. rapae, and determine if they engaged
in intraguild predation, in the absence of P. rapae. The cohort of predator species
selected as a result of these evaluations will then be used to test the hypotheses that
increasing predator species richness increases the mortality imposed on larval P.
rapae and that identity of the predators in assemblages influences the relationship
between species richness and levels of prey mortality (see chapter 4).
Methods
Objectives: (1) To determine if there is a significant difference in mean P. rapae
larval mortality imposed by each of the selected predator species compared to that in
a no-predator treatment. (2) To determine the per capita consumption by predators
which were found to consume 1
st
 instar P. rapae. (3) To determine whether the
predators of 1
st
 instar P. rapae engage in intraguild predation in the absence of prey.
Study System
Experiments were conducted from May to September in 2005 and 2006.
Collards (Vates variety) were grown from seed in a controlled environment in the
University of Maryland research greenhouse. Plants used for feeding trials and colony
maintenance were grown in 10cm square pots (multiple seeds were planted per pot to
ensure germination). Seeds were planted in MM510 soil (The Scotts Co.
 ?
,
26
Marysville, OH) treated with Multicote
?
, a controlled release fertilizer. Four flats of
15 pots (for a total of 60 pots) were planted every week.
Adult P. rapae were collected in the field, placed in glassine envelopes, kept
cool, and brought back to the lab to initiate a lab colony. P. rapae were kept in 60 X
60 X 60cm screen cages (BioQuip
?
, Rancho Dominguez, CA). Cages were placed
adjacent to a lab window, and provided with supplemental lighting, at ambient room
conditions (16:8 L:D, ~24?C and ~50% rH). Cages contained yellow sponges soaked
with honey water for nutrition and two collard plants on which adults could oviposit.
Collards were replaced daily, the eggs on them were allowed to hatch, some of the
larvae were used in experiments and others were reared to adults to maintain the
colony. Once the collard leaves had been completely skeletonized, the larvae were
removed with feather forceps and reared in Petri dishes with filter paper in groups of
five to ten (depending on their size) and fed fresh collard leaves, ad libitum. Upon
pupation, they were removed from the Petri dish and placed in 473ml plastic deli cups
(Solo Cup Co.
 ?
, Highland Park, IL) to allow space for the adults after emergence.
When the adults emerged they were used to maintain a lab colony in a 60 X 60 X
60cm cage.
An assessment of composition of the predator community in collards was
undertaken in 2004 (described in chapter 2). In 2005, adult predators known to exist
within the foliar microhabitat were evaluated in laboratory experiments to determine
if in fact they feed on 1
st
 instar P. rapae (Objective 1). The foliar predator assemblage
tested included Nabis roseipennis (Reuter), Coleomegilla maculata (DeGeer), Lygus
lineolaris (Palisot de Beauvois), Coccinella septempunctata (Linnaeus), Podisus
27
maculiventris (Say), Geocoris punctipes (Say), and Chauliognathus marginatus
(Fabricus). These predators, along with two numerically dominant predators found in
the epigeal microhabitat, Pardosa sp. and Pterostichus lucublandus (Say), were
collected in order to determine if they would feed on 1
st
 instar P. rapae. Predators
were field collected, placed in 29.6ml plastic cups (Solo Cup Co.
?
, Highland Park,
IL), kept cool and brought back to the lab to be used in experiments. Even though the
predators were originally collected in collards, they are generalists that are commonly
found in many cropping systems as well as non-managed systems. Thus, in order to
maximize collection an emphasis was placed on searching alfalfa, however some
individuals were collected in collards, sweet corn, and other vegetable and forage
crops. Sites in which predators were collected were located throughout central and
eastern Maryland. Collected predators were maintained individually in 473ml plastic
deli cups (Solo Cup Co.
 ?
, Highland Park IL) in the lab at ambient conditions (~24?C
and ~50%rH) until used in experiments. At the conclusion of each experiment, all
predators were placed back in 473ml plastic deli cups, reared with moist cotton and
fed P. rapae.
In 2006, those species that were confirmed to be predators of 1
st
 instar P.
rapae from objective 1, and were collected in sufficient numbers, were used in
experiments evaluating per capita consumption (objective 2) and intraguild predation
(objective 3). Per capita consumption was determined for C. maculata (a numerically
dominant predator), and C. septempunctata and P. maculiventris (two numerically
sub-dominant predators) (see chapter 2). Although N. roseipennis was collected in
moderate numbers in prior years, in 2006 it was not able to be collected in sufficient
28
numbers, so N. roseipennis was not included in the rest of the study. A P.
maculiventris colony was established with individuals collected through the use of
pheromone traps set out at Patuxent National Wildlife Refuge (Laurel, MD) and
Beltsville Agricultural Research Center (Beltsville, MD). Traps and pheromone were
obtained from Aldrich, J. R. (USDA-ARS, Beltsville, MD). P. maculiventris eggs
were also purchased through Biocontrol Network
?
 (Brentwood, TN) and reared to
adults. Once in the lab, adult predators were reared in cages with moist cotton and fed
Colorado potato beetle larvae obtained from a lab colony (maintained by G. Dively,
Entomology Department, UMD) as well as on black cutworms.  Coccinellid colonies
were kept in 25 X 25 X 25cm Plexiglas
?
 cages in the lab at ambient conditions
(~24?C and ~50%rH) until they were used in mesocosm experiments. All stages of P.
maculiventris were reared in 473ml plastic deli cups, at a density of 3-5 individuals
per cup to cut down on cannibalism, with moist cotton and a food source. C.
septempunctata were collected in alfalfa and other forage crops, small grains and
various vegetables Species known to be cannibalistic, such as larval coccinellids were
kept individually in 473ml plastic deli cups (Solo Cup Co.
 ?
, Highland Park, Il). A C.
maculata colony was established with individuals from USDA-ARS Biocontrol
Laboratory (Beltsville, MD) and supplemented with individuals collected in collards,
sweet corn, alfalfa, and other forage and vegetable crops. C. maculata were kept in 25
cm
3
 screen cages and fed artificial bee pollen obtained from the USDA-ARS
Biocontrol Laboratory (Beltsville, MD). C. septempunctata were kept in 25 cm
3
screen cages and fed P. rapae larvae, black cutworm larvae (Agrotis ipsilon
Hufnagel) obtained from Dow AgriSciences
?
 (Indianapolis, IN), or aphids collected
29
from alfalfa. At the conclusion of each experiment all living predators were returned
to their respective colonies.
Mesocosm Design
All experiments were conducted in laboratory mesocosms. Each mesocosm
consisted of a 10 cm
2
 pot containing a single collard plant, covered with a 1-gallon
nylon paint strainer bag (National/Ruskin, Inc.
 ?
, Hatfield, PA). The bags had an
elastic band, which secured them to the pots, with the aid of 15mm binder clips. Each
mesocosm contained a three-week-old collard plant. Ten P. rapae larvae from the lab
colony were haphazardly placed on the plant. In 2005, leaf segments on which larvae
were feeding were cut off and transferred to experimental plants. The larvae were
allowed 12 hours to move from the leaf segment to the experimental plant. If, after 12
hours, larvae remained on the leaf segments, they were moved onto the plant with a
small brush. In 2006, larvae were transferred with feather forceps (BioQuip
?
, Rancho
Dominguez, CA). Mortality of larvae was approximately the same using the two
methods, 12.5% in 2005 and 10.6% in 2006. Haphazardly selected adult predators
were removed from the colony, placed in 29.6ml plastic cups, and starved for 24
hours prior to running an experiment. When adding them to the mesocosms, predators
were placed on the soil.
In order to avoid the loss of degrees of freedom when using a repeated
measures analysis, novel plants, and P. rapae larvae were used for each experiment.
Novel predators (whenever possible) were randomly assigned for each experiment. If
a predator individual was used in multiple experiments the experimental treatment to
which it was assigned was re-randomized. The location within the environmental
30
chamber in which each mesocosm was placed was re-randomized for each
experiment.
Experimental Protocol
Objective (1): To determine if there is a significant difference in mean P. rapae larval
mortality imposed by each of the selected predator species compared to that in a no-
predator treatment.
Experiments in 2005 were designed to determine if mean larval P. rapae
mortality in the presence of each predator differed significantly from that in control
treatments in which no predator was added to the mesocosm. The design was
unbalanced because sample sizes were dependent on the number of each predator
species collected (Table 3.1). Mesocosms were randomly placed in an environmental
chamber (16:8 L:D, 28?/18?C, ~70% rH) for 24 hours. At the end of that time period
predators were removed and the number of living P. rapae larvae noted. 28 control
replicates, and at least four replicates of each predator treatment were run at a single
time (Table 3.1), when sufficient numbers of predators were collected. The response
variable in the experiment was the number of dead P. rapae larvae at the end of a 24-
hour time period (i.e., larval P. rapae mortality). A one-tailed one-way analysis of
variance (PROC MIXED, SAS
?
 1999) was used to compare mean larval P. rapae
mortality between treatments (predator species). Contrasts were used to determine if
mean larval P. rapae mortality for each predator species was greater than that of no-
predator control. The data satisfied the assumption of normality, however, the
variances were heterogeneous. In order to run the ANOVA it was necessary to group
31
the variances. A Bonferroni adjustment was made to control for experiment-wise
error rate.
Objective (2): To determine the per capita consumption by predators of 1
st
 instar P.
rapae.
Experiments conducted in 2006 determined per capita larval P. rapae
consumption of each predator species. Control treatments (in which no predator was
added to the mesocosm) were established to assess background mortality of P. rapae
larvae. Per capita consumption of P. rapae differs from P. rapae mortality in that per
capita consumption of P. rapae represents the mean mortality of P. rapae in the
presence of each predator species minus background mortality. Mesocosms were
randomly placed in an environmental chamber (16:8 L:D, 28?/18?C, ~70% rH) for 24
hours. At the end of that time period predators were removed and the number of
living P. rapae larvae noted. Up to 13 mesocosms (i.e., replicates) of each predator
treatment were run at a single time, when sufficient numbers of predators were
collected. A total of 15 replicates of each predator treatment and 16 controls were
tested. The response variable in the experiment, was the number of dead P. rapae
larvae at the end of a 24-hour time period, adjusted for the mortality observed in
controls (i.e. the no predator treatment). Per capita consumption represented the
difference between number of P. rapae larvae alive at the beginning (10) and at the
end of the time period, minus the mean number of larvae that died in the no-predator
treatment (an average of 1.06). The adjusted values represented the number of larvae
consumed by predators. A one-way analysis of variance (PROC MIXED, SAS
?
32
1999) was used to compare mean per capita consumption of predator species. The
data satisfied all of the assumptions of ANOVA and a Bonferroni adjustment was
made to control for experiment-wise error rate.
Objective (3): To determine whether the predators of 1
st
 instar P. rapae engage in
intraguild predation (IGP) in the absence of prey.
To test for IGP, the mesocosm protocol was used, but no P. rapae larvae were
introduced into each mesocosm. Instead, two predator individuals of differing species
were placed in the same mesocosm. The IGP mesocosms were subjected to the same
abiotic conditions noted above. After 24 and 48 hours the survival of the each
predator was noted. If both predators were still living after 48 hours it was determined
that no IGP occurred. However, if that was not the case, then I concluded that IGP did
occur. Control treatments, in which a single predator individual was placed in a
mesocosm, were used to determine background mortality, and were necessary for the
purpose of making comparisons.  Fifteen replicates of each predator combination, as
well as each control, were tested. Predator survival in each of the predator
combinations was compared to that of the corresponding control using a Chi-squared
test (PROC FREQ, SAS
?
 1999).  For tests in which 50% of the cells had expected
counts less than 5, the Chi-squared test may not be valid, so a Fisher?s exact test was
used to generate p-values.
33
Results
There was a significant difference in the mean P. rapae larval mortality
imposed by the collard agroecosystem predators tested (Figure 3.1; F = 10.5, p <
0.001). Mean 1
st
 instar P. rapae mortality was significantly greater in the presence of
C. maculata (mean mortality = 40%), C. septempunctata (mean mortality = 58%), N.
roseipennis (mean mortality = 34.4%), and P. maculiventris (mean mortality = 35%)
than in their absence (mean control mortality of 12.5% and p-value of p = 0.001, p <
0.001, p = 0.045, p = 0.001, respectively; Figure 3.1). Mean mortality of 1
st
 instar P.
rapae imposed by Chauliognathus marginatus  (mean mortality = 20%; p = 1.00),
Geocoris punctipes  (mean mortality = 20%; p = 1.00), Lygus lineolaris (mean
mortality = 7.1%; p = 1.00), Pardosa spp. (mean mortality = 6%; p  = 1.00) and
Pterostichus lucublandus (mean mortality = 16%; p = 1.00) was not significantly
different from controls.
The mean P. rapae mortality observed in the control treatment (1.06 out of
10) represented background mortality and this value was used to obtain an accurate
value for per capita consumption of each predator (Table 3.2). The per capita
consumption of 1
st
 instar P. rapae by P. maculiventris (5.1 out of 10) appeared to be
higher than that of C. maculata and C. septempunctata (2.9 and 3.1 out of 10,
respectively). However, these differences were not statistically significant (Figure
3.2; F = 2.83, p = 0.070).
There was evidence of intraguild predation in only one of the predators. In the
absence of another predator, all 15 C. maculata individuals remained alive through
the end of the 48-hour period (Figure 3.3). In the presence of C. septempunctata all
34
15 C. maculata survived. When paired with P. maculiventris, 12 of the 15 C.
maculata individuals survived, although this level of survival was not significantly
different from the control (?
2
 = 3.33, p = 0.224). In the absence of another predator,
13 of the 15 C. septempunctata individuals survived at the end of the 48-hour period
(Figure 3.3). In the presence of C. maculata, 13 C. septempunctata remained living,
although not a significant difference (?
2
 = 0.370, p = 1.00). When paired with P.
maculiventris there was a significant difference in survival, i.e., only 4 of the 15 C.
septempunctata individuals survived (?
2
 = 11.0, p < 0.001). In the absence of any
other predator, all 15 P. maculiventris individuals remained alive through the end of
the 48-hour period (Figure 3.3). There was no significant difference in P.
maculiventris in the presence of C. maculata, i.e., 13 of the 15 P. maculiventris
survived (?
2
 = 2.14, p = 0.483). When paired with C. septempunctata, all of the 15 P.
maculiventris individuals survived.
Discussion
The purpose of this study was to determine which of the predators found in
the foliar microhabitat during 2004 were in fact predators of larval P. rapae, and their
rate of consumption of larval P. rapae. Although large numbers of predators were
collected in the collard agroecosystem, only a small proportion of those collected are
predators of larval P. rapae. P. maculiventris, C. septempunctata, C. maculata, and
N. roseipennis are foliar predators that feed on larval P. rapae. Additionally, in this
system, L. lineolaris, G. punctipes, C. marginatus. Pardosa spp. and P. lucublanduds
were not predators of 1
st
 instars. Previous studies on predators found in crucifers have
35
also concluded that not all predators collected are predators of 1st instar P. rapae.
Ashby (1974), conducted feeding trials using potted cabbage plants and Schmaedick
and Shelton (2000) using preciptin tests (using many of the same species or
congeners of those I used), and found complimentary results, with only one
exception. Schmaedick and Shelton (2000) found that Pardosa milvina (Hentz)
consumed 0.90?0.38 (mean?SE) 1st instar P. rapae in 24 hours. However, no
statistical analyses were performed so it is impossible to make comparisons to what I
found. Nevertheless, these results enabled me to determine which of the predators
found in the collard agroecosystem actually feed on P. rapae and to identify the
subset of a predator assemblage most likely to affect the biological control of P.
rapae in the field.
 When the consumption rates of three confirmed predators of P. rapae, P.
maculiventris, C. septempunctata, C. maculata, were tested in mesocosms on potted
collard plants, P. maculiventris tended to consume more than the two species of
individuals of coccinellids (Table 3.2), although the difference turned out to not be
signifcant. The lack of signifcance is  consistent with the results from the 2005
experiments where per capita consumption was highest in C. septempunctata (Table
3.1). In fact in a previous study, C. maculata was found to consume 5.7?1.07
(mean?SE) 1st instar P. rapae in 24 hours (Schmaedick and Shelton 2000), which is
greater than that found for both C. septempunctata in 2005 and P. maculiventris in
2006. However, had the trials run for a longer period of time (e.g., 48 Hours), the
results may have been different in that P. maculiventris consumption may have
decrease because it?s searching behavior limits the number of prey it attacks
36
(Wiedenmann and O'Neil 1991). In order to identify the subset of a predator
assemblage most likely to affect P. rapae, it was necessary to determine the relative
effectiveness and differences in effectiveness among the predators that feed on P.
rapae. The numerically dominant predator (C. maculata) did not seem to vary in it?s
consumption of the target pest than that of the two numerical subdominant predators
(C. septempunctata and P. maculiventris). These results suggest that the assumption,
the most abundant predator is the most effective, may not always be true. Therefore
other factors, such as predator species richness or predator identity, may be
contributing to an assemblage?s effectiveness more than the relative abundances of
the predators in the assemblage.
Intraguild predation was found to occur in this system, however it was
asymetrical. When I measured intraguild predation between the three predators, the
only intraguild predator observed was P. maculiventris on C. septempunctata, its
intraguild prey. Intraguild predation has been shown to occur by C. maculata
(Schellhorn and Andow 1999), C. septempunctata (Agarwala and Dixon 1992) and P.
maculiventris (Mallampalli et al. 2002). In all of these studies the intraguild predation
was asymetrical, however, unlike in my study none of the intraguild prey were adults.
In using these three predator species, and by only using adults, I was able to limit the
potential for intraguild predation to occur when assembled in the mesocosms I used to
test the relationship between predator species richness and prey mortality (in chapter
4). Intraguild predation has been shown to effect predator?s ability to suppress pests
(Rosenheim et al. 1993). With this assemblage it was important to consider that only
P. maculiventris engages in intraguild predation and that C septempunctata is their
37
intraguild prey. P. maculiventris, being a numerically subdominant predator in the
assemblage, highlights the fact that tactics that are aimed at conserving more of the
assemblage may be more likely to increase negative interactions between predators in
the assemblage. These findings reinforce my speculation that conservation biological
control might be best accomplished by using targeted tactics. Those tactics should
focus on predators that don?t engage in detrimental interaction with other predators in
order to get the greatest impact on target pests.
In summary, I found that not all predators present in the agroecosystem
actually feed on the target pest, the relative effectiveness among the predators that
feed on the target pest appear to be equivalent and some, but not all of the predators
in the group that do consume the target pest engage intraguild predation. This
evidence supports the idea that in fact a subset of the generalist predators found in an
agroecosystem is responsible for suppression of the target pest. In making biological
control decisions to properly conserve the entire subset of predators responsible for
suppressing target pest populations, it is necessary to not only take in to account the
interactions occurring between the predators in the assemblage and the pest, but one
must also consider the negative interactions occurring between predators species in
the subset assemblage.
38
Table 3.1. Determination of predators of 1
st
 instar P. rapae. In 2005, comparisons
were made between mean P. rapae larval mortality imposed by each predator and that
in a no-predator treatment.
Treatmentn Mean Mortality (%)SEM
Control 28 12.5 0.324
Podisus maculiventris6 35.0 0.500
Coccinella septempunctata15 58.0 0.776
Coleomegilla maculata14 40.0 0.620
Nabis roseipennis9 34.4 0.603
Geocoris punctipes5 20.0 0.548
Lygus lineolaris7 7.10 0.286
Pardosa spp.5 6.00 0.400
Chauliognathus marginatus4 22.5 0.478
Pterostichus lucublandus5 16.0 0.748
39
Table 3.2. Per capita consumption of 1
st
 instar P. rapae. In 2006 the per capita
consumption of the three predators species collected in sufficient numbers, and that
were found to consume 1st instar P. rapae, was determined. Per capita consumption
represents the mean mortality in the presence of each predator species minus
background mortality. Therefore, per capita consumptions reflect adjusted mortalities.
The design was balanced, in that sample size was 15 for all treatments, unlike in
2005.
TreatmentMean Mortality (%)Per Capita
Consumption
SEM
Coleomegilla maculata39.3 2.870.480
Coccinella septempunctata42.0 3.140.825
Podisus maculiventris61.3 5.070.786
40
0
10
20
30
40
50
60
70
80
90
100
Control
Coleomegilla maculata
Coccinella septempunctata
Podisus maculiventris
Nabis roseipennis
Geocoris punctipes
Lygus lineolaris
Pardosa spp.
Chauliognathus marginatus
Pterostichus lucublanduds
Mean 
P. rapae
 Mortality (%)
Figure 3.1. Feeding trials to determine which predators feed on larval P. rapae.
Each predator treatment was compared to the no predator control. Significant
differences were found when C. maculata, C. septempunctata, P. maculiventris
and N. roseipennis were compared to the no predator control.
* Denotes p-values less than 0.05, and error bars represent standard error of the mean
**
 *
 *
41
Figure 3.2. Per capita consumption of 1
st
 instar P. rapae by each predator species.
There was no significant difference in consumption between any of the
predators. Error bars represent standard error of the mean.
0
1
2
3
4
5
6
7
8
9
10
CmacC7Pmac
Per Capita Consumption of 
P. rapae 
Larvae 
42
Figure 3.3. Test for Intraguild Predation. The first set of bars represents C. maculata
in the presence of C. septempunctata (left) control (center) and in the presence of
P. maculiventris (right). The second set of bars represents C. septempunctata in
the presence of C. maculata (left), control (center) and in the presence of P.
maculiventris (right). The third set of bars represents P. maculiventris in the
presence of C. maculata (left), control (center) and in the presence of C.
septempunctata (right).
* Denotes a p-value less than 0.05
0
10
20
30
40
50
60
70
80
90
100
 Survivial (%)
*
          w/C7   Cmac  w/Pmac               w/Cmac  C7    w/Pmac              w/Cmac  Pmac  w/C7
        C. maculata       C. septempunctata    P. maculiventris
43
Chapter 4: The Relationship Between Predator Species Richness
and Prey Mortality.
Introduction
Recent research on biodiversity suggest that increasing loss of biodiversity
can lead to a corresponding loss of ecosystem function (Chapin et al. 2000, Loreau
2000, Tilman 2000). More recently, the interest in the effect of biodiversity on
ecosystem function has resulted in more research on how the biodiversity of predator
assemblages (sensu Claridge 1987) affects their ability to suppress pests (Ives et al.
2005, Snyder et al. 2005, Wilby et al. 2005, Straub and Snyder 2006).  The increased
interest in the functioning of predator assemblages has highlighted the importance of
understanding how the species in a predator assemblage interact and ultimately
influence the assemblage?s ability to suppress pests. Research has shown that
increasing natural enemy species richness (a measure of biodiversity) can lead to
decreased biological control because of negative factors like interference, cannibalism
and intraguild predation (Rosenheim et al. 1993, Hochberg 1996, Denoth et al. 2002).
Debach (1974) went as far as to suggested that biological control would be most
effective if relatively few specialized natural enemies were employed as biological
control agents. That is, with agents that are able to survive within the same range of
the pest, capable of high reproductive capacity relative to the pest and highly efficient
at searching for the pest. The Species Assemblage Control Hypothesis (SACH)
epitomizes the opposing view. Central to SACH is the belief that an assemblage of
44
generalist predators, if conserved properly, can suppress associated prey populations
more effectively than a single predator species (Riechert and Lockley 1984,
Provencher and Riechert 1994, Riechert and Lawrence 1997). Riechert and Lockley
(1984) argue that certain characteristics of generalist predator assemblages (spiders in
this case) make the assemblage as a whole a more effective agent of natural control
than any single species.  They contend that a predator assemblage is self-regulating
because of factors like cannibalism and intraguild predation, lending stability in
periods of low prey availability. In times of high prey availability, the assemblage can
exhibit a strong numerical response through aggregation and reproduction. They go
on to suggest that as predator richness in an assemblage increases, a more diverse set
of foraging behaviors and sizes should be present, which should, in turn, enhance the
probability that prey of various sizes and species will be killed. This anticipated
advantage of greater species diversity is what has been termed a sampling effect
(Loreau et al. 2001). On the other hand, biological control theory predicts that
increasing predator richness should lead to an increase in potentially negative
predator-predator interactions, and limit effective suppression of pests (DeBach 1974,
Rosenheim et al. 1995). However there have been examples of predator assemblages
suppressing pests despite the occurrence of negative predator-predator interactions
(Lang 2003, Snyder and Ives 2003, Snyder et al. 2004).
The relationship between biodiversity of predator assemblages and pest
suppression is of great importance due to recent trends in agriculture that have led to
an increasing loss of predator biodiversity (Snyder et al. 2005). Increased biodiversity
has been shown to increase pest suppression (Cardinale et al. 2003, Snyder et al.
45
2006). In this study, I examined the relationship between biodiversity of a predator
assemblage, specifically species richness, and pest suppression (henceforth prey
mortality). In addition, I explored how predator species identity can alter the
relationship between species richness of a predator assemblage and prey mortality.
In general, if there are three levels of predator richness for an experimental
design, there are nine possible results (Figure 4.1). (A) There is no relationship
between predator richness and prey mortality (Figure 4.1a). (B) There is a direct
positive correlation between predator richness and prey mortality (Figure 4.1b). (C)
There is a direct correlation between predator richness and prey mortality up to a
point, after which there is a leveling off of mortality as richness increases (Figure
4.1c). (D) Predator richness is directly correlated with prey mortality up to a point,
after which they are negatively correlated (Figure 4.1d). (E) There is a direct negative
correlation between predator richness and prey mortality (Figure 4.1e). (F) Increasing
predator richness does not change prey mortality up to a point, after which increasing
richness leads to decreased mortality (Figure 4.1f). (G) Increasing predator richness
does not change prey mortality up to a point, after which increasing richness leads to
increased prey mortality (Figure 4.1g). (H) Predator richness is directly negatively
correlated with prey mortality up to a point, after which further increases in richness
does not change prey mortality (Figure 4.1h). (I) Predator richness is directly
negatively correlated with prey mortality up to a point, after which further increases
in richness leads to increased prey mortality (Figure 4.1i).
Testing of my hypotheses would help determine which of the potential
outcomes outlined may occur in collard agroecosystems in Maryland. Thus, I tested
46
the hypothesis that an increase in the species richness of an assemblage of predators
leads to an increase in larval Pieris rapae (Linnaeus) mortality imposed by the
assemblage. Further, I tested the hypothesis that predator species identity can alter the
relationship between species richness of a predator assemblage and P. rapae
mortality.
Methods
Study System
Experiments pertaining to this section took place from May to September
2006. Collard greens, Brassica oleracea var. acephala (Vates variety), were grown
from seed in a controlled environment at the University of Maryland research
greenhouse. Plants used for the assemblage manipulations and colony maintenance
were planted every week. For the assemblage manipulations, three collard plants were
planted 10 cm apart from one another in 30cm diameter plastic pots. More than one
seed was put in each pot to ensure germination. Sixteen pots were planted each week
to insure that there would be sufficient plants for the experiment. Collards for the
maintenance of P. rapae colonies were grown in 10cm
2
 pots. Four flats of 15 pots
(for a total of 60 pots) were planted every week.  Seeds were planted in MM510 soil
treated with Multicote
?
, a controlled release fertilizer.
Adults were collected in the field, placed in glassine envelopes, kept cool, and
brought back to the lab to initiate a lab colony of P. rapae. They were kept in 60 X 60
X 60cm screen cages (BioQuip
?
, Rancho Dominguez, CA). Cages were placed
adjacent to a lab window, but with supplemental lighting, at ambient room conditions
(16:8 L:D, ~24?C and ~50%rH). Cages contained yellow sponges soaked with honey
47
water for nutrition and two collard plants on which adults could oviposit. Collards
were replaced daily, the eggs on them were allowed to hatch, and some of the
resulting larvae used in experiments. Once the collard leaves had been completely
skeletonized, the larvae were removed with feather forceps. Remaining larvae were
reared in Petri dishes with filter paper in groups of five to ten (depending on their
size) and fed fresh collard leaves, ad libitum. Upon pupation, they were removed
from the Petri dish and placed in 473ml plastic deli cups (Solo Cup Co.
?
, Highland
Park, IL) to allow space for the adults after eclosion. When the adults emerged they
were used to maintain a lab colony in a 60 X 60 X 60cm cage.
Predator collections focused on the numerically dominant Coleomegilla
maculata (DeGeer), and two sub-dominant predators Coccinella septempunctata
(Linnaeus) and Podisus maculiventris (Say). These species were predators that were
members of the foliar predator assemblage on collards during the summer 2004 (see
chapter 2) and known to be predators of P. rapae (see chapter 3). Predators were
collected in the field, placed in 29.6ml plastic cups (Solo Cup Co.
?
, Highland Park,
IL), kept cool and brought back to the lab to be used in experiments. Even though the
predators were originally collected in collards, they are commonly found in many
cropping systems as well as non-managed systems. In order to maximize collection,
efforts targeted different crops for each predator. Sites in which predators were
collected were located primarily in Maryland, although some early season collection
took place in the North Carolina piedmont area. A C. maculata colony was first
established with individuals from USDA-ARS Biocontrol Laboratory (Beltsville,
MD) and supplemented with individuals collected in collards, sweet corn and alfalfa,
48
and other forage and vegetable crops. C. maculata were fed artificial bee pollen
obtained from the USDA-ARS Biocontrol Laboratory (Beltsville, MD). C.
septempunctata were collected in alfalfa, other forage crops, small grains, and various
vegetables. They were fed P. rapae larvae, black cutworm larvae (Agrotis ipsilon
Hufnagel) obtained from Dow AgriSciences
?
 (Indianapolis, IN), or aphids collected
from alfalfa.  The P. maculiventris colony was established with individuals collected
through the use of pheromone traps set out at Patuxent National Wildlife Refuge
(Laurel, MD) and Beltsville Agricultural Research Center (Beltsville, MD). Traps and
pheromone were obtained from Aldrich, J. R. (USDA-ARS, Beltsville, MD). P.
maculiventris eggs were also purchased through Biocontrol Network
?
 (Brentwood,
TN) and raised to adulthood. Once in the lab, adult coccinellids were reared with
moist cotton and their respective food source in 25 X 25 X 25cm Plexiglas
?
 cages in
the lab at ambient conditions at (~24?C and ~50%rH) until they were used in
mesocosm experiments. Species known to be cannibalistic, such as larval coccinellids
were kept individually in 473ml plastic deli cups (Solo Cup Co.
?
, Highland Park, Il).
All stages of P. maculiventris were reared in 473ml plastic deli cups, at a density of
3-5 individuals per cup to cut down on cannibalism, with moist cotton. P.
maculiventris were fed Colorado potato beetle larvae obtained from a lab colony
(maintained by G. Dively, Entomology Department, UMD) as well as black
cutworms. At the conclusion of each experiment all living predators were returned to
their respective colonies.
49
Mesocosm Design
All experiments were conducted in laboratory mesocosms. Mesocosms for
assemblage manipulations consisted of 30 cm diameter pots containing collard plants
covered with 5-gallon nylon paint strainer bags (National/Ruskin
?
 Inc. Hatfield, PA)
supported by a tomato trellis.  The bags had an elastic band, which secured them to
the pots.
Experimental Protocol
The experimental protocol for this study consisted of determining whether the
number of first instar P. rapae consumed increased significantly when in the presence
of assemblages with the same abundance of predator individuals but comprised of
different number of predator species. Thus, the design of this experiment was a
substitutive experimental design (Snyder et al. 2006), in that the absolute abundance
of predator individuals remained constant throughout all experimental treatments,
while the number of species was varied. This design allows for the isolation of the
effect that predator species richness on prey mortality while eliminating the
potentially confounding effects of total predator density and species composition (see
below for further details).
Each mesocosm contained three five-week-old plants (about 30cm tall). Ten
first instar P. rapae from the lab colony were placed on each of the three plants (30
per mesocosm). This density was higher than that found in the field (personal
observation) in order to emulate the average number of larvae consumed by the most
voracious predator species (5.8 P. rapae larvae/individual predator, see chapter 3). In
50
order to infest the plants in the mesocosms P. rapae larvae were gently transferred
using feather forceps (BioQuip
?
, Rancho Dominguez, CA). A treatment was then
randomly assigned to each mesocosm following the substitutive experimental design
(Snyder et al. 2006). In this design, the absolute number of predator individuals in
each treatment remained constant (six predator individuals) while species richness
was varied. Such a design, with three predator species yielded the following
treatments 1) six C. maculata individuals, 2) six C. septempunctata individuals, 3) six
P. maculiventris individuals, 4) three C. maculata individuals and three C.
septempunctata, 5) three C. maculata individuals and three P. maculiventris 6) three
C. septempunctata individuals and three P. maculiventris, and 7) two individuals of
each of the three predator species.  A no predator control was included to measure
background larval P. rapae mortality. Predators were taken from the lab colony and
randomly assigned to each appropriate treatment. Each individual predator was
starved for 24 hours prior to being placed into the mesocosm. The mesocosms were
randomly assigned a location within a growth chamber in the lab (set at 16L:8D,
24?C, ~70% rH). At the end of a 48-hour time period the predators were removed
from the mesocosms, and the number of living P. rapae larvae counted. All eight
treatments were replicated ten times over time for a total of 80 mesocosms. The
number of replicates of each treatment that were ran at a particular time depended on
the availability of each predator species, but no more than eight were run at a given
time. In order to avoid the loss of degrees of freedom when using a repeated measures
analysis, novel plants and P. rapae larvae were used for each replicate, and the
51
treatment to which predators were assigned was randomized. The location within the
chamber that each mesocosm was placed was re-randomized each time.
The response variable in the experiment, was the number of dead P. rapae
larvae at the end of a 48-hour time period, adjusted for the mortality observed in
controls (i.e., the no predator treatment). Adjusted mortality represented the
difference between number of P. rapae larvae alive at the beginning (30) and at the
end of the time period, minus the mean number of larvae that died in the no-predator
treatment (6.08). These data represented the number of larvae consumed by predators.
The mean P. rapae larval mortality observed among treatments with the same species
richness were pooled, and the means for each richness level were compared. Pooling
treatments was an appropriate approach because the variances were homogeneous.  A
one-tailed one-way analysis of variance (PROC MIXED, SAS
?
 1999) was used to
determine the effect of predator species richness on larval P. rapae mortality, in
which the pooled mean P. rapae mortality for the three one-predator treatments
combined, the pooled mean P. rapae mortality for the three two-predator treatments
combined, and the mean larval P. rapae mortality for the three-predator treatment
were compared. The data satisfied all of the assumptions of ANOVA and a
Bonferroni adjustment was made to control for experiment-wise error rate. The
following six sets of pair-wise comparisons were also made to determine the effect of
predator identity on the relationship between predator richness and prey mortality: (1)
predator treatments 1, 4 and 7 (from above) and (2) predator treatments 1, 5 and 7,
which represent the C. maculata containing treatments; (3) predator treatments 2, 4
and 7 and (4) predator treatments 2, 6 and 7, which represent the C. septempunctata
52
containing treatments; (5) predator treatments 3, 5 and 7, and (6) predator treatments
3, 6 and 7, which represent the P. maculiventris containing treatments. Bonferroni
adjustments were made for each set of pair-wise comparisons to control for
experiment-wise error rate.
Results
There was a significant relationship between species richness of an
assemblage of predators and prey mortality; however, the relationship was non-linear
(Figure 4.2). Increased predator species richness from one species to two species led
to a significant increase in mean larval P. rapae mortality (F = 14.7, p = 0.001)
however, there was no significant increase in mean larval P. rapae mortality when the
richness was increased from two species to three (F = 4.03, p = 0.727). In fact,
although the mean larval P. rapae mortality in the three species treatment was greater
than that of the one species treatments, that difference was not significant (F = 0.49, p
= 0.074) and thus the change in prey mortality from two to three species may indeed
represent a decline in prey mortality.
Predator species identity did affect the relationship between species richness
of a predator assemblage and prey mortality. In the first two set of comparisons,
where I focused on C. maculata-containing assemblages, as predator species richness
increased prey mortality also increased. However, the increase in prey mortality
appeared to level off in the three species assemblage (Figure 4.3). In comparison 1
(Figure 4.3a), there was no significant difference in mean larval P. rapae mortality
imposed by C. maculata (the single species ?assemblage?) and the two species
53
assemblage consisting of C. maculata and C. septempunctata (t = -2.16, p = 0.051).
However, there was a significant increase the mortality imposed by the three predator
species assemblages (t = -2.46, p = 0.025) compared to C. maculata alone, although
there was no significant differences imposed by the three species assemblage and the
assemblage comprised of C. maculata and C. septempunctata (t = -0.29, p = 1.00). In
comparison 2 (Figure 4.3b), mean larval P. rapae mortality imposed by the two
species (C. maculata and P. maculiventris) assemblage as well as the three species
(C. maculata, P. maculiventris and C. septempunctata) assemblage were significantly
greater than C. maculata alone (t = -3.83, p = 0.001 and t = -2.46, p = 0.025,
respectively). There was no significant difference imposed by the two and three
species assemblages (t = 1.38, p = 0.260).
The next two sets of comparisons focused on C. septempunctata-containing
assemblages. For this set of comparisons, the relationship between predator richness
and prey mortality differed from those noted above (Figure 4.4). In the first
comparison (Figure 4.4a), there was no relationship between predator richness and
prey mortality. That is, there was no significant difference in mean larval P. rapae
mortality imposed by C. septempunctata alone and the two species (C.
septempunctata and C. maculata) assemblages (t = -1.72, p = 0.135), or C.
septempunctata alone and the three species (C. septempunctata, C. maculata and P.
maculiventris) assemblages (t = -2.02, p = 0.072). However, the second comparison
(Figure 4.4b) differs in that there is a relationship between species richness and prey
mortality when one compares C. septempunctata alone and the two species (C.
septempunctata and P. maculiventris) assemblage (t = -2.65, p = 0.015). There was
54
no significant difference between the three species assemblage and C. septempunctata
alone (t = -2.02, p = 0.072) or when the two and three species assemblages were
compared. (t = 0.06, p = 1.00).
The final two sets of comparisons focused on P. maculiventris-containing
assemblages. There was no relationship between predator richness and prey mortality
(Figure 4.5). In the first comparison (Figure 4.5a), there was no significant difference
in the mean larval P. rapae mortality imposed by P. maculiventris alone, a two
species (P. maculiventris and C. maculata) assemblage, or a three species (P.
maculiventris, C. maculata and C. septempunctata) (t = -1.82, p = 0.110 and t = -0.44,
p = 1.00, respectively). There was also no significant difference in mortality imposed
by the two and three species assemblages (t = 1.38, p = 0.260). The second
comparison (Figure 4.5b) yielded the same results. That is, there was no significant
difference in the mean larval P. rapae mortality imposed by P. maculiventris alone, a
two species (P. maculiventris and C. septempunctata) assemblage (t = -1.08, p =
0.425), or a three species (P. maculiventris, C. maculata and C. septempunctata)
assemblage (t = -0.44, p = 1.00). There was also no significant difference in mortality
imposed by the two and three species assemblages (t = 0.64, p = 1.00).
Discussion
The results of the analyses supported the hypothesis that increased species
richness of an assemblage of predators can lead to an increase in prey mortality
imposed by the assemblage. However, this relationship was not found to be linear,
and dependent on the identity of the species in the assemblage. When all one species
assemblages were compared to all two and three species assemblages (Figure 4.2),
55
larval P. rapae mortality increased initially, but then leveled off at the highest level of
species richness. When put in the context of the nine possible outcomes (discussed in
the introduction) the relationship was best described by the lines in Figures 4.1c and
4.1d. That is, an increase in mortality imposed when a single species is compared to a
two species assemblage followed by either a leveling off of mortality (as in Figure
4.1c), or a decline in mortality imposed (as in Figure 4.1d). The results of statistical
analysis do not allow a clear distinction of these two interpretations of the changes
between two and three species assemblages.
Predator identity, however, does have an impact on the relationship between
species richness and prey mortality, and thus results vary across assemblages. I
explored how species identity affected the results through six sets of comparisons. In
some cases the results were ambiguous, as in the pooled results, such that the
relationship between richness and mortality was not clear. For example, in
comparisons in which C. septempunctata was the focal predator, the addition of the
third predator (i.e., C. maculata) did not lead to a significant increase in P. rapae
mortality (Figure 4.4b). In fact, mortality imposed by the three-predator assemblage
was not significantly different from that imposed by C. septempunctata alone. Thus
the relationship was that which is depicted by either Figure 4.1c or 4.1d. In contrast,
in the comparison in which C. maculata was the focal predator, the addition of C.
septempunctata to create a two species assemblage did not lead to a significant
change in P. rapae mortality, but when all three predators were present there was a
significant difference in mortality compared to C. maculata alone (Figure 4.3a). This
relationship between predator richness and prey mortality was best represented by the
56
relationship depicted in Figures 4.1c and 4.1g. For the other sets of comparisons, the
interpretation was more straightforward. In comparison 2 (Figure 4.3b), there was a
direct correlation between predator richness and prey mortality up to a point, after
which there was a leveling off of mortality as richness increased. For three of the
comparisons, 3, 5 and 6 (Figures 4.4a, 4.5a and 4.5b respectively), no relationship
existed between predator richness and prey mortality. The fact that such a variety of
conclusions can be made from these data suggests that the relationship between
predator richness and prey mortality can be mitigated by the identity of predator
species in the assemblage.
The idea, that species identity may dictate an assemblage?s effectiveness, has
been recently supported by Wilby (2005) and Straub and Snyder (2006). These
findings suggest there is an idiosyncratic nature to how effective assemblages of
predators can be depending upon the identity of the predators. This idea along with
the findings of my study, suggest that a thorough understanding of the predator
species in the assemblage being conserved, as well as an understanding of the
interactions between those species, are necessary for conservation biological control
to be effective.
The leveling off of, and in some cases the decrease in, prey mortality as
predator richness increased may be explained by considering the interactions
occurring between predator species. When multiple species occur in an assemblage,
interactions between the species are inevitable. The outcome of those interactions can
affect the functioning of the ecosystem in which the community resides. In terms of
predator assemblages, interactions can yield beneficial ecosystem functions [so-called
57
positive multiple predator effects (MPEs), sensu (Sih et al. 1998)] as well as
outcomes that are detrimental to the functioning of  ecosystems (i.e., negative MPEs).
Mechanisms by which positive MPEs arise include synergism (Losey and Denno
1998), facilitation (Bruno et al. 2003) and resource partitioning (Townsend and
Hildrew 1979). Facilitation and synergism differ in that, while facilitation benefits
one participant in the interaction, in synergistic predator-predator interactions both
predators are benefited. There exist far greater evidence of mechanisms for how
negative MPEs such as intraguild predation (Polis et al. 1989, Rosenheim et al. 1993)
and interference (Snyder and Wise 1999) occur. In order for an assemblage to
effectively suppress pests, the positive MPEs must outweigh the negative MPEs. In
fact, assemblages have been shown be effective at influencing pest numbers despite
the potentially negative effects of intraguild predation (Snyder and Ives 2003) and
interference (Lang 2003). In my study, I asked if increasing the number of predators
in an assemblage, despite the increased likelihood of negative MPEs occurring, can
still lead to a situation in which there is an increase in natural suppression of pests.
What I found was that increasing the species richness of a predator assemblage can
lead to a corresponding increase in prey mortality, but that relationship is not linear.
In fact, it may have been the case that at higher species richness prey mortality
decreased. In order to explain this pattern, further investigation into negative MPEs
other than intraguild predation (i.e. cannibalism and interference) would be necessary.
A recent meta-analysis looked at the effects of species richness of various
trophic groups including predators and their depletion of a resource, prey in the case
of predators (Cardinale et al. 2006). Although diverse ?polycultures? of predators on
58
average consumed more prey than ?monocultures,? the highest performing
?monoculture? was not statistically distinguishable from the ?polyculture.? This
suggests that a single predator could be more effective at suppressing pests than an
assemblage of predators. To the contrary, the Species Assemblage Control
Hypothesis suggests that an assemblage of predators can be more effective at
suppressing pest than a single natural enemy (Riechert and Lockley 1984, Provencher
and Riechert 1994, Riechert and Lawrence 1997). And, the results of my study show
that it?s not the entire assemblage of predators present in the agroecosystem that is
responsible for pest suppression, as suggested by SACH. Rather, it?s a subset of the
predator assemblage that is actually responsible for pest suppression, and that the
identity of predators in the assemblage influences how the assemblage functions. So
how do we reconcile the contradictory evidence emerging about the relationship
between biodiversity and biological control? In the future, experiments that look at
assemblages with greater numbers of predators in field conditions are needed to gain
further insight into the relationship biodiversity and biological control, and determine
how this relationship can be altered by predator species identity. It is clear that a
better understanding of the predator assemblage influencing a target pest, specifically
an understanding of predator-predator interactions as well as predator-prey
interaction, is needed before conservation decisions can be made that provide
effective pest suppression.
59
Figure 4.1. Generalized results graphs. All possible results for an experimental
design with three levels of species richness.
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
Richness
Mortality
a   b
   f
   c
  e
  h   g    i
   d
60
Figure 4.2. Relationship between predator richness and prey mortality. Adjusted P.
rapae mortality was pooled across all treatments with the same richness and then
compared between treatments with 1 species, 2 species and 3 species.
0
5
10
15
20
25
30
1 Species2 Species3 Species
Predator Richness
Mean Number of 
P. rapae
Eaten 
(adjusted mortality) 
  a
     b
 ab
61
Figure 4.3. C. maculata (Cmac) in one, two and three species assemblages. In figure
4.3a, treatments with C. maculata alone, treatments with C. maculata and C.
septempunctata (C7), and treatments with all three predators were compared.
Figure 4.3b differs in that the two-predator treatments contained C. maculata
and P. maculiventris (Pmac), however the same comparisons were made.
0
5
10
15
20
25
30
CmacCmac + PmacCmac + C7 + Pmac
Treatment
Mean Number of P. rapae Eaten 
(adjusted mortality)
3b
   a
   b
    b
0
5
10
15
20
25
30
CmacCmac + C7Cmac + C7 + Pmac
Treatment
Mean Number of 
P. rapae
Eaten 
(adjusted mortality)
3a
  a
    ab
  b
62
Figure 4.4. C. septempunctata (C7) in one, two and three species assemblages. In
figure 4.4a, treatments with C. septempunctata alone, treatments with C.
septempunctata and C. maculata (Cmac), and treatments with all three predators
were compared. Figure 4.4b differs in that two-predator treatments contained C.
septempunctata and P. maculiventris (Pmac), but same comparisons were made.
0
5
10
15
20
25
30
C7 C7 + CmacCmac + C7 + Pmac
Treatment
Mean Number of P. rapae Eaten 
(adjusted mortality)
4a
a
   a
 a
0
5
10
15
20
25
30
C7 C7 + PmacCmac + C7 + Pmac
Treatment
Mean Number of P. rapae Eaten 
(adjusted mortality)
4b
a
   b
ab
63
Figure 4.5. P. maculiventris (Pmac) in one, two and three species assemblages. In
figure 4.5a, treatments with P. maculiventris alone, treatments with P.
maculiventris and C. maculata (Cmac), and treatments with all three predators
were compared. Figure 4.5b differs in that the two-predator treatments contained
P. maculiventris and C. septempunctata, however same comparisons were made.
0
5
10
15
20
25
30
PmacPmac + CmacCmac + C7 + Pmac
Treatment
Mean Number of P. rapae Eaten 
(adjusted mortality)
5a
 a
     a
  a
0
5
10
15
20
25
30
PmacPmac + C7Cmac + C7 + Pmac
Treatment
Mean Number of P. rapae Eaten 
(adjusted mortality)
5b
 a
  a
  a
64
Appendix: Data from the assessment of the predator
assemblages found in the epigeal, foliar and aerial microhabitats
of the collard agroecosystem
Pitfall Samples
Heteroptera
RankAbundanceIDSpecies Family
1171 2.1.04Orius insidiosusAnthocoridae
822 2.1.02Nabis roseipennisNabidae
1181 2.1.05Barce fraternaReduviidae
1191 2.1.06Melanolestes picipesReduviidae
812 2.1.01Micracanthia humilisSaldidae
1201 2.1.08 Unknown
Hymenoptera
RankAbundanceIDSpecies Family
2 350 4.1.01Lasius alienusFormicidae
5 172
4.1.03,
4.1.12,
4.1.13Pheidole bicarinata vinelandicaFormicidae
8 106 4.1.04 Formicidae
1352 4.1.02Tetramorium caespitumFormicidae
1634 4.1.05Aphaenogaster sp.Formicidae
585 4.1.07Monomorium minimumFormicidae
595 4.1.09Aphaenogaster sp.Formicidae
942 4.1.10Tetramorium sp.Formicidae
952 4.1.15 Formicidae
1451 4.1.11Tetramorium sp.Formicidae
1471 4.1.16Formica (fusca) sp.Formicidae
1481 4.1.17Hypoponera sp.Formicidae
1491 4.1.18 Formicidae
1461 4.1.14 Mutilidae
Coleoptera
RankAbundanceIDSpecies Family
7 116 3.1.03Pterostichus lucublandusCarabidae
1443 3.1.15Bembidion semistrictumCarabidae
1732 3.1.14 Carabidae
65
2026 3.1.18Stenolophus ochropezusCarabidae
2323 3.1.06 Carabidae
2619 3.1.09 Carabidae
3510 3.1.07 Carabidae
3410 3.1.12 Carabidae
496 3.1.02 Carabidae
506 3.1.04 Carabidae
486 3.1.16 Carabidae
545 3.1.05 Carabidae
644 3.1.20 Carabidae
852 3.1.08 Carabidae
832 3.1.11 Carabidae
842 3.1.13 Carabidae
1211 3.1.01 Carabidae
1221 3.1.10 Carabidae
1231 3.1.17 Carabidae
1241 3.1.19 Carabidae
1251 3.1.21 Carabidae
3114
3.3.2,
3.3.1,
3.3.3Coleomegilla maculataCoccinellidae
1 575 3.2.05Amisch sp.Staphylinidae
1542 3.2.04Amisch sp.Staphylinidae
1832 3.2.06 Staphylinidae
2423 3.2.01 Staphylinidae
2816 3.2.03 Staphylinidae
3014 3.2.19 Staphylinidae
3211 3.2.15 Staphylinidae
3610 3.2.02 Staphylinidae
447 3.2.08 Staphylinidae
437 3.2.25 Staphylinidae
555 3.2.14 Staphylinidae
565 3.2.18 Staphylinidae
575 3.2.26 Staphylinidae
713 3.2.07 Staphylinidae
723 3.2.09 Staphylinidae
703 3.2.23 Staphylinidae
862 3.2.16 Staphylinidae
872 3.2.17 Staphylinidae
882 3.2.22 Staphylinidae
1261 3.2.10 Staphylinidae
1271 3.2.11 Staphylinidae
66
1281 3.2.12 Staphylinidae
1291 3.2.13 Staphylinidae
1301 3.2.20 Staphylinidae
1311 3.2.21 Staphylinidae
1321 3.2.24 Staphylinidae
1331 3.2.27 Staphylinidae
3710 3.4.07 Unknown
418 3.4.16 Unknown
516 3.4.06 Unknown
733 3.4.01 Unknown
743 3.4.13 Unknown
912 3.4.03 Unknown
922 3.4.04 Unknown
932 3.4.05 Unknown
892 3.4.10 Unknown
902 3.4.20 Unknown
1411 3.4.02 Unknown
1431 3.4.08 Unknown
1441 3.4.09 Unknown
1341 3.4.11 Unknown
1351 3.4.12 Unknown
1361 3.4.14 Unknown
1371 3.4.15 Unknown
1381 3.4.17 Unknown
1391 3.4.18 Unknown
1401 3.4.19 Unknown
1421 3.4.21 Unknown
Araneae
RankAbundanceIDSpecies Family
1163 1.2.02 Linyphiidae
2125 1.2.09 Linyphiidae
2223 1.2.06 Linyphiidae
2522 1.2.03 Linyphiidae
2914 1.2.07 Linyphiidae
3310 1.2.13 Linyphiidae
427 1.2.01 Linyphiidae
466 1.2.17 Linyphiidae
476 1.2.08 Linyphiidae
535 1.2.31 Linyphiidae
614 1.2.14 Linyphiidae
624 1.2.30 Linyphiidae
67
634 1.2.32 Linyphiidae
673 1.2.20 Linyphiidae
683 1.2.24 Linyphiidae
693 1.2.27 Linyphiidae
762 1.2.18 Linyphiidae
772 1.2.22 Linyphiidae
782 1.2.23 Linyphiidae
792 1.2.26 Linyphiidae
971 1.2.10 Linyphiidae
981 1.2.11 Linyphiidae
1001 1.2.19 Linyphiidae
1011 1.2.21 Linyphiidae
1021 1.2.25 Linyphiidae
1031 1.2.28 Linyphiidae
1041 1.2.29 Linyphiidae
1161 1.2.05 Linyphiidae
3 216 1.1.08Pardosa sp.Lycosidae
4 200 1.1.02Pardosa sp.Lycosidae
6 149 1.1.11Pardosa sp.Lycosidae
9 93 1.1.07Pardosa sp.Lycosidae
1090 1.1.01Pardosa sp.Lycosidae
1259 1.1.12Pardosa sp.Lycosidae
2718 1.1.05Pardosa sp.Lycosidae
389 1.1.14Pardosa sp.Lycosidae
399 1.2.15 Lycosidae
408 1.2.04 Lycosidae
456 1.1.04Pardosa sp.Lycosidae
525 1.1.09 Lycosidae
604 1.1.06Pardosa sp.Lycosidae
653 1.1.10 Lycosidae
663 1.1.13Pardosa sp.Lycosidae
961 1.1.03Pardosa sp.Lycosidae
991 1.2.16 Thomisidae
802 1.2.36 Unknown
1051 1.2.33 Unknown
1061 1.2.34 Unknown
1071 1.2.35 Unknown
1081 1.2.37 Unknown
1091 1.2.38 Unknown
1101 1.2.39 Unknown
1111 1.2.40 Unknown
1121 1.2.41 Unknown
68
1131 1.2.42 Unknown
1141 1.2.43 Unknown
1151 1.2.44 Unknown
Opiliones
RankAbundanceIDSpecies Family
1511 6.1.1 Unknown
Chilopoda
RankAbundanceIDSpecies Family
1927 5.1.1 Unknown
753 5.1.2 Unknown
1501 5.1.3 Unknown
  
Visual Samples
Heteroptera
RankAbundanceIDSpecies Family
4 11
2.1.01,
2.1.06,
2.1.08Lygus lineolarisMiridae
2 21 2.1.02Nabis roseipennisNabidae
153 2.1.03Podisus maculiventrisPentatomidae
401 2.1.11Euschistus servusPentatomidae
Hymenoptera
RankAbundanceIDSpecies Family
421 4.1.1Tetramorium sp.Formicidae
Diptera
RankAbundanceIDSpecies Family
431 5.1.1 Syrphidae
Coleoptera
RankAbundanceIDSpecies Family
252 3.4.1Chauliognathus marginatusCantharidae
1 102 3.3.1Coleomegilla maculataCoccinellidae
9 4 3.3.5Coccinella septempunctataCoccinellidae
411 3.4.2 Lampyridae
Araneae
RankAbundanceIDSpecies Family
8 4 1.4.1 Araneidae
69
113 1.2.02 Araneidae
5 5 1.1.02 Lycosidae
103 1.1.03 Lycosidae
261 1.1.01 Lycosidae
133 1.2.15 Salticidae
212 1.2.18 Salticidae
451 1.2.24 Salticidae
3 12 1.3.01 Tetragnathidae
391 1.5.1 Thomisidae
6 4 1.2.07 Unknown
7 4 1.2.17 Unknown
123 1.2.10 Unknown
143 1.3.02 Unknown
162 1.2.01 Unknown
172 1.2.03 Unknown
182 1.2.04 Unknown
192 1.2.12 Unknown
202 1.2.13 Unknown
222 1.2.21 Unknown
232 1.2.22 Unknown
242 1.2.26 Unknown
271 1.2.05 Unknown
281 1.2.06 Unknown
291 1.2.08 Unknown
301 1.2.09 Unknown
311 1.2.11 Unknown
321 1.2.14 Unknown
331 1.2.16 Unknown
341 1.2.19 Unknown
351 1.2.20 Unknown
441 1.2.23 Unknown
361 1.2.25 Unknown
371 1.2.27 Unknown
381 1.2.28 Unknown
  
Sweep Samples
Heteroptera
RankAbundanceIDSpecies Family
2 45 2.1.03Orius insidiosusAnthocoridae
143 2.1.17Jalysus wickhamiBerytidae
172 2.1.10Geocoris punctipesGeocoridae
1 318 2.1.02,
2.1.11,
2.1.04,
2.1.07
Lygus lineolarisMiridae
70
2.1.11,
2.1.04,
2.1.07
4 17 2.1.12Polymerus basalisMiridae
182 2.1.09Trigonotylus caelestialiumMiridae
251 2.1.01Nabis roseipennisNabidae
261 2.1.19Sinea sp. Reduviidae
134
2.1.06,
2.1.13Micracanthia humilisSaldidae
Hymenoptera
RankAbundanceIDSpecies Family
116
4.1.1,
4.1.2Monomorium minimumFormicidae
192 4.1.3Lasius alienusFormicidae
Coleoptera
RankAbundanceIDSpecies Family
3 37
3.1.3,
3.1.2Coleomegilla maculataCoccinellidae
202
3.1.5,
3.1.7Harmonia axyridisCoccinellidae
291 3.1.6 Carabidae
153 2.1.05Chauliognathus marginatusCantharidae
212 3.1.4 Staphylinidae
7 8 5.1.2 Unknown
301 3.1.01 Staphylinidae
Araneae
RankAbundanceIDSpecies Family
5 10 1.1.05 Unknown
6 9 1.1.08 Unknown
9 7 1.1.03 Unknown
107 1.1.04 Unknown
125 1.1.07 Unknown
232 1.1.02 Unknown
242 1.1.06 Unknown
271 2.1.16 Unknown
281 2.1.18 Unknown
311 1.1.10 Unknown
8 8 1.1.01 Tetragnathidae
222 1.1.09 Salticidae
71
Chilopoda
RankAbundanceIDSpecies Family
163 5.1.1 Unknown
72
References
Agarwala, B. K., and A. F. G. Dixon. 1992. Laboratory Study of Cannibalism and
Interspecific Predation in Ladybirds. Ecological Entomology 17:303-309.
Ashby, J. W. 1974. A Study of Arthropod Predation of Pieris rapae L. Using
Serological and Exclusion Techniques. The Journal of Applied Ecology
11:419-425.
Barbosa, P., editor. 1998. Conservation Biological Control Academic Press, San
Diego, CA.
Barbosa, P., A. Caldas, and S. E. Reichert. 2005. Species abundance distribution and
predator-prey interactions: theoretical and applied consequences. in P.
Barbosa and I. Castellanos, editors. Ecology of Predator-Prey Interactions.
Oxford University Press, Oxford.
Bruno, J. F., J. J. Stachowicz, and M. D. Bertness. 2003. Inclusion of facilitation into
ecological theory. Trends in Ecology & Evolution 18:119-125.
Caltagirone, L. E. 1981. Landmark Examples in Classical Biological-Control. Annual
Review of Entomology 26:213-232.
73
Cardinale, B. J., C. T. Harvey, K. Gross, and A. R. Ives. 2003. Biodiversity and
biocontrol: emergent impacts of a multi-enemy assemblage on pest
suppression and crop yield in an agroecosystem. Ecology Letters 6:857-865.
Cardinale, B. J., D. S. Srivastava, J. E. Duffy, J. P. Wright, A. L. Downing, M.
Sankaran, and C. Jouseau. 2006. Effects of biodiversity on the functioning of
trophic groups and ecosystems. Nature 443:989-992.
Chang, G. C., and P. Kareiva. 1999. The case for indigenous generalists in biological
control. Pages 103-115 in B. A. Hawkins and H. V. Cornell, editors.
Theoretical Approaches to Biological Control. Cambridge University Press,
Cambridge, UK.
Chapin, F. S., E. S. Zavaleta, V. T. Eviner, R. L. Naylor, P. M. Vitousek, H. L.
Reynolds, D. U. Hooper, S. Lavorel, O. E. Sala, S. E. Hobbie, M. C. Mack,
and S. Diaz. 2000. Consequences of changing biodiversity. Nature 405:234-
242.
Claridge, M. F. 1987. Insect assemblages - diversity, organization, and evolution. in J.
H. R. Gee and P. S. Giller, editors. Organization of communities. Past and
present. 27th Symposium of the British Ecological Society. Blackwell
Science, London.
74
DeBach, P., editor. 1964. Biological control of insect pests and weeds. Chapman &
Hall, London.
DeBach, P. 1974. Biological control by natural enemies. Cambridge University Press,
London ; New York, N.Y.
Dempster, J. P. 1967. The Control of Pieris rapae with DDT. I. The Natural Mortality
of the Young Stages of Pieris. The Journal of Applied Ecology 4:485-500.
Denoth, M., L. Frid, and J. H. Myers. 2002. Multiple agents in biological control:
improving the odds? Biological Control 24:20-30.
Dinter, A. 2002. Microcosm studies on intraguild predation between female erigonid
spiders and lacewing larvae and influence of single versus multiple predators
on cereal aphids. Journal of Applied Entomology-Zeitschrift Fur Angewandte
Entomologie 126:249-257.
Doutt, R. L. 1967. Biological control. . Pages 4-29 in W. W. Kilgore and R. L. Doutt,
editors. Pest Control: Biological, Physical and Selected Chemical Methods. .
Academic Press, New York, NY.
Eilenberg, J., A. Hajek, and C. Lomer. 2001. Suggestions for unifying the
terminology in biological control. Biocontrol 46:387-400.
75
Finke, D. L., and R. F. Denno. 2004. Predator diversity dampens trophic cascades.
Nature 429:407-410.
Harcourt, D. G. 1961. Spatial pattern of the imported cabbageworm, Pieris rapae
(L.)(Lepidoptera: Pieridae), on cultivated cruciferae. The Canadian
Entomologist 93:945-952.
Hochberg, M. E. 1996. Consequences for host population levels of increasing natural
enemy species richness in classical biological control. American Naturalist
147:307-318.
Hooks, C. R. R., R. R. Pandey, and M. W. Johnson. 2003. Impact of avian and
arthropod predation on lepidopteran caterpillar densities and plant
productivity in an ephemeral agroecosystem. Ecological Entomology 28:522-
532.
Howarth, F. G. 1991. Environmental Impacts of Classical Biological-Control. Annual
Review of Entomology 36:485-509.
Ives, A. R., B. J. Cardinale, and W. E. Snyder. 2005. A synthesis of subdisciplines:
predator-prey interactions, and biodiversity and ecosystem functioning (vol 8,
pg 102, 2005). Ecology Letters 8:782-782.
76
Jones, R. E. 1977. Search behaviour: a study of three caterpillar species. Behaviour
60:237-257.
Landis, D. A., S. D. Wratten, and G. M. Gurr. 2000. Habitat management to conserve
natural enemies of arthropod pests in agriculture. Annual Review of
Entomology 45:175-201.
Lang, A. 2003. Intraguild interference and biocontrol effects of generalist predators in
a winter wheat field. Oecologia 134:144-153.
Loreau, M. 2000. Biodiversity and ecosystem functioning: recent theoretical
advances. Oikos 91:3-17.
Loreau, M., S. Naeem, P. Inchausti, J. Bengtsson, J. P. Grime, A. Hector, D. U.
Hooper, M. A. Huston, D. Raffaelli, B. Schmid, D. Tilman, and D. A. Wardle.
2001. Ecology - Biodiversity and ecosystem functioning: Current knowledge
and future challenges. Science 294:804-808.
Losey, J. E., and R. F. Denno. 1998. Positive predator-predator interactions:
Enhanced predation rates and synergistic suppression of aphid populations.
Ecology 79:2143-2152.
77
Madsen, M., S. Terkildsen, and S. Toft. 2004. Microcosm studies on control of aphids
by generalist arthropod predators: Effects of alternative prey. Biocontrol
49:483-504.
Mallampalli, N., I. Castellanos, and P. Barbosa. 2002. Evidence for intraguild
predation by Podisus maculiventris on a ladybeetle, Coleomegilla maculata:
Implications for biological control of Colorado potato beetle, Leptinotarsa
decemlineata. Biocontrol V47:387-398.
Pearson, D. E., and R. M. Callaway. 2003. Indirect effects of host-specific biological
control agents. Trends in Ecology & Evolution 18:456-461.
Polis, G. A., C. A. Myers, and R. D. Holt. 1989. The Ecology and Evolution of
Intraguild Predation - Potential Competitors That Eat Each Other. Annual
Review of Ecology and Systematics 20:297-330.
Provencher, L., and S. E. Riechert. 1994. Model and Field-Test of Prey Control
Effects by Spider Assemblages. Environmental Entomology 23:1-17.
Riechert, S. E., and K. Lawrence. 1997. Test for predation effects of single versus
multiple species of generalist predators: spiders and their insect prey.
Entomologia Experimentalis Et Applicata 84:147-155.
78
Riechert, S. E., and T. Lockley. 1984. Spiders as Biological-Control Agents. Annual
Review of Entomology 29:299-320.
Rosenheim, J. A., H. K. Kaya, L. E. Ehler, J. J. Marois, and B. A. Jaffee. 1995.
Intraguild Predation Among Biological-Control Agents: Theory and Evidence.
Biological Control 5:303-335.
Rosenheim, J. A., L. R. Wilhoit, and C. A. Armer. 1993. Influence of Intraguild
Predation among Generalist Insect Predators on the Suppression of an
Herbivore Population. Oecologia 96:439-449.
Schellhorn, N. A., and D. A. Andow. 1999. Mortality of coccinellid (Coleoptera :
Coccinellidae) larvae and pupae when prey become scarce. Environmental
Entomology 28:1092-1100.
Schellhorn, N. A., and V. L. Sork. 1997. The impact of weed diversity on insect
population dynamics and crop yield in collards, Brassica oleraceae
(Brassicaceae). Oecologia 111:233-240.
Schmaedick, M. A., and A. M. Shelton. 2000. Arthropod predators in cabbage
(Cruciferae) and their potential as naturally occurring biological control
agents for Pieris rapae (Lepidoptera : Pieridae). Canadian Entomologist
132:655-675.
79
Sih, A., G. Englund, and D. Wooster. 1998. Emergent impacts of multiple predators
on prey. Trends in Ecology & Evolution 13:350-355.
Snyder, W. E., S. N. Ballard, S. Yang, G. M. Clevenger, T. D. Miller, J. J. Ahn, T. D.
Hatten, and A. A. Berryman. 2004. Complementary biocontrol of aphids by
the ladybird beetle Harmonia axyridis and the parasitoid Aphelinus asychis on
greenhouse roses. Biological Control 30:229-235.
Snyder, W. E., G. C. Chang, and R. P. Prasad. 2005. Conservation Biological
Control: Biodiversity Influences the Effectiveness of Predators. in P. Barbosa
and I. Castellanos, editors. Ecology of Predator-Prey Interactions. Oxford
University Press, Oxford.
Snyder, W. E., and A. R. Ives. 2003. Interactions between specialist and generalist
natural enemies: Parasitoids, predators, and pea aphid biocontrol. Ecology
84:91-107.
Snyder, W. E., G. B. Snyder, D. L. Finke, and C. S. Straub. 2006. Predator
biodiversity strengthens herbivore suppression. Ecology Letters 9:789-796.
Snyder, W. E., and D. H. Wise. 1999. Predator Interference and the Establishment of
Generalist Predator Populations for Biocontrol. Biological Control 15:283-
292.
80
Stiling, P. 2004. Biological control not on target. Biological Invasions 6:151-159.
Straub, C. S., and W. E. Snyder. 2006. Species identity dominates the relationship
between predator biodiversity and herbivore suppression. Ecology 87:277-
282.
Suenaga, H., and T. Hamamura. 2001. Occurrence of carabid beetles (Coleoptera :
Carabidae) in cabbage fields and their possible impact on lepidopteran pests.
Applied Entomology and Zoology 36:151-160.
Symondson, W. O. C., K. D. Sunderland, and M. H. Greenstone. 2002. Can generalist
predators be effective biocontrol agents? Annual Review of Entomology
47:561-594.
Tilman, D. 2000. Causes, consequences and ethics of biodiversity. Nature 405:208-
211.
Townsend, C. R., and A. G. Hildrew. 1979. Resource Partitioning by Two Freshwater
Invertebrate Predators with Contrasting Foraging Strategies. The Journal of
Animal Ecology 48:909-920.
Wiedenmann, R. N., and R. J. O'Neil. 1991. Searching behavior and time budgets of
the predator Podisus maculiventris. Entomologia Experimentalis Et Applicata
60:83-93.
81
Wilby, A., S. C. Villareal, L. P. Lan, K. L. Heong, and M. B. Thomas. 2005.
Functional benefits of predator species diversity depend on prey identity.
Ecological Entomology 30:497-501.