THE POPULATION AND COMMUNITY ECOLOGY OF SMALL FRESHWATER PONDS: ASSIGNING PROCESS TO PATTERN
BY
CHRISTOPHER J. HOLMES
DISSERTATION
Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biology with a concentration in Ecology, Ethology, and Evolution in the Graduate College of the University of Illinois at Urbana-Champaign, 2019
Urbana, Illinois
Doctoral Committee:
Professor Carla E. Cáceres, Chair and Director of Research Associate Professor Brian F. Allan Dr. Ephantus J. Muturi, United States Department of Agriculture Associate Professor James P. O’Dwyer Professor Andrew V. Suarez ABSTRACT
Ecologists have long been intrigued by patterns of spatial structuring for populations and communities inhabiting natural and, more recently, human-created ecosystems. Empirical and theoretical advancements have highlighted the importance of considering the effects of both historical (i.e., colonization history and priority effects) and contemporary processes (e.g., species sorting and interspecific interactions) when studying population and community dynamics. Multiple studies have documented that divergent population and community structures can arise in similar habitats when colonization history differs. For example, early colonists may hinder, inhibit, or in some cases facilitate colonization by later arriving taxa by altering the suitability of a habitat, especially in actively dispersing organisms. The importance of both abiotic and biotic factors on the establishment and subsequent success in a habitat has been well documented in a wide variety of taxa, though the relative importance of these processes has been shown to vary significantly among systems. Furthermore, the spatial distribution of patches in the landscape will shape the nature of these biotic interactions and thus have profound effects on local and regional processes. Given the complexity of these simultaneously acting factors, generating accurate predictions for the outcome of community and population structuring remains difficult for most systems.
In much of the developed world, human alteration of the landscape has necessitated the creation of safe and efficient stormwater management infrastructure. However, a by-product of this practice has included the development of newly created small ponds, which have been shown to harbor larval mosquitoes and other insects, crustacean zooplankton, and a wide range of other vertebrate and invertebrate organisms. Given their ubiquity and potential to harbor diverse communities, small stormwater ponds provide a unique opportunity in which to study the
ii mechanisms underlying the formation and dynamics of populations and communities. To this end, I use mosquito and zooplankton communities inhabiting newly created ponds as a model system to empirically and theoretically explore the factors underlying population and community structure. In Chapter 1, I use a stochastic and spatially explicit model to examine how pond network structure and the number and identity of ponds stocked, or removed, from the landscape contributes to overall patterns of metapopulation occupancy and robustness in a focal zooplankton species, Daphnia pulex. I parameterize this model with four-years of D. pulex occupancy data from a small network of 38 newly-constructed forested ponds at Svend O.
Heiberg Memorial Forest (Tully, NY, USA). I show that the location of patches stocked or removed from the pond network has contrasting effects on metapopulation occupancy and persistence. When centrally-located ponds were removed from the network, the metapopulation collapsed rapidly. However, when initially founding a metapopulation, the location of ponds stocked does not appear to play an important role. Furthermore, I introduce a simple differential equation model that qualitatively matches results predicted by the stochastic simulations, but is less time intensive and computationally expensive to analyze. Chapters 2 and 3 examine larval mosquito and zooplankton communities inhabiting subsets of a 37 stormwater pond network in
Champaign County, Illinois (USA) and provide insights as to the relative importance of the biotic and abiotic environment on the abundance and distribution of larval mosquitoes. In
Chapter 2, I show that interspecific variation in predator- and competitor- avoidance behavior during the initial colonization by ovipositing mosquitoes may explain the negative association between zooplankton and mosquitoes in a multi-year field survey. In Chapter 3, I use structural equation modeling to explore the direct and indirect effects of multiple biotic and abiotic factors on the larval abundance of three common species of culicine mosquitoes (Culex pipiens, Culex
iii restuans, and Aedes vexans). I found that the three species varied in response to these factors.
Predator abundance, which was driven by hydroperiod, was negatively correlated with Cx. pipiens abundance and positively correlated with Ae. vexans abundance. However, we found no variables that explained variation in the abundance of Cx. restuans. Combined, these studies highlight the complexity of ecological interactions that may occur in small ponds and how the relative importance of these interactions may vary among closely related species.
iv
ACKNOWLEDGEMENTS
My research journey would not have been possible without the generous and continuous support of many individuals in my life. Many years ago, Dr. Carla Cáceres took a chance on me and provided me, a young and naïve undergraduate at the time, the opportunity to conduct research in her lab. After mentoring me through three degrees at the University of Illinois in
Urbana-Champaign (UIUC), I am proud to owe much of my success and identity to Carla. My research journey hasn’t always been easy, but with Carla’s mentorship it has always been fun and enriching. I thank you, Carla, for your patience, support, advice, and willingness to allow me to grow into the scholar and person I am today. I would also like to thank my committee members, Brian Allan, Ephantus (Juma) Muturi, James O’Dwyer, and Andrew Suarez for their helpful comments, support, and guidance over the past several years. No matter the question, they were always willing to invest their time in helping me and my research.
I must also thank my wonderful wife, Jessica Holmes, for her “willingness” to review and provide comments on nearly every research document I have created. Without her never-ending support, I am not sure I would have had the strength to endure the rigors of graduate school. To my beautiful, intelligent, and curious daughter Ariella Holmes: as I write this text on this overcast spring evening, you have just accomplished a very significant milestone in your life – you have just crawled for the first time. Every day, you continue to impress your mother and I with your curious nature and strong determination to learn. With your whole life ahead of you, remember to always work hard, never give up, and live life to the fullest. With that being said, I must thank my grandparents (Kenneth and Linda Holmes) and parents (Kari and Nirmal Singh) for encouraging these same values in me throughout my life. And to my sister and brother,
v
Jazmyn and Matthew – thank you for the years of laughter and endless joy derived from beating you both at Mario Superstar Baseball on the Nintendo Gamecube.
To the current and former members of the Cáceres lab, I thank you for friendship, encouragement, and guidance over the years. In no particular order, this list includes but is not limited to: Tara Stewart, Lynette Strickland, Cameron Schwing, John W. Crawford, Matt
Schrader, Jelena Pantel, Ilona Menel and Ping Lee. The research presented herein required a significant amount of laboratory and field work, which would not have possible without assistance from several undergraduate assistants: Anna Osborn, Shalyn Keiser, Cameron
Schwing, Andrea Baldwin, Sana Khadri, Xorla Ocloo, Lauren Emerson, Ilona Menel, Liliana
Calderon, Hannah Wright, Ryan Smith, Frankie Rodriguez, and Kelly Hogan.
I am also thankful to Kimberly Schulz (Associate Professor, SUNY College of
Environmental Science and Forestry) and Zoi Rapti (Associate Professor, UIUC) for their numerous contributions to the research presented herein. I thank Kim for her guidance, support, and affording me the use of her lab and supplies during the summers of 2012, 2013, and 2014. I thank Zoi for her patience, support, and willingness to collaborate on the mathematics results presented in Chapter 1. A special shout-out and thanks is owed to Ken Paige for his advice, wisdom, support, and overall contributions to my growth as a scholar, academic professional, and antique collector.
This research would not have been possible without the generous financial support by the
Department of Animal Biology (UIUC), School of Integrative Biology (UIUC), College of
Liberal Arts and Sciences (UIUC), Graduate College (UIUC), Sigma Xi Foundation, and NSF awards received by Carla Cáceres and others. Last but not least, I owe thanks the Office of
Undergraduate Research, Office of the Provost, and School of Integrative Biology who provided
vi me with research and teaching assistantships during my graduate school career. This generous and continued support helped to keep me out of the local soup kitchens.
vii
Dedicated to my loving, supportive, and selfless wife, Jessica Holmes,
our beautiful and bright daughter Ariella Holmes, my grandparents Drs. Kenneth and Linda Holmes, and parents Dr. and Mrs. Nirmal and Kari Singh.
viii
TABLE OF CONTENTS
CHAPTER 1: HOW AND WHEN PATCH CENTRALITY AND NETWORK CONNECTIVITY AFFECT METAPOPULATION DYNAMICS IN SMALL FRESHWATER PONDS ...... 1
CHAPTER 2: NEGATIVE ASSOCIATION BETWEEN ZOOPLANKTON AND MOSQUITOES IN STORMWATER PONDS IS DRIVEN BY PRE- AND POST- COLONIZATION BEHAVIOR ...... 40
CHAPTER 3: PREDATION DIFFERENTIALLY STRUCTURES IMMATURE MOSQUITO ASSEMBLAGES IN STORMWATER PONDS ...... 73
APPENDIX A: SUPPLEMENTAL TABLE AND FIGURES ...... 108
ix
CHAPTER 1: HOW AND WHEN PATCH CENTRALITY AND NETWORK CONNECTIVITY AFFECT METAPOPULATION DYNAMICS IN SMALL FRESHWATER PONDS
ABSTRACT
Despite advances in metapopulation theory over the past several decades, recent studies have emphasized the difficulty in understanding and accurately predicting dynamics in natural systems. We attempt to address this knowledge gap through our coupling of metapopulation theory with occupancy data from a large-scale and multi-year field survey. Herein, we couple four years of population data for the freshwater zooplankter, Daphnia pulex inhabiting 38 newly established and semi-natural ponds in Upstate New York, with (1) a spatially explicit stochastic model and (2) a deterministic model where we have averaged the spatial dependencies. We show that the centrality of ponds (stocked or removed) has contrasting effects on metapopulation persistence when selecting ponds to initially stock vs. selecting which ponds to preserve. The metapopulation was not robust to the removal of centrally located ponds as the removal of these ponds resulted in rapid collapse of the metapopulation. However, when initially founding a metapopulation, the location of patches initially stocked did not play an important role in overall metapopulation occupancy. We introduce a quantity that contains all spatial information that can be used to predict the quasi-steady state of the stochastic simulations. Using this quantity, we then show how the output of our simple differential equation model matched the quasi-steady state of the stochastic simulations quite well, but only in networks characterized by high connectivity. The method we use is general enough to be applied in other systems for which presence-absence time-series data exists, and provide insights for habitat conservation and restoration efforts including how network spatial structure can drive spatiotemporal metapopulation dynamics.
1
INTRODUCTION
Models of metapopulation dynamics, especially those based on the work of Levins
(Levins 1969) and Hanski and collaborators (Hanski 1994, Hanski and Ovaskainen 2000) are numerous and well-studied (see Etienne and Heesterbeek 2000; Hanski and Ovaskainen 2003;
Vergara et al. 2016 and references therein). A metapopulation is defined as a set of "spatially separated" populations that interact through the migration of individuals among populations
(Levins 1969, Hanski 1998). As a result, patch occupancy is driven by local extinctions and recolonizations. This idea was formalized in 1969 by Richard Levins (Levins 1969), whose metapopulation model consisted of a single ordinary differential equation (ODE)
= ��(1 − �) − ��, (1) where p denotes the fraction of occupied patches in the network, m is the migration rate and e is the extinction rate for the patches (Levins 1969). Despite its simplicity, the Levins metapopulation model has influenced or served as the theoretical background for many subsequent metapopulation studies (Harrison 1991, Adler and Nuernberger 1994, Hanski and
Ovaskainen 2000, Etienne 2002, Etienne and Nagelkerke 2002, Keeling 2002), both theoretically
(Durrett and Levin 1994b, Green 1994, Bascompte and Solé 1996, Keymer et al. 2000, Black and McKane 2012, Thompson et al. 2016), and by fitting models to field or experimental data
(With and Crist 1995, Hanski et al. 1996, Nichols et al. 1998, Lafferty et al. 1999, Tyre et al.
2001, Molofsky and Ferdy 2005, Dorazio et al. 2010).
Recent empirical studies suggest that spatially-realistic models better represent the biological reality (Allen 2007, Anderson et al. 2015). Using a network modeling approach,
Gilarranz and Bascompte (2012) found significant effects of network spatial structure (i.e., the configuration of habitat patches) on patch occupancy dynamics and metapopulation persistence;
2 poorly connected patches were less likely to be encountered by dispersing propagules and thus failed to be initially colonized or recolonized following local extinction. Furthermore, Moilanen and Hanski (1995) found that their spatially-realistic model produced different dynamics from a spatially-explicit, but not realistic, cellular automata model of Melitaea cinxia butterflies.
Additionally, habitat patches are not created equally; some patches may have greater effects on metapopulation dynamics than others (Hanski and Ovaskainen 2000, Minor and Urban
2007, Bodin and Saura 2010). While habitat quality is often cited when evaluating patch importance, habitat connectivity can also play a crucial role in these dynamics (Moilanen and
Hanski 1998, Calabrese and Fagan 2004, Pressey et al. 2007). When patches are nonuniformly arranged in the landscape, the rate and probability of colonization depends on their physical locations in space (Hanski and Ovaskainen 2000, Ellner and Fussmann 2003, Gilarranz and
Bascompte 2012). This has led to the development of several connectivity indices (reviewed in
Pascual-Hortal and Saura 2006) which have been used to assess the relative importance of patches based on their contributions to landscape connectivity (Pascual-Hortal and Saura 2008).
Most notably, the graph-theory approach used by Urban and Keitt (2001) describes how reducing the distance among patches in a metapopulation can enhance patch persistence (i.e., finding the minimum spanning network tree).
These theoretical studies all highlight the importance of incorporating spatially-explicit dispersal into metapopulation theory and continuing to do so will help generate applied knowledge that conservation and restoration planners can use to identify patches most important for a species' long-term persistence. Furthermore, these modeling exercises may enhance restoration efforts by identifying optimal network structures that maximize metapopulation persistence (Schultz and Crone 2005, Vuilleumier et al. 2007).
3
Herein, we explore the effects of patch centrality and thus network connectivity on metapopulation dynamics through our coupling of metapopulation theory with field occupancy dynamics in a metapopulation of semi-natural freshwater ponds. Using a stochastic model based on the Levins framework (Levins 1969), we examine how the location of ponds initially stocked or removed from a metapopulation impacts site-specific occupancy dynamics and overall patterns of metapopulation occupancy and robustness (i.e., the ability of the metapopulation to resist regional extinction when patches are removed from the network). We then use spatial averaging methods to develop a simple ordinary differential equation (ODE) model; a method that is less computationally intensive to analyze and can be easily applied to other systems.
Using these methods, we ask:
(1) Does the threshold of initially occupied patches that precludes regional extinction depend
on patch centrality?
(2) How robust is the metapopulation to the removal of patches from the network and is
patch centrality important?
(3) How well does our simple ODE agree with the stochastic simulations?
POND NETWORK AND MODEL APPLICATION
Study system
The current study is motivated by previously published data on the occupancy patterns of the focal species Daphnia pulex in a network of 38 semi-natural ponds in Upstate New York
(Holmes et al. 2016a, 2016b). Daphnia inhabiting small freshwater ponds are an ideal system for examining metapopulation dynamics and have been the focus of numerous metapopulation studies (Shurin 2000, Pajunen and Pajunen 2003, Altermatt and Ebert 2008). Because ponds
4 have discrete borders, populations can be clearly delimited and easily monitored for extinction and recolonization events within and among years. Furthermore, modes and rates of dispersal in
Daphnia have been well documented (Shurin 2000, Pajunen and Pajunen 2003, Havel and
Shurin 2004). Daphnia are known to disperse via wind, rain, both large and small animal vectors
(Cáceres and Soluk 2002, Allen 2007), and can colonize new habitats very rapidly (<7 days;
Holmes et al. 2016a).While for other species, the validity of the Levins model has been questioned (Kritzer and Sale 2004), it has been argued that the Levins metapopulation paradigm fits Daphnia populations well (Harrison 1991, Bengtsson and Ebert 1998). In particular, colonizations and extinctions have been found to govern the distribution of Daphnia inhabiting rockpools (Bengtsson and Ebert 1998, Pajunen and Pajunen 2003).
Unlike many other metapopulation studies, the network under consideration by our study is small (< 40 patches) and consists of patches that are non-uniformly distributed over an area of about 1 km2 (Fig. 1.1). As a result, some patches are highly connected (i.e., increased closeness centrality) and others are less-connected (i.e., decreased closeness centrality). As a result, stochastic effects are expected to be prominent, similar to the effect of demographic stochasticity on the dynamics of small populations (see Hanski et al. 1996 for a discussion). As a reference, in other studies of spatial network structure and metapopulation persistence, the number of patches is in the order of thousands: 1,024 patches in Gilarranz and Bascompte (2012) and 3,335 patches in Fortuna et al. (2006) were considered.
The 38 ponds under examination in the field survey were created in the late-summer of
2010 through mechanical soil excavation of the landscape and were designed to be spatially arranged in one of three clustering layouts (nine, three, or one pond(s) per cluster) with each cluster delimited into arbitrary landscape "hexagons". The spatial clustering of ponds was
5 intentional and allowed for companion studies to address questions pertaining to amphibian habitat conservation and restoration (Youker-Smith et al. 2018). Following their construction, D. pulex populations were stocked in 27 out of 38 of the ponds to empirically examine the effects of stocking diversity on landscape population genetic diversity dynamics (Holmes et al. 2016b).
While many failed to establish, by May 2011, eleven ponds had detectable populations which served as sources of dispersing propagules in the metapopulation (Holmes et al. 2016b). Ponds were sampled bi-weekly from mid-May to mid-August in 2011 and 2012, and once in May of
2013 and 2014, providing fifteen time points of population data. More specific sampling methods can be found in (Holmes et al. 2016b). During an initial survey of the region in 2010, we found 29 pre-existing natural ponds near or within the pond infrastructure that may have served as additional sources for potential colonists. However, these ponds were not sampled during the 2011 – 2014 field survey and the inclusion of a regional colonization term did not significantly alter the patterns observed in the stochastic simulation (Fig. A1). As a result, we made the decision to omit the effects of these ponds from subsequent analyses presented herein.
Though patch size has been shown both empirically and theoretically to play an important role in population-level processes spanning extreme area ranges (25 - 550,500 m2 in
Fortuna et al. (2006), 1.1 - 8,674 m2 in Frisch et al. (2012), see also (Hanski 1994, Day and
Possingham 1995, Hill et al. 1996, Bender et al. 1998, North and Ovaskainen 2007, Prugh et al.
2008), it was not included as a variable in our study. Ponds in our network had little variation as they were designed to be uniform in size (with variation being introduced during excavation).
Area for each pond was estimated during a three-day period in May 2013 using measurements of the major and minor axes and ranged from 12 – 64 m2. A recent study by Frisch and colleagues of artificial and natural freshwater ponds found a significant effect of surface area on
6 colonization rates of cladoceran and copepod zooplankton (Frisch et al. 2012). At this scale, we suspect that patch size may play a less important role than the connectivity on metapopulation occupancy dynamics.
Stochastic model
We use a stochastic and spatially explicit model to examine the effects of network structure on spatial and temporal occupancy dynamics in our pond metapopulation. Similar models have been previously used in both theoretical (Roy et al. 2008) and empirical work
(Lafferty et al. 1999) to study metapopulations. Our stochastic model considers a network of N =
38 nonidentical patches (ponds) whose spatial configuration (location and distance) is based on the actual field configuration of ponds (Section: Study system; Fig. 1.1). The simulations begin with a subset of the 38 ponds being initially occupied. Unless otherwise specified, these initial conditions (IC) are the actual field occupancy data taken from the first sampling date in May
2011 (ponds 4, 7, 10, 12, 20, 24, 26, 32, 34, 35, 36). At a given time instance t, each pond i is either occupied (δi(t)=1) or vacant (δi(t)=0), depending on two stochastic processes: extinction and colonization. We assume that the extinction probability of pond i, denoted by pe(i), can vary between ponds (representing slight ecological differences among ponds) but is fixed for a given pond over time. Specifically, each pond's extinction probability pe(i) is drawn from the uniform distribution 0.05 < Pe < 0.25, where Pe is fixed.
The probability of colonization for a single pond i is determined by the between-pond colonization probabilities pc(i,j), where j ≠ i, which depend on distance and are described by the following equation:
7
With this probability kernel our network can be described as a weighted undirected network, where a link exists between two patches i and j if the distance between them is equal or lower than 1500 m. When they are connected, colonization can occur from i to j or j to i: we assume that colonization is independent of direction. Such networks have been used in the past in both theoretical (Urban and Keitt 2001, Gilarranz and Bascompte 2012) and empirical studies (Fortuna et al. 2006). The colonization kernel of Eq. 2 is adopted based on findings from Daphnia dispersal experiments (Shurin 2000, Havel and Shurin 2004, Allen 2007) and is shown in (Fig. 1.2).
The updating of the pond network is asynchronous meaning at each time t, the state of a single pond i, which is sampled from the discrete uniform distribution of the integers {1, ..., N}, changes according to the following two simple rules:
i. If i is occupied (i.e., δi(t) = 1), it can go extinct with probability pe(i) or remain
occupied with probability 1 − pe(i). This is implemented by first sampling a random
number r from the standard uniform distribution. If r < pe(i), i becomes empty;
otherwise it remains occupied. This method is similar to those employed by other
authors (Lafferty et al. 1999, Roy et al. 2008).
8
ii. If i is empty, it has the potential to be colonized by other occupied ponds in the
landscape. Again, a number r is drawn from the standard uniform distribution; if
� (�, �)� (�) r < , � − 1 ,
then i becomes occupied; otherwise, it remains empty.
In the model, 38 of these colonization/extinction updates correspond to the weekly time scale used in the field experiment. This algorithm is simulated for 5,000 - 50,000 updates (131 -
1,315 weeks) to compare model results to observed dynamics in the field study. We then replicate these simulations 1,000 - 10,000 times to obtain statistics about the state of each pond in the network. When studying the quasi-stationary state, to avoid transient dynamics, several thousands of initial updates are discarded and the fraction of occupied ponds is averaged over the remaining updates.
Parameterization of model using field data
As has been noted in other studies (Gilarranz and Bascompte 2012, Yackulic et al. 2015), there is an infinite combination of colonization and extinction probabilities that can produce a single quasi-stationary occupancy. Since the purpose of this work is not to fit the model to the field data but to explore metapopulation dynamics in small networks, parameters were selected in the following fashion.
We used the algorithm outlined in “Stochastic model” to generate patterns of occupancy for different combinations of extinction and colonization probabilities. Each pond i was assigned an extinction probability (pe(i)) that was drawn from the uniform distribution U(Pe - 0.05 , Pe +
0.05). Pe varied from 0.05 to 0.25 in our simulations. The colonization probability also varied,
9 and was determined by the maximum colonization probability, Pc in Eq. 2, which we allowed to range from 0.15 to 0.65. It was found that the combination (Pc, Pe) = (0.5, 0.12) yielded an occupancy of around 50% which is close to the mean occupancy of the field data (53%) and also yields spatial configurations that are positively correlated with the data (Pearson correlation
>40%, Figs. 1.3a and 1.3b). It is important to note that the selection of these parameter values reflect an assumption that our Daphnia metapopulation has reached its quasi-stationary state.
Ponds that were predicted by the model to have high occupancy (Fig. 1.3a) corresponded with ponds that were often occupied across the four-year field survey (Fig. 1.3b). The difference between the model-predicted and field occupancy ranged from 54% under-predicted by the model (pond 24) to 53% over-predicted by the model (pond 34), with 12 out of the 38 ponds falling within ± 25% (Fig. 1.3c). Overall metapopulation occupancy in the field ranged from
29% (May 2011) to 74% (May 2013; Fig. 1.3d).
The degree of each node (patch) in the network is equal to 37, since our model, as parameterized above, is equivalent to a complete graph (each pair of nodes is connected by a link). Hence, instead of using degree as a metric of connectivity, we adopt closeness centrality
(Estrada and Bodin 2008), namely the inverse sum of shortest distances to all other nodes from a focal node, as our connectivity metric.
RESULTS
Threshold of initially occupied ponds to preclude regional extinction
Using techniques from the theory of Markov chains, it has been well established that the expected time to regional extinction depends on the initial state of the metapopulation (Nåsell
2001, Grimm and Wissel 2004, Frank 2005). It is straightforward to see that in the deterministic
10
Levins model (Eq. 1) the zero steady state is unstable for < 1 and the population settles at the carrying capacity steady state �∗ = 1 − which is stable. In a closed system (i.e., no colonization from the region), the state where all ponds are vacant is an absorbing state; once all ponds become vacant, they will remain vacant forever, since there is no chance of recolonization.
Hence, in the stochastic simulations, this state is where the system will eventually settle after an exponentially long time. When the simulations are initiated with just a few occupied ponds, there is (due to stochasticity) a higher chance of getting to the absorbing state in a finite time.
To investigate what the threshold of initially occupied ponds that precludes regional extinction is and the role of well-connected patches (patches characterized by increased closeness centrality) we performed the following analysis. First, we simulated the metapopulation dynamics by assuming that nic, where nic varies from 1 to N, randomly chosen ponds were initially occupied. Ponds were randomly selected at the start of each simulation and each nic was simulated for 263 weeks (10,000 colonization/extinction updates) and was replicated 1,000 times. We then repeated the numerical experiment by initiating the simulation with the nic most (least) connected ponds, namely those whose closeness centrality is the largest
(smallest), being occupied. The same number of updates (10,000) and replicates (1,000) were simulated, as in the random case. The results are shown in Fig. 1.4. The average mean occupancy
(Fig. 1.4a) and the coefficient of variation (Fig. 1.4b) are shown as a function of nic. The coefficient of variation is defined as the ratio of the standard deviation over the mean, which makes the comparison of the three cases easy.
The two major insights that emerge are: (A) there is a threshold in the number of initially occupied patches nic above which closeness centrality does not affect the mean patch occupancy
(Fig. 1.4). For nic above this threshold, the coefficient of variation is very small and the three
11 ways of selecting the nic ponds are all equivalent. (B) The increased coefficient of variation can be used as a criterion to determine the minimum number of ponds required to be initially stocked.
The increased coefficient of variance is driven by stochastic effects that are more likely to push the metapopulation to the absorbing state. To investigate this, we generated histograms using the algorithm outlined in “Stochastic model”. Specifically, we simulated the occupancy dynamics for 131 weeks (5,000 colonization/extinction updates) and replicated this 10,000 times for each nic. We discarded the first 2,000 colonization/extinction updates, which ensures that the quasi-stationary state has been reached. Then we recorded the frequency (counts) with which each percent occupancy was observed.
One notices that for nic below a certain threshold, the histograms are bimodal (Fig. 1.4b).
Namely, the simulations settle either at the trivial (absorbing) state or at the nontrivial one predicted by the deterministic model. Such noise-induced bimodality has also been observed in other systems (Artyomov et al. 2007, To and Maheshri 2010) where in the deterministic version there exist no bistabilities. This bimodality explains the large coefficient of variation observed in
Fig. 1.4b.
Metapopulation persistence
Habitat patch networks in natural systems are dynamic (Fabritius et al. 2017); new habitats can be created (Chapman 2013), and existing habitats can become degraded (Mortelliti et al. 2010) or lost (Arntzen et al. 2017) over time. The dynamic nature of these networks can result in suboptimal networks (i.e., scenarios in which several habitats are uninhabitable and do not contribute to metapopulation dynamics) that may result in extinction of the metapopulation.
12
For freshwater invertebrates, such as Daphnia inhabiting vernal ponds, these uninhabitable periods can occur during seasonal pond desiccation (Pajunen and Pajunen 2003). Because most ponds failed to dry throughout the field survey (Holmes et al. 2016a), we did not examine the effects of removing ponds over time in our simulations; instead, we focus on varying network structure at the beginning of simulations and keeping these structures static throughout each simulation. To quantify the effect of habitat loss (patch removal) in a metapopulation, we focused on metatapopulation persistence, namely we studied how the quasi-stationary state varies as ponds are removed from our network. Ponds were removed in a random fashion, and also non-randomly, by selecting the nrem most or least connected ponds.
Two key trends emerge as we increase the number of ponds removed from the network
(decreasing the number of ponds in the network at the start of the model simulations). First, metapopulation occupancy decreases gradually in the case of random removal and the case when the most central ponds are removed; the opposite trend is observed in the removal of the least central ponds (Fig. 1.5). In this case, percent occupancy at the quasi-stationary state increases initially and starts decreasing only after more than 15 ponds have been removed. Even so, regional extinction occurs when nrem is almost 30. Hence, removing isolated ponds results in a better-connected remaining network with increased percent occupancy (Fig. 1.5).
Comparing the role of closeness centrality in selecting which ponds to initially stock vs. selecting which ponds to preserve, we notice a striking difference. Specifically, while closeness centrality did not play a crucial role in selecting which ponds to initially stock, it appears to be critical in selecting which ponds to preserve in order to avoid regional extinction.
13
Linking the Levins model to stochastic spatial simulations
While stochastic simulations can be used to explore the role of extinction and colonization probabilities and network structure in metapopulation dynamics, they are nevertheless relatively time intensive and computationally demanding (Morozov and Poggiale
2012). Hence, various approximations have been used to yield predictions in case networks are altered or dispersal and extinction processes are modified (Adler and Nuernberger 1994) and to compare intervention measures in case of habitat loss and other disturbances (Etienne and
Nagelkerke 2002). Deterministic or stochastic master equations approximating stochastic simulations may also be easier to analyze and require fewer parameters. In this section, we develop, through spatial averaging, a minimal ODE model and demonstrate that its predicted steady states are in very good agreement with the ones obtained from the stochastic simulations.
Based on the stochastic algorithm, if pond i is occupied at time t (δi (t) = 1), then it might become vacant with probability pe(i) or remain occupied with probability 1 - pe(i). Similarly, if pond i is vacant at time t (δi(t) = 0), then it might become colonized from pond j with probability pc(i,j) if pond j is occupied, namely if δj(t) = 1. Hence, if we denote the proportion of ponds occupied with
1 �(�) = � (�), � it follows by summing over all N ponds that
14 where Mij and Ei denote colonization and extinction rates, as in the original Levin’s model
(Levins 1969) given in (Eq. 1). Next, we set Ei = e assuming uniform extinction probabilities across all ponds. We also set Mij = Fm, where m is constant and F is calculated as follows. We find all pond distances and then bin them into classes that match the distances where the dispersal kernel experiences step-like decreases (Fig. 1.2b). We then calculate F as the weighted average of the probabilities:
∑ ( ) ( ) � − , (3) ∑ ( ) where NB = 6 is the number of bins in the histogram, B(k) is the height of the kth bin (see Fig.
1.2b), and M(k) are given in Eq. 2. We note that F is independent of Pc, as can be seen from Eqs.
2 and 3 and takes the value F = 0.5195. Finally, by setting Mij = Fm, we obtain:
Assuming that we are at equilibrium (Holmes et al. 2016a), we can then solve for the nontrivial steady state:
from which given the observed field occupancy p* = 0.5526, the ratio of extinction vs. colonization rates can be found to be = 0.2388.
We compared the resulting occupancy from our stochastic simulations to the non-trivial steady states predicted by the theoretical Levins model (Eq. 4) to examine the role of extinction and dispersal probabilities on metapopulation occupancy. Extinction probabilities pe(i) are
15 randomly drawn from the standard uniform distribution on the open interval (Pe - 0.05, Pe +
0.05), where 0.05 ≤ Pe ≤ 0.25 and dispersal probabilities pc(i,j) are given by Eq. 2, where 0.15 ≤
Pc ≤ 0.65.
The theoretical line (with slope 0.2388 obtained from Eq. 5) exactly overlaps with the region in the phase space where percent occupancy is p* (Fig. 1.6). This indicates that the deterministic version of the model is sufficient should the objective of a study be to find overall metapopulation occupancy. Specifically, when one correctly averages the spatial inhomogeneity of the network, a simple ODE model can yield the quasi-stationary state observed in the stochastic simulations.
The advantage of Eq. 5 is that it can be used to compare the effect of different network changes on percent occupancy. For instance, increasing habitat quality which is sometimes equivalent to decreasing e might be less preferable to varying the number N and connectivity F of patches. All this information is compactly contained in Eq. 5. Similar averaged measures have been used in previous studies of metapopulation persistence. For instance, in Grilli et al. (2015), the concept of metapopulation capacity (the leading eigenvalue of their dispersal matrix), first introduced in Hanski and Ovaskainen (2000), was used to compare metapopulation persistence in different fragmented habitats. Our expression (Eq. 5) is more comprehensive in the sense that it includes information on how both the number and the identity of removed ponds affect regional persistence. In contrast, one has to find the eigenvector corresponding to the leading eigenvalue of the dispersal matrix in order to link the identity of patches to the effect of their removal to metapopulation persistence (Hanski and Ovaskainen 2000, Grilli et al. 2015).
As an application of our spatial approximation, we used it to predict percent occupancy as the nrem least connected ponds are removed from the network. The results are shown Fig. 1.7.
16
We notice that the approximation works well until about the 15 least central ponds have been removed. Since the approximation is in essence a mean-field approximation, it works well initially because the remaining network is better connected, and as expected, breaks down when the network becomes small (consisting of around 23 ponds). On the other hand, the approximation does not work when the most central ponds are removed, since the remaining network is both too small and not well connected, so a continuous approximation fails.
DISCUSSION
Through the coupling of numerical simulations with field occupancy data for the freshwater zooplankton Daphnia pulex inhabiting a newly established pond metapopulation, we provide additional evidence for the importance of network connectivity on metapopulation dynamics, including occupancy and persistence (Durrett and Levin 1994a, Hanski and
Ovaskainen 2000, Roy et al. 2008, Gilarranz and Bascompte 2012). Results from our stochastic simulations show that the centrality of ponds manipulated in the network has contrasting effects when selecting ponds to initially stock versus selecting which ponds to preserve. We found that when creating a network of habitats, above a certain threshold the connectedness of ponds initially stocked does not affect metapopulation occupancy. However, when the goal is to preserve an existing network of habitat patches, preference should be given to preserving highly- connected patches in the landscape. While results from our stochastic simulations suggest that our metapopulation is fairly robust to the removal of ponds from the landscape, we found that the loss of more-connected ponds rapidly eroded metapopulation occupancy and persistence when compared to the removal of random and less-connected ponds. Furthermore, we present a spatially-averaged ordinary differential equation (ODE) model that was effective at predicting
17 overall occupancy dynamics that were given by the spatially explicit stochastic model. However, the extent to which the two agree depends on network structure; networks characterized by high connectivity showed very strong agreement.
Results from this study add to a long history of empirical and theoretical work highlighting the importance of patch connectedness on metapopulation dynamics (Hansson 1991,
Baguette and Van Dyck 2007, Planes et al. 2009, Chapman 2013, Eaton et al. 2014, Wang et al.
2015, Albert et al. 2017). This includes empirical work conducted in Daphnia metapopulations
(Altermatt and Ebert 2008, 2010, Havel et al. 2012). Unexpectedly, however, was the finding that the stocking of central ponds did not influence metapopulation occupancy. It is possible that other unmodeled factors may have played a more important role than the number or location of ponds stocked in our system. Factors not examined by our study included habitat patch quality, population density, and temporal variation in colonization and extinction parameters which have been shown to enhance metapopulation dynamics and persistence in other studies (Moilanen and
Hanski 1998, Sæther et al. 1999, Ovaskainen 2002, Anthes et al. 2003, Altermatt and Ebert 2010, but see Van Langevelde and Wynhoff 2009). For example, a study of multiple species by
Thomas et al. (2001) found that habitat quality better predicted local patch occupancy than patch isolation. For our stochastic simulations, we allowed extinction rates to randomly vary between ponds but did not attempt to attribute this variation to real ecological differences among ponds.
However, ecological differences likely exist; previously published work in this system has shown that the ponds differ in their community structure and in several measured abiotic variables (Holmes et al. 2016a). These differences may have had unmeasured effects on metapopulation dynamics observed in our system.
18
While the Daphnia system afforded a unique and significant opportunity to couple mathematical simulations with metapopulation dynamics in a large-scale and newly established field system, we acknowledge some limitations of our study and study system. Our model ignored the potential contributions of the Daphnia dormant egg bank and long-distance dispersal events on population occupancy dynamics in our metapopulation. In Daphnia, the production of desiccation-resistant dormant eggs (ephippia), which allow for dispersal through both time and space, has been shown to contribute to population dynamics (Hairston 1996, Cáceres 1997,
Hairston et al. 2002). These eggs can also allow for colonization by long-distance dispersal events (Havel and Shurin 2004). As a result, ponds without detectable populations may have persisted through the dormant egg bank, resulting in false absences in our field data.
Furthermore, ephippia production during late-summer may have allowed for spring D. pulex populations to be re-founded from overwintering ephippia (Pajunen and Pajunen 2003, Altermatt and Ebert 2008). At the start of the field survey, we documented 29 nearby ponds in the region
(and there were likely many more in the landscape) that may have provided additional colonists to our system. These ponds may have provided dispersing adult or dormant eggs which can be transported by wind, rain, or animal vectors (Cáceres and Soluk 2002). However, when outside dispersal was factored into our model, the results of the model were not significantly altered
(Fig. A1). The nature of our study system makes it difficult to disentangle the relative contributions of the dormant egg bank, long-distance dispersal events, and local overland dispersal on occupancy patterns. At this time, we cannot confidently determine whether the source of re-colonization propagules were from overland dispersal events or recolonization via the dormant egg bank. The use of genetic markers can and has been used to understand patterns of dispersal and colonization (Gagnaire et al. 2015, Geismar et al. 2015); however, the limited
19 genetic diversity of D. pulex in our metapopulation (8 distinct multi-locus clonal genotypes:
Holmes et al. 2016b) limits our ability to implement such methods.
Findings from our modeling exercises have important implications for habitat conservation and restoration efforts (Smith and Green 2005). In small freshwater ponds, climate and anthropogenic factors may affect habitat permeance and long-term persistence (Windmiller and Calhoun 2007, Vanschoenwinkel et al. 2009). As a result, these climatic and anthropogenic factors may, through altering patch quantity and connectivity, have significant effects on metapopulation persistence. Herein, we provided further evidence for the role of connectivity on metapopulation dynamics using the Daphnia system and introduce an analytical approach that can be easily applied to other systems for which presence-absence time-series and dispersal data exists, such as in amphibian species under extinction threat (Anderson et al. 2015). Similar to other studies, our metapopulation was sensitive to the spatial structure of the pond network; removing highly connected habitats from the landscape can lead to widespread extinction (Hess
1996). As a result, conservation efforts should be focused on highly-connected patches (Heller and Zavaleta 2009, Watson et al. 2017). This paper provides stochastic and ODE approaches that can be easily applied to other systems of interest to determine stocking and conservation strategies to enhance metapopulation persistence. While it is generally considered ideal to conserve or restore the maximum number of habitats to enhance metapopulation viability, results on metapopulation robustness from our stochastic simulations suggest that the (temporary or permanent) loss of a few patches from the landscape does not necessarily yield regional extinction. We encourage others with data in other systems, especially those in newly established metapopulations, to use and evaluate this approach to enhance our understanding of how this method can be applied to threatened species.
20
ACKNOWLEDGEMENTS
This research was supported by the United States National Science Foundation [DEB-
0947314, DEB-0947245, DEB-1120804, DUE-1129198, and DEB-1354407], the University of
Illinois Research Board [RB17060], and by grants from the University of Illinois at Urbana-
Champaign School of Integrative Biology and Department of Animal Biology. We thank Jessica
R. Holmes, James P. O'Dwyer and Andrew V. Suarez who kindly provided comments on this manuscript.
21
FIGURES
1 2 3 11 10 4 12 13 5
6 9 7 8
15 16 17 14 18 19 20 24 26 4 25 21 22
23
37 35
38 36 34
27 33 29 28 32 31 30
330m
Figure 1.1: Spatial configuration of the 38 semi-natural pond network at Svend O. Heiberg
Memorial Forest (Tully, NY, 42.774310, -76.085746) used in this study.
22
a 0.5
0.4
0.3
0.2 dispersal probability 0.1
0 0 200 400 600 800 1000 1200 1400 1600 1800 distance (m)
Figure 1.2: (a) An example of the colonization kernel with Pc = 0.5 given by Eq. 2 as a function of distance. (b) Histogram of all pairwise pond distances in our dataset using bin widths that reflect the dispersal kernel.
23
a 1 b 5 0.9 5
0.8 10 10 0.7
15 15 0.6
20 0.5 20 pond pond 0.4 25 25 0.3 30 30 0.2
0.1 35 35
0 0 10 20 30 40 50 60 70 80 90 100110120130140150160 20 40 60 80 100 120 140 weeks weeks
0.75 d ! 0.7
0.65
0.6 !"#$%&'($#)*+$#&, -./$(0$#&-**1234*5diffSimData 1.0 0.55
0.5 0.5 0.45 0.0 0.4 percent occupancy −0.5 0.35 mean mean + std mean - std 0.3 −1.0 data 0.25 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 weeks
Figure 1.3: (a) A heatmap of the simulated pond occupancy dynamics for each of the 38 ponds averaged over 160 simulated weeks. Each cell represents the probability of being occupied at each simulated week. (b) The observed field occupancy data for each pond (red corresponding to occupied and blue to empty) over the 15 sampling dates (May 2011 to May 2014) are scaled to show weekly occupancy dynamics. (c) The difference in occupancy between model predictions and average observed occupancy is shown for each pond (red corresponding to over-predicted occupancy by the model, and blue shows the model under-predicting occupancy). (d) Simulated occupancy from the model (± standard deviation) is plotted with observed field occupancy data from the 15 sampling dates.
24
0.6 a 0.55
0.5
0.45
0.4
0.35
0.3
percent occupancy 0.25 least central 0.2 most central random 0.15 0 5 10 15 20 25 30 35 40 number of initially occupied ponds (n ) ic
Figure 1.4: The effect of number of initially occupied ponds on regional occupancy of the simulated metapopulation. (a) The percent occupancy averaged over all simulations and (b) the coefficient of variation as a function of the number of initially occupied ponds nic. Three cases are considered: randomly chosen ponds (blue curve), least connected ponds (green curve) and most connected ponds (red curve). Histograms of the simulated metapopulation occupancy as nic varies appear as insets.
25
0.6
0.5
0.4
0.3
0.2 percent occupancy
least central 0.1 most central random 0 5 10 15 20 25 30 number of removed ponds (n ) rem
Figure 1.5: The effect of removing nrem ponds from the network of 38 ponds on regional occupancy of the simulated metapopulation. The percent occupancy for three different cases is shown: removal of least central ponds (green curve), removal of most central ponds (red line) and random pond removal (blue curve).
26
0.25 0.8
0.7 0.2 0.6 P e 0.5
0.15 0.4
0.3
0.1 0.2
0.1
0.05 0.2 0.3P 0.4 0.5 0.6 c
Figure 1.6: Percent occupancies at steady state as we vary the parameter Pc (horizontal axis) that controls the colonization probability and the parameter Pe (vertical axis) that controls the extinction probability.
27
0.6
0.5
0.4
0.3
0.2 percent occupancy incr. centr. decr. centr. 0.1 theor. decr. theor. decr. 0 5 10 15 20 25 30
number of removed ponds
Figure 1.7: Percent occupancies at the quasi-stationary state as a function of ponds removed from the network (nrem) for the case when the nrem least connected ponds are removed. The results from the stochastic simulations are the solid (green/red) curves and the theoretical results correspond to the dashed (green/red) curves.
28
REFERENCES
Adler, F. R., and B. Nuernberger. 1994. Persistence in patchy irregular landscapes. Theoretical
Population Biology 45:41–75.
Albert, C. H., B. Rayfield, M. Dumitru, and A. Gonzalez. 2017. Applying network theory to
prioritize multispecies habitat networks that are robust to climate and land-use change.
Conservation Biology 31:1383–1396.
Allen, M. R. 2007. Measuring and modeling dispersal of adult zooplankton. Oecologia 153:135–
143.
Altermatt, F., and D. Ebert. 2008. The influence of pool volume and summer desiccation on the
production of the resting and dispersal stage in a Daphnia metapopulation. Oecologia
157:441–452.
Altermatt, F., and D. Ebert. 2010. Populations in small, ephemeral habitat patches may drive
dynamics in a Daphnia magna metapopulation. Ecology 91:2975–2982.
Anderson, T. L., B. H. Ousterhout, W. E. Peterman, D. L. Drake, and R. D. Semlitsch. 2015. Life
history differences influence the impacts of drought on two pond-breeding salamanders.
Ecological Applications 25:1896–1910.
Anthes, N., T. Fartmann, G. Hermann, and G. Kaule. 2003. Combining larval habitat quality and
metapopulation structure - the key for succesful management of pre-alpine Euphydryas
aurinia colonies. Journal of Insect Conservation 7:175–185.
Arntzen, J. W., C. Abrahams, W. R. M. Meilink, R. Iosif, and A. Zuiderwijk. 2017. Amphibian
decline, pond loss and reduced population connectivity under agricultural intensification
over a 38 year period. Biodiversity and Conservation 26:1411–1430.
29
Artyomov, M. N., J. Das, M. Kardar, and A. K. Chakraborty. 2007. Purely stochastic binary
decisions in cell signaling models without underlying deterministic bistabilities.
Proceedings of the National Academy of Sciences of the United States of America
104:18958–18963.
Baguette, M., and H. Van Dyck. 2007. Landscape connectivity and animal behavior: functional
grain as a key determinant for dispersal. Landscape Ecology 22:1117–1129.
Bascompte, J., and R. V. Solé. 1996. Habitat fragmentation and extinction thresholds in spatially
explicit models. Journal of Animal Ecology 65:465–473.
Bender, D. J., T. A. Contreras, and L. Fahrig. 1998. Habitat loss and population decline: a meta-
analysis of the patch size effect. Ecology 79:517–533.
Bengtsson, J., and D. Ebert. 1998. Distributions and impacts of microparasites on Daphnia in a
rockpool metapopulation. Oecologia 115:213–221.
Black, A. J., and A. J. McKane. 2012. Stochastic formulation of ecological models and their
applications. Trends in Ecology and Evolution 27:337–345.
Bodin, Ö., and S. Saura. 2010. Ranking individual habitat patches as connectivity providers:
integrating network analysis and patch removal experiments. Ecological Modelling
221:2393–2405.
Cáceres, C. E. 1997. Temporal variation, dormancy, and coexistence: A field test of the storage
effect. Proceedings of the National Academy of Sciences 94:9171–9175.
Cáceres, C. E., and D. A. Soluk. 2002. Blowing in the wind: A field test of overland dispersal
and colonization by aquatic invertebrates. Oecologia 131:402–408.
Calabrese, J. M., and W. F. Fagan. 2004. A comparison-shopper’s guide to connectivity metrics.
Frontiers in Ecology and the Environment 2:529–536.
30
Chapman, M. G. 2013. Constructing replacement habitat for specialist and generalist molluscs-
the effect of patch size. Marine Ecology Progress Series 473:201–214.
Day, J. R., and H. P. Possingham. 1995. A stochastic metapopulation model with variability in
patch size and position. Theoretical Population Biology 48:333–360.
Dorazio, R. M., M. Kéry, J. A. Royle, and M. Plattner. 2010. Models for inference in dynamic
metacommunity systems. Ecology 91:2466–2475.
Durrett, R., and S. Levin. 1994a. The importance of being discrete (and spatial). Theoretical
Population Biology 46:363–394.
Durrett, R., and S. A. Levin. 1994b. Stochastic spatial models: a user’s guide to ecological
applications. Philosophical Transactions of the Royal Society B: Biological Sciences
343:329–350.
Eaton, M. J., P. T. Hughes, J. E. Hines, and J. D. Nichols. 2014. Testing metapopulation
concepts: effects of patch characteristics and neighborhood occupancy on the dynamics of
an endangered lagomorph. Oikos 123:662–676.
Ellner, S. P., and G. Fussmann. 2003. Effects of successional dynamics on metapopulation
persistence. Ecology 84:882–889.
Estrada, E., and Ö. Bodin. 2008. Using network centrality measures to manage landscape
connectivity. Ecological Applications 18:1810–1825.
Etienne, R. S. 2002. A scrutiny of the Levins metapopulation model. Comments on Theoretical
Biology 7:257–281.
Etienne, R. S., and J. A. P. Heesterbeek. 2000. On optimal size and number of reserves for
metapopulation persistence. Journal of Theoretical Biology 203:33–50.
31
Etienne, R. S., and C. J. Nagelkerke. 2002. Non-equilibria in small metapopulations: Comparing
the deterministic Levins model with its stochastic counterpart. Journal of Theoretical
Biology 219:463–478.
Fabritius, H., A. Jokinen, and M. Cabeza. 2017. Metapopulation perspective to institutional fit:
maintenance of dynamic habitat networks. Ecology and Society 22:30.
Fortuna, M. A., C. Gómez-Rodríguez, and J. Bascompte. 2006. Spatial network structure and
amphibian persistence in stochastic environments. Proceedings of the Royal Society B:
Biological Sciences 273:1429–1434.
Frank, K. 2005. Metapopulation persistence in heterogeneous landscapes: lessons about the
effect of stochasticity. The American Naturalist 165:374–388.
Frisch, D., K. Cottenie, A. Badosa, and A. J. Green. 2012. Strong spatial influence on
colonization rates in a pioneer zooplankton metacommunity. PloS ONE 7:e40205.
Gagnaire, P. A., T. Broquet, D. Aurelle, F. Viard, A. Souissi, F. Bonhomme, S. Arnaud-Haond,
and N. Bierne. 2015. Using neutral, selected, and hitchhiker loci to assess connectivity of
marine populations in the genomic era. Evolutionary Applications 8:769–786.
Geismar, J., P. Haase, C. Nowak, J. Sauer, and S. U. Pauls. 2015. Local population genetic
structure of the montane caddisfly Drusus discolor is driven by overland dispersal and
spatial scaling. Freshwater Biology 60:209–221.
Gilarranz, L. J., and J. Bascompte. 2012. Spatial network structure and metapopulation
persistence. Journal of Theoretical Biology 297:11–16.
Green, D. G. 1994. Connectivity and complexity in landscapes and ecosystems. Pacific
Conservation Biology 1:194–200.
32
Grilli, J., G. Barabás, and S. Allesina. 2015. Metapopulation persistence in random fragmented
landscapes. PLoS Computational Biology 11:e1004251.
Grimm, V., and C. Wissel. 2004. The intrinsic mean time to extinction: a unifying approach to
analysing persistence and viability of populations. Oikos 105:501–511.
Hairston Jr. N. G., and C. M. Kearns, 2002. Temporal dispersal: ecological and evolutionary
aspects of zooplankton egg banks and the role of sediment mixing. Integrative and
Comparative Biology 42:481–491.
Hairston Jr., N. G. 1996. Zooplankton egg banks as biotic reservoirs in changing environments.
Limnology and Oceanography 41:1087–1092.
Hanski, I. 1994. A practical model of metapopulation dynamics. The Journal of Animal Ecology
63:151–162.
Hanski, I. 1998. Metapopulation dynamics. Nature 396:41–49.
Hanski, I., A. Moilanen, and M. Gyllenberg. 1996. Minimum viable metapopulation size. The
American Naturalist 147:527–541.
Hanski, I., and O. Ovaskainen. 2000. The metapopulation capacity of a fragmented landscape.
Nature 404:755–758.
Hanski, I., and O. Ovaskainen. 2003. Metapopulation theory for fragmented landscapes.
Theoretical Population Biology 64:119–127.
Hansson, L. 1991. Dispersal and connectivity in metapopulations. Biological Journal of the
Linnean Society 42:83–103.
Harrison, S. 1991. Local extinction in a metapopulation context: an empirical evaluation.
Biological Journal of the Linnean Society 42:73–88.
33
Havel, J. E., and J. B. Shurin. 2004. Mechanisms, effects, and scales of dispersal in freshwater
zooplankton. Limnology and Oceanography 49:1229–1238.
Havel, J. E., J. B. Shurin, and J. R. Jones. 2012. Estimating dispersal from patterns of spread:
spatial and local control of lake invasions. Ecology 83:3306–3318.
Heller, N. E., and E. S. Zavaleta. 2009. Biodiversity management in the face of climate change: a
review of 22 years of recommendations. Biological Conservation 142:14–32.
Hess, G. R. 1996. Linking extinction to connectivity and habitat destruction in metapopulation
models. The American Naturalist 148:226–236.
Hill, J. K., C. D. Thomas, and O. T. Lewis. 1996. Effects of habitat patch size and isolation on
dispersal by Hesperia comma butterflies: implications for metapopulation structure. The
Journal of Animal Ecology 65:725–735.
Holmes, C. J., S. Figary, K. L. Schulz, and C. E. Cáceres. 2016a. Effects of diversity on
community assembly in newly formed pond communities. Ecosphere 7:e01377.
Holmes, C. J., J. H. Pantel, K. L. Schulz, and C. E. Cáceres. 2016b. Initial genetic diversity
enhances population establishment and alters genetic structuring of a newly established
Daphnia metapopulation. Molecular Ecology 25:3299–3308.
Keeling, M. J. 2002. Using individual-based simulations to test the Levins metapopulation
paradigm. Journal of Animal Ecology 71:270–279.
Keymer, J. E., P. A. Marquet, J. X. Velasco-Hernández, S. A. Levin, J. X. Velasco-Hernández,
and S. A. Levin. 2000. Extinction thresholds and metapopulation persistence in dynamic
landscapes. The American Naturalist 156:478–494.
Kritzer, J. P., and P. F. Sale. 2004. Metapopulation ecology in the sea: from Levins’ model to
marine ecology and fisheries science. Fish and Fisheries 5:1–10.
34
Lafferty, K. D., C. C. Swift, and R. F. Ambrose. 1999. Extirpation and recolonization in a
metapopulation of an endangered fish, the tidewater goby. Conservation Biology 13:1447–
1453.
Van Langevelde, F., and I. Wynhoff. 2009. What limits the spread of two congeneric butterfly
species after their reintroduction: quality or spatial arrangement of habitat? Animal
Conservation 12:540–548.
Levins, R. 1969. Some demographic and genetic consequences of environmental heterogeneity
for biological control. Bulletin of the Entomological Society of America 15:237–240.
Minor, E. S., and D. L. Urban. 2007. Graph theory as a proxy for spatially explicit population
models in conservation planning. Ecological Applications 17:1771–1782.
Moilanen, A., and I. Hanski. 1995. Habitat destruction and coexistence of competitors in a
spatially realistic metapopulation model. Journal of Animal Ecology 64:141–144.
Moilanen, A., and I. Hanski. 1998. Metapopulation dynamics: effects of habitat quality and
landscape structure. Ecology 79:2503–2515.
Molofsky, J., and J.-B. Ferdy. 2005. Extinction dynamics in experimental metapopulations.
Proceedings of the National Academy of Sciences 102:3726–3731.
Morozov, A., and J. C. Poggiale. 2012. From spatially explicit ecological models to mean-field
dynamics: the state of the art and perspectives. Ecological Complexity 10:1–11.
Mortelliti, A., G. Amori, and L. Boitani. 2010. The role of habitat quality in fragmented
landscapes: a conceptual overview and prospectus for future research. Oecologia 163:535–
547.
Nåsell, I. 2001. Extinction and quasi-stationarity in the verhulst logistic model. Journal of
Theoretical Biology 211:11–27.
35
Nichols, J. D., T. Boulinier, J. E. Hines, K. H. Pollock, and J. R. Sauer. 1998. Estimating rates of
local species extinction, colonization, and turnover in animal communities. Ecological
Applications 8:1213–1225.
North, A., and O. Ovaskainen. 2007. Interactions between dispersal, competition, and landscape
heterogeneity. Oikos 116:1106–1119.
Ovaskainen, O. 2002. The effective size of a metapopulation living in a heterogeneous patch
network. The American Naturalist 160:612–628.
Pajunen, V. I., and I. Pajunen. 2003. Long-term dynamics in rock pool Daphnia
metapopulations. Ecography 26:731–738.
Pascual-Hortal, L., and S. Saura. 2006. Comparison and development of new graph-based
landscape connectivity indices: towards the priorization of habitat patches and corridors for
conservation. Landscape Ecology 21:959–967.
Pascual-Hortal, L., and S. Saura. 2008. Integrating landscape connectivity in broad-scale forest
planning through a new graph-based habitat availability methodology: Application to
capercaillie (Tetrao urogallus) in Catalonia (NE Spain). European Journal of Forest
Research 127:23–31.
Planes, S., G. P. Jones, and S. R. Thorrold. 2009. Larval dispersal connects fish populations in a
network of marine protected areas. Proceedings of the National Academy of Sciences
106:5693–5697.
Pressey, R. L., M. Cabeza, M. E. Watts, R. M. Cowling, and K. A. Wilson. 2007. Conservation
planning in a changing world. Trends in Ecology and Evolution 22:583–592.
36
Prugh, L. R., K. E. Hodges, A. R. E. Sinclair, and J. S. Brashares. 2008. Effect of habitat area
and isolation on fragmented animal populations. Proceedings of the National Academy of
Sciences 105:20770–20775.
Roy, M., K. Harding, and R. D. Holt. 2008. Generalizing Levins metapopulation model in
explicit space: models of intermediate complexity. Journal of Theoretical Biology 255:152–
161.
Sæther, B. E., S. Engen, and R. Lande. 1999. Finite metapopulation models with density-
dependent migration and stochastic local dynamics. Proceedings of the Royal Society B:
Biological Sciences 266:113–118.
Schultz, C. B., and E. E. Crone. 2005. Patch size and connectivity thresholds for butterfly habitat
restoration. Conservation Biology 19:887–896.
Shurin, J. B. 2000. Dispersal limitation, invasion resistance, and the structure of pond
zooplankton communities. Ecology 81:3074–3086.
Smith, M. A., and D. M. Green. 2005. Dispersal and the metapopulation paradigm in amphibian
ecology and conservation: Are all amphibian populations metapopulations? Ecography
28:110–128.
Thomas, J. A., N. A. D. Bourn, R. T. Clarke, K. E. Stewart, D. J. Simcox, G. S. Pearman, R.
Curtis, and B. Goodger. 2001. The quality and isolation of habitat patches both determine
where butterflies persist in fragmented landscapes. Proceedings of the Royal Society B:
Biological Sciences 268:1791–1796.
Thompson, C. J., E. Shtilerman, and L. Stone. 2016. A discrete Markov metapopulation model
for persistence and extinction of species. Journal of Theoretical Biology 404:391–397.
37
To, T. L., and N. Maheshri. 2010. Noise can induce bimodality in positive transcriptional
feedback loops without bistability. Science 327:1142–1145.
Tyre, A. J., H. P. Possingham, and D. B. Lindenmayer. 2001. Inferring process from pattern: Can
territory occupancy provide information about life history parameters? Ecological
Applications 11:1722–1737.
Urban, D., and T. Keitt. 2001. Landscape connectivity: A graph-theoretic perspective. Ecology
82:1205–1218.
Vanschoenwinkel, B., A. Hulsmans, E. De Roeck, C. De Vries, M. Seaman, and L. Brendonck.
2009. Community structure in temporary freshwater pools: disentangling the effects of
habitat size and hydroregime. Freshwater Biology 54:1487–1500.
Vergara, P. M., A. Saravia-Zepeda, N. Castro-Reyes, and J. A. Simonetti. 2016. Is
metapopulation patch occupancy in nature well predicted by the Levins model? Population
Ecology 58:335–343.
Vuilleumier, S., C. Wilcox, B. J. Cairns, and H. P. Possingham. 2007. How patch configuration
affects the impact of disturbances on metapopulation persistence. Theoretical Population
Biology 72:77–85.
Wang, S., B. Haegeman, and M. Loreau. 2015. Dispersal and metapopulation stability. PeerJ
3:e1295.
Watson, D. M., V. A. J. Doerr, S. C. Banks, D. A. Driscoll, R. van der Ree, E. D. Doerr, and P.
Sunnucks. 2017. Monitoring ecological consequences of efforts to restore landscape-scale
connectivity. Biological Conservation 206:201–209.
38
Windmiller, B., and A. J. K. Calhoun. 2007. Conserving Vernal Pool Wildlife in Urbanizing
Landscapes. Pages 233–251 in A. J. K. Calhoun and P. G. DeMaynadier, editors. Science
and Conservation of Vernal Pools in Northeastern North America. CRC Press, Boca Raton,
FL, USA.
With, K. A., and T. O. Crist. 1995. Critical thresholds in species’ responses to landscape
structure. Ecology 76:2446–2459.
Yackulic, C. B., J. D. Nichols, J. Reid, and R. Der. 2015. To predict the niche, model
colonization and extinction. Ecology 96:16–23.
Youker-Smith, T. E., P. H. Boersch-Supan, C. M. Whipps, and S. J. Ryan. 2018. Environmental
drivers of ranavirus in free-living amphibians in constructed ponds. EcoHealth 15:608–618.
39
CHAPTER 2: NEGATIVE ASSOCIATION BETWEEN ZOOPLANKTON AND MOSQUITOES IN STORMWATER PONDS IS DRIVEN BY PRE- AND POST- COLONIZATION BEHAVIOR
ABSTRACT
Interspecific variation in behavior can influence the organization and dynamics of ecological communities, especially in human-altered landscapes. Disentangling the relative importance of these individual behaviors on community structure can be difficult given the presence of interacting predators, competitors, and prey. We coupled the results of a field survey of 37 stormwater ponds with laboratory experiments to examine whether the behavior of ovipositing mosquitoes combined with the preferential consumption (i.e., selectivity) by a copepod predator explained the distribution of larval mosquito populations in the field.
Communities were characterized in the summers of 2014 – 2017 and we observed a negative association between larval mosquitoes and zooplankton. We sought to explain this pattern using two laboratory experiments; in the first, we examined the oviposition behavior of Aedes and
Culex mosquitoes in response to the presence or absence of zooplankton. We found that Culex pipiens, but not Culex restuans or Aedes mosquitoes, avoided oviposition in habitats that contained (or previously contained) zooplankton. In the second experiment, we examined the behavior of a predatory copepod (Acanthocyclops) to determine if the presence of alternative prey influenced consumption of Aedes and Culex larvae. Acanthocyclops spp. selectively grazed on first-instar mosquito larvae even in the presence of alternative prey (Daphnia). We conclude that behavior of both ovipositing mosquitoes and a copepod predator shaped the distribution of larval mosquitoes in the field. This study further highlights the important role that an organism’s behavior can play in shaping the composition and dynamics of communities in natural and human-constructed habitats.
40
INTRODUCTION
Anthropogenic changes are occurring at an unprecedented rate resulting in a massive and unplanned ecological experiment affecting species richness and diversity (Andrén 1994, Grimm et al. 2000, McKinney 2002, Alberti 2005, Venter et al. 2016). Associated with these changes are the creation of new habitats that are suitable for novel communities to assemble (e.g., ponds, parks, streams; McDonnell and Pickett 1990, Hassall 2014, Lepczyk et al. 2017, Johnson et al.
2018). The assembly of these habitats is often influenced by a combination of the arrival of passively-dispersed individuals and those that exhibit habitat-selection behavior (Lima and
Zollner 1996, Bowler and Benton 2005, Cahill and McNickle 2011). For example, organisms exhibiting habitat-selection behavior may avoid colonizing habitats previously colonized by species that reduce their fitness; this pattern has been observed in response to predators and competitors in multiple systems (Resetarits 2001, Vonesh and Blaustein 2010, Lovari et al. 2013,
Pintar and Resetarits 2017). The behavior of interacting species post-colonization also contributes to patterns of community structure (Östman and Chase 2007, Vonesh et al. 2009).
For example, through the preferential consumption of one prey type over another, selective predators may enhance coexistence for non-selected prey at the expense of selected-prey populations (e.g., predator-mediated release from interspecific competition: Kesavaraju et al.
2008, Juliano et al. 2010, Ryabov et al. 2015). Predator selectivity becomes especially interesting in small aquatic habitats because predation has been shown to play a major role in the structuring of communities in these systems (Wellborn et al. 1996, Cottenie and De Meester 2004).
Aquatic invertebrates inhabiting small and constructed freshwater habitats (e.g., ponds built for the management of stormwater) are an ideal system to examine the effects of behavior on the spatial distribution of individual species. Common residents of these habitats include
41 amphibians, zooplankton, and insects, many of which exhibit high levels of dispersal (Schneider and Frost 1996, Cohen and Shurin 2003, Pinel-Alloul and Mimouni 2013, Holmes et al. 2016b).
Included in this list are larval mosquitoes which serve as both prey to multiple predators
(Blaustein and Chase 2007, Juliano 2009, Webb and Bashir 2013) and competitors to other planktonic grazers such as Daphnia (Duquesne et al. 2011). Stormwater ponds can contain both vertebrate (e.g., amphibians, birds, and fish) and invertebrate (e.g., insects and crustacean zooplankton) predators (Williams et al. 2003, Scheffer et al. 2006). These predators and competitors, both directly through predation and competition for limited resources, and indirectly through modifying behavior, decrease the abundance and alter the distribution of larval mosquitoes (Grill and Juliano 1996, Calliari et al. 2003, Knight et al. 2004, Blaustein and Chase
2007, Awasthi et al. 2015). For example, ovipositing females of the genera Culex and Culiseta commonly avoid the physical presence and chemical signal of predatory notonectids (Kiflawi et al. 2003, Blaustein et al. 2004, Silberbush et al. 2014). Furthermore, high densities of the competitive grazer Daphnia magna has been shown to limit oviposition by Cx. pipiens
(Duquesne et al. 2011). These indirect effects may have a significant influence on the life history characteristics of an organism (Schmitz et al. 1997, Walsh and Reznick 2008, Reznick et al.
2012, Chandrasegaran et al. 2018). However, these indirect effects of predation and competition are not well known for all predators and competitors of larval mosquitoes, including predatory copepods commonly found in small aquatic systems, including stormwater ponds (but see
Schmitz et al. 2004, Kesavaraju et al. 2008, and Juliano et al. 2010).
Our main objective was to examine the effects of behavior on the distribution of larval mosquitoes in stormwater ponds. First, we conducted a multi-year field survey to characterize the invertebrate communities inhabiting 37 stormwater ponds. Results from this field survey
42 were then coupled with two behavioral experiments; the first examining the oviposition behavior of Aedes and Culex mosquitoes in response to the presence of zooplankton communities
(including both predators and competitors). We predicted that Culex and Aedes would avoid oviposition in habitats that contained (or previously contained) these zooplankton assemblages.
In the second experiment, we examined selectivity of a predatory copepod (Acanthocyclops spp.) to determine if the presence of alternative prey options influenced patterns of Aedes and Culex consumption. We selected Daphnia as the alternative prey option, as they commonly inhabit small freshwater ponds, including those known to harbor larval mosquitoes (Dodson and Silva-
Briano 1996, Ortells et al. 2014, Holmes et al. 2016a). Specifically, we predicted that the predator would preferentially feed on early-stage mosquito larvae over Daphnia, thereby providing a possible mechanistic explanation for the negative correlation between mosquitoes and zooplankton.
MATERIALS AND METHODS
Field Survey
We characterized the invertebrate communities of 37 stormwater ponds in Champaign
County, IL, USA. Ponds were sampled throughout the summer months (May-August) from 2014
– 2017, resulting in 23 sampling periods. Because not all ponds contained enough water to be sampled at each sampling period, we collected a total of 280 samples over the four-year period.
To sample invertebrate communities, 3 L of pond water was collected haphazardly from each pond using a standard mosquito dipper (350 ml; BioQuip Products, Rancho Dominguez, USA).
Pond water was then filtered through a 70 µm sieve and all animals were preserved in 95%
EtOH. Individuals were identified to the lowest taxonomic level possible using Merritt et al.
43
(1996), Darsie and Ward (2005), and Haney (2013). Each sample was scanned for rare taxa; taxa with fewer than 300 individuals were counted completely, and those with over 300 individuals were subsampled (a minimum of three, 2 ml subsamples following whole sample dilution to 100 ml). For those that were subsampled, whole sample abundance for each taxa was estimated by N
= 100 * (n1+n2+n3)/6 where n1, n2, n3 are the three subsampled abundances.
We hypothesized that patterns observed in the field survey would be explained by both the pre-colonization behavior of ovipositing mosquitoes in response to the presence of a resident zooplankton community and the post-colonization behavior of a predatory copepod in the presence of alternative prey options. To test this, we conducted two separate behavioral experiments: 1) an oviposition choice experiment and 2) a predator choice experiment.
Oviposition Choice Experiment
To examine how the presence of a resident zooplankton community influences oviposition behavior in species of both Aedes and Culex, binary choice assays were conducted with gravid female mosquitoes. The four species used in the oviposition experiment reflected those whose gravid females could be readily collected from the field at the time of the experiment: Aedes triseriatus, Aedes japonicus, Culex pipiens, and Culex restuans. The fully factorial binary choice experiment included aquatic habitats containing the following treatments: zooplankton, kairomone (zooplankton removed immediately prior to assay), and control with the following binary combinations: control/zooplankton, control/kairomone, and zooplankton/kairomone.
Seven days in advance of the experiment, zooplankton and control stock solutions were made using a combination of 75% lake and 25% pond waters that were each filtered through a 1
44
µm glass microfiber filter. The zooplankton stock received regionally collected zooplankton at the following densities: Ostracods – 50 ind L-1, Copepods – 30 ind L-1, Ceriodaphnia spp. – 50 ind L-1. Densities of zooplankton used in the experiment were field-averaged values from the field survey in 2014 and 2015 (Fig. 2.1). Control stocks received no added zooplankton. Both stocks were stored at 15°C for seven days prior to the start of the experiment and received algae supplements of 3 x 104 cells ml-1 Ankistrodesmus falcatus (2 mg C L-1) on days 1 and 4. Because kairomones are chemical cues that degrade over time, the kairomone treatment was made by filtering out the zooplankton using a 70 µm sieve immediately prior to the start of the binary choice assays (< 30 minutes).
The afternoon prior to the binary choice assays, six grass infusion baited CDC gravid traps (John W. Hock Company, Gainesville, FL) containing approximately 3.78 L of grass infusion, were placed at Trelease Woods (Urbana, IL, USA; 40.129606, -88.143649) to collect gravid mosquitoes. Traps were checked early the following morning to ensure maximum survival of the adult mosquitoes. Gravid females were then returned to the lab and Culex were aspirated into experimental cages (24” x 24” x 24”) containing up to 30 individuals per cage.
Because it is difficult to enumerate oviposition events when more than one Aedes is present, gravid Aedes females were aspirated individually into separate 8” x 8” x 8” cages. Small (500 ml) black square “cups” were filled with 250 ml from their respective stock infusion (control, zooplankton, or kairomone), randomized, and placed in opposite corners of the cages to maximize the distance between each cup. Because infusions were sampled from larger stock solutions, the environment of each treatment cup may not have been identical across trials (e.g., zooplankton abundance and composition in the zooplankton treatment cups may have varied among trials). Furthermore, given the potential importance of copepod predation on these results,
45 we ensured each zooplankton treatment cup contained at least one copepod predator of the genus
Acanthocyclops. Cups used in the Aedes cages were lined with germination paper so that eggs could be collected and hatched for species identification. All cages were checked after 24 hours, and egg rafts (Culex), and germination papers (Aedes) were collected, distributed into separate
50 ml conical tubes, and hatched. Hatched larvae were grown to 4th instar and individuals were identified to species using Darsie and Ward (2005).
Predator Choice Experiment
We investigated the consumption behavior of a predatory copepod to examine the potential role of this predator in driving patterns observed in the field survey. Specifically, we examined how the presence of alternative prey influenced the rate of larval mosquito consumption using 24-hour predation choice assays in which single Acanthocyclops copepods were exposed to equal densities of neonate Daphnia (< 12 hours old) and first-instar mosquito larvae (Culex or Aedes; < 12 hours old). Two species of Aedes (Aedes albopictus and Aedes aegypti) were used because we had access to large numbers of recently laid eggs from laboratory colonies. One species of Culex (Cx. pipiens) was used because egg rafts were readily available in the field at the time of the experiment. We also calculated predation rates on each prey group individually by exposing some copepod predators to only one prey type. No-predator control trials were simultaneously conducted to determine baseline mortality of the prey.
Copepods identified to the genus Acanthocyclops using Haney et al. (2013) were collected from a local permanent pond. The day prior to the start of the trials, Culex egg rafts were collected from the field using grass infusion-baited 18.9 L buckets. Eggs of Ae. albopictus and Ae. aegypti were provided by the Medical Entomology Laboratory at the Illinois Natural
46
History Survey (Champaign, IL, USA) from laboratory-reared cultures (> 20 generations). Eggs and egg rafts were hatched at 25°C into a deionized water/yeast mixture. One set of trials for Ae. albopictus and Ae. aegypti was conducted in a 100 ml mixture of 75% lake and 25% pond waters that were each filtered through 1 µm glass microfiber filter. We conducted an additional set of trials for Ae. albopictus and Cx. pipiens in 40 ml again using a mixture of 75/25% lake and pond water. Total volume was reduced in the second set of trials due to the limited access to experimental animals and egg rafts from the field. All cyclopoid predators were starved for 12 –
24 hours at 20°C prior to prey introductions. Even though the volume of water was not consistent between trials (100 ml and 40 ml), the densities of total prey were equal across experimental containers (1 prey ml-1). Though, this difference may have led to different prey encounter rates
(Witt and Cáceres 2004, Turesson and Brönmark 2007).
The day of the experiment, neonate Daphnia (< 12 hours old) and first instar mosquito larvae (< 12 hours old) were placed in experimental containers (150 ml beakers and 50 ml conical tubes for the 100 ml and 40 ml trials, respectively). Daphnia used in this study ranged from 0.52 to 0.74 mm (mean: 0.65 +/- 0.01). While we did not measure mosquito larvae prior to the start of the experiment, previous studies have documented that first instar mosquito larvae can range in length from 0.5 mm to 1.5 mm (Bar and Andrew 2013). Control and single-prey treatments were stocked with 100% of a single prey (100 or 40 Daphnia or mosquito larvae for the 100 ml and 40 ml trials, respectively), and the prey choice treatments received 50% of each prey (50 Daphnia and 50 mosquito larvae or 20 Daphnia and 20 mosquito larvae for the 100 ml and 40 ml trials, respectively). Following inoculation with a predator, all experimental containers were held at 20°C for 24 hours. Following the 24-hour experimental period, predators were
47 removed and all remaining prey were filtered through a 70 µm sieve and preserved in 95% EtOH until samples were counted. Any partially consumed prey were recorded as “consumed”.
Analyses
Two logistic regressions were used to investigate potential associations between zooplankton and mosquito larvae in the field. The first examined the association between total zooplankton abundance and the presence of mosquito larvae; the second examined the association between cyclopoid copepod abundance and the presence of mosquito larvae. We also used a Kruskal-Wallis test to compare mosquito density across zooplankton density divided into categories of low (0 – 100 ind L-1), medium (101 – 300 ind L-1), and high (300+ ind L-1: Fig.
A2). To determine if the physical presence of zooplankton, or their chemical signal, influenced oviposition behavior of Aedes and Culex mosquitoes, Fisher’s exact tests were conducted separately for each treatment combination (control/zooplankton, kairomone/control, and kairomone/zooplankton). Each oviposition event represented a unique replicate. To reduce type
I (false-positive) error, a Bonferroni correction was used to adjust (α) for each comparison. Due to low sample sizes, both species of Aedes were only assayed on the control/zooplankton treatment and were pooled together for statistical analyses.
To determine whether Acanthocyclops spp. exhibited prey preference, we calculated the
Manly-Chesson selectivity index (α) for each prey-choice trial (Manly 1974, Chesson 1978,
1983). This index is widely used in ecological studies examining prey-choice behavior
(Mittelbach 1988, Nilsson and Bronmark 2000, León and Bjorndal 2002, Järv et al. 2011) and is calculated with the formula:
� = ( / ) � = 1,2, … �, (6) ∑( / )
48 where ri is the proportion of prey item i consumed, pi is the proportion of prey item i in the environment, and m is the number of prey taxa in the environment. The selectivity index (α) ranges from 0 (complete avoidance of prey type) to 1 (only prey type consumed; i.e., highly selective). With two prey types, the threshold for random feeding (i.e., no choice) is 0.5. To determine whether Acanthocyclops copepods exhibited selectivity behavior for each of the three mosquito species, we conducted separate one-sample t-tests for each species to determine significant deviations from the threshold of no prey selectivity (0.5). A Bonferroni correction was again used to adjust α for each comparison to reduce type I (false-positive) error.
RESULTS
Field Survey
We observed at least 12 mosquito species belonging to seven genera (Uranotaenia sp.,
Culiseta inornata, Orthopodomia sp., Anopheles quadrimaculatus, Psorophora columbiae,
Aedes vexans, Aedes atropalpus, Aedes japonicus, Culex pipiens, Culex restuans, Culex erraticus, Culex territans) inhabiting the 37 stormwater ponds throughout the four-year field survey. Zooplankton were ubiquitous and abundant throughout the landscape (all but one of the
284 samples contained at least one zooplankton individual) and were recorded at all ponds at least once throughout the four years of sampling. Larval mosquito populations were observed in
39.4% of the 284 samples collected with four of the 37 ponds being devoid of mosquito larvae across all sample periods. Larval mosquito density was significantly lower in high-zooplankton density ponds than in both medium- and low- zooplankton density ponds (Fig. A2: Kruskal-
Wallis, χ2 = 7.3, df = 2, P = 0.03). Results from the logistic regression revealed that the probability of encountering mosquito larvae was negatively correlated with total zooplankton
49 abundance (Fig. 2.1A, F1,277 = 7.58, P = 0.006). Furthermore, no mosquito larvae were observed at zooplankton densities exceeding 2,077 individuals per liter (Fig. 2.1A). However, this result was not driven by copepods; we found no correlation between mosquito larvae
(presence/absence) and cyclopoid copepod density (Fig. 2.1B, F1,278 = 0.11, P = 0.74).
Oviposition Choice Experiment
We collected four mosquito species from two genera for use in the oviposition choice experiment: Aedes triseriatus, Aedes japonicus, Culex pipiens, and Culex restuans). Mosquito species varied in their oviposition behavior in response to the presence of zooplankton. Gravid
Cx. pipiens significantly avoided oviposition in cups containing a zooplankton assemblage (Fig.
2.2A: χ2 = 9.80, df = 1, P = 0.002) and kairomones from these assemblages (Fig 2.2A: χ2 = 6.85, df = 1, P = 0.01), instead choosing to oviposit in control cups. When given the choice between cups containing zooplankton assemblages and cups containing kairomones, Cx. restuans exhibited no preference (Fig 2.2A: χ2 = 0.02, df = 1, P = 0.90). There was no treatment combination in which gravid Cx. restuans demonstrated oviposition choice (Fig 2.2B: control/zooplankton: χ2 = 0.05, df=1, P = 0.82, control/kairomone: χ2 = 2.77, df = 1, P = 0.10, kairomone/zooplankton: χ2 = 0.24, df = 1, P = 0.63). We found that gravid Aedes mosquitoes exhibited no choice between control and zooplankton cups (Fig 2.2C: χ2 = 0.06, df = 1, P =
0.80). In three out of the 16 Aedes trials, the gravid females dispersed their eggs in both containers, a behavior referred to as skip-oviposition. These three skip-oviposition events occurred exclusively in Ae. triseriatus.
50
Predator Choice Experiment
In the predator-free controls, 4.5% (± 0.9% SE) of animals were lost to mortality after the
24-hour period of the behavioral assays. There was evidence that Acanthocyclops copepods consumed either mosquito larvae or Daphnia neonates in all predator-containing treatments (Fig.
2.3). In the single-prey assays, 35% (± 3.9% SE) of total prey were consumed following the 24- hour experimental period (Fig. 2.3A), with no significant difference in the number of prey consumed between Daphnia-only and mosquito-only trials (t-test, t = 0.59, P = 0.56).
Consumption of mosquito larvae ranged from 16% to 85% in Ae. albopictus, 7% to 34% in Ae. aegypti of total prey, and 25% (no range) in Cx. pipiens, whereas, consumption of Daphnia ranged from 0% to 69% of total prey (Fig. 2.3A).
When exposed to both prey types, Acanthocyclops predators consumed similar total amounts of prey when compared to the single-prey trials (35.8% ± 4.0% of total prey). Individual predators exhibited variation in the strength of prey selectivity; Manly-Chesson α ranged from
0.33 to 0.95 in Ae. albopictus, 039 to 0.77 in Ae. aegypti, and 0.67 to 0.77 in Cx. pipiens (Fig.
2.3B). Despite this variation, there was evidence that when given the choice, Acanthocyclops predators preferred both Cx. pipiens (t-test, t = 9.53, df = 3, P = 0.001) and Ae. albopictus (t-test, t = 5.10, df =14, P < 0.001) over Daphnia (Fig. 2.3B). However, this pattern was not observed when Acanthocyclops copepods were given the choice between Ae. aegypti and Daphnia (t-test, t
= 0.95, df = 3, P = 0.21: Fig. 2.3B).
DISCUSSION
We found a negative and significant correlation between mosquito larvae and zooplankton in the stormwater ponds we surveyed. Results from the behavioral experiments
51 suggest this pattern may be driven by both the behavior of ovipositing mosquitoes and post- colonization behavior of the larval mosquito predators. However, these behaviors varied among species. While Culex pipiens avoided oviposition in habitats that contained, or previously contained, zooplankton assemblages, Culex restuans and Aedes did not exhibit this preference. In our laboratory predator-choice experiment, the copepod predator, Acanthocyclops spp., selectively grazed on first-instar mosquito larvae in the presence of alternative prey options
(neonate Daphnia). Our results show that interspecific variation in behavior can play a significant role in the organization and dynamics of mosquito communities in natural and constructed habitats.
Negative relationships between larval mosquitoes and other taxa in both natural and artificial habitats have been documented previously (Stav et al. 1999, Arav and Blaustein 2006,
Vonesh et al. 2009). For example, in a survey of UK wetlands, Golding et al. (2015) found that mosquito larvae (Anopheles maculipennis and Culex modestus) were negatively correlated with certain predators (ditch shrimp of the genus Palaemonetes and fish [Pisces]), but not others
(Coleoptera and Odonata). Results from the predator-choice experiment suggest that predatory copepods may play an important role in the patterns observed in the field study. However, the lack of correlation between copepods and larval mosquitoes in the field suggests they may not be the only species underlying this pattern. Predators and competitors may differ in their effects on a given system. For example, Silberbush and Resetarits (2017) found that not all predatory fish elicited an anti-predator response from ovipositing Culex mosquitoes. On the other hand, in a survey of small ephemeral ponds, Carver et al. (2010) found no negative relationship between mosquito density and any single taxonomic group, instead they found that mosquito density decreased as predator richness and the total density of other taxa increased.
52
In our system, the negative association between larval mosquitoes and zooplankton appears to be driven by a combination of site-avoidance behavior by ovipositing mosquitoes and potential post-colonization interactions of mosquito larvae with copepod predators. The ability to detect and avoid environmental cues that signal sub-optimal habitat conditions relies on both the successful detection and appropriate response to sensory information and can have significant advantages for an individual’s fitness (Chivers and Smith 1998, Apfelbach et al. 2005, Preisser et al. 2005, Liesenjohann et al. 2013). Extensive work in this area has shown that, more broadly, organisms can use a combination of visual, auditory, tactile, and chemosensory cues to detect threats in and assess fitness prospects of an environment (Bentley and Day 1989, Chivers and
Smith 1998, Kats and Dill 1998, Hemmi 2005, Miyakawa et al. 2010). The interplay of the physical, chemical and physiological factors influencing mosquito oviposition behavior has been reviewed in Bentley and Day (1989). Previous studies have shown that a wide range of mosquito taxa avoid oviposition in aquatic habitats containing predators, suggesting that avoidance during habitat selection may play an important role in the negative association observed in the field survey (Stav et al. 1999, Resetarits 2001, Angelon and Petranka 2002, Blaustein et al. 2004,
Vonesh and Blaustein 2010).
In our oviposition experiment, Cx. pipiens strongly avoided ovipositing in habitats that contained, or previously contained, zooplankton communities (including at least one copepod predator). Results from the predator choice experiment demonstrate the consequences of failing to avoid these threats for Cx. pipiens. Even in the presence of alternative prey, a single
Acanthocyclops spp. individual consumed Cx. pipiens larvae at rates exceeding 11 individuals per day. This finding is consistent with others who have shown that copepods are important predators of larval Cx. pipiens with a single copepod individual consuming > 6 mosquito larvae
53 per day (Blaustein and Margalit 1994, Calliari et al. 2003). Though we visually confirmed the presence of at least one Acanthocyclops predator in each zooplankton treatment of our oviposition trials, we are unable to determine whether avoidance behavior was attributed to the presence of the predator, competitors, or a combination of the two.
We found that Cx. restuans and Aedes spp. did not avoid habitats that contained zooplankton assemblages or kairomones, a result that we cannot fully explain. A meta-analysis of mosquito oviposition studies by Vonesh and Blaustein (2010) showed a stronger effect size for predator avoidance behavior in Cx. restuans than in Cx. pipiens. It is worth noting, however, that species varied significantly in their response to different predators. While multiple predators were included in this review, including multiple copepod predators, studies examining oviposition in response to Acanthocyclops spp. were absent. Similar to Cx. pipiens, Cx. restuans can inhabit a wide range of larval habitats, including those developed for stormwater management (Irwin et al. 2008, Gardner et al. 2012). In our field survey, we found that these two commonly colonized the same ponds, albeit at separate times. This suggests that Cx. restuans were likely exposed to the same or similar threats as Cx. pipiens. The use of alternative threat- avoidance behaviors has been observed in many animal groups (insects: Juliano 2009,
Hernandez and Peckarsky 2014, fish: Power 1984, Romare and Hansson 2003, birds: Norrdahl and Korpimaki 1998, snails: Levri et al. 2012, zooplankton: Zaret and Suffern 1976, also reviewed in: A.S. Griffin 2006), and may address this risk in Cx. restuans. For example, Ferrari et al. (2008) observed that Cx. restuans reduced activity in the presence of conspecific alarm cues (i.e., chemical cues exuded by larval mosquitoes exposed to salamander predators).
In a meta-analysis, Vonesh and Blaustein (2010) noted significant variation in oviposition response across species of Aedes with some species strongly avoiding predators (Aedes
54 taeniorhynchus), some being attracted to predators (Aedes triseriatus), and others showing no effect in either direction (Aedes aegypti and Aedes albopictus). Because Aedes lay eggs above the water line, the ability to detect predation cues may be lessened relative to other taxa which lay directly on the water surface (Vonesh and Blaustein 2010). However, in many of these studies presented for Aedes spp., multiple gravid females had access to each oviposition treatment (Torres-Estrada et al. 2001, Van Dam and Walton 2008). As a result, skip oviposition
(i.e., a single gravid mosquito dispersing her eggs across multiple containers) may have occurred but would be difficult to assess. We observed three occurrences (out of 16 oviposition events) of skip oviposition in the Aedes trials. This behavior is considered a bet-hedging strategy; by spreading risk across multiple habitats, mothers may enhance overall fitness at the expense of obtaining maximum fitness (Colton et al. 2003, Fonseca et al. 2015).
The copepod predator, Acanthocyclops spp., preferentially grazed on mosquito larvae in the presence of alternative prey (Li and Li 1979, Andreadis and Gere 1992). Our findings conflict with results by Andreadis and Gere (1992) who found that consumption of Aedes by
Acanthocyclops vernalis significantly declined in the presence of alternative prey (newly hatched copepod nauplii). While some studies examine predation in the absence of alternative prey options (Miura and Takahashi 1988, Kay et al. 1992, Lounibos et al. 1993), the inclusion of non- mosquito prey may modify rates of mosquito consumption by larval predators (Blaustein 1998,
Lundkvist et al. 2003, Kumar and Rao 2003). Results from our study show that in a simple predator/two-prey system, two species of larval mosquitoes (Ae. albopictus and Cx. pipiens) are preferentially consumed over an alternative zooplankton for the copepod predator
Acanthocyclops spp. However, mosquitoes inhabiting small freshwater ponds may co-occur with multiple predators and competitors that do not share this behavior (Blaustein and Chase 2007,
55
Tranchida et al. 2009, Duquesne et al. 2011, Awasthi et al. 2015, Rowbottom et al. 2015).
Exposure to predators within these aquatic environments may also elicit predator-avoidance behavior (e.g., foraging time, location in the water column, movement: Zaret and Suffern 1976,
Grill and Juliano 1996). Other unmeasured factors may also underlie this preference behavior
(e.g., differences in capture time or prey nutrition: Li and Li 1979); however, at this time we are unable to account for these in the patterns observed in our laboratory experiment.
We showed that the negative association between mosquito larvae and zooplankton observed in a survey of stormwater ponds were potentially explained by avoidance behavior of adult mosquitoes combined with post-colonization behavior of a copepod predator. However, this behavior was not consistent across all studied species, thereby providing further evidence that organisms exhibit interspecific variation in these important behavioral traits. For mosquito larvae inhabiting small stormwater ponds, the abundance, diversity, and distribution of crustacean zooplankton appear to play an important role in shaping the distribution of some species. When considering the design and management of stormwater ponds, efforts to promote crustacean zooplankton richness and diversity (e.g., enhancing landscape connectivity to regional ponds) may decrease mosquito populations by decreasing the quality of their juvenile habitats.
As humans continue to modify the environment, an understanding of the ecological mechanisms underlying patterns of colonization and community structure in these new human-altered landscapes are key to predicting how this development will influence global patterns of biodiversity. We encourage further research on such topics in this system; these constructed and semi-natural environments provide an excellent opportunity to examine contemporary patterns of community assembly in newly established habitats.
56
ACKNOWLEDGEMENTS
We first thank Andrew Mackay who provided us with study sites that were used in the field survey. We also thank our undergraduate research assistants, Cameron Schwing, Sana
Khadri, and Xorla Ocloo, as well as Ilona Menel and Ping Lee for their field and laboratory assistance on this project. We thank Brian F. Allan, Jessica R. Holmes, Ephantus J. Muturi,
James P. O’Dwyer, Tara E. Stewart, Lynette R. Strickland, and Andrew V. Suarez who kindly provided comments on multiple drafts of this manuscript. This research was funded by the
United States National Science Foundation [DEB-1754115], a grant through the University of
Illinois Institute for Sustainability, Energy, and Environment (iSEE), and through grants from the
University of Illinois at Urbana-Champaign School of Integrative Biology and Department of
Animal Biology.
57
FIGURES
A B 1 1 1.0 Genus 1.0 Genus
Aedes presence presence Anopheles Aedes Culex Anopheles 0.8 0.8 Culex
Oviposition experiment density 0.6 0.6 0.5 0.5 Mosquito Mosquito 0.4 0.4 Probability of presence Probability of presence 0.2 0.2
● ●●●●●●●●●●●● ●●●●●●●●● ● ● ● ●●●●●●●●● ●● ● ●●●●● ● ● ● 0 ●● 0 ●●●●●●●●●●●●●● ●● ● ●●●● ●●● ●● ● ●●●●●●● ● ● ●●●●● ● 0.0 ●●●●●●● ● ●● ●● ● ● ● 0.0 ●●● absence absence 0 500 1500 2500 3500 4500 0 500 1500 2500 3500 4500 Zooplankton Density (ind/L) Copepod Density (ind/L)
Figure 2.1: The presence/absence of mosquito larvae from the field survey are plotted as a function of (A) zooplankton density and (B) cyclopoid copepod density. The red line shows the probability of occurrence of mosquito larvae as a function of zooplankton density as fitted by the significant logistic regression analysis. The arrow indicates the zooplankton density used for the oviposition choice experiment (130 individuals L-1).
58
A Cx. pipiens oviposition choice expt.B Cx. restuans oviposition choice expt. N = 54 N = 75 N = 68 N = 20 N = 13 N = 17 100% 100%
75% * 75% * % in Z % in Z 50% 50% % in K % in C % in C % in K % Oviposition events % Oviposition events 25% 25%
0% 0%
AedesControl oviposition − Kairomone choice − Kairomone expt. − Control − Kairomone − Kairomone − C Zooplankton Control Zooplankton Zooplankton Control Zooplankton N = 16 100% Treatments Treatments
75%
50% % in Z % in C % Oviposition events 25%
0%
Control − Kairomone − Kairomone − Zooplankton Control Zooplankton Treatments
Figure 2.2: Percentage of oviposition events in each binary choice assay is reported for each treatment combination (control, zooplankton infusion, kairomone infusion [zooplankton removed immediately prior to 24 hour trial]). Stacked bars represent the percentage of oviposition events for cups in each treatment combination summed across replicates (relative frequency of oviposition events in each cup are color-coded with the following colors: zooplankton [black bars], kairomone
[grey bars], and control [white bars]). Each set of combinations was conducted for two species of
Culex mosquitoes (A: Culex. pipiens, B: Culex restuans) and C: Aedes mosquitoes. The number of replicates (N; i.e., the number of oviposition events) is shown above each stacked bar. Significance using Fisher’s exact tests are indicated at the P = 0.016 (0.05/3) for each pairwise combination with an asterisk (*). 59
1.00 A 1.00 B * *
0.75 0.75 ● ● ● N = 15 N = 4 ●
0.50 N = 3 N = 10 0.50 N = 4
● Chesson index − % Consumed
● ● 0.25 ● ● 0.25
N = 5 N = 1 Manly N = 5 N = 5
0.00 0.00 Ae. aegypti Ae. albopictus Cx. pipiens Daphnia spp. Ae. aegypti Ae. albopictus Cx. pipiens Prey Species Mosquito Species
Figure 2.3: A) Percentage of prey consumed is plotted (+/- SE) for single-prey trials (Aedes aegypti,
Aedes albopictus, Culex pipiens, and Daphnia spp.). Trials conducted in 100 ml and 40 ml are represented with black and grey points, respectively. B) Predator preference (Manly-Chesson selection index, α) is plotted (+/- SE) for predator-choice experiment in which a single
Acanthocyclops spp. is exposed to equal proportions of mosquito (Ae. aegypti, Ae. albopictus, or Cx. pipiens) and Daphnia spp. individuals. The dashed line at y = 0.5 represents no preference by the predator. The number of replicates (N) is shown aside each trial and significance using one-sample t- tests is indicated at P = 0.016 (0.05/3) for each species with an asterisk (*).
60
REFERENCES
Alberti, M. 2005. The effects of urban patterns on ecosystem function. International Regional
Science Review 28:168–192.
Andreadis, T. G., and M. A. Gere. 1992. Laboratory evaluation of Acanthocyclops vernalis and
Diacyclops bicuspidatus thomasi (Copepoda: Cyclopidae) as predators of Aedes canadensis
and Ae. stimulans (Diptera: Culicidae). Journal of Medical Entomology 29:974–979.
Andrén, H. 1994. Effects of habitat fragmentation on birds and mammals of suitable habitat: a
review landscapes with different proportions. Oikos 71:355–366.
Angelon, K. A., and J. W. Petranka. 2002. Chemicals of predatory mosquitofish (Gambusia
affinis) influence selection of oviposition site by Culex mosquitoes. Journal of Chemical
Ecology 28:797–806.
Apfelbach, R., C. D. Blanchard, R. J. Blanchard, R. A. Hayes, and I. S. McGregor. 2005. The
effects of predator odors in mammalian prey species: a review of field and laboratory
studies. Neuroscience and Biobehavioral Reviews 29:1123–1144.
Arav, D., and L. Blaustein. 2006. Effects of pool depth and risk of predation on oviposition
habitat selection by temporary pool dipterans. Journal of Medical Entomology 43:493–497.
Awasthi, A. K., J. C. Molinero, C. H. Wu, K. H. Tsai, C. C. King, and J. S. Hwang. 2015.
Behavioral changes in mosquito larvae induced by copepods predation. Hydrobiologia
749:113–123.
Bar, A., and J. Andrew. 2013. Morphology and morphometry of Aedes aegypti larvae. Annual
Review & Research in Biology 3:1–21.
Bentley, M. D., and J. F. Day. 1989. Chemical ecology and behavioral aspects of mosquito
oviposition. Annual Review of Entomology 34:401–421.
61
Blaustein, L. 1998. Influence of the predatory backswimmer, Notonecta maculata, on
invertebrate community structure. Ecological Entomology 23:246–252.
Blaustein, L., and J. M. Chase. 2007. Interactions between mosquito larvae and species that share
the same trophic level. Annual Review of Entomology 52:489–507.
Blaustein, L., M. Kiflawi, A. Eitam, M. Mangel, and J. E. Cohen. 2004. Oviposition habitat
selection in response to risk of predation in temporary pools: mode of detection and
consistency across experimental venue. Oecologia 138:300–305.
Blaustein, L., and J. Margalit. 1994. Differential vulnerability among mosquito species to
predation by the cyclopoid copepod, Acanthocyclops virdis. Israel Journal of Zoology
40:55–60.
Bowler, D. E., and T. G. Benton. 2005. Causes and consequences of animal dispersal strategies:
relating individual behaviour to spatial dynamics. Biological Reviews of the Cambridge
Philosophical Society 80:205–225.
Cahill, J. F., and G. G. McNickle. 2011. The behavioral ecology of nutrient foraging by plants.
Annual Review of Ecology, Evolution, and Systematics 42:289–311.
Calliari, D., K. Sanz, M. Martínez, G. Cervetto, M. Gómez, and C. Basso. 2003. Comparison of
the predation rate of freshwater cyclopoid copepod species on larvae of the mosquito Culex
pipiens. Medical and Veterinary Entomology 17:339–342.
Carver, S., H. Spafford, A. Storey, and P. Weinstein. 2010. The roles of predators, competitors,
and secondary salinization in structuring mosquito (Diptera: Culicidae) assemblages in
ephemeral water bodies of the wheatbelt of Western Australia. Environmental Entomology
39:798–810.
62
Chandrasegaran, K., S. Rao Kandregula, S. Quader, and S. A. Juliano. 2018. Context-dependent
interactive effects of nonlethal predation on larvae impact adult longevity and body
composition. PLoS ONE 13:e0192104.
Chesson, J. 1978. Measuring preference in selective predation. Ecology 59:211–215.
Chesson, J. 1983. The estimation and analysis of preference and its relationship to foraging
models. Ecology 64:1297–1304.
Chivers, D. P., and R. J. F. Smith. 1998. Chemical alarm signalling in aquatic predator-prey
systems: a review and prospectus. Ecoscience 5:338–352.
Cohen, G. M., and J. B. Shurin. 2003. Scale-dependence and mechanisms of dispersal in
freshwater zooplankton. Oikos 103:603–617.
Colton, Y. M., D. D. Chadee, and D. W. Severson. 2003. Natural skip oviposition of the
mosquito Aedes aegypti indicated by codominant genetic markers. Medical and Veterinary
Entomology 17:195–204.
Cottenie, K., and L. De Meester. 2004. Metacommunity structure: synergy of biotic interactions
as selective agents and dispersal as fuel. Ecology 85:114–119.
Van Dam, A. R., and W. E. Walton. 2008. The effect of predatory fish exudates on the
ovipostional behaviour of three mosquito species: Culex quinquefasciatus, Aedes aegypti
and Culex tarsalis. Medical and Veterinary Entomology 22:399–404.
Darsie, R. J., and R. Ward. 2005. Identification and geographical distribution of the mosquitoes
of North America, north of Mexico. Page Mosquito Systematics. 2nd edition. University
Press of Florida, Gainsville, FL, USA.
63
Dodson, S. I., and M. Silva-Briano. 1996. Crustacean zooplankton species richness and
associations in reservoirs and ponds of Aguascalientes State, Mexico. Hydrobiologia
325:163–172.
Duquesne, S., I. Kroeger, M. Kutyniok, and M. Liess. 2011. The potential of cladocerans as
controphic competitors of the mosquito Culex pipiens. Journal of Medical Entomology
48:554–560.
Ferrari, M. C. O., F. Messier, and D. P. Chivers. 2008. Variable predation risk and the dynamic
nature of mosquito antipredator responses to chemical alarm cues. Chemoecology 17:223–
229.
Fonseca, D. M., L. R. Kaplan, R. A. Heiry, and D. Strickman. 2015. Density-dependent
oviposition by female Aedes albopictus (Diptera: Culicidae) spreads eggs among containers
during the summer but accumulates them in the fall. Journal of Medical Entomology
52:705–712.
Gardner, A. M., G. L. Hamer, A. M. Hines, C. M. Newman, E. D. Walker, and M. O. Ruiz.
2012. Weather variability affects abundance of larval Culex (Diptera: Culicidae) in storm
water catch basins in suburban Chicago. Journal of Medical Entomology 49:270–276.
Golding, N., M. A. Nunn, and B. V Purse. (2015). Identifying biotic interactions which drive the
spatial distribution of a mosquito community. Parasites & Vectors 8:367.
Griffin, A. S. 2006. Social learning about predators: a review and prospectus. Learning &
Behavior 31:131–140.
Grill, C. P., and S. A. Juliano. 1996. Predicting species interactions based on behaviour:
predation and competition in container-dwelling mosquitoes. The Journal of Animal
Ecology 65:63–76.
64
Grimm, N. B., J. Morgan Grove, S. T. A. Pickett, and C. L. Redman. 2000. Integrated
approaches to long-term studies of urban ecological systems. BioScience 50:571.
Haney, J. F. et al. 2013. “An-Image-based Key to the Zooplankton of North America” version
4.0 released 2010. University of New Hampshire Center for Freshwater Biology
Hassall, C. 2014. The ecology and biodiversity of urban ponds. Wiley Interdisciplinary Reviews:
Water 1:187–206.
Hemmi, J. M. 2005. Predator avoidance in fiddler crabs: 2. The visual cues. Animal Behaviour
69:615–625.
Hernandez, S. A., and B. L. Peckarsky. 2014. Do stream mayflies exhibit trade-offs between
food acquisition and predator avoidance behaviors? Freshwater Science 31:124–133.
Holmes, C. J., S. Figary, K. L. Schulz, and C. E. Cáceres. 2016a. Effects of diversity on
community assembly in newly formed pond communities. Ecosphere 7:e01377.
Holmes, C. J., J. H. Pantel, K. L. Schulz, and C. E. Cáceres. 2016b. Initial genetic diversity
enhances population establishment and alters genetic structuring of a newly established
Daphnia metapopulation. Molecular Ecology 25:3299–3308.
Irwin, P., C. Arcari, J. Hausbeck, and S. Paskewitz. 2008. Urban wet environment as mosquito
habitat in the upper Midwest. EcoHealth 5:49–57.
Järv, L., J. Kotta, I. Kotta, and T. Raid. 2011. Linking the structure of benthic invertebrate
communities and the diet of native and invasive fish species in a brackish water ecosystem.
Annales Zoologici Fennici 2450:129–141.
65
Johnson, A. L., D. Borowy, and C. M. Swan. 2018. Land use history and seed dispersal drive
divergent plant community assembly patterns in urban vacant lots. Journal of Applied
Ecology 55:451–460.
Juliano, S. A. 2009. Species interactions among larval mosquitoes: context dependence across
habitat gradients. Annual Review of Entomology 54:37–56.
Juliano, S. A., L. P. Lounibos, N. Nishimura, and K. Greene. 2010. Your worst enemy could be
your best friend: predator contributions to invasion resistance and persistence of natives.
Oecologia 162:709–718.
Kats, L. B., and L. M. Dill. 1998. The scent of death: chemosensory assessment of predation risk
by prey animals. Ecoscience 5:361–394.
Kay, B. H., C. P. Cabral, A. C. Sleigh, M. D. Brown, Z. M. Ribeiro, and A. W. Vasconcelos.
1992. Laboratory evaluation of Brazilian Mesocyclops (Copepoda: Cyclopidae) for
mosquito control. Journal of Medical Entomology 29:599–602.
Kesavaraju, B., K. Damal, and S. A. Juliano. 2008. Do natural container habitats impede invader
dominance? Predator-mediated coexistence of invasive and native container-dwelling
mosquitoes. Oecologia 155:631–639.
Kiflawi, M., L. Blaustein, and M. Mangel. 2003. Oviposition habitat selection by the mosquito
Culiseta longiareolata in response to risk of predation and conspecific larval density.
Ecological Entomology 28:168–173.
Knight, T. M., J. M. Chase, C. W. Goss, and J. J. Knight. 2004. Effects of interspecific
competition, predation, and their interaction on survival and development time of immature
Anopheles quadrimaculatus. Journal of Vector Ecology 29:277–284.
66
Kumar, R., and T. Rao. 2003. Predation on mosquito larvae by Mesocyclops thermocyclopoides
(Copepoda: Cyclopoida) in the presence of alternate prey. International Review of
Hydrobiology 88:570–581.
León, Y. M., and K. A. Bjorndal. 2002. Selective feeding in the hawksbill turtle, an important
predator in coral reef ecosystems. Marine Ecology Progress Series 245:249–258.
Lepczyk, C. A., M. F. J. Aronson, K. L. Evans, M. A. Goddard, S. B. Lerman, and J. S. Macivor.
2017. Biodiversity in the city: fundamental questions for understanding the ecology of
urban green spaces for biodiversity conservation. BioScience 9:799–807.
Levri, E. P., A. N. Dubensky, A. S. Mears, and C. A. Opiela. 2012. Interpopulation variation in
predator avoidance behavior of a freshwater snail to the same predator. Canadian Journal of
Zoology 90:616–623.
Li, J. L., and H. W. Li. 1979. Species-specific factors affecting predator-prey interactions of the
copepod Acanthocyclops vernalis with its natural prey. Limnology and Oceanography
24:613–626.
Liesenjohann, M., T. Liesenjohann, R. Palme, and J. A. Eccard. 2013. Differential behavioural
and endocrine responses of common voles (Microtus arvalis) to nest predators and resource
competitors. BMC Ecology 13:33.
Lima, S. L., and P. A. Zollner. 1996. Towards a behavioral ecology of ecological landscapes.
Trends in Ecology and Evolution 11:131–135.
Lounibos, L. P., N. Nishimura, and R. L. Escher. 1993. Fitness of a treehole mosquito:
influences of food type and predation. Oikos 66:114–118.
67
Lovari, S., I. Minder, F. Ferretti, N. Mucci, E. Randi, and B. Pellizzi. 2013. Common and snow
leopards share prey, but not habitats: competition avoidance by large predators? Journal of
Zoology 291:127–135.
Lundkvist, E., J. Landin, M. Jackson, and C. Svensson. 2003. Diving beetles (Dytiscidae) as
predators of mosquito larvae (Culicidae) in field experiments and in laboratory tests of prey
preference. Bulletin of Entomological Research 93:219–226.
Manly, B. F. J. 1974. A model for certain types of selection experiments. Biometrics 30:281–
294.
McDonnell, M. J., and S. T. Pickett. 1990. Ecosystem structure and function along gradients: an
unexploited urban-rural opportunity for ecology. Ecology 71:1232–1237.
McKinney, M. L. 2002. Urbanization, biodiversity, and conservation. BioScience 52:883–890.
Merritt, R. W., K. W. Cummins, and M. B. Berg, editors. 1996. An Introduction to the Aquatic
Insects of North America. Page Kendall Hunt Publishing Company. 4th edition. Dubuque,
IA, USA.
Mittelbach, G. G. 1988. Competition among refuging sunfishes and effects of fish density on
littoral zone invertebrates. Ecology 69:614–623.
Miura, T., and R. M. Takahashi. 1988. A laboratory study of predation by damselfly nymphs,
Enallagma civile, upon mosquito larvae, Culex tarsalis. Journal of the American Mosquito
Control Association 4:129–131.
Miyakawa, H., M. Imai, N. Sugimoto, Y. Ishikawa, A. Ishikawa, H. Ishigaki, Y. Okada, S.
Miyazaki, S. Koshikawa, R. Cornette, and T. Miura. 2010. Gene up-regulation in response
to predator kairomones in the water flea, Daphnia pulex. BMC Developmental Biology
10:45.
68
Nilsson, P. A., and C. Bronmark. 2000. Prey vulnerability to a gape-size limited predator:
behavioural and morphological impacts on northern pike piscivory. Oikos 88:539–546.
Norrdahl, K., and E. Korpimaki. 1998. Fear in farmlands: how much does predator avoidance
affect bird community structure? Journal of Avian Biology 29:79–85.
Ortells, R., J. Vanoverbeke, G. Louette, and L. De Meester. 2014. Colonization of Daphnia
magna in a newly created pond: founder effects and secondary immigrants. Hydrobiologia
723:167–179.
Östman, Ö., and J. M. Chase. 2007. Predator selectivity alters the effect of dispersal on
coexistence among apparent competitors. Oikos 116:387–394.
Pinel-Alloul, B., and E. A. Mimouni. 2013. Are cladoceran diversity and community structure
linked to spatial heterogeneity in urban landscapes and pond environments? Hydrobiologia
715:195–212.
Pintar, M. R., and W. J. Resetarits. 2017. Relative predation risk and risk of desiccation co-
determine oviposition preferences in Cope’s gray treefrog, Hyla chrysoscelis. Oecologia
184:423–430.
Power, M. E. 1984. Depth distributions of armored catfish: predator-induced resource
avoidance? Ecology 65:523–528.
Preisser, E. L., D. I. Bolnick, and M. F. Benard. 2005. Scared to death? The effects of
intimidation and consumption in predator-prey interactions. Ecology 86:501–509.
Resetarits, W. J. 2001. Colonization under threat of predation: avoidance of fish by an aquatic
beetle, Tropisternus lateralis (Coleoptera: Hydrophilidae). Oecologia 129:155–160.
Reznick, D. N., R. D. Bassar, J. Travis, and F. Helen Rodd. 2012. Life-history evolution in
guppies VIII: the demographics of density regulation in guppies (Poecilia reticulata).
69
Romare, P., and L. A. Hansson. 2003. A behavioral cascade: top-predator induced behavioral
shifts in planktivorous fish and zooplankton. Limnology and Oceanography 48:1956–1964.
Rowbottom, R., S. Carver, L. A. Barmuta, P. Weinstein, D. Foo, and G. R. Allen. 2015.
Resource limitation, controphic ostracod density and larval mosquito development. PLoS
ONE 10:1–14.
Ryabov, A. B., A. Morozov, and B. Blasius. 2015. Imperfect prey selectivity of predators
promotes biodiversity and irregularity in food webs. Ecology Letters 18:1262–1269.
Scheffer, M., G. J. Van Geest, K. Zimmer, E. Jeppesen, M. Søndergaard, M. G. Butler, M. A.
Hanson, S. Declerck, and L. De Meester. 2006. Small habitat size and isolation can promote
species richness: second-order effects on biodiversity in shallow lakes and ponds. Oikos
112:227–231.
Schmitz, O. J., A. P. Beckerman, and K. M. O’Brien. 1997. Behaviorally mediated trophic
cascades: effects of predation risk on food web interactions. Ecology 78:1388–1399.
Schmitz, O. J., V. Krivan, and O. Ovadia. 2004. Trophic cascades: the primacy of trait-mediated
indirect interactions. Ecology Letters 7:153–163.
Schneider, D. W., and T. M. Frost. 1996. Habitat duration and community structure in temporary
ponds. Journal of the North American Benthological Society 15:64–86.
Silberbush, A., and W. J. Resetarits. 2017. Mosquito female response to the presence of
larvivorous fish does not match threat to larvae. Ecological Entomology 42:595–600.
Silberbush, A., I. Tsurim, Y. Margalith, and L. Blaustein. 2014. Interactive effects of salinity and
a predator on mosquito oviposition and larval performance. Oecologia 175:565–75.
Stav, G., L. Blaustein, and J. Margalith. 1999. Experimental evidence for predation risk sensitive
oviposition by a mosquito, Culiseta longiareolata. Ecological Entomology 24:202–207.
70
Torres-Estrada, J. L., M. H. Rodríguez, L. Cruz-López, and J. I. Arredondo-Jimenez. 2001.
Selective oviposition by Aedes aegypti (Diptera: culicidae) in response to Mesocyclops
longisetus (Copepoda: Cyclopoidea) under laboratory and field conditions. Journal of
Medical Entomology 38:188–92.
Tranchida, M. C., M. V. Micieli, A. Macia, and J. J. Garcia. 2009. Native argentinean cyclopoids
(Crustacea: Copepoda) as predators of Aedes aegypti and Culex pipiens (Diptera: Culicidae)
mosquitoes. Revista de Biologia Tropical 57:1059–1068.
Turesson, H., and C. Brönmark. 2007. Predator-prey encounter rates in freshwater piscivores:
effects of prey density and water transparency. Oecologia 153:281–290.
Venter, O., E. W. Sanderson, A. Magrach, J. R. Allan, J. Beher, K. R. Jones, H. P. Possingham,
W. F. Laurance, P. Wood, B. M. Fekete, M. A. Levy, and J. E. M. Watson. 2016. Sixteen
years of change in the global terrestrial human footprint and implications for biodiversity
conservation. Nature Communications 7:12558.
Vonesh, J. R., and L. Blaustein. 2010. Predator-induced shifts in mosquito oviposition site
selection: a meta-analysis and implications for vector control. Israel Journal of Ecology &
Evolution 56:263–279.
Vonesh, J. R., J. M. Kraus, J. S. Rosenberg, M. Jonathan, J. Shoshana, and J. M. Chase. 2009.
Predator effects on aquatic community assembly: disentangling the roles of habitat selection
and post-colonization processes. Oikos 118:1219–1229.
Walsh, M. R., and D. N. Reznick. 2008. Interactions between the direct and indirect effects of
predators determine life history evolution in a killifish. Proceedings of the National
Academy of Sciences 105:594–599.
71
Webb, C., and M. Bashir. 2013. Managing mosquitoes in constructed freshwater wetlands. Pages
137–160 in S. Paul, editor. Workbook for managing urban wetlands in Australia. 1st ed.
Sydney Olympic Park Authority, Sydney, Australia.
Wellborn, G. A., D. K. Skelly, and E. E. Werner. 1996. Mechanisms creating community
structure across a freshwater habitat gradient. Annual Review of Ecology and Systematics
27:337–363.
Williams, P., M. Whitfield, J. Biggs, S. Bray, G. Fox, P. Nicolet, and D. Sear. 2003.
Comparative biodiversity of rivers, streams, ditches and ponds in an agricultural landscape
in Southern England. Biological Conservation 115:329–341.
Witt, A. M., and C. E. Cáceres. 2004. Potential predator-prey relationships between Bythotrephes
longimanus and Cercopagis pengoi in Southwestern Lake Michigan. Journal of Great Lakes
Research 30:519–527.
Zaret, T. M., and J. S. Suffern. 1976. Vertical migration in zooplankton as a predator avoidance
mechanism. Limnology and Oceanography 21:804–813.
72
CHAPTER 3: PREDATION DIFFERENTIALLY STRUCTURES IMMATURE MOSQUITO ASSEMBLAGES IN STORMWATER PONDS
ABSTRACT
Understanding the factors underlying the abundance and distribution of species requires the consideration of a complex suite of interacting biotic and abiotic factors operating on multiple spatial and temporal scales. Larval mosquitoes inhabiting small human-constructed ponds represent a unique opportunity to investigate the relative importance of these structuring mechanisms while simultaneously generating applied knowledge on mosquito control. We conducted a multi-year field survey of 32 stormwater ponds in Central Illinois (Champaign
County, IL, USA). From each pond, we collected data on pond structure type and hydroperiod, the presence/absence of cattails (Typha spp.), and measures of total nitrogen, phosphorus, and organic carbon, and chlorophyll a. We characterized the communities of crustacean zooplankton and aquatic insects and assigned these taxa into two main groups: predators and competitors of larval mosquitoes. Structural equation modeling was used to explore the direct and indirect effects of these biotic and abiotic factors on larval density for three species of culicine mosquitoes (Culex pipiens, Culex restuans, and Aedes vexans). Hydroperiod had an indirect negative effect on Cx. pipiens density. However, this effect was mediated by predator density; more permanent ponds had more predators, which therefore reduced the density of Cx. pipiens larvae. Ae. vexans density was positively correlated with predator density. We found no predictor variables that explained variation in Cx. restuans density. Herein, we show that the relative importance of these biotic and abiotic factors varies among species of culicine mosquitoes inhabiting stormwater ponds.
73
INTRODUCTION
Understanding the processes underlying the abundance and distribution of species is a central goal in community ecology (Gleason 1927, Clements 1936, Hutchinson 1961, MacArthur and Wilson 1967, Diamond 1975, Weiher and Keddy 2001). Decades of theoretical and empirical work have identified the importance of environmental filtering and biotic interactions as contemporary structuring mechanisms underlying species’ distributions (Menge and Olson
1990, Keddy 1992, Kraft et al. 2015). The effects of the abiotic environment may manifest prior to, or following, colonization of an organism in a particular habitat. Prior to colonization, the abiotic environment may facilitate or limit diversity by altering site-selection behavior (Bentley and Day 1989, Price 2010). Following colonization, physically unfavorable environments may limit the long-term persistence of susceptible individuals (i.e., habitat filtering; Jones 2001,
Maire et al. 2012). Furthermore, biotic interactions (e.g., competition and predation) may further constrain community structure, prior to, or following, colonization by directly or indirectly reducing, or eliminating, some species from a habitat (Lynch 1979, Inouye et al. 1980). While one or a few of these processes may emerge as particularly important in explaining distributional patterns in certain systems, populations and communities are the product of multiple simultaneously acting factors (Martin 2001, Resetarits et al. 2005, Agrawal et al. 2007,
HilleRisLambers et al. 2011, Weiher et al. 2011, Götzenberger et al. 2012). Further complicating our understanding of species distributions is the fact that closely related species may exhibit variation in their response to environmental factors, making the outcome of colonization processes difficult to predict in nature (Paine 1966, Tilman 1994, Dupré and Ehrlén 2002,
Arrington et al. 2005).
74
Larval mosquitoes inhabiting small stormwater ponds (i.e., retention ponds, detention ponds, and drainage ditches) provide a unique opportunity to examine the relative importance of multiple simultaneous structuring mechanisms on population dynamics and how these mechanisms differentially affect species. Stormwater ponds are an ideal system given their small size, discrete borders, simple and tractable species assemblages, and their ability to support mosquito larvae (Gingrich et al. 2006, Hassall and Anderson 2015). In most of the developed world, the creation of small and variable ponds resulting from stormwater management practices has increased the amount of suitable habitats available for mosquito breeding and larval development (Karpiscak et al. 2004, Tixier et al. 2011). The larval period is an important life stage in the mosquito life cycle (Ferguson et al. 2010), as it serves as a key link between the ecology of the pre-larval stages (i.e., adult oviposition behavior, embryonic development, and hatching success) and the production of pupal and adult life-stages, the latter of which can vector a multitude of human diseases (Keating et al. 2004, Carlson et al. 2004).
Structural equation modeling (SEM) has emerged as a powerful statistical analysis that allows for the simultaneous evaluation of the plausible relationships in multivariate datasets to generate causal inferences about observed patterns and mechanisms of interest (Arhonditsis et al.
2006, Joseph and Preston 2016). Based on hypothesized relationships established a priori, SEM uses a combination of confirmatory factor analysis (a form of factor analysis that assesses how well the actual data fit a pre-specified factor structure) and multiple regression to reproduce the covariance structure of the observed data to simultaneously assess competing hypotheses
(Ullman 2007). Similar to path analysis, SEM provides parameter estimates of the direct and indirect effects between observed variables. As a result, SEM is a powerful and flexible statistical approach that can allow ecologists to statistically disentangle the relative importance of
75 multiple direct and indirect effects on complex ecological patterns, such as community structure
(Alsterberg et al. 2013).
Variable selection
We used SEM to test two hypotheses that relate the abundance of mosquito larvae to biotic and abiotic predictor variables in small stormwater ponds. Previous work in this system has revealed the importance of both biotic interactions and the abiotic environment in shaping the abundance and distribution of mosquito larvae (Juliano 1998, Carver et al. 2010, Coon et al.
2016, Bohenek et al. 2017). Larval mosquitoes coexist with a wide variety of other aquatic plant and animal species, including crustacean zooplankton, that can impact mosquito population dynamics in these ponds (Griswold and Lounibos 2005, Pinel-Alloul and Mimouni 2013).
Specifically, predation and competition by zooplankton have been identified as important factors regulating mosquito populations in field and laboratory studies (Knight et al. 2004, Blaustein and
Chase 2007, Vonesh and Blaustein 2010, Duquesne et al. 2011). Given their dynamics are intimately linked to that of their food resources (e.g., bacteria and algae), mosquito larvae may be indirectly influenced by those abiotic factors (e.g., inorganic nutrient availability, hydroperiod, etc.) that drive the quality, abundance, and accessibility of these resources (Merritt et al. 1992). The organic matter produced by macrophytic vegetation commonly found in these stormwater ponds (e.g., cattails) may also stimulate mosquito population growth (Walton and
Jiannino 2005, Mackay et al. 2016), as this matter serves as an important food resource for mosquito larvae. SEM provides a suitable framework to address and disentangle the relative importance of these factors from field-collected data, a task difficult to accomplish
76 experimentally given the abundance of and potential complex interactions among these multiple structuring factors.
The aims of this study were to 1) determine whether measurable direct and indirect effects of the larval rearing environment could explain larval culicine abundance in a set of 32 stormwater ponds and 2) determine how these effects differ between three abundant taxa. We tested the following hypotheses:
1) Increased hydroperiod (i.e., retention ponds and permanent drainage ditches)
would foster greater densities of predators and competitors, resulting in lower
densities of larval mosquitoes.
2) Increased levels of inorganic nutrients (i.e., total nitrogen (TN) and total
phosphorus (TP)) and food resources added through the decomposition of organic
matter from cattails (Typha spp.) would have a positive indirect effect on larval
abundance by increasing the abundance of algal (chlorophyll a) and bacterial food
resources (total organic carbon as a proxy).
We tested these hypotheses in three different mosquito species (Culex pipiens, Culex restuans, and Aedes vexans) and expected that species would vary based on differences in ecology and dispersal behavior (Vonesh and Blaustein 2010).
MATERIALS AND METHODS
Field survey
We surveyed 32 stormwater ponds in Champaign County, IL, USA during the summers
(May – August) of 2014 – 2017. For the purpose of statistical analyses, ponds were categorized according to their design for stormwater management: retention ponds, ditches, and detention
77 ponds (Wanielista and Yousef 1993, National Research Council 2008). Retention ponds are designed to collect and retain stormwater runoff (i.e., high permanence), detention ponds to collect and temporarily store stormwater, and drainage ditches to transport stormwater (can be designed to be either temporary or permanent). Differences in these designs resulted in variation in overall size and pond permanence (i.e., hydroperiod), which we captured through a weekly hydroperiod survey from May – August of 2016 and 2017. A single hydroperiod value (ranging from 0 = always empty to 1 = permanent) was calculated for each pond by averaging the number of times a pond was empty (0) or with water (1) across the 26 observation dates in the hydroperiod survey.
Invertebrate communities were sampled every two weeks throughout the summers (May
– August) of 2014 – 2017; in total we had 23 sampling periods across all four years. To characterize communities, 3 L of pond water was collected haphazardly from the edge of each pond using a standard 350 ml mosquito dipper, filtered through a 70 µm sieve, and all animals were preserved in 95% EtOH. Samples were only collected if each pond contained at least 3 L of water. Individuals were identified to the lowest taxonomic level possible using Merritt et al.
(1996), Darsie and Ward (2005), and Haney (2013). Each sample was scanned for rare taxa; taxa with fewer than 300 individuals were counted completely, and those with over 300 individuals were subsampled (a minimum of three, 2 ml subsamples following whole sample dilution to 100 ml). For those that were subsampled, whole sample abundance for each taxa was estimated by N
= 100 * (n1+n2+n3)/6 where n1, n2, n3 are the three subsampled abundances. Due to the ephemeral nature of these ponds, we were unable to collect samples from all ponds on all sample dates; in total we collected 280 samples over the four-year period.
78
The three mosquito species that were most abundant during the field survey were selected for use as our single response variables in the three separate structural equation models (Aedes vexans, Culex pipiens, and Culex restuans). Non-mosquito taxa were divided into two groups, predators and competitors, based on the taxon’s previously documented, or supposed, relationships to larval mosquitoes by the authors (Table A1; Baldwin et al. 1955, Quiroz-
Martinez et al. 2007, Shaalan and Canyon 2009, Duquesne et al. 2011). The presence or absence of cattails (Typha spp.) was used in the SEM as cattails were the dominant emergent macrophytes in the ponds during the field survey. Using a visual survey during one sampling period in July 2016, the presence or absence of cattails (Typha spp.) was recorded for all 32 stormwater ponds. Visual surveys were conducted alongside bi-weekly community sampling in
2016 and 2017, and we found no changes to these initial observations.
Total nitrogen (TN), phosphorus (TP), and organic carbon (TOC) measurements were collected from bi-weekly water column samples for 11 sampling periods across 2015 and 2017.
Measures of TOC, which served as a proxy for bacterial biomass, and TN, an important nutrient source for the growth of phytoplankton, were obtained for each pond using a TOC/TN-VCPH analyzer (Shimadzu Corporation, Kyoto, Japan). TP measurements, another important nutrient source for plankton, were collected using the molybdate, ascorbic acid, and antimony method outlined in Wetzel and Likens (2000). Chlorophyll a concentrations (Chl a) were obtained by first filtering a known volume of pond water through Whatman GF/F glass fiber filters, extracting the chlorophyll using 100% ethanol for 24 hours, and measuring fluorescence on a
Turner Biosystems Trilogy fluorometer (Turner Biosystems, Sunnyvale, CA, USA; Sartory and
Grobbelaar 1984). For each environmental variable above, measurements were averaged across
79 all sampling periods to produce a single value for each pond (Table 3.1). We used mean values for environmental variables in which missing data were encountered.
Data analysis
Structural equation modeling (SEM) is a statistical method closely related to regression analysis but is superior for complex datasets as SEM tests the relationships between all variables simultaneously (Bollen 2005, Grace 2006). Because this approach does not allow for us to examine temporal relationships among these variables, temporal samples for each variable (e.g.,
TN) were collapsed (using methods in “Field survey”) into a single measurement for each pond for use in the SEM. We used SEM to evaluate the direct and indirect effects of the rearing environment on larval populations of Ae. vexans, Cx. pipiens, and Cx. restuans mosquitoes inhabiting stormwater ponds. The causal network structure illustrated in Figure 3.1 denotes the hypothesized relationships between variables influencing larval mosquito density. This structure was specified based on previous relationships documented in the literature (for justification of variable selection and the proposed structure, please see “variable selection”). This structure was determined a priori and was fixed across analyses for all three mosquito species (Cx. pipiens, Cx. restuans, and Ae. vexans). In the model, we included three exogenous predictor variables
(structure type [retention pond, detention pond, or drainage ditch], TN [mean], and TP [mean]), six measured endogenous variables (hydroperiod, cattail [presence or absence], chlorophyll a
[mean], total organic carbon [mean], predator density, competitor density, and a single response variable (mosquito density)). Natural-log transformations were sufficient to linearize the relationships between taxa count data for analyses in the SEM. To this end, predator, competitor, and mosquito densities were summed across all sample dates for each pond and the totals were
80 natural log-transformed. Three separate SEM’s were conducted, one for each species of mosquito. The three SEM analyses were conducted using R package lavaan (version 0.5-23:
Rosseel 2012). We tested the adequacy of these models to fit the data using maximum likelihood estimation and χ2 tests as outlined in (Grace 2006). In SEM analysis, a good model fit is indicated by a non-significant χ2 parameter estimate, which demonstrates good quantitative agreement between observed and model covariance matrices. For all three species, we found that our conceptual model adequately described the data, which allowed us to draw conclusions about the nature of these modeled interactions (Cx. pipiens: χ2 = 27.8, df = 22, p = 0.18, Cx. restuans:
χ2 = 28.9, df = 22, p = 0.15, Ae. vexans: χ2 = 27.9, df = 22, p = 0.18).
RESULTS
We found that relatively few factors explained patterns of larval mosquito density in the field across the wide range of biotic and abiotic variables examined in our study and that these factors differed among mosquito species. Of the 20 paths present in the model, we found five significant interactions across two models (Culex pipiens and Aedes vexans) and four significant interactions across the third (Culex restuans; Fig. 3.1). Furthermore, the importance of these factors differed for each species of mosquito. Our models explained a significant percentage of the variance in larval mosquito density (R2 = 26%, 35%, and 43% of the variance in Cx. pipiens,
Cx. restuans, and Ae. vexans density, respectively). Of the 10 variables examined, predator density was the sole direct predictor of Cx. pipiens and Ae. vexans densities, whereas Cx. restuans had no significant predictor variables (Fig. 3.1). We found a negative and indirect effect of hydroperiod on Cx. pipiens density; however, this was driven by an increase in predator density with increasing hydroperiod (z = -1.96, p = 0.05; Figs. 3.1A and 3.2A). This pattern was
81 reversed for Ae. vexans, with larvae being more abundant in ponds with high densities of predators (z = 2.05, p = 0.04; Figs. 3.1C and 3.2C). However, it is important to note that this result was primarily driven by three ponds with high densities of both predators and Ae. vexans larvae (Fig. 3.2B. We found no effect of predator density on Cx. restuans (z = 0.50, p = 0.62;
Figs. 3.1B and 3.2B).
Across all three SEM’s, mosquito predator and competitor densities were positively correlated with hydroperiod (predator: z = 2.35, p = 0.02; competitor: z = 3.7, p < 0.001; Figs.
3.1 and 3.3). Chlorophyll a (Chl a), which served as our proxy for algal resources, was positively correlated with competitor density (z = 4.06, p < 0.001; Fig. 3.1). Additionally, we found a positive correlation between total organic carbon (TOC), which served as our proxy for bacterial resources, and total phosphorus (TP; z = 5.60, p < 0.001; Fig. 3.1). The lack of correlation between TOC and Chl a further highlights the suitability of TOC as a proxy for bacterial abundance; TOC was not driven by the abundance of phytoplankton (Fig. 3.3).
DISCUSSION
Our findings highlight the importance of hydroperiod and predation in predicting the density of larval mosquitoes inhabiting stormwater ponds. Furthermore, we show how the importance of these factors can vary significantly among mosquito species. For Culex pipiens, a longer pond hydroperiod indirectly drove a decrease in larval density. This pattern was driven by more permanent ponds having, on average, higher predator densities than temporary ponds.
Though we found a significant effect of predator density on Aedes vexans density, it was in the opposite direction of what we expected (Ae. vexans were positively correlated with predator density). Culex restuans density was predicted neither by predator density nor hydroperiod.
82
Hydroperiod has been shown to be an important factor controlling larval mosquito populations (Pritchard and Scholefield 1983, Schäfer and Lundström 2006, Ellis et al. 2006), as well as other invertebrate and vertebrate populations inhabiting small aquatic habitats (Chase
2007, McCauley et al. 2008, Anderson et al. 2015). However, we have shown this effect to be primarily mediated by an increase in predator density, a finding replicated in other studies
(Batzer 1996, Resetarits 1996, Chase and Knight 2003, Rubbo et al. 2011; but see Westby and
Juliano 2017). A longer hydroperiod can facilitate greater abundances and a more diverse assemblage of both predators and competitors (including crustaceans, insects, and vertebrates) in ponds (McCauley et al. 2008), with both potentially suppressing larval mosquito and other prey populations. Though we partially support this claim through our findings that hydroperiod had a negative indirect effect on Cx. pipiens density (i.e., more permanent ponds had fewer larvae), results from our study show that mosquito species can vary in this pattern as we found no effect of either hydroperiod or predator density on Cx. restuans density, and a positive association between predator and Ae. vexans densities. Westby and Juliano (2017) found that the effect of hydroperiod on mosquito larval abundance was not mediated by predator abundance. However,
Westby and Juliano found no difference in predator abundance across hydroperiod treatments, a difference we did observe in our system. Furthermore, Westby and Juliano only examined a single predator (Toxorhynchites rutilus); previous work has widely documented interspecific variation in predator behavior. As a result, the identity of the predator may affect the outcome of these interactions (Carver et al. 2010, Silberbush and Resetarits 2017).
The direct and indirect effects of predation have been widely documented for multiple mosquito species (Vonesh and Blaustein 2010). The consumption of larvae by insect and crustacean predators can significantly suppress mosquito populations, especially in small habitats
83
(Calliari et al. 2003, Mandal et al. 2008). However, in our study, this effect was inconsistent across mosquito species. Previous work in this system has shown that predators exhibit selectivity behavior; the copepod predator Acanthocyclops spp. preferentially grazed on both
Culex and Aedes larvae in the presence of alternative crustacean prey (Chapter 2). It is possible that the patterns we observed in our field survey were driven by either selectivity of the predator(s) present in our system, or by different mosquito species varying in their anti-predator behaviors (Roberts 2014). Prey may avoid colonizing habitats that contain predators (Stav et al.
1999, Resetarits 2001, Angelon and Petranka 2002, Vonesh and Blaustein 2010). Vonesh and
Blaustein (2010) showed extensive interspecific variation in this anti-predator colonization behavior in ovipositing mosquitoes. We suspect pre-colonization behavior is playing a major role in the patterns observed in our study; in previous work, we found that Cx. pipiens but not Cx. restuans or Aedes spp. avoided ovipositing in habitats that contained zooplankton predators and competitors, as well as their chemical cues (Chapter 2).
The cost of failing to avoid predators can be high for ovipositing mosquitoes; we previously documented that predation rates by a single Acanthocyclops spp. individual can exceed 11 mosquito larvae per day (Chapter 2; 6.1 individuals per day in Calliari et al. 2003).
Given the ubiquity and abundance of this predator in the landscape, larval consumption may have also driven this negative relationship between Cx. pipiens and predator densities.
Alternative behaviors that allow larvae to avoid predators following colonization may have allowed Cx. restuans and Ae. vexans to coexist with predators in these habitats, though there is limited evidence for this type of behavior in Ae. vexans. In Cx. restuans, Ferrari et al. (2008) observed that larvae reduced activity in the presence of conspecific alarm cues (i.e., chemical cues exuded by larval mosquitoes exposed to salamander predators). Furthermore, the positive
84 association between Ae. vexans and predator densities may also be an artifact of both being positively associated with an unmeasured factor in our study. For example, others have shown that Aedes spp. can be associated with multiple variables that we did not measure in our system
(e.g., canopy coverage and composition, surface area, water turbidity, water depth, and substrate type; Beier et al. 1983, Li et al. 2014). Although we suggest that predation plays an important, albeit differential, role in determining the abundance of the three species examined, we cannot rule out the potential effects of unmeasured environmental variables on the observed field associations.
We found no correlation between resource abundance (i.e., algal and bacterial) and mosquito density for any of the three species examined and conclude that resources are likely not limiting mosquito populations in these ponds. Stormwater ponds can be nutrient rich compared to other aquatic systems (Wu et al. 2006). While we found a wide range in the concentration of
TP and TN, average values for each were high and all were above the minimum value to support the growth of freshwater algae (Table 3.1; Grover 1989, Correll 1998) and also well exceed EPA guidelines for lake and reservoir waters (Cullum et al. 2006). Additional nutrient enrichment may have occurred following decomposition of aquatic macrophytes (Berkelhamer and Bradley
1989). In a study of Culex mosquitoes in stormwater ponds, MacKay et al. (2016) found that plant clippings deposited following mowing resulted in an increase in larval abundance. While management may have varied among ponds, no macrophytes were mowed or removed across all four years of study. Our findings suggest that larval mosquito populations are likely not limited by the availability of algae and bacteria in the stormwater ponds presented herein.
The biotic and environmental variables selected for use in the SEM have been previously documented for their importance in determining the abundance and distribution of mosquito
85 larvae (Khawaled et al. 1989, Urabe et al. 2002, Bohonak and Jenkins 2003, Hillebrand 2005,
Duquesne et al. 2011, Schriever 2015). However, variables that we did not include in the model may have had important consequences for these dynamics as well. Other unexamined factors that may be important in predicting mosquito dynamics include, canopy coverage and composition, surrounding and submerged vegetation, water turbidity, pond shape and surface area, water depth, and substrate type (Beier et al. 1983, Li et al. 2014, Gardner et al. 2015). For example, surrounding and submerged vegetation (e.g., duckweed, hyacinth, black willow, invasive shrubs) may reduce or enhance habitat attractiveness to graving female mosquitoes and its ability to support larval survival and development (Angerilli 1980, Ameen et al. 1999, Gardner et al.
2015). Furthermore, pond surface area may also affect the abundance and distribution of mosquitoes in the landscape, as well as their predators and competitors (though, the effect of surface area may have been partially captured by structure type in the SEM [e.g., retention ponds were always larger than drainage and detention ponds]; Fischer and Schweigmann 2004, Frisch et al. 2012). Specifically, larger ponds may facilitate higher species richness and overall abundance by increasing colonization by both active and passively dispersing invertebrates
(King et al. 1996, Rundle et al. 2002, Vanschoenwinkel et al. 2009, Frisch et al. 2012). Because these small aquatic systems are influenced by a multitude of factors operating at various spatial and temporal scales, we encourage further investigation of potential factors that may explain this inter-specific variation in mosquito dynamics.
Our study did not examine the spatio-temporal dynamics of these interactions, but previous research has shown the importance of historical processes and context in shaping the abundance and distribution of organisms in nature (Holt 1993, Kardol et al. 2007, Fukami 2015).
We used an averaging-snapshot approach (e.g., multiple dates were averaged to create a single
86 value for each variable for each pond) to gain insights as to how, more generally, the complex interplay between the biotic and abiotic environment drives larval mosquito populations in the field; ponds were sampled on multiple dates but we collapsed this temporal variation by averaging (or summing) each variable over time. This is a method commonly employed by ecological studies that use SEM, as these analyses are data intensive and may be significantly impacted by missing data (Laughlin 2011, Prugh and Brashares 2012). However, we acknowledge the potential effects of historical contingency on the observed patterns in the field survey and encourage future empirical studies to consider temporal processes when seeking to disentangle these various structuring mechanisms.
Stormwater ponds present a unique opportunity to study the ecological processes underlying the dynamics of populations and communities in semi-natural systems. The complex interplay between the biotic and abiotic environment processes structuring natural populations and communities has been shown through extensive empirical and theoretical work in a variety of systems (Holyoak et al. 2005, Agrawal et al. 2007, Weiher et al. 2011, Mittelbach and
Schemske 2015). Previous work in small aquatic habitats have provided mixed results regarding the relative importance of select biotic and abiotic processes on dynamics of focal populations and whole communities (Schneider and Frost 1996, King et al. 1996, Westby and Juliano 2017).
Using structural equation modeling, found that the three species examined were differentially shaped by the various ecological factors examined. While our results have provided valuable insights to the biotic and abiotic processes structuring three species of larval mosquito populations in stormwater ponds, the field of vector ecology would benefit from further studies on the complex multivariate relationships between populations of larval mosquitoes and their
87 complex and dynamic environments across a wider range of natural and artificial aquatic habitats
(e.g., tree holes, containers, catch basins, natural ponds, etc.).
ACKNOWLEDGEMENTS
We first thank Andrew J. Mackay who provided us with study sites that were used in the field survey. We thank Gevan Behnke for conducting the total organic carbon and total nitrogen measurements. Yuan Yubai provided invaluable assistance with the structural equation modeling results. Donald R. Schoolmaster Jr. provided useful R functions and codes to execute SEM analyses and visualize SEM diagrams. We thank our undergraduate research assistants, Cameron
Schwing, Sana Khadri, and Xorla Ocloo, as well as technicians Ilona Menel and Ping Lee for their field and laboratory assistance on this project. We thank Jessica R. Holmes who kindly provided comments on multiple drafts of this manuscript. This research was funded by the
United States National Science Foundation [DEB-1754115], a grant through the University of
Illinois Institute for Sustainability, Energy, and Environment, and through grants from the
Department of Animal Biology and School of Integrative Biology at the University of Illinois at
Urbana-Champaign.
88
TABLE AND FIGURES
Coordinates: Latitude (DMS) - Hydroperiod Total nitrogen Total phosphorus Total organic Chlorophyll a Structure Cattails Longitude (DMS) (0 - 1) (mg L-1) (mg L-1) carbon (mg L-1) (µg L-1) 40°7'52.1" N - 88°14'22.8" W Detention Pond 0.08 0.76 0.5 13 0.59 Absent 40°7'50.3" N - 88°14'26.9" W Detention Pond 0.08 0.72 0.31 10.13 1.54 Absent 40°7'58.6" N - 88°15'1.1" W Detention Pond 0.08 0.56 0.05 9.34 9.53* Present 40°5'47.6" N - 88°13'59.0" W Detention Pond 0.15 1.39 0.2 16.51 5.83 Present 40°5'45.7" N - 88°13'58.4" W Detention Pond 0.31 2.12 (1.58 - 2.66) 1.3 14.09 (7.62 - 20.57) 9.53* Present 40°8'23.0"N - 88°11'36.7" W Detention Pond 0.46 1.32 (1.25 - 1.4) 0.36 (0.11 - 0.52) 17.07 (16.4 - 17.92) 7.37 (2.64 - 12.09) Present 40°7'53.6" N - 88°9'46.8" W Drainage Ditch 0.46 1.38* 0.35* 10.22* 9.53* Absent 40°4'59.8" N - 88°13'56.6" W Drainage Ditch 0.67 2.28 (0.51 - 4.18) 0.36 (0.19 - 0.46) 5.83 (4.53 - 7.7) 3.63 (1.88 - 6.29) Present 40°7'7.1" N - 88°16'57.1" W Drainage Ditch 0.77 0.6 (0.57 - 0.63) 0.37 6.77 (5.34 - 8.19) 3 Present 40°5'39.3" N - 88°14'29.9" W Detention Pond 0.83 0.53 (0.46 - 0.6) 0.07 8.66 (5.26 - 12.05) 9.53* Present 40°6'48.4" N - 88°9'56.2" W Drainage Ditch 0.85 0.87 (0.39 - 1.34) 0.39 (0.13 - 0.78) 9.3 (4.01 - 13.16) 1.65 Present 40°5'3.6" N - 88°13'8.8" W Drainage Ditch 0.85 1.92 (0.94 - 5.08) 0.77 (0.25 - 1) 10.43 (6.65 - 18.74) 9.6 (9.16 - 10.03) Absent 40°7'15.4" N - 88°18'25.5" W Detention Pond 0.92 0.51 0.35* 18.26 9.53* Present 40°6'48.2" N - 88°19'0.3" W Drainage Ditch 0.92 0.62 (0.42 - 0.72) 0.26 (0.09 - 0.56) 7.22 (4.26 - 9.23) 17.34 (9.42 - 25.27) Present 40°6'56.0" N - 88°11'3.1" W Drainage Ditch 0.92 0.55 (0.33 - 1.11) 0.18 (0.04 - 0.44) 7.82 (3.52 - 11.48) 25.27 Present 40°6'55.6" N - 88°9'45.9" W Drainage Ditch 0.92 2.29 (1.01 - 5.47) 0.97 (0.48 - 2.02) 17.05 (14.87 - 19.84) 6.07 (1.83 - 10.3) Absent 40°6'32.8" N - 88°10'39.2" W Retention Pond 1 3.23 (0.45 - 9.89) 1.08 (0.18 - 2.21) 28.09 (8.39 - 94.59) 18.75 (2.96 - 34.55) Present 40°4'6.4" N - 88°18'52.3" W Retention Pond 1 0.64 (0.35 - 1.07) 0.07 (0.04 - 0.1) 7.86 (5.39 - 10.72) 9.53* Absent 40°6'45.9" N - 88°9'57.8" W Detention Pond 1 1.01 (0.5 - 1.87) 0.08 (0.02 - 0.18) 5.64 (3.19 - 8.61) 1.12 (0.17 - 1.84) Present 40°6'47.4" N - 88°9'57.2" W Drainage Ditch 1 1.62 (0.41 - 2.84) 0.34 (0.09 - 0.7) 5.75 (3.27 - 9.58) 32.72 (0.41 - 95.65) Present 40°6'46.9" N - 88°9'39.2" W Drainage Ditch 1 0.89 (0.59 - 1.33) 0.46 (0.08 - 0.9) 5.76 (4.47 - 6.97) 19.05 (5.8 - 34.83) Present 40°4'52.7" N - 88°11'18.0" W Retention Pond 1 1.5 (0.67 - 4.92) 0.34 (0.03 - 0.78) 9.49 (6.18 - 13.25) 8.28 (2.95 - 13.61) Absent 40°7'53.2" N - 88°17'9.5" W Drainage Ditch 1 0.89 (0.45 - 1.63) 0.11 (0.05 - 0.17) 6.38 (3.14 - 7.64) 5.25 Absent 40°8'0.8" N - 88°17'14.9" W Retention Pond 1 0.68 (0.42 - 1.34) 0.07 (0.05 - 0.1) 6.42 (4.01 - 8.96) 22.84 (18.61 - 27.07) Absent 40°7'31.0" N - 88°14'59.6" W Detention Pond 1 2.52 (1.97 - 3.28) 0.37 (0.29 - 0.45) 16.83 (7.43 - 22.8) 7.81 Absent 40°5'48.3" N - 88°13'54.9" W Detention Pond 1 1.52 (0.82 - 2.92) 0.31 (0.05 - 0.84) 9.8 (4.79 - 30.78) 0.39 (0.38 - 0.4) Absent 40°7'58.7" N - 88°15'3.1" W Detention Pond 1 0.9 (0.65 - 1.58) 0.1 (0.06 - 0.13) 6.4 (5.29 - 7.52) 9.53* Present 40°4'21.0" N - 88°18'16.3" W Detention Pond 1 4.36 (0.84 - 8.61) 0.17 (0.05 - 0.38) 5.56 (1.32 - 8.54) 9.53* Absent 40°6'48.7" N - 88°9'47.9" W Drainage Ditch 1 1.03 (0.61 - 1.44) 0.33 (0.11 - 0.94) 6.93 (4.96 - 9.08) 9.53* Present 40°8'0.6" N - 88°6'51.0" W Drainage Ditch 1 1.49 (0.44 - 2.56) 0.12 (0.03 - 0.37) 5.59 (3.14 - 7.54) 4.91 Present 40°4'55.9" N - 88°11'12.5" W Retention Pond 1 1.58 (0.89 - 2.61) 0.29 (0.18 - 0.41) 10.48 (6.06 - 15.56) 9.53* Absent 40°4'58.4" N - 88°13'57.3" W Drainage Ditch 1 2 (0.44 - 4.22) 0.33 (0.14 - 0.56) 8.25 (4.95 - 15.66) 6.65 (1.78 - 13.04) Present
Table 3.1: Coordinates and environmental data are shown for each pond. Data ranges are reported alongside mean values for total nitrogen, total phosphorus, total organic carbon, and chlorophyll a. Missing values were imputed using column averages and are designated with an asterisk (*).
89
Figure 3.1: Structural equation models used to assess the direct and indirect effects of pond environment on larval mosquito populations. Three separate models were run for each species of
90
Figure 3.1 (cont.) mosquito (A. Culex pipiens, B. Culex restuans, C. Aedes vexans). The model uses pond structure type [categorical], total nitrogen (TN) [mean], and total phosphorus (TP) [mean], as the three exogenous predictor variables and 7 measured endogenous variables (hydroperiod, cattail
[presence or absence], chlorophyll a [mean], total organic carbon [mean], predator density, competitor density. Numbers on bolded arrows are standardized path coefficients and are only shown for significant paths (P < 0.05).
91
A B 6 6
● 5 5 ● 4 4 ● 3 3
● ● 2 2 ● ● ● ● ● ● ● ● ● ● Mosquito density (ln ind / L) 1 Mosquito density (ln ind / L) 1 ● ● ●● ● ● ●
● ● ● ● ● 0 0 ● ● ●● ● ● ● ●●●● ● ● ● ● ● ●● ● ● ● ●●●● ●●● ●● ●● C0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 6 Predator density (ln ind / L) Predator density (ln ind / L) ● 5 4 3 ●
● 2 Mosquito density (ln ind / L) 1 ● ● ● ● ●
0 ● ● ● ● ● ● ● ● ●●● ●●●● ●●● ●●
0 1 2 3 4 5 6 7
Predator density (ln ind / L)
Figure 3.2: Mean mosquito density (ind / L) plotted against mean predator density (ind / L) for
A. Culex pipiens, B. Culex restuans, and C. Aedes vexans for each pond. All variables were natural log-transformed. SEM predicted a significant negative relationship for Cx. pipiens, and a significant positive relationship for Ae. vexans. However, the latter result was primarily driven by three ponds having high densities of both predators and mosquitoes.
92
restuans Culex.pipiens Culex pipiens
CulexCulex.pipiens pipiens Culex.restuans Culex
CulexCulex.restuans restuans Aedes vexans Aedes.vexans
AedesAedes.vexans vexans Competitors
Competitors Predators
Predators * * Structure
Structure TN
TN TP
TP Cattails Cat.Tails
Cat.TailsCattails TOC TOC * Hydroperiod Hydroperiod * * ChlA ChlA *
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
Figure 3.3. Bivariate correlation heat map between variables included in all three structural equation models. Blue and white indicate a negative or no correlation; red indicate a positive correlation. Shown are the correlations between Culex pipiens density (natural log-transformed),
Culex restuans density (natural log-transformed), Aedes vexans density (natural log- transformed), mosquito competitor density (natural log-transformed), mosquito predator density
(natural log-transformed), structure (retention pond, detention pond, or drainage ditch), total
93
Figure 3.3 (cont.) nitrogen (TN), total phosphorus (TP), cattails (presence/absence), total organic carbon (TOC; proxy for bacterial resources), hydroperiod (ranging from 0 [never containing water] to 1 [always containing water]), and chlorophyll a (ChlA; proxy for algal resources). Significant correlations from SEM analyses are boxed and indicated with an asterisk (*).
94
REFERENCES
Agrawal, A. A., D. D. Ackerly, F. Adler, A. E. Arnold, C. Cáceres, D. F. Doak, E. Post, P. J.
Hudson, J. Maron, K. A. Mooney, M. Power, D. Schemske, J. Stachowicz, S. Strauss, M. G.
Turner, and E. Werner. 2007. Filling key gaps in population and community ecology.
Frontiers in Ecology and the Environment 5:145–152.
Alsterberg, C., J. S. Eklöf, L. Gamfeldt, J. N. Havenhand, and K. Sundbäck. 2013. Consumers
mediate the effects of experimental ocean acidification and warming on primary producers.
Proceedings of the National Academy of Sciences of the United States of America
110:8603–8.
Ameen, M., M. I. Hossain, and M. A. Chowdhury. 1999. Integrated mosquito management in
Dhaka city: promising non-chemical components. Pages 447–456 Proceedings of the 3rd
International Conference on Urban Pests.
Anderson, T. L., B. H. Ousterhout, W. E. Peterman, D. L. Drake, and R. D. Semlitsch. 2015. Life
history differences influence the impacts of drought on two pond-breeding salamanders.
Ecological Applications 25:1896–1910.
Angelon, K. A., and J. W. Petranka. 2002. Chemicals of predatory mosquitofish (Gambusia
affinis) influence selection of oviposition site by Culex mosquitoes. Journal of Chemical
Ecology 28:797–806.
Angerilli, N. P. D. 1980. Influences of extracts of freshwater vegetation on the survival and
oviposition by Aedes aegypti (Diptera: Culicidae). The Canadian Entomologist 112:1249–
1252.
95
Arhonditsis, G. B., C. A. Stow, L. J. Steinberg, M. A. Kenney, R. C. Lathrop, S. J. McBride, and
K. H. Reckhow. 2006. Exploring ecological patterns with structural equation modeling and
Bayesian analysis. Ecological Modelling 192:385–409.
Arrington, D. A., K. O. Winemiller, and C. A. Layman. 2005. Community assembly at the patch
scale in a species rich tropical river. Oecologia 144:157–167.
Baldwin, W. F., H. G. James, and H. E. Welch. 1955. A study of predators of mosquito larvae
and pupae with a radio-active tracer. The Canadian Entomologist 87:350–356.
Batzer, D. P. 1996. Ecology of insect communities in nontidal wetlands. Annual Review of
Entomology 41:75–100.
Beier, J. C., C. Patricoski, M. Travis, and J. Kranzfelder. 1983. Influence of water chemical and
environmental parameters on larval mosquito dynamics in tires. Environmental Entomology
12:434–438.
Bentley, M. D., and J. F. Day. 1989. Chemical ecology and behavioral aspects of mosquito
oviposition. Annual Review of Entomology 34:401–421.
Berkelhamer, R. C., and T. J. Bradley. 1989. Mosquito larval development in container habitats:
the role of rotting Scirpus californicus. J Am Mosq Control Assoc 5:258–260.
Blaustein, L., and J. M. Chase. 2007. Interactions between mosquito larvae and species that share
the same trophic level. Annual Review of Entomology 52:489–507.
Blaustein, L., and J. Margalit. 1994. Differential vulnerability among mosquito species to
predation by the cyclopoid copepod, Acanthocyclops virdis. Israel Journal of Zoology
40:55–60.
96
Bohenek, J. R., M. R. Pintar, T. M. Breech, and W. J. Resetarits. 2017. Patch size influences
perceived patch quality for colonising Culex mosquitoes. Freshwater Biology 62:1614–
1622.
Bohonak, A. J., and D. G. Jenkins. 2003. Ecological and evolutionary significance of dispersal
by freshwater invertebrates. Ecology Letters 6:783–796.
Bollen, K. A. 2005. Structural Equation Models. Pages 5269–5278 in P. Armitage and T. Colton,
editors. Encyclopedia of Biostatistics. 2nd ed. Wiley, New York, NY, USA.
Calliari, D., K. Sanz, M. Martínez, G. Cervetto, M. Gómez, and C. Basso. 2003. Comparison of
the predation rate of freshwater cyclopoid copepod species on larvae of the mosquito Culex
pipiens. Medical and Veterinary Entomology 17:339–342.
Carlson, J., J. Keating, C. M. Mbogo, S. Kahindi, and J. C. Beier. 2004. Ecological limitations on
aquatic mosquito predator colonization in the urban environment. Journal of Vector
Ecology 29:331–339.
Carver, S., H. Spafford, A. Storey, and P. Weinstein. 2010. The roles of predators, competitors,
and secondary salinization in structuring mosquito (Diptera: Culicidae) assemblages in
ephemeral water bodies of the wheatbelt of Western Australia. Environmental Entomology
39:798–810.
Chase, J. M. 2007. Drought mediates the importance of stochastic community assembly.
Proceedings of the National Academy of Sciences 104:17430–17434.
Chase, J. M., and T. M. Knight. 2003. Drought-induced mosquito outbreaks in wetlands.
Ecology Letters 6:1017–1024.
Clements, F. E. 1936. Nature and structure of the climax. Journal of Ecology 24:252–284.
97
Coon, K. L., M. R. Brown, and M. R. Strand. 2016. Mosquitoes host communities of bacteria
that are essential for development but vary greatly between local habitats. Molecular
Ecology 25:5806–5826.
Correll, D. L. 1998. The role of phosphorus in the eutrophication of receiving waters: a review.
Journal of Environment Quality 27:261–266.
Cullum, R. F., S. S. Knight, C. M. Cooper, and S. Smith. 2006. Combined effects of best
management practices on water quality in oxbow lakes from agricultural watersheds. Soil
and Tillage Research 90:212–221.
Darsie, R. J., and R. Ward. 2005. Identification and geographical distribution of the mosquitoes
of North America, north of Mexico. Mosquito Systematics. 2nd edition. University Press of
Florida, Gainsville, FL, USA.
Diamond, J. M. 1975. Assembly of species communities. (M. L. C. and J. M. Diamond, Ed.)
Ecology and Evolution of Communities. Harvard University Press, Cambridge,
Massachusetts, USA.
Dupré, C., and J. Ehrlén. 2002. Habitat configuration, species traits and plant distributions.
Journal of Ecology 90:796–805.
Duquesne, S., I. Kroeger, M. Kutyniok, and M. Liess. 2011. The potential of cladocerans as
controphic competitors of the mosquito Culex pipiens. Journal of Medical Entomology
48:554–560.
Ellis, A. M., L. P. Lounibos, and M. Holyoak. 2006. Evaluating the long-term metacommunity
dynamics of tree hole mosquitoes. Ecology 87:2582–2590.
98
Ferguson, H. M., A. Dornhaus, A. Beeche, C. Borgemeister, M. Gottlieb, M. S. Mulla, J. E.
Gimnig, D. Fish, and G. F. Killeen. 2010. Ecology: a prerequisite for malaria elimination
and eradication. PLoS Medicine 7:e1000303.
Ferrari, M. C. O., F. Messier, and D. P. Chivers. 2008. Variable predation risk and the dynamic
nature of mosquito antipredator responses to chemical alarm cues. Chemoecology 17:223–
229.
Fischer, S., and N. Schweigmann. 2004. Culex mosquitoes in temporary urban rain pools:
seasonal dynamics and relation to environmental variables. Journal of Vector Ecology
29:365–373.
Frisch, D., K. Cottenie, A. Badosa, and A. J. Green. 2012. Strong spatial influence on
colonization rates in a pioneer zooplankton metacommunity. PloS ONE 7:e40205.
Fukami, T. 2015. Historical contingency in community assembly: integrating niches, species
pools, and priority effects. Annual Review of Ecology Evolution and Systematics 46:1–23.
Gardner, A. M., B. F. Allan, L. A. Frisbie, and E. J. Muturi. 2015. Asymmetric effects of native
and exotic invasive shrubs on ecology of the West Nile virus vector Culex pipiens (Diptera:
Culicidae). Parasites & Vectors 8:329.
Gingrich, J. B., R. D. Anderson, G. M. Williams, L. O’Connor, and K. Harkins. 2006.
Stormwater ponds, constructed wetlands, and other best management practices as potential
breeding sites for West Nile virus vectors in Delaware during 2004. Journal of the
American Mosquito Control Association 22:282–291.
Gleason, H. A. 1927. Further views on the succession-concept. Ecology 8:299–326.
99
Götzenberger, L., F. de Bello, K. A. Bråthen, J. Davison, A. Dubuis, A. Guisan, J. Lepš, R.
Lindborg, M. Moora, M. Pärtel, L. Pellissier, J. Pottier, P. Vittoz, K. Zobel, and M. Zobel.
2012. Ecological assembly rules in plant communities—approaches, patterns and prospects.
Biological Reviews 87:111–127.
Grace, J. B. 2006. Structural Equation Modeling and Natural Systems. Cambridge University
Press. Cambridge University Press, New York, NY, USA.
Griswold, M. W., and L. P. Lounibos. 2005. Competitive outcomes of aquatic container Diptera
depend on predation and resource levels. Annals of the Entomological Society of America
98:302–308.
Grover, J. P. 1989. Phosphorus-dependent growth kinetics of 11 species of freshwater algae.
Limnology and Oceanography 34:341–348.
Haney, J. F. et al. 2013. “An-Image-based Key to the Zooplankton of North America” version
4.0 released 2010. University of New Hampshire Center for Freshwater Biology
Hassall, C., and S. Anderson. 2015. Stormwater ponds can contain comparable biodiversity to
unmanaged wetlands in urban areas. Hydrobiologia 745:137–149.
Hillebrand, H. 2005. Light regime and consumer control of autotrophic biomass. Journal of
Ecology 93:758–769.
HilleRisLambers, J., P. B. Adler, W. S. Harpole, J. M. Levine, and M. M. Mayfield. 2011.
Rethinking community assembly through the lens of coexistence theory. Annual Review of
Ecology, Evolution, and Systematics 43:227–248.
100
Holt, R. D. 1993. Ecology at the mesoscale: the influence of regional processes on local
communities. Pages 77–88 Species Diversity in Ecological Communities: Historical and
Geographical Perspectives.
Holyoak, M., M. A. Leibold, and R. D. Holt. 2005. Metacommunities: Spatial dynamics and
ecological communities. University of Chicago Press.
Hutchinson, G. E. 1961. The paradox of the plankton. The American Naturalist 95:137–145.
Inouye, R. S., G. S. Byers, and J. H. Brown. 1980. Effects of predation and competition on
survivorship, fecundity, and community structure of desert annuals. Ecology 61:1344–1351.
Jones, J. 2001. Habitat selection studies in avian ecology: a critical review. The Auk 118:557–
562.
Joseph, M. B., and D. L. Preston. 2016. Integrating occupancy models and structural equation
models to understand species occurrence. Ecology 97:765–775.
Juliano, S. A. 1998. Species introduction and replacement among mosquitoes: Interspecific
resource competition or apparent competition? Ecology 79:255–268.
Kardol, P., N. J. Cornips, M. M. L. Van Kempen, J. M. T. Bakx-Schotman, and W. H. Van Der
Putten. 2007. Microbe-mediated plant-soil feedback causes historical contingency effects in
plant community assembly. Ecological Monographs 77:147–162.
Karpiscak, M. M., K. J. Kingsley, R. D. Wass, F. A. Amalfi, J. Friel, A. M. Stewart, J. Tabor,
and J. Zauderer. 2004. Constructed wetland technology and mosquito populations in
Arizona. Journal of Arid Environments 56:681–707.
Keating, J., K. Macintyre, C. M. Mbogo, J. I. Githure, and J. C. Beier. 2004. Characterization of
potential larval habitats for Anopheles mosquitoes in relation to urban land-use in Malindi,
Kenya. International Journal of Health Geographics 3:9.
101
Keddy, P. A. 1992. Assembly and response rules: two goals for predictive community ecology.
Journal of Vegetation Science 3:157–164.
Khawaled, K., M. S. Mulla, and A. Zaritsky. 1989. Distribution and abundance of algae in
mosquito developmental sites. Bulletin of the Society for Vector Ecology 14:71–80.
King, J. L., M. A. Simovich, and R. C. Brusca. 1996. Species richness, endemism and ecology of
crustacean assemblages in northern California vernal pools. Hydrobiologia 328:85–116.
Knight, T. M., J. M. Chase, C. W. Goss, and J. J. Knight. 2004. Effects of interspecific
competition, predation, and their interaction on survival and development time of immature
Anopheles quadrimaculatus. Journal of Vector Ecology 29:277–284.
Kraft, N. J. B., P. B. Adler, O. Godoy, E. C. James, S. Fuller, and J. M. Levine. 2015.
Community assembly, coexistence and the environmental filtering metaphor. Functional
Ecology 29:592–599.
Laughlin, D. C. 2011. Nitrification is linked to dominant leaf traits rather than functional
diversity. Journal of Ecology 99:1091–1099.
Li, Y., F. Kamara, G. Zhou, S. Puthiyakunnon, C. Li, Y. Liu, Y. Zhou, L. Yao, G. Yan, and X.
G. Chen. 2014. Urbanization increases Aedes albopictus larval habitats and accelerates
mosquito development and survivorship. PLoS Neglected Tropical Diseases 8:e3301.
Lynch, M. 1979. Predation, competition, and zooplankton community structure: an experimental
study. Limnology and Oceanography 24:253–272.
MacArthur, R. H., and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton
University Press, Princeton, New Jersey, USA.
102
MacKay, A. J., E. J. Muturi, M. P. Ward, and B. F. Allan. 2016. Cascade of ecological
consequences for West Nile virus transmission when aquatic macrophytes invade
stormwater habitats. Ecological Applications 26:219–232.
Maire, V., N. Gross, L. Börger, R. Proulx, C. Wirth, L. da S. Pontes, J. F. Soussana, and F.
Louault. 2012. Habitat filtering and niche differentiation jointly explain species relative
abundance within grassland communities along fertility and disturbance gradients. New
Phytologist 196:497–509.
Mandal, S. K., A. Ghosh, I. Bhattacharjee, and G. Chandra. 2008. Biocontrol efficiency of
odonate nymphs against larvae of the mosquito, Culex quinquefasciatus Say, 1823. Acta
Tropica 106:109–114.
Martin, T. E. 2001. Abiotic vs. biotic influences on habitat selection of coexisting species:
climate change impacts? Ecology 82:175–188.
McCauley, S. J., C. J. Davis, R. A. Relyea, K. L. Yurewicz, D. K. Skelly, and E. E. Werner.
2008. Metacommunity patterns in larval odonates. Oecologia 158:329–342.
Menge, B. A., and A. M. Olson. 1990. Role of scale and environmental factors in regulation of
community structure. Trends in Ecology and Evolution 5:52–57.
Merritt, R. W., K. W. Cummins, and M. B. Berg, editors. 1996. An Introduction to the Aquatic
Insects of North America. Kendall Hunt Publishing Company. 4th edition. Dubuque, IA,
USA.
Merritt, R. W., R. H. Dadd, and E. D. Walker. 1992. Feeding behavior, natural food, and
nutritional relationships of larval mosquitoes. Annual Review of Entomology 37:349–376.
Mittelbach, G. G., and D. W. Schemske. 2015. Ecological and evolutionary perspectives on
community assembly. Trends in Ecology and Evolution 30:241–247.
103
National Research Council. 2008. Urban Stormwater Management in the United States. National
Academy Press, Washington, D.C.
Paine, R. T. 1966. Food web complexity and species diversity. The American Naturalist 100:65–
75.
Pinel-Alloul, B., and E. A. Mimouni. 2013. Are cladoceran diversity and community structure
linked to spatial heterogeneity in urban landscapes and pond environments? Hydrobiologia
715:195–212.
Price, N. 2010. Habitat selection, facilitation, and biotic settlement cues affect distribution and
performance of coral recruits in French Polynesia. Oecologia 163:747–758.
Pritchard, G., and P. J. Scholefield. 1983. Survival of Aedes larvae in constant area ponds in
Southern Alberta (diptera: Culicidae). The Canadian Entomologist 115:183–188.
Prugh, L. R., and J. S. Brashares. 2012. Partitioning the effects of an ecosystem engineer:
kangaroo rats control community structure via multiple pathways. Journal of Animal
Ecology 81:667–678.
Quiroz-Martinez, H., and A. Rodriguez-Castro. 2007. Aquatic insects as predators of mosquito
larvae. Journal of the American Mosquito Control Association 23:110–117.
Resetarits, W. J. 1996. Oviposition site choice and life history evolution. American Zoologist
36:205–215.
Resetarits, W. J. 2001. Colonization under threat of predation: avoidance of fish by an aquatic
beetle, Tropisternus lateralis (Coleoptera: Hydrophilidae). Oecologia 129:155–160.
104
Resetarits, W. J., C. A. Binckley, and D. R. Chalcraft. 2005. Habitat selection, species
interactions, and processes of community assembly in complex landscapes: a
metacommunity perspective. Pages 374–398 in M. Holyoak, M. A. Leibold, and R. D. Holt,
editors. Metacommunities: Spatial Dynamics and Ecological Communities. University of
Chicago Press, Chicago, Illinois, USA.
Roberts, D. 2014. Mosquito larvae change their feeding behavior in response to kairomones from
some predators. Journal of Medical Entomology 51:368–374.
Rosseel, Y. 2012. Lavaan: an R package for structural equation modeling and more. Journal of
Statistical Software 48:1–36.
Rubbo, M. J., J. L. Lanterman, R. C. Falco, and T. J. Daniels. 2011. The influence of amphibians
on mosquitoes in seasonal pools: Can wetlands protection help to minimize disease risk?
Wetlands 31:799–804.
Rundle, S. D., A. Foggo, V. Choiseul, and D. T. Bilton. 2002. Are distribution patterns linked to
dispersal mechanism? An investigation using pond invertebrate assemblages. Freshwater
Biology 47:1571–1581.
Sartory, D. P., and J. U. Grobbelaar. 1984. Extraction of chlorophyll a from freshwater
phytoplankton for spectrophotometric analysis. Hydrobiologia 114:177–187.
Schäfer, M. L., and J. O. Lundström. 2006. Different responses of two floodwater mosquito
species, Aedes vexans and Ochlerotatus sticticus (Diptera : Culicidae), to larval habitat
drying. Journal of Vector Ecology 31:123–128.
Schneider, D. W., and T. M. Frost. 1996. Habitat duration and community structure in temporary
ponds. Journal of the North American Benthological Society 15:64–86.
105
Schriever, T. A. 2015. Food webs in relation to variation in the environment and species
assemblage: a multivariate approach. PLoS ONE 10:1–18.
Shaalan, E. A. S., and D. V. Canyon. 2009. Aquatic insect predators and mosquito control.
Tropical Biomedicine 26:223–261.
Silberbush, A., and W. J. Resetarits. 2017. Mosquito female response to the presence of
larvivorous fish does not match threat to larvae. Ecological Entomology 42:595–600.
Stav, G., L. Blaustein, and J. Margalith. 1999. Experimental evidence for predation risk sensitive
oviposition by a mosquito, Culiseta longiareolata. Ecological Entomology 24:202–207.
Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75:2–16.
Tixier, G., M. Lafont, L. Grapentine, Q. Rochfort, and J. Marsalek. 2011. Ecological risk
assessment of urban stormwater ponds: literature review and proposal of a new conceptual
approach providing ecological quality goals and the associated bioassessment tools.
Ecological Indicators 11:1497–1506.
Ullman, J. B. 2007. Structural equation modeling. Pages 676–780 in B. G. Tabachnick and L. S.
Fidell, editors. Using Multivariate Statistics. Fifth edition. Allyn and Bacon, Boston, MA,
USA.
Urabe, J., M. Kyle, W. Makino, T. Yoshida, T. Andersen, and J. J. Elser. 2002. Reduced light
increases herbivore production due to stoichiometric effects of light/nutrient balance.
Ecology 83:619–627.
Vanschoenwinkel, B., A. Hulsmans, E. De Roeck, C. De Vries, M. Seaman, and L. Brendonck.
2009. Community structure in temporary freshwater pools: disentangling the effects of
habitat size and hydroregime. Freshwater Biology 54:1487–1500.
106
Vonesh, J. R., and L. Blaustein. 2010. Predator-induced shifts in mosquito oviposition site
selection: a meta-analysis and implications for vector control. Israel Journal of Ecology &
Evolution 56:263–279.
Walton, W. E., and J. A. Jiannino. 2005. Vegetation management to stimulate denitrification
increases mosquito abundance in multipurpose constructed treatment wetlands. Journal of
the American Mosquito Control Association 21:22–27.
Wanielista, M. P., and Y. A. Yousef. 1993. Stormwater Management. Stormwater Management.
John Wiley & Sons, Inc., New York, NY, USA.
Weiher, E., D. Freund, T. Bunton, A. Stefanski, T. Lee, and S. Bentivenga. 2011. Advances,
challenges and a developing synthesis of ecological community assembly theory.
Philosophical Transactions of the Royal Society B: Biological Sciences 366:2403–2413.
Weiher, E., and P. Keddy. 2001. Ecological Assembly Rules: Perspectives, Advances, Retreats.
Cambridge University Press, Cambridge, United Kingdom.
Westby, K. M., and S. A. Juliano. 2017. No detectable role for predators mediating effects of
aquatic habitat size and permanence on populations and communities of container-dwelling
mosquitoes. Ecological Entomology 42:439–448.
Wetzel, R. G., and G. E. Likens. 1991. Inorganic nutrients: nitrogen, phosphorus, and other
nutrients. Pages 81–105 in R. G. Wetzel and G. E. Likens, editors. Limnological Analyses.
Springer, New York, NY, USA.
Wu, J., S. Yu, and R. Zou. 2006. A water quality based approach for watershed wide BMP
strategies. Journal of the American Water Resources Association 42:1193–1204.
107
APPENDIX A: SUPPLEMENTAL TABLE AND FIGURES
Competitors Predators Suborder Family Species Species Cladocera Bosminidae Bosmina spp. Chydoridae Acroperus spp. Alona spp. Chydorus spp. Eurycercus spp. Daphniidae Ceriodaphnia spp. Daphnia spp. Scapholeberis mucronata Simocephalus serrulatus Simocephalus vetulus Ilyocryptidae Ilyocryptus spp. Moinidae Moina spp. Sididae Diaphanosoma spp. Sida spp. Macrothricidae Macrothricidae
Order Family Species Species Calanoida Diaptomidae Skistodiaptomus spp. Cyclopoida Cyclopoidae Ectocyclops phaleratus Acanthocyclops spp. Eucyclops agilis Orthocyclops modestus Eucyclops elegans Microcyclops rubellus Paracyclops popei Tropocyclops prasinus mexicanus Unknown cyclopoid Harpacticoida -- Harpacticoida
Order Order Other crustaceans Ostracoda Amphipoda
Class Order Family Family Insecta Coleoptera Dytiscidae Diptera Chironomidae Chaoboridae Hemiptera Corixidae Notonectidae Odonata Anisoptera (suborder) Zygoptera (suborder)
Table A1: Categorization of each invertebrate taxa used in the structural equation.
108
0.9
0.85
0.8
0.75
0.7
0.65
percent occupancy 0.6 mean mean+std 0.55 mean-std determ. 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6
regional colonization rate
Figure A1: The effect of regional colonization is shown for the stochastic simulations (blue) ± standard deviation and the ODE (pink). Percent occupancy increases with regional colonization for our network of N = 38 ponds.
109
500 400 300 200 Mosquito density (ind/L) 100 0
low medium high 0−100 101−350 350+ Zooplankton density (ind/L) Figure A2: Larval mosquito density (ind/L) is plotted against zooplankton density (divided into three density categories: low [0-100 ind/L], medium [101-350 ind/L], and high [350+ ind/L]).
110