Consequences of Linguistic Uncertainty for Insect Management

Home , Acephate, Aphytis melinus, Chrysoperla rufilabris, Hippodamia washingtoni, Liviidae, Sogatella

WATCH YOUR LANGUAGE

THERESA M. CIRA, KATHY QUICK, AND ROBERT C. VENETTE

JA'CRISPY/ISTOCK

258 AMERICAN ENTOMOLOGIST | WINTER 2019 his is a call to entomologists to consider the uncertainty introduced into our works through the language we use. Albert Einstein remarked, “Every- thing depends on the degree to which words and word-combinations correspond to the world of im- Tpression,” which makes language a “dangerous source of error and deception” (Hawking 2007). Scientists usually research and deliberately choose the language they use to propose new concepts or terminology. Often, however, terms slip into the lexicon through less systematic means. Words may seem to have meanings so obvious that broad understanding of the term may be taken for granted, but individual contexts are diverse, and to assume that a word’s meaning is unchanging across time and space is to overlook the uncertainties of language. Thus, communication through even the most common entomological vocabulary can fail. Language is a mutable, uncertain, and imperfect way of representing the world. Because language is integral to science and yet inherently imprecise, it is important for scientists to recognize and address uncertainty in the language of our works.

Uncertainty, as defined by the Society a phenomenon is repeatedly observed and for Risk Analysis (2015), is “not knowing the individuals agree that measurements con- true value of a quantity or the future con- verge, to some degree, on a specific point, sequences of an activity,” or “imperfect or more certainty about the true nature of the incomplete information/knowledge about phenomenon can be asserted. However, a hypothesis, a quantity, or the occurrence that agreement assumes a common use of of an event.” This state of not knowing language to describe phenomena. arises from a lack of knowledge, inherent Language is a “system for describing THERESA M. CIRA, variability in systems, and/or inconsisten- perceptions” (Zadeh 2004); however, lan- KATHY QUICK, AND cies introduced by language (i.e., epistemic guage is dynamic and in a constant state of ROBERT C. VENETTE uncertainty, aleatory uncertainty, and lin- change. In 2000, there were approximately guistic uncertainty, respectively; Hoffman 1,022,000 words in the English lexicon, and Hammonds 1994, Regan et al. 2002, nearly double the 544,000 words in 1900 Society for Risk Analysis 2015). All types (Michel et al. 2011). Not only do new words of uncertainty can impede effective deci- arise and unused words fall away, but one sion-making (Carey and Burgman 2008, word may have many meanings (polysemy), Milner-Gulland and Shea 2017). Certainty and the meanings of words change over about natural phenomena increases time (semantic change). Britton (1978) ana- through the convergence and agreement lyzed the 100 most frequently used English of perceptions among people and across words for polysemy and found, conserva- time (Eckhardt 1981), but will never be fully tively, that 32% of these words had more achieved (Popper 2012). The scientific pro- than one meaning, and the median number cess primarily addresses epistemological and of meanings for an ambiguous word was aleatory sources of uncertainty. The empiri- three. More than half (52%) of the English cal nature of the scientific process is, in part, lexicon was found to be completely undoc- what makes scientific knowledge useful. If umented in the Oxford English Dictionary,

AMERICAN ENTOMOLOGIST | WINTER 2019 259 so-called “dark matter” (Michel et al. 2011). Table 1. The five types of linguistic uncertainty according to Regan et al. (2008), with entomologically Similar challenges affect communication in relevant examples. any language. Ambiguity When a word has more than The term “significant” has multiple Within the context of pest management, one meaning and the intended meanings (e.g., statistically the uncertainty of some terms has been meaning is not stated. significant, or subjectively perceived noted or examined. Horowitz et al. (2009) to be important), and often the introduce their book on biorational control intended meaning is not stated. of pests by noting the confusion caused by a Context dependency When the context a statement The sentence, “Syrphid flies are lack of a standardized definition of the term should be understood in is not predators” lacks context because it “biorational.” Eilenberg et al. (2001) found stated. does not specify which life stage (the that different disciplines that encompass larva) is predatory. biological control research use different Underspecificity When the necessary degree To say “the specimen was found terms, which has led to confusion among of specificity to understand a east of Minneapolis” is underspecific people such as extension agents and regula- statement is not stated. because it would not be possible tory officials. Crump et al. (1999) identified to determine exactly where the confusing terminology related to biological specimen was found. weed control and pointed out that the con- Indeterminacy of When the meaning of a term In the 16th century the term “pest” notations of certain words, such as “chemi- theoretical terms changes over time. arose from the term pestilence, or cal” and “synthetic,” could affect the public a fatal epidemic, specifically the bubonic plague, and “pest” referenced trust and, ultimately, the fate of one type the plague or a contagious disease. of control over another. They concluded Through semantic change, “pest” has that allowing terms to be defined individu- since come to mean noxious people ally in each paper, rather than insisting on or things (Onions et al. 1966). a common definition, complicates com- Vagueness When a sharp boundary cannot In the spectrum of visible light, it is munication and increases confusion, and be drawn around the meaning difficult to draw a sharp boundary that “Clear, meaningful terminology is one of a term. around the specific wavelengths that essential step in gaining public confidence.” constitute the color red. The occurrence and consequences of linguistic uncertainty in entomology have rarely been considered, meriting atten- Fig. 1. Relative frequency of use over time of the term “broad-spectrum insecticide” compared to other tion as potentially significant barriers to tri-grams (three-word phrases) in books scanned by Google. the development of concepts and effective communication within the discipline. Do ambiguous and indeterminate terms provide flexibility, or do they increase confusion? Table 1, a summary of a taxonomy of linguistic uncertainty proposed by Regan et al. (2008), illustrates this point with entomologically relevant examples. We intend to illustrate how linguistic uncertainty may affect the strength, validity, reproducibility, and long-term stability of research. Our assertion that linguistic uncertainty is com- monplace and consequential is illustrated through an analysis of how insecticides have been described and categorized in entomological publications, particularly with respect to broad-spectrum insecticides. enables us to call an ant that has lost two Insecticides are categorized and legs an insect. This individually unique and described in a variety of ways. In some Case Study: Insecticide Categoriza- vast prior knowledge builds the matrix for cases, categories have been structured by tion and Broad-Spectrum Insecticides a person’s mental representation of a typi- governments, non-governmental certi- Categories can be shortcuts to convey cal member in a category, and guides where fying agencies, or research communities meaning without extensive description. they believe the delimiting bounds of the (Table 2). Regulating bodies can outline The concept of a specific category, such category should be (Taylor 1995). Individual the category or set of categories and/or as “insect,” is not independent from a per- understandings of a category are never certify which insecticides are in different son’s acquaintance with members of that identical, yet, in general, the overlap of indi- categories. In addition to formalized lists category. Although a dictionary definition vidually created meanings for a category is of categories, less-structured categories of “insect” indicates an animal that has six substantial enough to communicate with are commonly used to describe or classify legs, familiarity with insects and the world an adequate level of precision and accuracy. insecticides (e.g., “biorational,” “botanical,”

260 AMERICAN ENTOMOLOGIST | WINTER 2019 “soft,” “persistent”). These less-structured Table 2. Categories of insecticides defined or certified by an organization, with aspects of linguistic categorization schemes sometimes make uncertainty noted. sense and are useful. However, there is neither consensus about what these cate- Terms Regulating Characteristics on which gories mean nor a consistent set of guide- body* membership is based lines for placing different insecticides in Reduced risk US EPA Risk in comparison to alternatives (USEPA 2016) given categories. No categorization system captures the entire profile of an insecticide, Organic OMRI Physical components of insecticide (OMRI 2016) from its chemical makeup to its effects on various living organisms. Even so, assign- Insect physiological systems affected; mode of action Group, subgroup, classification informs the symptomology, speed of ing an insecticide to a category frames chemical class, active IRAC action, and other properties of the actives therein research results and implications, which, in ingredient turn, can strongly influence decision-mak- (IRAC 2017) ing and insect management practices. Level of harm (e.g., Harm to natural enemies, where harm is evaluated as harmless, slightly either mortality or a reduction in beneficial capacity It is important to understand, acknowl- IOBC edge, and consider the mechanisms used harmful) (i.e., parasitism) compared to a control (Hassan et al. 1994) to create categories for insecticides when choosing which insecticide categories to Containing naturally occurring substances that control use in scientific writing. pests (biochemical pesticides), microorganisms that Biopesticide US EPA control pests (microbial pesticides), or substances produced by plants containing added genetic material Materials and Methods (plant-incorporated protectants) (USEPA 2017) To investigate linguistic uncertainty in the categorization of insecticides, we explored *Acronyms: US EPA = United States Environmental Protection Agency; OMRI = Organic Materials Review Institute; IRAC = Insecticide one unregulated category in detail: Resistance Action Committee; IOBC = International Organization for Biological Control “broad-spectrum.” We chose this term because of the prevalence and long history of its use in the entomological literature Implicit Definitions of “Broad-Spectrum evaluative use (i.e., associating a negative (Fig. 1). We surveyed literature containing Insecticide” or positive suite of characteristics). In the the term “broad-spectrum” in conjunction Using the 60 recent papers referred to in evaluative-use category, we qualitatively with the word “insecticide,” with the follow- the previous paragraph, we recorded terms assessed positive and negative valences, ing questions in mind: that conveyed an inferred meaning to the assuming a human-centric, profit-driven • How is “broad-spectrum insecticide” category of “broad-spectrum insecticide.” perspective (see supplementary material for explicitly defined? These terms were taken from the sentence details on these methods). • How is “broad-spectrum insecticide” in which the phrase “broad-spectrum” was implicitly defined? used, or from the preceding or following Results and Discussion • Which insecticides are considered sentences (see supplementary material for We strongly emphasize that in this case “broad-spectrum”? details on these methods). study, we are not criticizing the cited liter- • What is the utility of the category ature or disparaging the language choices “broad-spectrum insecticide”? Membership in the “Broad-Spectrum of individual authors. Rather, we use the These questions helped us determine Insecticide” Category whole set of papers that matched our search whether the category “broad-spectrum From each of the aforementioned papers, criteria to draw attention to the collective insecticide” was clear, consistent, and useful. two things were recorded: 1) the categoriza- problem that arises from one linguistically tion of the insecticide(s) (groups, subgroups/ uncertain term. Explicit Definitions of “Broad-Spectrum chemical classes, and active ingredients), as Insecticide” defined by the Insecticide Resistance Action Explicit Definitions We sought the earliest publication of the Committee (IRAC); and 2) whether or not The term “broad-spectrum insecticide” term “broad-spectrum” in relation to insec- the insecticide(s) was/were considered has been integrated into the entomologi- ticide; our hypothesis was that at the time “broad-spectrum” based on the explicit and/ cal lexicon without a unifying definition, of its first use, a definition would have been or implicit language. whether in the oldest literature or in more proposed. We also searched a subset of stan- recent papers. The use of the term mark- dard entomological references in print and Utility of the “Broad-Spectrum Insecticide” edly increased around 1950 (Fig. 1). We on the Internet. Additionally, using 60 recent Category found no seminal paper that coined the (2006–2016) peer-reviewed papers from the We used thematic coding and categorizing term “broad-spectrum insecticide.” The top 10 entomology journals that employed to evaluate how the term “broad-spectrum earliest record in the electronically search- the term “broad-spectrum” to describe an insecticide” was used. The categories cov- able and verifiable literature that described insecticide, we examined their cited work for ered the full range of uses, with minimal an insecticide as “broad-spectrum” (Cooper explicit definitions (see supplementary mate- overlap or difficulty differentiating between 1958) was written as if the term was already rial at https://doi.org/10.1093/ae/tmz068 for them: 1) descriptive use; 2) comparative use understood in the lexicon. None of the 60 details about these methods). (i.e., in relation to another category); and 3) recent papers or the cited works within

AMERICAN ENTOMOLOGIST | WINTER 2019 261 Table 3. Subset of definitions for the term “broad-spectrum” in relation to insecticides. the term are also vague (i.e., defined with words that do not specify clear boundar- Term Definition ies). Using terms such as “wide range” in the Broad-spectrum Relating to pesticides, killing a wide range of target organisms (Torre- definition of “broad-spectrum insecticide” Bueno et al. 1989). leaves unanswered questions. Where is the Broad-spectrum pesticide Chemicals that affect pests and beneficial organisms (Gordh and cutoff for “wide”? More than one species? Headrick 2011). More than one family? More than 10 fam- A pesticide that kills a large number of unrelated species (University ilies? Is the range of affected insects deter- of California Agriculture and Natural Resources Statewide IPM mined by anthropocentric means, such as Program 2016). the destruction of pests versus beneficial Broad-spectrum Broad-spectrum insecticides are effective against all insects, even the insects? Within the explicit definitions we insecticide good ones (National Pesticide Information Center 2015). found, additional vague terms were used: Many insecticides are general purpose or wide range killers. These “affect,” “large number,” “unrelated,” “effec- “broad spectrum” pesticides are used when several different kinds tive,” “good insects,” “killer,” and “several of insects are a problem. One chemical can kill them all. No broad- kinds of insects” (Table 3). spectrum insecticide kills all insects; each varies as to the kinds of “Broad-spectrum” as a category contains insects it controls (Cornell University Cooperative Extension 2012). further linguistic uncertainty because it is Insecticides affect various insects differently. An insecticide that is inherently a relative term: “broad” in rela- lethal against one group may have little, if any, effect on another. tion to what? Relativity is not specifically Generally, however, most insecticides kill many kinds of insects: addressed in the taxonomy used by Regan broad-spectrum insecticides are those that kill more kinds than others et al. (2008). However, the relative nature (Ware and Whitacre 2004). of the word “broad” increases the likeli- Notable references that lack definitions of “broad-spectrum insecticide” includeEncyclopedia of Entomology (Capinera 2008) and a hood of under-specificity and “indetermi- foundational integrated pest management publication, The Integrated Control Concept (Stern et al. 1959). nacy of theoretical terms” (i.e., semantic change). When authors do not explicitly state what “broad” is in relation to, or which insecticides they consider to be or not be “broad-spectrum,” the term is under-speci- fied. Additionally, the relative nature of the term “broad-spectrum” makes it unstable over time, increasing the chance of semantic change. As different, and possibly more selective, insecticides become available across time, the boundary of “broad” can shift. New information may change what constitutes “destruction,” and values may change what constitutes “wide.” These fluc- tuations may cause readers in the future to see the boundaries of the category differently than the authors intended.

Fig. 2. Terms A) associated with and B) distinguished from the “broad-spectrum insecticide” category. Implicit Definitions Size of the word represents relative frequency of its use in the 60 recent papers we surveyed, com- When an explicit definition is lacking, pared with the other words in the figure. terms associated and dissociated with “broad-spectrum” could assist a reader in them provided an explicit definition of 2016), the term is defined in different and, building an inferred definition. However, the term “broad-spectrum” in reference at times, conflicting ways (Table 3). The contextual cues to intuit the meaning of to insecticidal activity or as an insecticidal ambiguous nature of the term “broad-spec- “broad-spectrum insecticide” among the category. In the expanded search for defi- trum insecticide” was neither acknowledged papers surveyed were diverse and, at times, nitions of “broad-spectrum insecticide” in nor addressed in any of the papers surveyed. contradictory (Fig. 2). Our inventory of a subset of standard entomological refer- Authors did not specify which definition terms associated with and distinguished ences (Torre-Bueno et al. 1989, Capinera of “broad-spectrum insecticide” should be from “broad-spectrum” provides useful 2008, Gordh and Headrick 2011), pest applied to their work, so it seems that writ- information about the matrix of mental management books (Ware and Whitacre ers take for granted a uniform understand- representations of the category and allows 2004), and extension websites (Cornell ing of “broad-spectrum,” or they simply us to compare the delimiting bounds of the University Cooperative Extension 2012, accept the level of uncertainty associated category “broad-spectrum” across papers National Pesticide Information Center 2015, with the term. (Fig. 2). Terms associated and dissociated University of California Agriculture and Not only is the term “broad-spectrum with “broad-spectrum insecticide” were not Natural Resources Statewide IPM Program insecticide” ambiguous, but definitions of mutually exclusive. For example, the term

262 AMERICAN ENTOMOLOGIST | WINTER 2019 “reduced-risk insecticide” was often disso- Table 4. The insecticide classes, active ingredients, and trade names stated to be broad-spectrum ciated from “broad-spectrum”; however, or not broad-spectrum in the 60 recent papers we reviewed. From this list, membership in the broad-spectrum insecticide category can be created. the terms “synthetic” and “contact,” when associated with “broad-spectrum,” can also Insecticide groups: describe “reduced-risk insecticides” (Fig. 2). It is not possible for someone to build a log- Considered to Considered to be broad-spectrum NOT be broad-spectrum ically consistent definition of “broad-spectrum insecticide” based on what the current acephate Gusathion acetamiprid* literature infers “broad-spectrum insecti- acetamiprid* imidacloprid* buprofezin cides” to be. aldicarb lambda-cyhalothrin chlorantraniliprole* azadirachtin Maldison diamides* Membership in the Category azinphos-methyl methomyl emamectin benzoate Several nouns among the papers surveyed Besiege methyl bromide flubendiamide were described as “broad-spectrum,” and, beta-cypermethrin naled imidacloprid* in other cases, as a category distinguished bifenthrin neonicotinoids* indoxacarb from “broad-spectrum”; these terms are carbamates organochlorines methoxyfenozide organophosphates shown in bold in Table 4. We found two rea- carbaryl neonicotinoids* phorate sons for this discrepancy: chlorantraniliprole* oxadiazine phosmet pymetrozine 1. The term “broad-spectrum” was used pri- chlorfenapyr chlorpyrifos pyrethroid pyriproxyfen marily to modify the term “insecticide” or cyfluthrin pyrethrum spinetoram* a synonymous term (e.g., a specific insec- cypermethrin spinetoram* spinosad ticidal subgroup/chemical class or active diamides* spinosyns sulfoxaflor ingredient). In fewer instances, the term dinotefuran Temik thiacloprid “broad-spectrum” was used to modify endosulfan Thimet thiamethoxam activity in relation to a specific insecti- esfenvalerate α-cypermethrin cide, subgroup/chemical class, or active * These insecticide groups were considered in some papers to be broad-spectrum insecticides and in other papers to not be broad- ingredient. The relationship between spectrum insecticides (i.e., they appear in both columns). “broad-spectrum” used as an adjective (e.g., “broad-spectrum activity”) and “broad-spectrum” used to define a cate- . Examples of discrepancies in the literature between “broad-spectrum” insecticidal activity and gory (e.g., “broad-spectrum insecticide”) Table 5 “broad-spectrum” insecticidal category for a given active ingredient. In most cases, discrepancies occur was not clear (Table 5). It seems counter- across papers. intuitive that an insecticide described as having “broad-spectrum activity” would Active ingredient Insecticidal activity Insecticidal category not be considered a “broad-spectrum Spinetoram “Like other spinosyns, it Reduced-risk (Sial and Brunner insecticide.” However, only a narrow has broad-spectrum activity 2010, Roubos et al. 2014, Hamby group of insecticides stated to have causing hyperexcitation of the et al. 2015) to identify products “broad-spectrum activity” were not cate- nervous system by activating with low toxicity or short duration gorized as “broad-spectrum insecticides” the Da6 subunit of the nicotinic effects on biological control agents. (i.e., spinetoram, chlorantraniliprole, acetylcholine receptor while In total, 14 insecticides were acetamiprid). Yet the question remains, maintaining a safe environmental evaluated using treated petri dishes and toxicological profile” (Hamby and four commercially available when does something with “broad-spec- et al. 2015) natural enemies (Aphidius colemani trum activity” become a “broad-spectrum Viereck, Orius insidiosus [Say], insecticide”? Chrysoperla rufilabris [Burmeister], 2. In other cases, such as with the IRAC and Hippodamia convergens [Guerin- insecticide subgroup neonicotinoids and Meneville] some active ingredients in this subgroup, Chlorantraniliprole “Provides potent and broad- Anthranilic diamide group (Wang et discrepancies in categorization reflect spectrum activity within the insect al. 2010) the linguistic uncertainty of “broad-spec- order Lepidoptera” (Wang et al. Selective (Whalen et al. 2016) trum” as a category. Ambiguity and 2010) Reduced risk (Sial and Brunner 2010, vagueness allow authors to subjec- Hamby et al. 2015) tively determine the boundaries of the Acetamiprid “Exhibited broad-spectrum Broad-spectrum (Rill et al. 2008) “broad-spectrum insecticide” category insecticidal activity that included Reduced-risk (Roubos et al. 2014) (Table 6). Typically, three to four IRAC toxicity to predatory beetles, Not a “conventional broad-spectrum” subgroups, including neonicotinoids, predaceous and omnivorous bugs, (Wise et al. 2010) green lacewings, predatory flies, were listed when an author gave exam- and several moth species” (Rill et ples of “broad-spectrum insecticides” al. 2008) (e.g., Sisterson et al. 2008, Hamby et al. 2015, Morrison et al. 2016), or it was

AMERICAN ENTOMOLOGIST | WINTER 2019 263 Table 6. Categorization of neonicotinoid active ingredients.

Active ingredient Insecticide category Excerpt Acetamiprid Broad-spectrum “The broad-spectrum neonicotinoid acetamiprid ….” (Rill et al. 2008) Compared against “The reduced-risk products, acetamiprid and spinetoram, were as toxic as some of the broad-spectrum broad-spectrum insecticides.” (Roubos et al. 2014) “For H. convergens, indoxacarb, acetamiprid, imidacloprid, and the broad-spectrum insecticides caused significant acute effects.” (Roubos et al. 2014) Distinguished from “The neonicotinoids thiacloprid and acetamiprid demonstrated strong larvicidal and ovicidal activity conventional broad- but were somewhat weaker adulticides than the conventional broad-spectrum compounds.” (Wise et spectrum al. 2010) Dinotefuran Broad-spectrum “Dinotefuran is the third neonicotinoid which possesses a broad spectrum and systemic insecticidal activity.” (Mu et al. 2016) Imidacloprid Broad-spectrum “Sprays of Azadirachtin (Neem), Tropane (Datura), Spirotetramat, Spinetoram, and broad-spectrum Imidacloprid were evaluated to control ACP in spring and summer on 10-year-old “Kinow” Citrus reticulata Blanco trees producing new growth.” (Khan et al. 2014) Compared against “We compared three insecticides (imidacloprid, thiamethoxam, and pymetrozine) to a broad-spectrum broad-spectrum insecticide (lambda-cyhalothrin) and an untreated control.” (Ohnesorg et al. 2009) Thiacloprid Distinguished from “The neonicotinoids thiacloprid and acetamiprid demonstrated strong larvicidal and ovicidal activity conventional broad- but were somewhat weaker adulticides than the conventional broad-spectrum compounds.” (Wise et spectrum al. 2010)

Thiamethoxam Compared against “We compared three insecticides (imidacloprid, thiamethoxam, and pymetrozine) to a broad-spectrum broad-spectrum insecticide (lambda-cyhalothrin) and an untreated control.” (Ohnesorg et al. 2009)

stated directly that neonicotinoids are any value-based connotations or in compar- For example, if a study stated a compari- “broad-spectrum” (Moser and Obrycki ison to anything else, 33.3% of the time; e.g., son of imidacloprid to a “broad-spectrum 2009). Conversely, neonicotinoids were “Within these areas, use of broad-spectrum insecticide,” the framing eases a reader also distinguished as an alternative to or insecticides is higher in Fresno County than into thinking that imidacloprid is not substitute for “broad-spectrum insecti- in Madera County” (Sisterson et al. 2008). In “broad-spectrum” (Table 4). When the cides” (Chuang and Hou 2008) or as pref- 40.4% of cases, “broad-spectrum” was used boundaries of an insecticidal category are erable to “broad-spectrum insecticides” in relation to or in opposition with another ambiguous, vague, and/or relative, subjec- (Ohnesorg et al. 2009). This trend held category; e.g., “The impact of imidacloprid tive decisions about how the category should for specific active ingredients within this on predatory hemipterans was shown to be be applied are required. The high degree of subgroup, as well (Table 6). less than that of a broad-spectrum insecti- flexibility for defining “broad-spectrum” cat- Table 4 represents the first attempt, cide” (Ohnesorg et al. 2009). egory membership and lack of explicit defi- to our knowledge, to create a systematic “Broad-spectrum” was used to connote nitions made it impossible, in many cases, inventory of insecticides categorized as the positive or negative value of an insec- to discern the mechanisms authors were “broad-spectrum” or characterized as dis- ticide with desirable traits in 5.5% of uses using to categorize insecticides. Individuals tinct from “broad-spectrum” (i.e., they are and undesirable traits in 16.1% of uses. For designate which insecticides are or are not not considered “broad-spectrum”). This example, the statement “Foliar sprays of “broad-spectrum,” and apparently there has analysis allows us to see how consistently broad-spectrum insecticides applied during been no discussion about dissimilar cate- membership to the “broad-spectrum” cat- tree dormancy kill the majority of over- gorizations among papers. Subjective judg- egory is assigned, without necessarily pro- wintering adults leaving few to infest new ment is used to determine “broad-spectrum viding clarity about what belongs in the shoots in spring” (Khan et al. 2014) was insecticide” category membership, but this category or the rules for categorization. considered to be desirable. The statement has not been acknowledged. “Broad-spectrum insecticides can negatively The second most common usage of the Utility of the “Broad-Spectrum Insecticide” impact the natural enemies associated with term “broad-spectrum” was to describe Category aphids” (Tran et al. 2016) was considered a noun unattached to any value-based The term “broad-spectrum” was used 255 to be undesirable. A small portion of uses connotations. An author may think that times in the 60 recent papers we surveyed. (4.7%) encompassed more than one of our describing an insecticide as “broad-spec- We categorized its appearance into three evaluative categories. trum” is increasing clarity; however, using types of use: 1) stand-alone description; 2) a The term “broad-spectrum” was used “broad-spectrum” as a descriptor decreases term in opposition or in relation to another most often in comparison to another insec- certainty, if an author does not define category; or 3) a term connoting positive or ticidal category. Establishing these compar- the term. There is no commonly under- negative value to an insecticide. We found isons enables authors to imply membership stood and accepted meaning of this term that the term “broad-spectrum” was used and non-membership within a category, in our field. Additionally, we were unable as a stand-alone descriptor, unattached to without actually categorizing an insecticide. to find a list of insecticides considered to

264 AMERICAN ENTOMOLOGIST | WINTER 2019 be “broad-spectrum” (i.e., Table 4). The in which category membership was deter- third and fourth most-common use of mined (e.g., crop system) should continue to “broad-spectrum” was to connote unde- reduce linguistic uncertainty. Individually sirable and desirable qualities, respectively, determined definitions maintain the flex- indicating that, at least during 2006–2016, AN AUTHOR MAY ibility of the term while addressing writer- the category “broad-spectrum” generally to-reader miscommunication. However, carried more negative connotations. Thus, THINK THAT multiple definitions reduce a reader’s ability when framing research, it is possible for to compare across studies that use different an author to use the term intentionally or DESCRIBING AN definitions. unintentionally to alter readers’ perceptions Even greater reduction in linguistic of risk and benefit. In our inventory of terms INSECTICIDE AS uncertainty around the term “broad-spec- used in association with “broad-spectrum” BROAD-SPECTRUM trum” could result from choosing one (Fig. 2), many terms with negative valences definition for the term, which should (e.g., “poison,” “toxicant”) were associated IS INCREASING eliminate ambiguity. Without a paper that with “broad-spectrum” and placed in oppo- coins a term, the authority to dictate the sition with alternatives described in positive CLARITY; HOW- “right” definition is unclear. Strong insti- terms (e.g., “softer chemistries,” “less harm- tutional organization and power likely ful to humans,” “less toxic”). Thus, to cate- EVER, USING would be needed to make such a reform. gorize something as “broad-spectrum” often BROAD-SPECTRUM Like Horowitz et al. (2009), we believe the insinuates that it is undesirable. United States Environmental Protection AS A DESCRIPTOR Agency (EPA), or a similar regulating body Are You (Un)certain? that can independently assess existing and A Broad Spectrum of Uncertainty DECREASES future insecticides, should be involved and Our case study highlights the consequences CERTAINTY, IF AN could make category membership criteria of ignoring the linguistic uncertainty widely available, or could create a list of of a commonly used and long-standing AUTHOR DOES NOT insecticides they determine to be members term. Despite analyzing only a small frac- of this category, similar to how the EPA cur- tion of the papers that included the term DEFINE THE TERM. rently defines “reduced-risk” insecticides “broad-spectrum insecticide,” we found (USEPA 2016). However, the consequences disagreement among the papers in explicit problematic when the rationale, premise, of such an extreme linguistic change merit and implicit definitions of “broad-spectrum conclusions, and/or implications of a paper broader discussion among the entomologi- insecticide.” We found that “broad-spec- rely heavily on the “broad-spectrum insec- cal community. A useful measure for deter- trum” as a category of insecticides is ticide” category. Combined, these linguistic mining the level of linguistic uncertainty ambiguous, vague, relative, and subjective: uncertainties interfere with clear commu- would be for a writer to question whether a linguistic uncertainties that result in con- nication and complicate attitudes about reader at any point in time would be able to fusion about the bounds of this insecticidal insecticides. use the set of criteria outlined for a category category. This has led to specific insec- Entomologists need to grapple with to determine whether a new insecticide fits ticides and groups of insecticides being the question, “Which, if any, linguistic in that category or not. described as both “broad-spectrum” and not uncertainty should be reduced for the “broad-spectrum,” difficulty in interpret- broad-spectrum insecticide category?” We Certainty in an Uncertain World ing and comparing results across studies, describe potential options in this article, Some authors have argued strongly for and instability of the term’s meaning over but we believe the answer must derive from reductions in linguistic uncertainty (e.g., time. Of particular concern was the unac- a broader discussion among entomolo- Peters 1991, Herrando-Pérez et al. 2014), knowledged contradictory categorization of gists. At one extreme, the linguistic uncer- believing the quality and efficiency of sci- neonicotinoid insecticides. This linguistic tainty could be accepted as is. Absent any entific discourse decrease with greater uncertainty is highly consequential for the changes to the meaning or use of the term linguistic uncertainty. Others believe that quality of communication among scientists “broad-spectrum insecticide,” contradict- lexical richness is good and that prescrip- and for the legitimacy of our science to be ing categorization will likely continue, the tively standardizing terminology is bad used in public and policy dialogue. boundaries of the category will remain dif- (Hodges 2008). Hodges (2008) stated that It seems that writers and readers are ficult to determine, and highly subjective precise definitions are rarely necessary and likely unaware of the miscommunication determination of membership in this cate- can hinder scientific advances by preventing that occurs when this linguistically uncer- gory will persist. To reduce some, but not all, certain lines of questioning, which ornithol- tain term is used. By not defining the term linguistic uncertainty, the “broad-spectrum ogists seem to agree with (Fraser et al. 2015). in a paper, authors are relying on read- insecticide” category could be explicitly Even when well-defined terminology exists er-created definitions of “broad-spectrum defined in each individual context or paper and writers are aware of the terms, they insecticide,” which decreases certainty in in which it is used. If vague words are used may not use them (Kritsky 2001). We found communication because readers’ contexts in the definition, further clarification will that the linguistic uncertainty that arises (e.g., cultural, linguistic, historical) may be needed to address that type of linguistic when categorizing and describing insecti- differ from the writers’ contexts. This lin- uncertainty. Additionally, clarifying use of cides illustrates a tension between reduced guistic uncertainty becomes particularly the word broad by specifying the context quality of information and more succinct

AMERICAN ENTOMOLOGIST | WINTER 2019 265 communication, and could have both pos- inevitably arises in our communications. To itive and negative consequences. improve communication across time and Linguistically uncertain words can space, entomologists should mindfully con- increase readers’ power to define and inter- sider where and why linguistically uncertain pret research. Without a definition, readers words are being used in our field. apply their own choice of definition and ... ALLOWING understanding of a term, which may or may TERMS TO Acknowledgments not match the writers’ intended meaning. We thank Dr. William D. Hutchison, Dr. In some cases, this may be desired, such as BE DEFINED Robert L. Koch, and Dr. Amy C. Morey when flexibility to extrapolate results to (University of Minnesota) for reviewing a myriad contexts is needed. Allowing readers INDIVIDUALLY previous version of this manuscript for a to take liberties interpreting the meaning of Doctoral Dissertation Fellowship. We also research could encourage creative new lines IN EACH PAPER, thank Benjamin Gronowski for copy-edit- of questioning or discovery of new under- RATHER THAN ing assistance. standings. However, allowing for variable understandings could allow readers’ biases INSISTING ON Theresa M. Cira, Department of Entomology, to arise in unexpected ways. Flexibility can A COMMON University of Minnesota, 219 Hodson Hall, 1980 also provide the ability to misappropriate Folwell Ave., St. Paul, MN 55108, USA results in scientifically unsound ways; e.g., DEFINITION, pushing an agenda by using the value-laden Kathy Quick, Humphrey School of Public Af- connotations of uncertain words. It is COMPLICATES fairs, University of Minnesota, 235 Humphrey important to consider the level of subjective interpretation that words allow. COMMUNICATION School, 301 19th Ave S, Minneapolis, MN 55455, Language can be used to shape the USA framing and understanding of knowledge. AND INCREASES Commonly, results are compared to a neu- CONFUSION ... Robert C. Venette, USDA Forest Service, tral reference point in order to judge the Northern Research Station, 1561 Lindig Ave. St. outcome as positive or negative; changing Paul, MN, 55108, USA the reference point can alter the perception of gain or loss (Tversky and Kahneman 1981). carry greater uncertainty and how that DOI: 10.1093/ae/tmz068 Thus, the reference points (e.g., insecticidal uncertainty may affect the interpretation of categories) that scientists choose can change their work. Within a paper, the writer and References Cited readers’ perceptions of risk or benefit, a shift the reader (usually never having met each Britton, B.K. 1978. Lexical ambiguity of words that can critically reduce or undermine the other) must come to a shared understand- used in English text. Behavior Research validity of the information being communi- ing of language and meaning. Developing a Methods 10: 1–7. cated (Jamieson 2017). Science is not wholly shared context about the linguistic meaning Capinera, J.L. 2008. Encyclopedia of entomology, 2nd ed. Springer Science + Business Media, an objective discipline, but when linguistic of words, preferably by explicitly defining Berlin, Germany. uncertainty is used deliberately to conceal key terms, assists this process. The defini- Carey, J.M., and M.A. Burgman. 2008. Linguistic biases in scientific writing, it threatens the tion can be stated with reference to one or uncertainty in qualitative risk analysis and credibility of the science. Linguistically more foundational sources (if they are con- how to minimize it. Annals of the New York uncertain terms, when recognized as such, sistent) or with recognition of the range of Academy of Science 1128: 13–17. can be used to highlight areas where more definitions in the literature and an explana- Chuang, Y.-Y., and R.F. Hou. 2008. Effectiveness empirical work is needed (Ostry et al. 2011), tion about the author’s understood defini- of attract-and-kill systems using methyl euge- where expert disagreement exists (Table 4), tion. Similarly, we suggest that editors and nol incorporated with neonicotinoid insec- or where other types of inquiry (e.g., val- peer-reviewers assess manuscripts’ linguis- ticides against the oriental fruit fly (Diptera: ues-based analysis) are more appropriate. tic uncertainty (e.g., undefined terms, vague Tephritidae). Journal of Economic Entomology 101: 352–359. Highly technical and certain language terms, insufficient context) along with our Cooper, G.S. 1958. Trends in Canadian ento- may inhibit understanding of scientific long-standing practice of assessing epis- mology: industry. Annual Report of the work by other stakeholder groups (NRC temic uncertainty (i.e., experimental design, Entomological Society of Ontario 89: 38–41. 1989). In an effort to make the science more analysis, reporting sources of error). Cornell University Cooperative Extension. 2012. accessible or applicable in decision-making We hope this paper increases awareness Types of pesticides. (psep.cce.cornell.edu/ contexts, it may be necessary to translate of linguistic uncertainty regarding cate- Tutorials/core-tutorial/module13/index.aspx). scientific results into less certain terms. gorization of insecticides, and also arms Crump, N.S., E.J. Cother, and G.J. Ash. 1999. Linguistically uncertain statements may readers with the ability to examine their Clarifying the nomenclature in microbial weed make scientific work more generalizable own areas of expertise for sources of lin- control. Biocontrol Science and Technology 9: across different contexts, but leaving out guistic uncertainty. By facilitating recogni- 89–97. Eckhardt, W. 1981. Limits to knowledge. important clarifications or assumptions tion of this type of uncertainty, we want to Knowledge: Creation, Diffusion, Utilization 3: may lead to unsound extrapolations. encourage discussion and problem-solving 61–81. Writers must understand which words among entomologists, because uncertainty Eilenberg, J., A. Hajek, and C. Lomer. 2001.

266 AMERICAN ENTOMOLOGIST | WINTER 2019 Suggestions for unifying the terminology in Applied Ecology 54: 2063–2068. Sisterson, M.S., R. Yacoub, G. Montez, E.E. biological control. BioControl 46: 387–400. Morrison, W.R., D.-H.H. Lee, B.D. Short, A. Grafton-Cardwell, and R.L. Groves. 2008. Fraser, H., G.E. Garrard, L. Rumpff, C.E. Hauser, Khrimian, and T.C. Leskey. 2016. Establishing Distribution and management of citrus in M.A. Mccarthy, and D. Sol. 2015. Consequences the behavioral basis for an attract-and-kill California: implications for management of inconsistently classifying woodland birds. strategy to manage the invasive Halyomorpha of glassy-winged sharpshooter. Journal of Frontiers in Ecology and Evolution 3: 61–81. halys in apple orchards. Journal of Pest Science Economic Entomology 101: 1041–1050. Gordh, G., and D. Headrick. 2011. A dictionary of 89: 81–96. Society for Risk Analysis. 2015. Society for risk entomology, 2nd ed. CABI, Wallingford, UK. Moser, S.E., and J.J. Obrycki. 2009. Non-target analysis glossary. (www.sra.org/sites/default/ Hamby, K.A., N.L. Nicola, F.J.A. Niederholzer, and effects of neonicotinoid seed treatments; mor- files/pdf/SRA_glossary_20150622.pdf). F.G. Zalom. 2015. Timing spring insecticide tality of coccinellid larvae related to zoophy- Stern, V., R. Smith, R. van den Bosch, and K. applications to target both Amyelois trans- tophagy. Biological Control 51: 487–492. Hagen. 1959. The integrated control concept. itella (Lepidoptera: Pyralidae) and Anarsia lin- Mu, X.C., W. Zhang, L.X. Wang, S. Zhang, K. Hilgardia 29: 81–100. eatella (Lepidoptera: Gelechiidae) in almond Zhang, C.F. Gao, and S.F. Wu. 2016. Resistance Taylor, J.R. 1995. Linguistic categorization, second orchards. Journal of Economic Entomology monitoring and cross-resistance patterns of ed. Oxford University Press, Oxford, UK. 108: 683–693. three rice planthoppers, Nilaparvata lugens, Torre-Bueno, J.R. de la, S.W. Nichols, G.S. Tulloch, Hassan, S.A., F. Bigler, H. Bogenschütz, E. Boller, Sogatella furcifera and Laodelphax striatellus and R.T. Schuh. 1989. The Torre-Bueno glossary J. Brun, J.N.M. Calis, J. Coremans-Pelseneer, to dinotefuran in China. Pesticide Biochemistry of entomology. The New York Entomological C. Duso, A. Grove, U. Heimback, et al. 1994. and Physiology 134: 8–13. Society, in cooperation with the American Results of the sixth joint pesticide testing National Pesticide Information Center. 2015. Museum of Natural History, New York, NY. programme of the IOBC/WPRS-working Insecticides. npic.orst.edu/ingred/ptype/insec- Tran, A.K., T.M. Alves, and R.L. Koch. 2016. group—pesticides and beneficial organisms. ticide.html Potential for sulfoxaflor to improve conser- Entomophaga 39: 107–119. NRC (National Research Council Committee vation biological control of Aphis glycines Hawking, S. 2007. A stubbornly persistent illu- on Risk Perception and Communication). (Hemiptera: Aphididae) in soybean. Journal of sion: the essential scientific works of Albert 1989. Improving risk communication. National Economic Entomology 109: 2105–2114. Einstein. Running Press, Philadelphia, PA. Academies Press, Washington, DC. Tversky, A., and D. Kahneman. 1981. The fram- Herrando-Pérez, S., B.W. Brook, and C.J.A. Ohnesorg, W.J., K.D. Johnson, and M.E. O’Neal. ing of decisions and the psychology of choice. Bradshaw. 2014. Ecology needs a convention 2009. Impact of reduced-risk insecticides on Science 211: 453–458. of nomenclature. Bioscience 64: 311–321. soybean aphid and associated natural ene- USEPA (United States Environmental Protection Hodges, K.E. 2008. Defining the problem: termi- mies. Journal of Economic Entomology 102: Agency). 2016. Conventional reduced risk nology and progress in ecology. Frontiers in 1816–1826. pesticide program. (https://www.epa.gov/ Ecology and the Environment 6: 35–42. OMRI (Organic Materials Review Institute). 2016. pesticide-registration/conventional-re- Hoffman, F.O., and J.S. Hammonds. 1994. OMRI Lists. (www.omri.org/omri-lists). duced-risk-pesticide-program#status). Propagation of uncertainty in risk assessments: Onions, C.T., G.W.S. Friedrichsen, and R.W. USEPA (United States Environmental the need to distinguish between uncertainty Burchfield. 1966. The Oxford dictionary of Protection Agency). 2017. What are bio- due to lack of knowledge and uncertainty due English etymology. Oxford University Press, pesticides? (https://www.epa.gov/ to variability. Risk Analysis 14: 707–712. London, UK. ingredients-used-pesticide-products/ Horowitz, A.R., P.C. Ellsworth, and I. Ishaaya. Ostry, M.E., R.C. Venette, and J. Juzwik. 2011. what-are-biopesticides). 2009. Biorational pest control—an overview, Decline as a disease category: is it helpful? University of California Agriculture and Natural pp. 1–20. In I. Ishaaya and A.R. Horowitz. Phytopathology 101: 404–409. Resources Statewide IPM Program. 2016. Biorational control of arthropod pests. Springer Peters, R.H. 1991. A critique for ecology, 1st ed. Glossary. (http://ipm.ucanr.edu/PMG/glossary. Science + Business Media, B.V, Wallingford, UK. Cambridge University Press, New York, NY. html). IRAC (Insecticide Resistance Action Committee). Popper, K. 2012. In search of a better world: lec- Wang, X., X. Li, A. Shen, and Y. Wu. 2010. Baseline 2017. The IRAC mode of action classification. tures and essays from thirty years. Routledge, susceptibility of the diamondback moth (www.irac-online.org/modes-of-action/). Abingdon, UK. (Lepidoptera: Plutellidae) to chlorantraniliprole Jamieson, K.H. 2017. The need for a science of sci- Regan, H.M., M. Colyvan, and M.A. Burgman. in China. Journal of Economic Entomology 103: ence communication: communicating science’s 2002. A taxonomy and treatment of uncer- 843–848. values and norms. In K.H. Jamieson, D.M. Kahan, tainty for ecology and conservation biology. Ware, G.W., and D.M. Whitacre. 2004. The pes- and D. Scheufele (eds.). The Oxford Handbook Ecological Applications 12: 618–628. ticide book, 6th ed. Meister Publishing Co., of the Science of Science Communication. Rill, S.M., E.E. Grafton-Cardwell, and J.G. Morse. Willoughby, OH. Oxford University Press, Oxford, UK. 2008. Effects of two insect growth regulators Whalen, R.A., D.A. Herbert, S. Malone, T.P. Kuhar, Khan, A.A., M. Afzal, J.A. Qureshi, A.M. Khan, and and a neonicotinoid on various life stages of C.C. Brewster, and D.D. Reisig. 2016. Effects A.M. Raza. 2014. Botanicals, selective insecti- Aphytis melinus (Hymenoptera: Aphelinidae). of diamide insecticides on predators in soy- cides, and predators to control Diaphorina citri BioControl 53: 579–587. bean. Journal of Economic Entomology 109: (Hemiptera: Liviidae) in citrus orchards. Insect Roubos, C.R., C. Rodriguez-Saona, R. Holdcraft, 2014–2019. Science 21: 717–726. K.S. Mason, and R. Isaacs. 2014. Relative toxic- Wise, J.C., P.E. Jenkins, R. Vander Poppen, and Kritsky, G. 2001. Periodical revolutions and the ity and residual activity of insecticides used in R. Isaacs. 2010. Activity of broad-spectrum early history of the “locust” in American cicada blueberry pest management: mortality of nat- and reduced-risk insecticides on various life terminology. American Entomologist 47: 1–4. ural enemies. Journal of Economic Entomology stages of cranberry fruitworm (Lepidoptera: Michel, J.-B., Y.K. Shen, A.P. Aiden, A. Veres, M.K. 107: 277–285. Pyralidae) in highbush blueberry. Journal of Gray, The Google Books Team, J.P. Pickett, Sial, A.A., and J.F. Brunner. 2010. Assessment Economic Entomology 103: 1720–1728. D. Hoiberg, D. Clancy, P. Norvig, et al. 2011. of resistance risk in obliquebanded leafroller Zadeh, L.A. 2004. Precisiated natural language Quantitative analysis of culture using millions (Lepidoptera: Tortricidae) to the reduced-risk (PNL). AI Magazine 25: 74–91. of digitized books. Science 331: 176–82. insecticides chlorantraniliprole and spineto- Milner-Gulland, E.J., and K. Shea. 2017. Embracing ram. Journal of Economic Entomolology 103: uncertainty in applied ecology. Journal of 1378–1385.

AMERICAN ENTOMOLOGIST | WINTER 2019 267