Accepted Article

Received Date : 29-Jun-2013 Revised Date : 30-Sep-2013 Accepted Date : 02-Oct-2013 Article type

: Opinion

Molecular Ecology – for MTI-2 Special Issue

OPINION

Corresponding author: Matthew H. Greenstone USDA-ARS-IIBBL, Room 214, Building 011A, BARC West 10300 Baltimore Avenue Beltsville, Maryland 20705 U.S.A. Tel:

301 504 5139

Fax:

301 504 5104

Email: [email protected]

The Detectability Half-life in Arthropod Predator-Prey Research: What it is, Why We Need it, How to Measure it, and How to Use it by

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/mec.12552 This article is protected by copyright. All rights reserved.

Accepted Article

MH Greenstone, U.S.D.A. – Agricultural Research Service, Invasive Insect Biocontrol and Behavior Laboratory, 10300 Baltimore Avenue, Beltsville, Maryland 20705, USA ME Payton, Department of Statistics, 301 MSCS Building, Oklahoma State University, Stillwater, Oklahoma 74078, USA DC Weber, U.S.D.A. – Agricultural Research Service, Invasive Insect Biocontrol and Behavior Laboratory, 10300 Baltimore Avenue, Beltsville, Maryland 20705, USA AM Simmons, U.S.D.A. – Agricultural Research Service, U.S. Vegetable Laboratory, 2700 Savannah Highway, Charleston, South Carolina 29414, USA

Keywords: biological control, ELISA, molecular gut-content analysis, PCR, predator-prey interactions

Running title:

Detectability half-life in predator-prey research

Abstract

Molecular gut-content analysis enables detection of arthropod predation with minimal disruption of ecosystem processes. Most assays produce only qualitative results, with each predator testing either positive or negative for target prey remains. Nevertheless they have yielded important insights into community processes. For example, they have confirmed the longhypothesized role of generalist predators in retarding early-season buildup of pest populations

This article is protected by copyright. All rights reserved.

Accepted Article

prior to the arrival of more specialized predators and parasitoids, and documented the ubiquity of secondary and intraguild predation. However, raw qualitative gut-content data cannot be used to assess the relative impact of different predator taxa on prey population dynamics: they must first be weighted by the relative detectability periods for molecular prey remains for each predator-prey combination. If this is not done, interpretations of predator impact will be biased toward those with the longest detectabilities. We review the challenges in determining detectability half-lives, including unstated assumptions that have often been ignored in the performance of feeding trials. We also show how detectability half-lives can be used to properly weight assay data to rank predators by their importance in prey population suppression, and how sets of half-lives can be used to test hypotheses concerning predator ecology and physiology. We use data from 32 publications, comprising 97 half-lives, to generate and test hypotheses on taxonomic differences in detectability half-lives, and discuss the possible role of the detectability half-life in interpreting qPCR and next-generation sequencing data.

Introduction

Molecular gut-content analysis has revolutionized the study of arthropod predator-prey interactions, enabling detection of large numbers of feeding events with minimal disruption to ongoing population and community processes (Stuart & Greenstone 1990; Symondson 2002). Throughout its nearly seventy-year history, the field has striven to adopt state-of-the-art technologies, and been animated by a continual search for greater sensitivity and specificity. The first molecular gut-content assay was the tube-precipitin test (Brooke & Proske 1946), which was not improved upon until the Ouchterlony assay was adopted by some researchers

This article is protected by copyright. All rights reserved.

Accepted Article

(e.g., Frank 1967); as precipitating assays, both exhibit inherently low sensitivities (Greenstone 1996). The Ouchterlony’s sole advantage is the ability to distinguish false positives. Only precipitating assays were available until the introduction of enzyme-linked immunosorbent assay (ELISA; Fichter & Stephen 1981; Ragsdale et al. 1981), which yields sensitivities in the 1 – 10 ng protein range, translating to 10-5 animal equivalents (a.e.) or lower. Improvements in

antibody purification and production led to the highest specificities ever achieved by molecular gut-content assays (Ragsdale et al. 1981; Greenstone & Morgan 1989). Finally, adoption of polymerase chain reaction (PCR) assays at the end of the last century (Zaidi et al. 1999) ushered in the age of DNA-based assays, with sensitivities on the order of 10-7 a.e. These have

not entirely supplanted ELISA, because DNA cannot provide the stage-specificity of monoclonal antibody (MAb)-based assays. Nevertheless their relatively low cost, standardized protocols, and the ease of designing primers for any system, have enabled a larger number of practitioners to employ them, and made species-specific molecular gut-content assays potent and well respected tools for arthropod ecological research.

Raw molecular gut-content data are qualitative – they are read as either positive or negative and because one typically cannot know the size, stage, number, and time since consumption of the prey item or items represented by detected remains, they provide no quantitative information on predator impact on prey populations (Greenstone 1996). Nevertheless raw data have been used to make important ecological inferences. For example, there had been only circumstantial evidence on the role of generalist predators in annual crops. Data from qualitative assays have now confirmed the hypothesis that generalist predators can subsist on alternate non-target prey before the arrival of a pest, then switch to it and thereby retard population buildup before the arrival of more specialized predators and parasitoids (Harwood et al. 2007a; Boreau de Roincé et al. 2013; Firlej et al. 2013). They have also provided estimates of predator diet breadth and

This article is protected by copyright. All rights reserved.

Accepted Article

selectivity (Blankenship & Yayonos 2005; Lundgren & Weber 2010; Schmidt et al. 2012; von Berg et al. 2012; Chapman et al. 2013; Davey et al. 2013); shown that scavenging and secondary predation occur commonly in nature (Harwood et al. 2001; Calder et al. 2005; Foltan et al. 2005; Juen & Traugott 2005; Sheppard et al. 2005); that intraguild predation, including predation on already-parasitized hosts, is ubiquitous (Chacon et al. 2008; Traugott & Symondson 2008; Gagnon et al. 2011a; Moreno-Ripoll et al. 2012; Traugott et al. 2012); that cannibalism may make a significant contribution to prey population suppression (Sigsgaard et al. 2002); and that the incidence of predation may have unsuspected spatial influences (Greenstone 1983; Opatovsky et al. 2013). Finally, they have made it possible to delineate food webs in complex invertebrate communities, including subterranean ones (Juen & Traugott 2007; Lundgren et al. 2009; Chapman et al. 2013; Davey et al. 2013; Etzinger et al. 2013; Pianezolla

et al. 2013).

The remaining challenge of molecular gut-content analysis is to transform qualitative assay results into quantitative data that can be used to infer the impact of predation on prey population dynamics. This was not a problem when precipitating assays were the state-of-the-art: so insensitive were these assays that their detectability limits for prey in a predator’s gut tended to be shorter than the estimated mean intervals between prey captures, so that an assay positive represented a single prey item. This precision made it possible to confirm that the ‘all-other’ category of prey mortality in life table studies could safely be assigned to predation (e.g., see Dempster 1960).

Today, with the widespread use of assays capable of detecting 1 - 10 ng of protein (ELISA) and tens or hundreds of copies of DNA sequence (PCR), we can no longer make the same

This article is protected by copyright. All rights reserved.

Accepted Article

assumption. We could recover this advantage by making contemporary assays less sensitive, but in so doing we would miss many predation events.

Impetus for this paper

Early ELISA data suggested large, consistent differences among taxa in the length of time after feeding during which a prey item can be detected in the gut of a predator. Spiders and staphylinid beetles in particular seemed to have longer ‘maximum detectability times,’ or ‘Dmax’

(Sunderland et al. 1987), than some other predators. If this were true, assay results would have to be weighted to remove the bias, favoring those with long detectability times, that would otherwise confound interpretations of assay results. While Dmax was a useful concept for

discussing these differences, it substituted an absolute limit for what is really a continuous variable with significant variation. This provided the impetus for formulating a more realistic tool for weighting gut-content assay data.

In this paper we discuss such a tool, the detectability half-life, which can help us to gauge relative predator impacts on prey population dynamics. Given the investments in time, labor, and materials required to derive detectability half-lives, we must be clear about why we need to know the half-lives of prey remains in predators in an ecosystem, how to calculate them, and how to use them to interpret gut-content data. Here we suggest how it should be measured, and show how it can be used to rank different members of the predator complex by their impacts on a prey population; we also show how it can be used to pose and test some taxonomic and

This article is protected by copyright. All rights reserved.

Accepted Article

physiological hypotheses. Finally, we discuss how developments in quantitative PCR and nextgeneration sequencing may affect the need for, and utility of, the detectability half-life.

History and nomenclature

The detectability half-life was introduced in a paper showing that the proportion of paper wasps fed a homogenate of late-instar corn earworm detectable in the gut by a MAb-based ELISA declined exponentially with time (Greenstone & Hunt 1993). The paper defined the detectability half-life as the time after feeding at which prey remains could be detected in only half of the assayed predators. The concept was extended to DNA-based assays with a one-aphid twopredator system, in a paper that also included half-life confidence intervals and statistical testing for differences between them (Chen et al. 2000). Shortly thereafter, Payton et al. (2003) stressed the importance of using probit or logistic protocols to calculate detectability half-lives, and provided further guidance for testing sets of half-lives from a given predator-prey system for statistical differences.

The detectability half-life has been widely adopted by the predator-prey research community to characterize gut-content data derived from both serological and DNA-based assays (see Table 1). However, two nomenclatural issues have confused the meaning of the detectability half-life. First, roughly half of the papers reporting it have given it a slightly or very different name, with more than a half dozen variants (Table 1); we return to this point in the Discussion. Second, authors have sometimes reported an entirely different half-life, that of the amount of analyte detected by a quantitative assay, i.e., ng of target prey protein by quantitative ELISA, or target

This article is protected by copyright. All rights reserved.

Accepted Article

DNA-sequence copy number by quantitative PCR (qPCR) or a surrogate by prey DNA dilution. When plotted over time these values follow time-courses of decay similar to those of the proportions positive in the binary qualitative assays, and are well fitted by the same statistical models (Fichter & Stephen 1981, 1984; Lövei et al. 1987, 1990; Symondson & Liddell 1993, 1995; Hagler & Naranjo 1997; Symondson et al. 2000; Harwood et al. 2001, 2004, 2007b; Calder et al. 2005; Weber & Lundgren 2009; Schmidt et al. 2012). Both types of curves are used to calculate the time since feeding at which only half of the variable plotted can be detected: the detectability of half of the meals consumed at time t = 0 in the first case; and the detectability of half of the original amount of the analyte consumed at time t = 0 in the second. These two analogously calculated times are the half-lives of the different variables being measured by the respective assays; they are not the same. And though these half-lives are certainly quantitative, neither by itself is informative of the impact of predators on prey population dynamics.

This paper concerns detectability half-lives only; relatively little use has been made to date of analyte half-lives to infer impacts of predators on prey population dynamics (but see Discussion).

Calculating the detectability half-life

To calculate a detectability half-life, we must perform an extensive feeding trial. It begins with a large number of predators that have been held without food for a period of time to ensure that they do not contain any prey remains that could produce a positive in the gut-content assay, and

This article is protected by copyright. All rights reserved.

Accepted Article

that they are hungry enough to feed readily. This takes one or two weeks for spiders and one to several days for insect predators (see references in Table 1). One then feeds the predators target prey, divides them into groups to be killed at known intervals following cessation of feeding, kills them at those intervals, and then quickly preserves them, preferably by placement into very cold 70 – 95% EtOH and thence posthaste to a freezer no warmer than -200C and

ideally -800C to retard further digestion (King et al. 2008; Weber & Lundgren 2009). At each step in the feeding trial, the investigator faces a number of choices, and may run the risk of violating certain unstated assumptions. We lay out these choices and assumptions next.

Predator and prey sources One wishes to simulate what happens in nature, yet also to produce replicable data with low variability. When obtaining animals for feeding trials, the best of both worlds would be established colonies of both predator and prey that are frequently replenished with fresh field material to avoid fixation of laboratory-selected traits. Established colonies enable predators to be run in cohorts of similar physiological state - motivation to feed, condition of the gut, time since last moult, etc. - and prey to be provided in cohorts as identical as possible in attractiveness to predators and quantity of the analyte contained. This is not always possible, and the investigator must do his or her best to standardize the animals in a trial whether derived from culture or field collection. Cannibalistic predators, such as wandering spiders, are especially difficult to maintain in culture, and are usually obtained from the field. But many other classes of predators are also difficult to rear, or lack established rearing protocols, and end up being field-collected as well.

This article is protected by copyright. All rights reserved.

Accepted Article

The next question is how many predators to feed for each interval and, ipso facto, how many predators total to set up at time t = 0. The confidence one has in the results will depend upon sample size, so in principal the more animals the better. But the number available at t = 0 will depend upon the vagaries of colony production or field collection, and the number that can be accommodated at t = 0 will depend upon the availability of supplies and labor. Finally, even when one begins with sufficient animals for a large number at each interval post-feeding, a sizeable fraction invariably fail to feed; further, mortality sometimes takes a high toll on the fed population during the course of the experiment (e.g., Ma et al. 2005). Either or both of these eventualities may necessitate changes in allocation of animals to time-interval groups as the experiment progresses, resulting in smaller group sizes than planned. Should this happen, the best one can do is to make a best guess at what the half-life will turn out to be and maximize the numbers of animals in the intervals surrounding that time point; this will assure the highest confidence in the calculated half-life estimate.

Wildly varying numbers of animals have been used per interval, with 10 being the mode and 20 the maximum (Fig. 1; data for this and all subsequent figures were derived from the papers in Table 1). Again, the larger the numbers used, the greater the confidence in the half-life estimate, and the greater the likelihood of detecting differences among them. To provide an estimate of the cost of choosing smaller sample sizes, we performed Monte Carlo simulations of the half-lives for 10- and 20-animal-per-interval feeding trials, using a published dataset of seven predator taxa feeding on the same prey species (Greenstone et al. 2010). For both sample-size scenarios, a probit regression was fit to the individual simulated datasets and respective half-lives were calculated. We repeated this for a total of 10,000 replicates. To compare the two sampling protocols, we computed means and standard deviations (SDs) of the half-lives and ratios of the standard deviations (SD for n=10 divided by SD for n=20). The

This article is protected by copyright. All rights reserved.

Accepted Article

estimated mean half-lives are essentially identical, but the SDs of the 10-sample half-lives are half-again as large as those of the 20-sample sample set (Table 2). These results demonstrate the sizeable negative tradeoff from reducing sample sizes in feeding trials.

An unstated assumption of all half-life feeding trials is that each predator consumes the same quantity of the analyte because, all other conditions being equal, one expects the half-life to be longer if the amount in the gut at time t = 0 is greater. This expectation has been verified experimentally (Hagler & Naranjo 1997; Hoogendoorn & Heimpel 2001; King et al. 2010; Waldner et al. 2013). The assumption is met if the prey item is small enough that a predator can consume all of it at one sitting. Cohorts of single eggs with known time since oviposition will contain essentially identical quantities of analyte, and be small enough for many predators to consume in a single feeding bout. The same will be true for many predators feeding on a single animal much smaller than themselves, as for example aphids, thrips, or collembola are apt to be for many predators. Another alternative, if the predator will take non-living prey and provided that a prey individual can be subdivided accurately and without significant leakage, is a quickly consumable fixed fraction of a prey individual (e.g., King et al. 2010; Sint et al. 2011; VirantDobertel et al. 2011). Unfortunately, the assumption of a fixed quantity of prey consumed at time

t = 0 has not always been met (Fig. 2).

Chen et al. (2000) introduced the practice of providing alternate prey, later referred to as ‘chaser prey’ (Weber & Lundgren 2009), to predators already fed the target and waiting to be killed, in order to ’maintain normal metabolism.’ Their concern stemmed from the knowledge that some predators, e.g., spiders, greatly depress their metabolic rates when starved (Anderson 1970), which could alter rates of digestion. The supposed importance of this for half-life studies rests

This article is protected by copyright. All rights reserved.

Accepted Article

on the assumption that predators are not starving in nature. That assumption has been tested and found to be poorly supported in some instances (Bilde & Toft 1998; Wilder 2011) but not all of them (Greenstone 1978), and not at all times and places for a given predator species (Wilder 2011). Further, most predators are polyphagous, and must perforce often contain remains of more than one prey species in the gut. In fact all studies employing contemporary assays for two or more prey species have found multiple species in the guts of some fraction of fieldcollected predators (Greenstone 1979; Symondson & Lidell 1993; Hagler & Naranjo 1994a, b; Agustí et al. 2003b; Harper et al. 2005, 2006; Pons 2006; Harwood et al. 2007b; King et al. 2010; Eitzinger & Traugott 2011; Gagnon et al. 2011a,b). Dietary mixing, including the use of chaser prey, has varying influences on the detectability of the target prey analyte, ranging from none (Lövei et al.1987), to only weakly and non-systematically (Lövei et al. 1990), to negative (Weber & Lundgren 2009). Given that the vast majority of arthropod predators are polyphagous, using a chaser seems reasonable and prudent, but since doing so sometimes increases the half-life (Fournier et al. 2006; Harper et al. 2005), further research on the nutritional status of

predators in the field is warranted.

Environmental conditions

The temperature at which predators are held may strongly influence prey detectability (McIver 1981; Sopp & Sunderland 1989; Hagler & Naranjo 1997; Hagler & Cohen 1990; Hosseini et al. 2008; von Berg et al. 2008; Kobayashi et al. 2011), and must therefore be considered in the

design of feeding trials. If one wishes to simulate natural conditions, holding the animals in an incubator that closely tracks the daily temperature cycle in the field (Chen et al. 2000; Greenstone et al., 2007, 2010), or one that switches from mean day-time to mean night-time temperature (Greenstone 1979; Kerzicnik et al. 2012) is best. Almost half of researchers have

This article is protected by copyright. All rights reserved.

Accepted Article

used this approach, while others have used an approximate 24-h mean; some, however, seem to have defaulted to what was available (Fig. 3). If instead one wishes to compare predator halflives on the basis of, say, size or taxon for predators occupying different habitats at different times, a common fixed temperature regime would be preferable.

Gut-content analysis and curve-fitting

Following the feeding trial, predators are processed to extract prey-species-specific remains, and the extracts run through a serological or DNA-based assay. These have been almost exclusively ELISA or PCR assays, very sensitive techniques with well-established and reproducible protocols. When designing an ELISA for gut-content analysis, the critical decision is choice of the antigen target, and boils down to the developmental or taxonomic level of specificity desired, with species, stage, and more inclusive levels all possible with MAb-based assays (e.g., Greenstone & Morgan 1989; Sigsgaard et al.2002; Harwood et al. 2007b). The critical decisions for PCR are the choice of genomic region and the design of primers. Most arthropod predator-prey researchers have used mitochondrial DNA, especially Cytochrome Oxidase I and II. Both contain hundreds or thousands of copies per cell, and both (and especially COI) have been used as barcodes, providing enormous databases of alreadyvalidated species-specific sequences to choose from. These barcoding sequences, however, tend to be large, and if used entire would quickly become unamplifiable in the gut. One needs sequences that are short enough to remain detectable for some time in the gut but long enough to contain sufficient variability to discriminate closely related species. This puzzle was actively researched during the first five years of development of DNA-based molecular gut-content methods (Agustí et al. 1999, 2003a; Zaidi et al. 1999; Chen et al. 2000; Hoogendoorn &

This article is protected by copyright. All rights reserved.

Accepted Article

Heimpel 2001), and the consensus among practitioners was that amplicons on the order of 200 – 300 bp in size struck the right balance. Nevertheless arthropod molecular ecologists have continued to study this issue (Fig. 4). Amplicon size and detectability half-life are inversely correlated in a given predator-prey system, but there is no significant relationship between them (p > 0.05) when data from all available studies are plotted (Fig. 5). Nor is there a significant relationship between these two variables within the Araneae, Coleoptera, or Heteroptera (data not shown).

Following assay, the predators in each post-feeding interval are scored for the proportion positive for the antigen or DNA sequence targeted by the assay, and the data are fit to an appropriate model. Since assay results are either positive or negative, they should only be fit to models designed for binary data. With very few exceptions, plotted half-life data-points manifestly decline logistically, and according to canonical statistical practice should be fit with probit or logit models, or variations of these procedures. When reporting half-life results, it is important to also report fiducial limits, both for the reporting of variability, and for testing for differences in half-lives (Payton et al. 2003).

Uses for the detectability half-life

The use of the detectability half-life to weight the proportions of prey positive for different predator taxa attacking the same pest was first proposed for a one-prey, two-predator system by Chen et al. (2000). Using primers for a species-specific 198 bp fragment of Cytochrome Oxidase II of the corn leaf aphid, Rhopalosiphum maidis (Fitch), the authors determined that the

This article is protected by copyright. All rights reserved.

Accepted Article

half-life of detectability for a single aphid consumed by a predator maintained under simulated field temperature conditions was 3.95 h in the lacewing Chrysoperla plorabunda (Fitch), and 8.78 h in the ladybeetle Hippodamia convergens Guerin. This means that a single R. maidis is 2.2 times more likely to be detected in an individual of H. convergens than it is in an individual of C. ploribunda. An investigator looking only at the proportions positive for the prey in each predator, but lacking knowledge of these statistically different half-lives, would be misled into believing that the impact of the ladybeetle relative to the lacewing is much greater than it is. This is easily solved by using the ratio of half-lives as a multiplier (Chen et al. 2000). We subsequently demonstrated for a seven-predator single-prey system how the detectability halflife of any member of a predator complex can be used as the reference half-life for determining the relative ranks of all predators’ impacts on a prey population (Greenstone et al. 2010). Since the vast majority of predators are polyphagous or stenophagous, the half-lives of different prey in gut content analyses of predators will have to be considered as larger predator and prey complexes are treated (Gagnon et al. 2011a; Chapman et al. 2013; Davey et al. 2013).

Half-life data may also be used to test ecological and physiological hypotheses. It was hypothesized more than twenty-five years ago that spiders have intrinsically longer half-lives than other predatory arthropods (Greenstone 1983; Sunderland et al. 1987). In a paper

comparing half-lives for L. decemlineata eggs in two predators, Coleomegilla maculata (de Geer) and Podisus maculiventris (Say), the first with chewing and the second with sucking mouthparts (Greenstone et al. 2007), we rashly hypothesized on the basis of this single statistical difference that all sucking insects have longer detectability half-lives than all chewing insects, and that sucking insects share metabolic adaptations with spiders, which, besides having sucking mouthparts and long detectability half-lives, have lower basal metabolic rates than comparably-sized insects (Anderson 1970; Greenstone & Bennett 1980).

This article is protected by copyright. All rights reserved.

Accepted Article

The availability of a large number of half-lives makes it possible to test these two hypotheses. When the 97 half-lives from the 32 studies catalogued in Table 1 are plotted by taxon, we find that spiders do, indeed, have the longest recorded half-lives, as do many predatory heteropterans, which also have sucking mouth parts. However, they also exhibit some of the shortest half-lives (Fig. 6). According to this simple meta-analysis, the hypotheses that spiders and predatory heteropterans have intrinsically longer prey detectability half-lives than other arthropod predators is not supported.

Discussion

We have reviewed progress in the field and indicated how detectability half-life protocols must be standardized to enable rigorous measurement and comparability across laboratories. In particular, it is essential to standardize the size of the meal consumed at t = 0: not doing so yields a non-reproducible estimate that, by definition, does not estimate the half-life. We further suggest that much time, effort, and supplies could be saved by dispensing with further studies of ideal amplicon size: amplicons in the 200 – 300 bp range will usually provide the necessary species-specificity without being immediately digested to the point of non-detectability. Further, using amplicons at least 200 bp in size will reduce the likelihood of detecting scavenged prey, since the shorter sequences resulting from decay will be less likely to be detected and incorrectly attributed to predation (cf. Foltan et al. 2005).

The power and accessibility of molecular gut-content assays have tempted some practitioners to make ecological inferences on the basis of differences in the raw proportions of predators

This article is protected by copyright. All rights reserved.

Accepted Article

positive for target prey remains in the gut (e.g., Hagler & Naranjo 1994b; Zhang et al. 2007). As we have shown, sufficient data have accumulated to demonstrate that different predator taxa often differ statistically in detectability half-lives for a given prey. Further, different prey may have statistically different half-lives in a given predator (Gagnon et al. 2011a). It is therefore not defensible to make inferences concerning predator effectiveness as agents of prey mortality on the basis only of raw molecular gut-content assay data: they must first be weighted to compensate for differences in half-lives.

The importance of weighting is illustrated by two field studies of multiple-predator single-prey systems in which half-life-weighting dramatically altered the relative rankings of predators in their impact on the prey population (Greenstone et al. 2010; Gagnon et al. 2011b). When combined with prey and predator population data, such half-life-weighted assay data from fieldcollected predators enable ranked assessments of the importance of each predator population in the field, providing a powerful tool for designing predator management protocols to maximize pest population suppression by biocontrol (Szendrei et al. 2010).

Although half-lives of many predatory taxa have been determined, there are some surprising omissions (Fig. 7). We have exhibited an inordinate fondness for predatory beetles, but the staphylinids are strangely absent, and only one cantharid has been examined despite their abundance and ubiquity in many habitats. Spiders have also been popular, but only four families of this diverse and omnipresent order have been examined. With respect to other major predator groups, there are only two chrysopids and one each mite and centipede; the opilionids, another order with abundant and ubiquitous members, are not represented at all. The range of prey taxa treated reflects the diets of predators in the systems chosen for study (Fig. 8).

This article is protected by copyright. All rights reserved.

Accepted Article

‘Detectability half-life’ is the term originally coined for the time-point under discussion (Greenstone & Hunt 1993). Other terms have been used, some of them close cousins, others very different (Table 1). One family of terms, the ‘median detectability time’ and its variants, is formally identical to the detectability half-life. We advocate the use of ‘detectability half-life,’ both for reasons of consistency and clarity, and because it more evocatively captures the dynamic nature of the phenomenon being measured: the disappearance with time of the detectability of prey remains (terms like ‘half-life of prey DNA’ [Romeu-Dalmau et al. 2012] should not be used, since they connote disappearance of the consumed analyte, not of the detectability of remains). Some suggest that the term ‘half-life’ is not appropriate, because unlike, say, a radioisotope, the detectability half-life is a mass-less entity. However, not only is it very common to speak colloquially of mass-less entities like ideas, buzz-words, and careers as having half-lives: analogous concepts are also used in scholarly contexts, for example in the advertising literature (Grass & Wallace 1969), where the impact of an advertising campaign has a half-life that can be measured, and used to manage the campaign to improve its efficacy (Naik 1999).

Lack of a significant overall correlation between amplicon size and detectability half-life (Fig. 5) is surprising, given its invariant presence within a given predator-prey system (Zaidi et al. 1999; Chen et al. 2000; Hoogendoorn & Heimpel 2001; Agustí et al. 2001, 2003a; Juen & Traugott

2005; von Berg et al. 2008; King et al 2010). We hypothesize that this is due to differences in the nuclease diversity in the guts of different predators, differences in the susceptibilities of amplified sequences to digestion, or some combination of the two. Differences in assay sensitivity may also play a role (Chen et al. 2000; Agustí et al. 2001; de Leon et al. 2006; Kuusk et al. 2008; Gagnon et al. 2011b). Alternatively, differences in feeding-trial protocols, e.g., in number and size of prey consumed, holding temperature, and so on, may have swamped any relationship.

This article is protected by copyright. All rights reserved.

Accepted Article

Lack of support for long-held beliefs about taxonomically-based intrinsic differences in half-life (Fig. 6) is also surprising. Although differences in protocols could have affected this, the enormous range of half-lives for spiders, from fewer than 10 to more than 100 hours, and to some extent also for heteropterans, makes this extremely unlikely. An unfortunate corollary is that we seem unable at this point to predict the magnitude or even direction of differences in half-lives among taxa, and are therefore unable to abandon the expense and tedium of determining half-lives for the foreseeable future.

One thing we should do is to standardize the number of predator individuals per time interval in a feeding trial, a quantity that has varied widely (Fig.1). To insist on small sample sizes because large ones are too difficult, tedious, or time-consuming, is short-sighted, for it stints on data analysis after so much time and other resources were spent on sample collection, extraction, and amplification. If we accept smaller sample sizes we will lessen our confidence in half-life estimates, and in our ability to detect differences between them (Table 2). We advocate a sample size of 20, which was sufficient to statistically distinguish all half-life pairs in a suite of predators attacking one prey species in a thoroughgoing field study (Greenstone et al. 2010).

A complication that must be taken into account is the possible occurrence of scavenging and secondary predation (Harwood et al. 2001; Calder et al. 2005; Foltan et al. 2005; Juen & Traugott 2005; Sheppard et al. 2005), which will lead to false positives and thereby inflate the estimates of proportions of predators positive for a given prey species. Characteristics of the predator complex and differential half-lives of primarily and secondarily consumed prey may rule out the likelihood of detecting secondary predation (Harwood et al. 2001). Otherwise, the extent of secondary predation can be estimated by parallel gut-content analysis of intraguild predation

This article is protected by copyright. All rights reserved.

Accepted Article

(Sheppard et al. 2005) and a portion of the positives thereby removed from the analysis. Where scavenging is likely, one would have to estimate the availability, and rates of disappearance of, target prey carcasses (Foltan et al. 2005); alternatively, one could obtain an ‘index of carrion availability’ by gut-content analysis of obligate scavengers (Sunderland 1966), provided such animals are present in the system.

Absent improvements in our understanding of the relationship between the quantity of material in the gut and its relationship to feeding history, qualitative assays are apt to remain the foundation of molecular gut-content analysis for the foreseeable future. Quantitative PCR does enable exquisite measurements of quantity of analyte which, under laboratory conditions, correlate with the amount of prey consumed at t = 0 and the time since consumption of a fixed amount of prey (Weber & Lundgren 2009). But qPCR, like conventional PCR, cannot simultaneously supply information on prey size, number, and time since feeding of remains detected by a positive assay in a field-collected animal. Perhaps qPCR incorporating primers for different-sized amplicons will yield inroads into the time-since-feeding problem (Hoogendoorn & Heimpel 2001). Further, if the calculations of proportions positive among the taxa in the predator complex were based solely on the largest amplicons, half-lives might be made sufficiently short to place a limit on the possible number of prey represented by a positive.

An exciting prospect for the near future is the adoption of next-generation sequencing (NGS) and PCR (Gómez-Polo et al. 2013), which can be expected to facilitate more complete descriptions of more arthropod food webs. Nevertheless, at least as currently constructed, any food web delineated by molecular gut-content assay, including NGS-PCR, will represent simply a map of trophic links, with no clear route to measures of the strength of those links. Even if all

This article is protected by copyright. All rights reserved.

Accepted Article

of the biases inherent in NGS-PCR (Pompanon et al. 2012) can be calibrated and compensated for, something like the detectability half-life will be needed to weight the relative strengths of predator-prey links for differential loss of detectability.

Assay data weighting with the detectability half-life is an important tool but only the first step toward reaching the Holy Grail of molecular gut-content analysis: estimates of the number of prey individuals eaten per unit time by each member of the predator complex. If this is ever to be achieved it will take a concerted modeling effort that includes the myriad variables impinging on the amount and detectability of prey remains in a predator’s gut at any given time; this will, in turn, require us to collect the kinds of data required by the model, concurrently with the predators for assay. Preliminary models of this type have been written (Sopp et al. 1992; Naranjo & Hagler 2001). Refinements may well include the quantity of analyte remaining. If so, estimates of that other half-life, the analyte half-life, may finally come into their own.

CONCLUSIONS

We have entered a golden age in molecular gut-content analysis of predation, with a critical mass of investigators and time-tested assay protocols available at relatively affordable cost. The last few decades have seen breathtaking strides in sensitivity and specificity. With the development of qPCR and NGS, we are poised to enter new realms of data gathering. However, our ability to make use of molecular gut-content data has not kept pace with technical developments, and we risk overextending the data beyond what the techniques can support. Our purpose in this paper was to highlight the contribution that one refinement, the detectability

This article is protected by copyright. All rights reserved.

Accepted Article

half-life, can bring to this process. Above all we have tried to demonstrate that the half-life is not a mere construct or artifact: it is a demonstrable, universal phenomenon that we must measure and apply to correct our raw data, just as we must deal with false positives from scavenging and secondary predation. Further, we stress the importance of standardizing the numerous factors involved in deriving detectability half-lives, and in having large sample sizes to enable the detection of differences in detectability for different predator-prey pairs; this is simply good science.

We are the first to admit that the detectability half-life is not the final answer to the problem of achieving ecologically meaningful quantitation of gut-content data. Rather, it is one indispensable tool for continuing progress toward our goal. That will likely involve the collection of more kinds of data than heretofore, in support of comprehensive models relating gut contents to rates of predation

ACKNOWLEDGMENTS

We thank James Harwood, Anita Juen, Michael Keller, Bill Symondson, and Michael Traugott for alerting us to recent papers from their laboratories that we might have missed, and for clarifying some points in their publications. To them and to the many other colleagues who have discussed molecular gut-content analysis with us, we are deeply indebted. We also thank the anonymous reviewers and Editor for their suggestions, which greatly improved the manuscript.

This article is protected by copyright. All rights reserved.

Accepted Article

References

Agustí N, de Vicente MC, Gabarra R (1999) Development of sequence amplified characterized region (SCAR) markers of Helicoverpa armigera: a new polymerase chain reaction-based technique for predator gut analysis. Molecular Ecology 8, 1467-1474.

Agustí N, Unruh TR, Welter SC (2003a) Detecting Cacopsylla pyricola (Hemiptera: Psyllidae) in predator guts using COI mitochondrial markers. Bulletin of Entomological Research 93, 179185.

Agustí N, Shayler P, Harwood D, Vaughan IP, Sunderland KD, Symondson WOCS (2003b) Collembola as alternative prey sustaining spiders in arable ecosystems: prey detection within predators using molecular markers. Molecular Ecology 12, 3467-3475.

Anderson JF (1970) Metabolic rates of spiders. Comparative Biochemistry and Physiology 33, 51-72.

Bilde T, Toft S (1998) Quantifying food limitation of arthropod predators in the field. Oecologia 115, 54-58.

Blankenship IE, Yayanos AA (2005) Universal primers and PCR of gut contents to study marine invertebrate diets. Molecular Ecology 14, 891-899.

Boreau de Roincé C, Lavigne C., Mandarin J-F, Rollard C, Symondson WOC (2013) Earlyseason predation on aphids by winter-active spiders in apple orchards. Bulletin of Entomological Research 103, 148-154.

Brooke MM, Proske HO (1946) Precipitin test for determining natural insect predators of immature mosquitoes. Journal of the National Malaria Society 5, 45-56.

This article is protected by copyright. All rights reserved.

Accepted Article

Calder CR, Harwood JD, Symondson WOC (2005) Detection of scavenged material in the guts of predators using monoclonal antibodies: a significant source of error in measurement of predation? Bulletin of Entomological Research 95, 1-6.

Chacón JM, Landis DA, Heimpel GE (2008) Potential for biotic interference of a classical biological control agent of the soybean aphid. Biological Control 46, 216-225.

Chapman EG, Schmidt JM, Welch KD, Harwood JD (2013) Molecular evidence for dietary selectivity and pest suppression potential in an epigeal spider community in winter wheat. Biological Control 65, 72-86.

Chen Y, Giles KL, Payton ME, Greenstone MH (2000) Identifying key cereal aphid predators by molecular gut analysis. Molecular Ecology 9, 1887-1898.

Davey JS, Vaughan IP, King RA, Bell JR, Bohan DA, Bruford MW, Holland JM, Symondson WOC (2013) Intraguild predation in winter wheat: prey choice by a common epigeal carabid consuming spiders. Journal of Applied Ecology 50, 271-279.

Dempster JP (1960) A quantitative study of the predators on the eggs and larvae of the broom beetle, Phytodecta olivacea Forster, using the precipitin test. Journal of Animal Ecology 29, 149-1667.

Eitzinger B, Traugott M (2011) Which prey sustains cold-adapted invertebrate generalist predators in arable land? Examining prey choices by molecular gut-content analysis. Journal of Applied Ecology 48, 591-599.

This article is protected by copyright. All rights reserved.

Accepted Article

Eskelson MJ, Chapman EG, Archbold DD, Obrycki JJ, Harwood JD (2011) Molecular identification of predation by carabid beetles on exotic and native slugs in a strawberry agroecosystem. Biological Control 56, 245-253.

Eitzinger B, Micic A, Körner M, Traugott M, Scheu S (2013) Unveiling soil food web links: New PCR assays for detection of prey DNA in the gut of soil arthropod predators. Soil Biology & Biochemistry 57, 943-945.

Fichter BL, Stephen WP (1981) Time-related decay in prey antigens ingested by the predator Podisus maculiventris (Hemiptera: Pentatomidae) as detected by ELISA. Oecologia 51, 404407.

Fichter BL, Stephen WP (1984) Time-related decay in prey antigens ingested by arboreal spiders as detected by ELISA. Environmental Entomology 13, 1583-1587.

Firlej A, Doyon J, Harwood JD, Brodeur J (2013) A multi-approach study to delineate interactions between carabid beetles and soybean aphids. Environmental Entomology 42, 89-96.

Foltan P, Sheppard S, Konvicka M, Symondson WOC (2005) The significance of facultative scavenging in generalist predator nutrition: detecting decayed prey in the guts of predators using PCR. Molecular Ecology 14, 4147-4158.

Fournier V, Hagler JR, Daane KM, de Léon JH, Groves R (2008) Identifying the predator complex of Homalodisca vitripennis (Hemipter: Cicadellidae): A comparative study of the efficacy of an ELISA and PCR gut content assay. Oecologia 157, 629-640.

Fournier V, Hagler J, Daane K, de Léon J, Groves RL, Costa HS, Henneberry TJ (2006) Development and application of a glassy-winged sharpshooter egg-specific predator gut content ELISA. Biological Control 37, 108-118.

This article is protected by copyright. All rights reserved.

Accepted Article

Frank JH (1967) A serological method used in the investigation of the predators of the pupal stage of the winter moth, Operophtera brumata (L.) (Hydromeniidae). Quaestiones Entomologicae 3, 95-105.

Gagnon AÈ, Heimpel GE, Brodeur J (2011a) The ubiquity of intraguild predation among predatory arthropods. PLoS ONE 6, e28061.

Gagnon AÈ, Doyon J, Heimpel GE, Brodeur J (2011b) Prey DNA detection success following digestion by intraguild predators: influence of prey and predators. Molecular Ecology Resources 11, 1022-1032.

Gómez-Polo P, Alomar O, Castañé C, Lundgren JG, Piñol J, Agustí N (2013) Understanding predation of Orius majuscules (Reuter) (Hemiptera: Anthocoridae): a comparison of conventional PCR, real-time PCR, and next-generation sequencing technologies. Proceedings of the 2nd International Symposium on the Molecular Detection of Trophic Interactions, 13-17 May 2013, Lexington, Kentucky, USA.

Grace RC, Wallace WH (1969) Satiation effects of TV commercials. Journal of Advertising Research 9, 3-8.

Greenstone MH (1978) The numerical response to prey availability of Pardosa ramulosa (McCook) (Araneae: Lycosidae) and its relationship to the role of spiders in the balance of nature. Symposia of the Zoological Society of London 42, 183-193.

Greenstone MH (1979) Spider feeding behaviour optimises dietary essential amino acid composition. Nature 181, 501-503.

Greenstone MH (1983) Site-specificity and site-tenacity in a wolf spider: a serological dietary analysis. Oecologia 56, 79-83.

This article is protected by copyright. All rights reserved.

Accepted Article

Greenstone MH (1996). Serological analysis of arthropod predation: past, present and future. In: The Ecology of Agricultural pests - Biochemical Approaches (eds. Symondson WOC, Liddell E) pp. 265-300, Chapman and Hall, London, UK.

Greenstone MH, Bennett AF (1980) Foraging strategy and metabolic rate in spiders. Ecology 61, 1255–1259.

Greenstone MH, Morgan CE (1989) Predation on Heliothis zea (Lepidoptera: Noctuidae): an instar-specific ELISA for stomach analysis. Annals of the Entomological Society of America 82, 45-49.

Greenstone MH, Hunt JH (1993) Determination of prey antigen half-life in Polistes metricus using a monoclonal antibody-based immunodot assay. Entomologia Experimentalis et Applicata 68, 1-7.

Greenstone MH, Rowley DR, Weber DC Hawthorne DJ (2007) Feeding mode and prey detectability half-lives in molecular gut-content analysis: An example with two predators of the Colorado potato beetle. Bulletin of Entomological Research 97, 201-209.

Greenstone MH, Szendrei Z, Rowley DL, Payton ME, Weber DC (2010) Choosing natural enemies for conservation biological control: use of the prey detectability half-life to rank key predators of Colorado potato beetle. Entomologia Experimentalis et Applicata 136, 97-107.

Hagler JR, Cohen AC (1990) Effects of time and temperature on digestion of purified antigen by Geocoris punctipes (Hemiptera: Lygaeidae) reared on artificial diet. Annals of the Entomological Society of America 83, 1177-1180.

Hagler JR, Naranjo SE (1994a) Determining the frequency of heteropteran predation on sweet potato whitefly and pink bollworm using multiple ELISAs. Entomologia Experimentalis et Applicata 72, 59-66.

This article is protected by copyright. All rights reserved.

Accepted Article

Hagler JR, Naranjo SE (1994b) Qualitative survey of two coleopteran predators of Bemisia tabaci (Homoptera: Aleyrodidae) and Pectinophora gossypiella (Lepidoptera: Gelechiidae) using a multiple prey gut content ELISA. Biological Control 83, 193-197.

Hagler JR, Naranjo SE (1997) Measuring the sensitivity of an indirect predator gut content ELISA: detectability of prey remains in relation to predator species, temperature, time, and meal size. Biological Control 9, 112-119.

Harper GL, King RA, Dodd CS, Harwood JD, Glen DM, Bruford MW, Symondson WOC (2005) Rapid screening of invertebrate predators for multiple prey DNA targets. Molecular Ecology 14, 819-827.

Harper GL, Sheppard SK, Harwood JD, Read DS, Glen DM, Bruford MW, Symondson WOC (2006) Evaluation of temperature gradient gel electrophoresis for the analysis of prey DNA within the guts of invertebrate predators. Bulletin of Entomological Research 96, 295-304.

Harwood JD, Phillips SW, Sunderland KD, Symondson WOC (2001) Secondary predation: quantification of food chain errors in an aphid-spider-carabid system using monoclonal antibodies. Molecular Ecology 10, 2049-2057.

Harwood JD, Sunderland KD, Symondson WOC (2004) Prey selection by linyphiid spiders: molecular tracking of the effects of alternative prey on rates of aphid consumption in the field. Molecular Ecology 13, 3549-3560.

Harwood JD, Desneux N, Yoo HJS, Rowley DL, Greenstone MH, Obrycki JJ, O’Neil RJ (2007a) Tracking the role of alternative prey in soybean aphid predation by Orius insidiosus: a molecular approach. Molecular Ecology 16, 4390-4400.

Harwood JD, Bostrom MR, Hladilek EE, Wise DH, Obrycki JJ (2007b) An order-specific monoclonal antibody reveals the impact of alternate prey on spider feeding behavior in a complex food web. Biological Control 41, 397-407.

This article is protected by copyright. All rights reserved.

Accepted Article

Hoogendoorn M, Heimpel G (2001) PCR-based gut content analysis of insect predators: using ribosomal ITS-1 fragments from prey to estimate predation frequency. Molecular Ecology 10, 2059-2067.

Hosseini R, Schmidt O, Keller MA (2008) Factors affecting detectability of prey DNA in the gut contents of invertebrate predators: a polymerase chain reaction-based method. Entomologia Experimentalis et Applicata 126, 59-66.

Jamarillo J, Chapman EG, Vega FE, Harwood JD (2010) Molecular diagnosis of a previously unreported predator-prey association in coffee: Karyothrips flavipes Jones (Thysanoptera: Phlaeothripidae) predation on the coffee berry borer. Naturwissenschaften 97, 291 -298.

Juen A, Traugott M (2005) Detecting predation and scavenging by DNA gut-content analysis: a case study using a soil insect predator-prey system. Oecologia 142, 344-352.

Juen A, Traugott M (2007) Revealing species-specific trophic links in soil food webs: molecular identification of scarab predators. Molecular Ecology 16, 1545-1557.

Kerzicnik LM, Chapman EG, Harwood JD, Peairs FB, Cushing PE (2012) Molecular characterization of Russian wheat aphid consumption by spiders in winter wheat. Journal of Arachnology 40, 71-77.

King RA, Read DS, Traugott M, Symondson WOC (2008) Molecular analysis of predation: a review of best practice for DNA-based approaches. Molecular Ecology 17, 947-963.

King RA, Vaughan IP, Bell JR, Bohan DA, Symondson WOC (2010) Prey choice by carabid beetles feeding on an earthworm community analysed using species- and lineage-specific PCR primers. Molecular Ecology 19, 1721-1732.

This article is protected by copyright. All rights reserved.

Accepted Article

Kobayashi T, Takada M, Takagi S, Yoshioka A, Washitani I (2011) Spider predation on a mirid pest in Japanese rice fields. Basic and Applied Ecology 12, 532-539.

Kuusk A-K, Cassel-Lundhagen A, Kvarheden A, Ekbom B (2008) Tracking aphid predation by lycosid spiders in spring-sown cereals using PCR-based gut-content analysis. Basic and Applied Ecology 9, 718-725

Lövei GL, Sopp P, Sunderland KD (1987) The effect of mixed feeding on the digestion of the carabid Bembidion lampros. Acta Pathologica Entomologica Hungarica 22, 403-407.

Lövei GL, Sopp P, Sunderland KD (1990) Digestion rate in relation to alternative feeding in three species of polyphagous predators. Ecological Entomology 15, 293-300.

Lundgren JG, Weber DC (2010) Changes in digestive rate of a predatory beetle over its larval stage: implications for dietary breadth, Journal of Insect Physiology 56, 431-437.

Lundgren JG, Ellsbury ME, Prischmann DA (2009) Analysis of the predator community of a subterranean herbivorous insect based on polymerase chain reaction. Ecological Applications 19, 2157-2166.

Ma J, Li D, Keller M, Schmdt O, Feng X (2005) A DNA marker to identify predation of Plutella xylostella (Lep., Plutellidae) by Nabis kinbergii (Hem., Nabidae) and Lycosa sp. (Araneae, Lycosidae). Journal of Applied Entomology 129, 330-335,

McIver JD (1981) An examination of the utility of the precipitin test for evaluation of predatorprey relationships. Canadian Entomologist 113, 194-201.

McMillan S, Kuusk A-K, Cassel-Lunhagen A, Ekbom B (2007) The influence of time and temperature on molecular gut-content analysis: Adalia bipunctata fed Rhopalosiphum padi. Insect Science 14, 353-358.

This article is protected by copyright. All rights reserved.

Accepted Article

Moreno-Ripoll R, Gabarra R, Symondson WOC, King RA, Agustí N (2012) Trophic relationships between predators, whiteflies and their parasitoids in tomato greenhouses: a molecular approach. Bulletin of Entomological Research 102, 415-423

Naik PA (1999) Estimating the half-life of advertisements. Marketing Letters 10, 351-362. Naranjo SE, Hagler JR (2001) Toward the quantification of predation with predator gut immunoassays: a new approach integrating functional response behavior. Biological Control 20, 175-189.

Opatovsky I, Chapman EG, Weintraub PG, Lubin Y, Harwood JD (2013) Molecular characterization of the differential role of immigrant and agrobiont generalist predators in pest suppression. Biological Control 63, 25-30.

Payton ME, Greenstone MH, Schenker N (2003) Overlapping confidence intervals or standard error intervals: What do they mean in terms of statistical significance? 6pp. Journal of Insect Science 3, 34, Available online: insectscience.org/3.34.

Pianezzola E, Roth S, Hatteland BA (2013) Predation by carabid beetles on the invasive slug Arion vulgaris in an agricultural semi-field experiment. Bulletin of Entomological Research 103, 225-232.

Pompanon F, Deagle BE, Symondson WOC, Brown DS, Jarman SJ, Taberlet P (2012) Who is eating what: diet assessment using next generation sequencing. Molecular Ecology 21, 1931-1950.

Pons J (2006) DNA-based identification of preys from non-destructive, total DNA extraction using arthropod universal primers. Molecular Ecology Notes 6, 623-626.

Ragsdale DW, Larson AD, Newsome LD (1981). Quantiative assessment of the predators of Nezara viridula eggs and nymphs within a soybean agroecosystem using an (ELISA). Environmental Entomology 10, 402-405.

This article is protected by copyright. All rights reserved.

Accepted Article

Read DS, Sheppard SK, Bruford MW, Glen DM, Symondson WOC (2006) Molecular detection of predation by soil micro-arthropods on nematodes. Molecular Ecology 15, 1963-1972.

Romeu-Dalmau C, Piñol J, Agustí N (2012) Detecting aphid predation by earwigs in organic citrus orchards using molecular markers. Bulletin of Entomological Research 102, 566-572.

Schmidt JM, Harwood JD, Rypstra AL (2012) Foraging activity of a dominant epigeal predator: molecular evidence for the effect of prey density on consumption. Oikos 1212, 1715-1724.

Sheppard SK, Bell J, Sunderland KD, Fenlon J, Skervin D, Symondson WOC (2005) Detection of secondary predation by PCR analyses of the gut contents of invertebrate generalist predators. Molecular Ecology 14, 4461-468.

Sigsgaard L, Greenstone MH, Duffield SJ (2002). Egg cannibalism in Helicoverpa armigera Hübner (Lepidoptera: Noctuidae) on sorghum and pigeonpea. BioControl 47,151-165.

Sint DA, Raso LA, Kaufmann R, Traugott M (2011) Optimizing methods for PCR-based analysis of predation. Molecular Ecology Resources 11, 795-801.

Sopp PI, Sunderland KD (1989) Some factors affecting the detection period of aphid remains in predators using ELISA. Entomologia Experimentalis et Applicata 51, 11-20.

Sopp PI, Sunderland KD, Fenlon JS, Wratten SD (1992) An imporved quantitative method for estimating invertebrate predation in the field using an enzyme-linked immunosrbent assay (ELISA). Journal of Applied Ecology 29, 295-302.

Stuart MK, Greenstone MH (1990) Beyond ELISA: a rapid, sensitive, specific immunodot assay for identification of predator stomach contents. Annals of the Entomological Society of America 83,1101-1107.

This article is protected by copyright. All rights reserved.

Accepted Article

Sunderland KD (1996). Progress in quantifying predation using antibody techniques. In: The Ecology of Agricultural pests - Biochemical Approaches (eds. Symondson WOC, Liddell E) pp. 419-455, Chapman and Hall, London, UK.

Sunderland KD, Crook NE, Stacey DL, Fuller BJ (1987). A study of feeding by polyphagous predators on cereal aphids using ELISA and gut dissection. Journal of Applied Ecology 24, 907-933.

Symondson WOC (2002) Molecular identification of prey in predator diets. Molecular Ecology 11, 627-641.

Symondson WOC, Liddell JE (1993) Differential antigen decay rates during digestion of molluscan prey by carabid predators. Entomologia Experimentalis et Applicata 69, 277 - 287.

Symondson WOC, Liddell JE (1995) Decay rates for slug antigens within the carabid predator Pterostichus melanarius monitored with a monoclonal antibody. Entomologia Experimentalis et Applicata 75, 245 – 250.

Symondson WOC, Erickson ML, Liddell JE, Jayawardena KGI (1999) Amplified detection, using a monoclonal antibody, of an aphid-specific epitope exposed during digestion in the gut of a predator. Insect Biochemistry and Molecular Biology 29, 873-883.

Symondson WOC, Glen DM, Erickson ML, Liddell JE, Langdopn CJ (2000) Do earthworms help to sustain the slug predator Pterostichus melanarius (Coleoptera: Carabidae) within crops? Investigations using monoclonal antibodies. Molecular Ecology 9, 1279-1292.

Szendrei Z, Greenstone MH, Payton ME, Weber DC (2010) Molecular gut-content analysis of a predator assemblage reveals the effect of habitat manipulation on biological control in the field. Basic and Applied Ecology 11, 153-161.

Traugott M, Symondson WOC (2008) Molecular analysis of predation on parasitized hosts. Bulletin of Entomological Research 98, 239-247.

This article is protected by copyright. All rights reserved.

Accepted Article

Traugott M, Bell JR, Raso L, Sint D, Symondson WOC (2012) Generalist predators disrupt parasitoid aphid control by direct and coincidental intraguild predation. Bulletin of Entomological Research 102, 223-231.

Virant-Doberlet M, King A, Polajnar J, Symondson WOC (2011) Molecular diagnostics reveal spiders that exploit prey vibrational signals used in sexual communication. Molecular Ecology 20, 2204-2216.

von Berg K, Traugott M, Symondson WOC, Scheu S (2008) The effects of temperature on detection of prey DNA in two species of carabid beetle. Bulletin of Entomological Research 98, 263-269.

von Berg K, Traugott M, Scheu S (2012) Scavenging and active predation in generalist predators: A mecocosm study employing DNA-based gut content analysis. Pedobiologia 55, 1-5.

Waldner T, Sint D, Juen A, Traugott M (2013) The effect of predator identity on post-feeding prey DNA detection success in soil-dwelling macro-invertebrates. Soil Biology and Biochemistry. 63, 116-123.

Weber DC, Lundgren JG (2009) Detection of predation using qPCR: Effect of prey quantity, elapsed time, chaser diet, and sample preservation on detectable quantity of prey DNA. 12pp. Journal of Insect Science 9:41, available online: insectscience.org/9.41.

Waldner T, Traugott M (2012) DNA-based analysis of regurgitates: a non-invasive approach to examine the diet of invertebrate consumers. Molecular Ecology Resources 12, 669-675.

Waldner T, Sint D, Juen A, Traugott M (2013) The effect of predator identity on post-feeding DNA detection success in soil-dwelling macro-invertebrates. Soil Biology & Biochemistry 63, 116-123.

This article is protected by copyright. All rights reserved.

Accepted Article

Wilder SM (2011) Spider nutrition: An integrative perspective. Advances in Insect Physiology 40, 87-136.

Zaidi RH, Jaal Z, Hawkes NJ, Hemingway J, Symondson WOC (1999) Can multiple-copy sequences of prey DNA be detected amongst the gut contents of invertebrate predators? Molecular Ecology 8, 2081-2087.

Zhang G-F, Lü Z-C, Wan F-H (2007) Detection of Bemisia tabaci remains in predator guts using a sequence-characterized amplified region marker. Entomologia Experimentalis et Applicata 123, 81-90.

Figure legends

Figure 1. Frequencies of studies employing different numbers of predators per interval postfeeding in half-life feeding trials. Data are from the publications listed in Table 1.

Figure 2. Frequencies of studies employing different numbers and types of prey or their tissues in half-life feeding trials. The left-most categories satisfy the assumption of equal meal size at t = 0, with precision decreasing right-ward; those to the right of the gap cannot, by definition, satisfy the assumption. Data are from the publications listed in Table 1.

Figure 3. Frequencies of studies employing different holding temperatures for predators in halflife feeding trials, by predator taxon. RT = room temperature. Data are from the publications listed in Table 1.

This article is protected by copyright. All rights reserved.

Accepted Article

Figure 4. Frequencies of amplicon sizes used in PCR assays to determine detectability halflives. Data are from the publications listed in Table 1. OTHERS comprise two chrysopids and one each acarine, collembolan, dermapteran, geophilid, and thysanopteran.

Figure 5. Relationship of amplicon size and detectability half-life (T50). Data are from the

publications listed in Table 1.

Figure 6. Frequencies of detectability half-lives (T50), by predator taxon. OTHERS comprise two chrysopids and one each acarine, collembolan, dermapteran, geophilid, thysanopteran, and vespid.

Figure 7. Taxonomic distribution of arthropod predators for which half-lives have been determined. Data are from the publications listed in Table 1.

Figure 8. Taxonomic distribution of arthropod prey for which half-lives have been determined. Data are from the publications listed in Table 1.

This article is protected by copyright. All rights reserved.

Accepted Article

Author Contributions

This paper grew out of long-term collaborations among the authors. It embodies Mark Payton’s expertise in statistics; and Matthew Greenstone’s, Donald Weber’s, and Alvin Simmons’s in analysis of predator-prey interactions, arthropod rearing, and biological control. Table 1. Molecular gut-content studies employing the detectability half-life, or conceptually similar metrics, since its formulation by Greenstone & Hunt (1993). bp = amplicon size, base pairs; Ch= chaser prey used; N/t = number of predators fed per time interval; PrTx = prey taxon or taxa; PrdTx = predator taxon or taxa; T = temperature (0C) at which animals were held;T50 =

detectability half-life in hours.

Nomenclature

Meal

PrTx

PrdT x

T

C bp h

N /t

N o

720

Detectability half-life

Ad lib homogenate

Noct

Vesp

Fld

Median prey detection interval

1, 3, 6, or 10 eggs

Gele

Cocc

1535

1 egg

Aph

Detectability half-life

1 – 5 nymphs

Detection half-life

744

Anth

4.0, 8.8

188, 271

22

Plut RT

Car

16

N o

Aph

Car, Lin

Cica

Cocc, Chry

Immunod ot

Expone ntial

Greenstone & Hunt 1993

ELISA

Expone ntial

Hagler & Naranjo 1997

Probit

Chen et al. 2000

16

25

Specific PCR

Specific PCR

Linear Agustí et al. 2003

N o

17.5, 95.9

Probit Specific PCR

Ma et al. 2005 Linear

26.388.5

Mutiplex PCR

110, 245

Harper et al. 2005 Linear

14 3 or 5 eggs

Referen ce

78242 510

ad lib. aphids

Mod el

275 812

Aph, Agr, Ario, Lum

20.9, 24.1

N o Lyc, Nab

3 aphids; ad lib slug or earthworm

0.918.0

20

5 1 or 2 caterpillars

19.4

Assa y

198 Fld

Psyl

Half-life for detection

Median detection time/period

NA N o

Y es

Detectability half-life

Detection half-life

Cocc, Chry

NA

T50

24.159.8

N

This article is protected by copyright. All rights reserved.

Specific PCR

Logit

Sheppard et al. 2005

o

Accepted Article

Detection period

Detectability half life

Detectability half-life

Decay rate for prey DNA

Prey DNA detection success

Detection half-life

Time for median detection success

ad lib. aphids

Aph, Lum

Car

Hete, Rhab, Stei,

1 egg Chry

Acar, Coll

20 11.8, 34.4

25 Y es

ad lib. nematodes

NA*

332

ELISA

16

Fld

Fournier et al. 2006

Expone ntial

Harper et al. 2006

Probit

Read et al. 2006

14 TGGE 12.0

Cocc, Pen

Expone ntial

N o

150203

10

8.6

Specific PCR

20 214 1 egg or larval piece

16 Scar

Car

1 aphid

N o 291

343

331

10

7.0, 50.9

Specific PCR

Not fit Greenstone et al. 2007

10 Aph

Y es

Cocc

3 eggs

18

Specific PCR

Logit Juen & Traugott 2007

25 Cica

Chr, Cocc, Red

1 caterpillar

N o

4.9 197

20

Specific PCR

624

Logit/Lin ear McMillan et al. 2007

Plut 11.0 Cocc, Lyc, Nab

1 aphid

Estimated time for 50% prey detection success

N o

Logistic

293 Specific PCR

5

15

Aph

Lyc

1 egg

25

N o

17.149.6

331 1012

Curc

3.7

145 ~ 0.01 g tissue

16

Lum

Y es

Car Agr, Ario

Car

1 adult Mir

109, 310

16, 25

Detectability half-life

9.619.7

Y es Chry

246

Fld

Specific PCR

Probit

Specific PCR

Kuusk et al. 2008

Binomial Jaramillo et al. 2010

8

Lyc, Tetr

1 egg

Hosseini et al. 2008

10

N o

Median detection time

24.6

116256

14

Probit

15

Median detection time

2 slugs

Fournier et al. 2008

Specific PCR

Detectability half-life

Thy

Probit

4

19.7, 22.4

Specific PCR

Logistic King et al. 2010

Half-life of detection

~0.05g tissue Agr

Detectability half-life

Car, Cocc, Pen

Probit

N o 214

21

20

Hatteland et al. 2011 Probit

1-5 eggs Cocc

Prey DNA detectability success

1/3 beetle larva; 1 cricket

Gryl, Tene

Car

Cocc

24

10, 15

N o

175

10 7.084.4

N o

105137

½ adult

Cica

Specific PCR Logit

Kobayashi et al. 2011

10 5.4, 7.2

Probability of a 50% detection success

Median detection time

36.5193.7

Mutliplex PCR

Specific PCR

Logit

Greenstone et al. 2010

10 Car, Lyc

16

Y es

116612

5.219.3

This article is protected by copyright. All rights reserved.

Specific PCR

Linear Eskelson et al.

1 aphid

Aph

Ther

24/ 20

Accepted Article

Molecular half-life

Y es 3-5 aphids

Aph

2011

6 289, 348

Lyc, Tetr

30.079.2

Specific PCR

Probit Gagnon et al. 2011b

8 25

Half-life of prey DNA

1 beetle larva

Scar

N o

Derm

227 10

65.8, 72.8

Specific PCR

Expone ntial Sint et al. 2011

16

50% Prey detection success 224 1 collembolan

Coll, Plat

Car 24

N o

Logit 10

2.0, 4.2

Specific PCR

Virant-Doberlet et al. 2011

Detectability half-life

1 spider

Lin

127853

Lin, Tetr 16

Median detection time

23.8

N o

Specific PCR Logit

ad lib. aphids

Aph

180

Car 20

810

¼ beetle larva

Scar

Specific PCR 145318

Car 6, 16

10

Geo, Hist

Specific PCR 180

N o

246

Logit

Romeu-Dalmau et al. 2012

9.5, 32.0

N o Car, Can, Elat,

Kerzicnik et al. 2012

21.930.0

N o

Prey DNA detectability success

Fifty per-cent detection probability

Probit 10

Logit Waldner & Traugott 2012

19.742.6 Specific PCR

127891

Chapman et al. 2013

12.3 N o

Specific PCR 11.497.8

Davey et al. 2013

N o

Specific PCR

Firlej et al 2013

N o

Specific PCR

Waldner et al. 2013

N o

Prey Taxa: Agr = Agriolimacidae; .Aph = Aphidae; Ario = Arionidae; Chry = Chrysomelidae; Cicad = Cicadellidae; Cocc = Coccinellidae; Coll = Collembola; Curc = Curculionidae; Gele = Gelechiidae; Gryl = Gryllidae; Hete = Heterorhabditidae; Lin = Linyphiidae; Lum = Lumbricidae; Mir = Miridae; Noct = Nocutidae; Plat = Platygastridae; Plut = Plutellidae; Psyl = Psyllidae; Rhab = Rhabditidae; Sacr = Scarabeaidea; Stei = Steinemematidae; Tene = Tenebrionidae

Predator taxa: Acar = Acarina; Anth=Anthocoridae; Can = Cantharidae; Car = Carabidae; Chr = Chrysopidae; Cocc = Coccinellidae; Coll = Collembola; Derm = Dermaptera; Elat = Elateridae; Geo = Geophilidae; Hist = Histeridae; Lin = Linyphiidae; Lith = Lithobiidae; Lyc = Lycosidae; Nab = Nabidae;

This article is protected by copyright. All rights reserved.

Accepted Article

Pen = Pentatomidae; Red = Reduviidae; Tet = Tetragnathidae; Ther = Theridiidae; Thy = Thysanoptera; Vesp = Vespidae

**NA = Not applicable because serological assay

***Fld = field fluctuating temperatures; 24/20 = day/night temperatures

`Table 2. Mean, standard deviation (SD), and SD10/SD20 ratios for 20- and 10-animal-per-postfeeding interval simulations.

Species, stage

20-animal

10-animal

Mean

SD

Mean

Perillus bioculatus, adult

60.93

6.31

61.60 10.49

1.66

Perillus bioculatus, nymph

84.55

6.60

84.40

9.67

1.47

Podisus maculiventris, adult

17.62

2.54

17.45

3.70

1.46

Podisus maculiventris, nymph

51.68

5.90

52.34

9.23

1.56

Coleomegilla maculata, adult

26.52

3.23

26.57

4.82

1.49

Coleomegilla maculata, larva

7.00

0.79

6.94

1.13

1.43

Lebia grandis, adult

8.60

0.66

8.62

0.96

1.45

This article is protected by copyright. All rights reserved.

SD

SD10/SD20

Accepted Article This article is protected by copyright. All rights reserved.

Accepted Article This article is protected by copyright. All rights reserved.

Accepted Article This article is protected by copyright. All rights reserved.

Accepted Article This article is protected by copyright. All rights reserved.

The detectability half-life in arthropod predator-prey research: what it is, why we need it, how to measure it, and how to use it.

Molecular gut-content analysis enables detection of arthropod predation with minimal disruption of ecosystem processes. Most assays produce only quali...
352KB Sizes 0 Downloads 0 Views