There is an abundance of resources out there to help with referencing, many of which focus on the technicalities, such as the best style of referencing to use or the best software. However, there is a lack of guidance on how to interweave references into the narrative, which articles to include or exclude, whether to cite a review or a series of original articles, and how to deal with articles that contradict your argument. This post goes some way to bridging this gap.
In 2017, my colleagues and I published an article entitled “Dose-dependent behavioural fever responses in desert locusts challenged with the entomopathogenic fungus Metarhizium acridum” in Scientific Reports (click here for article link). This article took more than a year to write and was peer-reviewed by a key player in the field, and another anonymous reviewer.
Here, I will take you through each citation used in the article, describing why we chose the articles we did in relation to best practice. I will also discuss some of the reviewer comments we received in relation to the references and how we dealt with them.
Dose-dependent behavioural fever responses in desert locusts challenged with the entomopathogenic fungus Metarhizium acridum
Fever body temperatures are an adaptive response to infection in many species, and they can be achieved by physiological and/or behavioural means such as basking (Hart, 1988; Kluger et al., 1996).
This is a very broad statement and so we chose a seminal review on the topic (Hart, 1988), which has been cited 1798 times (according to Google Scholar, as on Jan 7th, 2019), and another well-received review (Kluger et al., 1996), in which key studies on many different species are referenced, as is stated. By citing these references, we not only direct the reader towards two important pieces of literature, which will give them a solid foundation in the topic, but also acknowledge two giants in the field, and show reviewers we have foundational knowledge on the topic.
Ectotherms rely exclusively on behavioural strategies to achieve fever, and these have been described across a diverse range of taxa including reptiles, amphibians, and insects (Kluger et al., 1996).
As we start to narrow in on the topic, we reference Kluger et al. (1996) again. This is because, in contrast to Hart (1998), who discusses the many different behavioural changes in response to infection, Kluger et al. (1996) focusses on fever only and lists key studies across different animal orders, as is stated.
Behavioural fever can decrease mortality and morbidity in infected animals (Louis et al., 1986; Carruthers et al., 1992; Blanford and Thomas, 2001; Elliot et al., 2002).
There are many studies that have proven this statement across a range of taxa. However, it is not practical or good practice to list them all. Therefore, in this case, we chose those studies most closely linked to our study (i.e., each of these studies use animals from the order Orthoptera) and which have been cited frequently.
This occurs because high body temperatures are suboptimal for pathogen growth (Inglis et al., 1996; Elliot et al., 2002) , and increase the mortality of some pathogens (Carruthers et al., 1992) ; high body temperatures also enhance several aspects of host immune function (Ouedraogo et al., 2003; Boltaña et al., 2013). 
 Both these references support this statement, showing reduced mycosis in grasshoppers/locusts kept at fever temperatures. A key element of effective referencing is concisely summarising the key findings from multiple articles. In this case, the concise statement “This occurs because high body temperatures are suboptimal for pathogen growth” is better than listing the specific results, e.g., “A low prevalence of Beauveria bassiana mycosis was observed in inoculated Melanoplus sanguinipes nymphs exposed to a continuous temperature of 35 and 40°C, whereas continuous exposure to 30°C did not have a significant effect on disease development”.
 We added this study in response to a reviewer comment (see Box 1). Among its many findings, this article showed that fever temperatures completely eliminated Entomophaga grylli from the clearwinged grasshopper, Camnula pellucida. It is important to state this, as it is generally accepted that orthopterans cannot eliminate Metarhizium infection through fever; hence why we say, “of some pathogens”. This is balanced referencing, i.e., providing all sides to an argument.
 In most articles on this topic, the Ouedraogo et al. (2003) article is cited in support of this statement, as it was the first to examine innate immune responses, such as hemocyte kinetics and phagocytosis, during fever in this host–pathogen pair. When you frequently see an article cited in a particular way it is easy to become complacent and accept it as fact. However, always remember to continue to search the literature for new developments in an area. New studies may have come out that support, or indeed contradict, your argument. Boltaña et al. (2013) strongly supports the argument made by Ouedraogo et al. (2003), showing the upregulation of anti-viral genes during behavioural fever in zebrafish. Again, note the concise way in which we summarise these two articles.
[New paragraph] Despite their benefit for fighting infection, fever body temperatures are also costly (Casadevall, 2016).
Again, this is a very broad statement, and given that it is relevant to the field of medicine, numerous articles could be referenced. We chose Casadevall (2016), as this Pearl—a “concise, practical, and educational insight into topics that span the pathogen field”—summarises and contextualises the debate to date on whether fever is beneficial or costly. This Pearl has only been cited 7 times (according to Google Scholar, as on Jan 9th, 2019); however, its author is a distinguished professor with many publications and citations in this field. I am not suggesting you cite an article based on the authors’ credentials; you yourself should critically assess the merit of each article you are considering citing.
Elevated body temperatures have been associated with increased metabolic rate (Muchlinski, 1985; Sherman and Stephens, 1998)  , reduced growth rates (Boorstein and Ewald, 1987) , as well as defects in egg development (Anderson et al., 2013) .
 After starting the paragraph with a broad statement, we then list the original research articles that support the statement. You’ll note that these articles are relatively old. However, they both definitively prove the statement from the perspective of “the costs of fever”, much like our article. You may read elsewhere that you should not include articles older than 5–6 years. However, it is better to be flexible with such rules. For example, it is fully acceptable to cite older breakthrough works when introducing a topic, and such rules are generally not universal across subject areas, e.g., fast-moving fields (e.g., medical sciences) will have a higher turnover rate than slower-moving fields (e.g., mathematics).
 Again, this article is comparatively old. However, it definitively proves the statement, again from the perspective of “the costs of fever,” and has been cited 133 times (according to Google Scholar, as on Jan 30th, 2019). This is also an elegant study, in that the narrative is clear (i.e., costs and benefits of fever) and it informed my writing style, i.e., how to present findings as a story.
 This is a relatively recent article on the costs of fever in fungal-infected insects. Notice how we interweave all five articles into one concise coherent statement.
Note that we added the appropriate citations directly after each specific statement. It would be poor practice to cite these references as follows:
Elevated body temperatures have been associated with increased metabolic rate, reduced growth rates, as well as defects in egg development (Muchlinski, 1985; Sherman and Stephens, 1998; Boorstein and Ewald, 1987; Anderson et al., 2013).
Furthermore, basking to achieve fever body temperatures confers additional costs in terms of missed feeding and mating opportunities, and increased predation risk (Otti et al., 2012).
To further support our argument, we cite a relatively recent study focussed on the costs of fever from a different perspective – behaviour.
These costs accrue quickly as body temperature rises, such that the temperature-fitness curves of healthy animals are often asymmetrical, with fitness declining more steeply with increasing temperature than with decreasing temperature (Martin and Heuy, 2008).
This citation provides an excellent example of the benefits of reading widely and outside your specific topic. We have already provided ample support to show that fever is costly. However, with this citation, we go a step further, and cite a high-impact article that supports our argument mathematically.
As a consequence, and because of their inability to thermoregulate perfectly, healthy ectotherms tend to adopt body temperatures slightly below the theoretical optimum, in order to avoid the disproportionate costs of unintentionally straying into higher body temperatures (Martin and Heuy, 2008).
We needed two longish sentences to get our point across here, but we judged that this was necessary to explain the concept fully.
Sick individuals likely face a similar requirement to balance costs and benefits and avoid excessively high fever body temperatures, especially since behavioural fever responses are often unable to fully clear infection (Elliot et al., 2002; Ouedraogo et al., 2004) (but see Carruthers et al., 1992).
As commented on earlier, one area of contention is whether or not animals can fully clear infections using fever (see Box 1). We, along with Elliot et al. (2002) and Ouedraogo et al. (2004), did not find evidence to suggest that locusts could rid themselves of Metharizium infection; however, we then cite the Carruthers et al. (1992) article, which showed that fever temperatures completely eliminated Entomophaga grylli from the clearwinged grasshopper, Camnula pellucida. Note the concise way we dealt with this contradictory article, i.e., “(but see Carruthers et al., 1992)”. Often this is all that is needed to present a balanced argument (or address a reviewer comment).
In these cases, fever body temperatures might be expected to be modulated according to the severity or type of infection (Gardner and Thomas, 2002).
We discovered this article at the writing stage. In other words, we had already formulated and tested our hypothesis. However, this article is the first, as far as we are aware, to directly propose the hypothesis that we tested in this article and we therefore duly cited it. This shows the importance of conducting a literature search for each statement before submission, to ensure you give credit where credit is due.
[New paragraph] In support of the adaptive deployment of fever responses, the occurrence and magnitude of behavioural fever are known to vary within and between insect species (Stahlschmidt and Adamo, 2013).
We start this paragraph with another broad statement, citing a meta-analysis of behavioural fever studies. Using meta-analysis articles can be an excellent resource for big-picture conclusions and for summarising multiple articles at once.
Crickets, Acheta domesticus, elicited behavioural fever responses when infected with thermo-susceptible parasites (e.g. Rickettsiella grylli), but not thermotolerant ones (e.g. Serratia marcescens, and the parasitoid fly Ormia ochracea) (Adamo, 1998).
Again, after starting the paragraph with a broad statement, we then cite an article that provides specific proof for the statement. As well as supporting our argument, by providing extra details here, we help the reader get an idea of what we mean by varying fever responses.
House flies, Musca domestica, infected with a higher dose of the fungus Beauveria bassiana exhibited higher-intensity fever responses than flies infected with a lower dose, putatively limiting fever costs (Anderson et al., 2013).
This is an excellent article from big players in the field and was a source of inspiration for this study. We reference this article throughout the entire manuscript, as it is most closely associated with our article. We reference it here as we narrow in on the specific aims of our article.
The physiological fevers of humans may also vary in magnitude according to the severity of infection (Kluger et al., 1996).
As we are entering the field of fever in the context of human health, there are numerous resources we could cite to support this statement; however, we decided it was best to reference Kluger et al. (1996) again, wherein the reader can access details of the most important foundational articles on the topic.
Thus, the adaptive deployment of fever may be a widespread phenomenon.
Note that we have not cited any article in relation to this statement. While there are numerous articles on the costs and benefits of fever, few articles have looked at the adaptive deployment of fever (adjusting fever to maximise the difference between these costs and benefits). In this study, we present additional support for this concept. Hence, we hypothesise that this phenomenon may be more widespread than originally thought.
[New paragraph] In this study, we investigate whether desert locust, Schistocerca gregaria, behavioural fever responses differ in intensity according to the severity of Metarhizium acridum infection. Metarhizium acridum is a specialist pathogen of locusts and grasshoppers and the basis of commercial biopesticides used for their control (Lomer et al., 2001).
This is a comprehensive review on the topic of the commercial biocontrol of locusts using entomopathogenic fungi (it has been cited 468 times, according to Google Scholar, as on Jan 15th, 2019), giving readers a solid source with which to assess the application of the findings of this study.
Whilst numerous laboratory and field trials have demonstrated the efficacy of these biopesticides, the speed of kill following application is highly variable (see van der Valk, 2007, and references therein).
This is a technical report from the United Nations, describing various M. acridum-application field trials conducted in Africa. It is an excellent resource, putting all the laboratory findings discussed so far in the context of locust biocontrol.
The prevailing theory is that environmental temperature leads to this variability, by allowing or preventing effective behavioural fever responses (Elliot et al., 2002).
This is the fourth time we have cited this article. Although the Elliot et al. (2002) study was not the first to suggest this theory, it was the first to definitively show the adaptive value of fever (i.e., infected locusts allowed to fever could go on the produce viable offspring), and therefore, is considered a breakthrough article on this topic.
Thus, additional knowledge of locust behavioural fever responses to Metarhizium infection may also help inform the application of biopesticides for locust control.
Key points from Part 1 (Introduction)
- Every statement (apart from the aims and original hypothesis) is supported by a least one article.
- We read widely and critically assessed each article, only including those articles we considered to be of sufficient quality.
- We started each paragraph with a broad-scope review, and then cited specific original research articles as we narrowed in on our point.
- We duly cited any articles that contradicted our argument.
- We placed the appropriate citations directly after each specific statement in a sentence, rather than grouped together at the end.
- We cited all (past and present) big players in the field.
- For broad statements, for which numerous articles could be cited, we cited the articles most similar to ours.
- We conducted a literature search for each statement before submission, to see if any recent articles supported/contradicted our findings.
- For introductory statements, we concisely summarised the key findings of multiple articles into a coherent whole.
How to reference (Part 2) to follow shortly.