Tests of search image and learning in the wild: insights from sexual conflict in damselflies
Data files
Feb 04, 2022 version files 172.70 KB
-
Analyses_DRYAD.txt
6.73 KB
-
Counted_Eggs_2018_DRYAD.xlsx
17.91 KB
-
FFMF.csv
6.77 KB
-
FFStacked.csv
18.69 KB
-
MF_ALL_STACKED_DRYAD_2021.xlsx
74.42 KB
-
MF.csv
5.03 KB
-
MF2.csv
1.73 KB
-
MR.csv
29.21 KB
-
OVO-FOLLOW_2018_DRYAD.xlsx
12.21 KB
Abstract
Search image formation, a proximal mechanism to maintain genetic polymorphisms by negative frequency-dependent selection, has rarely been tested under natural conditions. Females of many non-territorial damselflies resemble either conspecific males or background vegetation. Mate-searching males are assumed to form search images of the majority female type, sexually harassing it at rates higher than expected from its frequency, thus selectively favoring the less common morph. We tested this and how morph coloration and behavior influenced male perception and intersexual encounters by following marked Ischnura elegans and noting their reactions to conspecifics. Contrary to search image formation and associative learning hypotheses, although males encountered the minority, male-like morph more often, sexual harassment and clutch size were similar for both morphs. Prior mating attempts or copulas with morphs did not affect a male’s subsequent reaction to them; males rarely attempted matings with immature females or males. Females mated early in the day, reducing the opportunity for males to learn their identity beforehand. Once encountered, the male-like morph was more readily noticed by males than the alternative morph, which, once noticed was more likely to receive mating attempts. Flexible behavior gave morphs considerable control over their apparency to males, influencing intersexual encounters. Results suggested a more subtle proximal mechanism than male learning maintains these color polymorphisms and call for inferences of learning to be validated by behavior of wild receivers and their signalers.
We followed marked focal males and color polymorphic focal females in the wild for as long as possible, recording all interactions with conspecifics and the behaviors during those interactions. Females are either colored like males (andromorphs, A-morphs) or are more similar in coloration to background vegetation (heteromorphs, H-morphs). We inferred perception and intent of focal individuals from their behavioral responses to conspecifics.
Additionally we did censuses of individuals to measure the frequency of female color morphs in the population, and of copula pairs to understand the availability of potential mates across the day. We followed egg-laying females to see whether they got harassed by males, and collected eggs from females directly after copula ended, to determine whether one morph had higher clutch size than the other.
We used GLMM models to determine:
1) if males encountered the majority H-morph more often than expected given its frequency (70%) in the population.
2) if males correctly recognized conspecifics (i.e. H- and A-morphs) as potential mates and recognized immature females and males as inapproriate for mating.
3) to determine if focal female morphs differed in their behavioral responses to males
We used GLM models to determine:
4) to test whether males perceived the two morph types differntly
5) to test whether the two morph types received different rates of sexual harassment by males
6) to test for unexpected deviations between intersexual encounters of focal males and focal females
We analyzed the data on focal male behavior using a markov model to identify behavioral transition states of focal males, and then tested those transitions to see if they agreed with the expected assumption that males recognized conspecifics based on their past experience with them (i.e. associatively learned or formed search images to recognize color polymorphic females as mates, and whether they learned to avoid to males and immature females as potential mates.
Notes on variables are in all excel files. For GLM and GLM analyses, R-code is submitted under a file entitled 'Analysis_Dryad'.