Cuticular hydrocarbons (CHCs) form a layer on the insect cuticle, serving as the exoskeleton's desiccation barrier. Female crickets are demonstrably attracted to males with more short-chain CHCs. However, longer-chain CHCs are better at desiccation prevention, as they have higher melting points. Therefore, males can invest in CHCs as sexual signals, but may run the risk of desiccation.
I tested the hypothesis that regional desiccation risk constrains sexual selection by assessing differences in male CHC profiles across the Cook and Hawaiian Island populations. A limited data set by Simmons et al. 2014 shows that the Hilo population of crickets on the big island of Hawaii, which experiences heavy rainfall, have a greater proportion of short-chain CHCs than those from less precipitous Hawaiian Islands. Similar analysis of the biogeographic patterns of CHC profiles across precipitation gradients has been done in Drosophila melanogaster in North America (Rajpurohit et al. 2017) and Drosophila serrata and D. melanogaster in Australia (Frentiu and Chenoweth 2010). I expanded this analysis across the crickets’ range and analyzed two other metrics for desiccation risk: humidity and temperature. I predicted that crickets in areas with low desiccation risk would exhibit more attractive CHCs. I also predicted that crickets in the parasitized Hawaiian Islands, where the cost of singing is higher, will invest more in chemical signals and have more attractive CHCs than the crickets in the non-parasitized Cook Islands.
Methods: We collected and freeze-killed male and female crickets on five islands: Aitutaki (n = 23 F, 23 M), Mangaia (23 F, 22 M), Rarotonga (n = 20 F, 23 M) (Cook Islands), the big island of Hawaii (n = 20 F, 20 NW M), and Oahu (n = 20 F, 20 NW M, 21 FW M) (Hawaiian Islands). NW represents normal- wing males, and FW represents flatwings, both of which are currently only found in Hawaii. At each sampling site, we measured temperature and humidity levels with a HOBO Onset data logger. I verified that our findings were consistent with historical weather data in each area by comparing to the National Oceanographic and Atmospheric Administration (NOAA) database. Following well-established methods, I extracted CHCs using n-hexane and analyzed them with an Agilent Gas Chromatography Mass Spectrometer fitted with a DB-wax column (Blows and Allan 1998, Thomas and Simmons 2009a). The resulting chromatogram yields a CHC profile for each cricket that shows the presence and abundance of different compounds, identifiable by retention time. In collaboration with chemist Stephen Harvey at the University of Minnesota Mass Spectrometry Department, I identified individual compounds and their chemical configurations through the NIST library database.
I analyzed the data three ways, using proportional data in each because absolute peaks can be subject to experimental errors (Blows and Allan 1998, Savarit and Ferveur 2002). Analysis 1: I calculated the areas of each peak in the chromatogram and generated proportions of the area of short (<32 carbons) to long (≥ 32 carbons) chains (Simmons et al. 2014). Lower proportions were considered less attractive profiles, as they indicate a higher presence of long chain CHCs; higher proportions were considered more attractive, as they indicate a higher presence of short chain CHCs (Simmons et al. 2014). Analysis 2: I divided the sum of areas of short (<32) chains in a given sample over the sum of all peak areas in that sample. I then compared the percent of short chains found in each cricket across populations. Analysis 3: Following the multivariate methods established by Neems and Butlin (1995), I calculated proportional peaks by dividing the area of each peak in a sample by the sum of all peak areas in that sample. I then created log contrasts by dividing an arbitrarily chosen peak by the proportional peak area and taking the log of the new variables to solve the problem of non-independence caused by using proportions (Blows and Allan 1998). I used Log (1 + x), because not all crickets had every compound. I then conducted principal components analyses on these new variables. With the proportions in the first two analyses, I used linear models to assess relationships between a) all CHCs, b) male CHCs, c) female CHCs, and d) NW & FW CHCs and average, min, max, daytime, and nighttime precipitation, temperature, and humidity.
Results: Cook Island males had higher short to long chain ratios (t-test, p = 2.2 x 10-16, t = 12.36, df = 228.89) (Figure 1b) and higher proportions of short chains than Hawaiian males (t-test, p = 2.2 x 10- 16, t = 12.806, df = 230.4) (Figure 1c). We found no other differences in proportion of short to long chains between sexes (t-test, p=0.14, t=1.4811, df=225.55) or within regions (Fisher’s LSD: Aitutaki-Mangaia p=0.57, Aitutaki-Rarotonga p=0.82, Mangaia-Rarotonga p=0.73, Hilo-Oahu p=0.19).There was also no difference in proportion of short chains by sex (t-test, p=0.086, t=1.724, df=224.86) or within region (Fisher’s LSD: Aitutaki-Mangaia p=0.068, Aitutaki-Rarotonga p=0.140, Mangaia-Rarotonga p=0.743, Hilo-
Oahu p=0.649). We found no relationships between CHC attractiveness and desiccation risk (Table 1).
The Cook Islands experience higher temperatures than Hawai’i, and males suffer no parasitization risk from singing. Consequently, we expected a comparatively higher amount of less attractive but more protective long-chain CHCs in the Cook Islands. Surprisingly, Cook Island males have higher proportions of short to long chain and more attractive short-chain CHCs than Hawaiian males. We also find no discernible relationship between our metrics for CHC attractiveness and desiccation risk. There are a few potential explanations for this. While there are regional precipitation, temperature, and humidity differences between regions, both Hawai’i and the Cook Islands are sub-tropical areas. It is possible that the consistent precipitation and humidity of the tropics, regardless of inter-island variation, does not pose enough of a desiccation risk to cause the tradeoffs that we expected. There is another possibility that individuals in the Cook Islands are able to allocate resources to both song and attractive CHCs (Reznick et al. 2000; “big houses, big cars” principle), while Hawaiian crickets cannot. Depending on the currency involved in song and CHC production, it may be possible that the energetic expense of one does not come at a deficit to the other, as traditional life history theory predicts. Alternatively, the divergence in CHC profiles may be related to drift or other neutral processes. I analyze all these relationships in much more detail in my manuscript.