Kyle Finn

University of Pretoria | UP · Department of Zoology and Entomology

I'm assessing geneflow, relatedness and dispersal patterns in social mole-rats. They live in groups (4-15 animals) with high reproductive skew, 1 reproductive female and 1-2 reproductive males, with overlapping generations. The other group members are usually offspring of the breeders and don't contribute to breeding. Some unrelated immigrants do occur. I'm using SPAGeDI to estimate dispersal distances. Now some papers recommend removing offspring in genetic analyses so that you don't include pre-dispersal individuals. So I did that for Fst, relatedness coeff, and isolation by distance. For dispersal distances initially I thought to use just the reproductive individuals, but my sample size for females is 32 so I used all adults regardless of reproductive status getting me 84 males and 97 females. I used a population density equal to the expected number of breeding individuals per group in my study area, not the total number of individuals I captured. The values for males did not converge (no matter what density I used), and the predicted neighborhood size for females is 3 times than what I observed through captures. The predicted neighborhood size for both sexes combined is fairly accurate.

So a few questions: 1) when estimating dispersal distances through genetic methods do I limit my dataset to just reproductive individuals or use all the individuals even if there is overlapping generations? 2) specifically for SPAGeDi the population density you specify, is it the number of breeders per sq km or the total number of individuals (census density)? 3) assuming i used an accurate population density for the adult females, is the neighborhood size accurate? Thank you in advance

I have widely inaccurate spatial distances in my "Pairwise Spatial and genetic distances written in column form" table output from SPAGeDI 1.5. For example SPAGeDi measures the distance between points -29.325094, 29.707921 and -29.322181, 29.707929 as 0.001178km (Spatial dist (km) column), but they are actually +/- 320m apart! I have entered the lat and long as decimal degrees (Lat = -29.322710, Long = 29.716432) in columns labeled X and Y. I have also specified Num Coords in the second line of my file as "-2". Are the spatial distances something else entirely and not the distance between GPS points?

Hello all, I am analyzing body mass between small mammals and testing if it varies between sexes, reproductive status (breeding/non-breeding), and season (wet/dry). Would it be alright to have one variable called reproductive class as a factor with 4 variables: reproductive male, reproductive female, non-reproductive male, non-reproductive female. Is it possible to then use this in a linear model like body mass ~ reproductive class * season ?

Does sex and reproductive status have to be separate terms?

Additionally if I have significant differences between males and females, can I group the two sexes together to see if body mass overall changes with season?

Thank you

Hi all,

Due to quarantine my uni is closed and I can't get any hands on help with analysis for a manuscript. My study is looking at dispersal distances of small mammals in two populations and how dispersal distances are affected by population density. It seems pretty straight forward except that my study sites are two different shapes and one of my co-authors thinks that I should find a way to statistically correct for this difference since my results could simply be due to animals migrating outside of the boundaries for study site B and therefore dispersal distances are shorter than what actually occurs.

Study site A: mostly square with area of 3km2, with a low population density and higher mean dispersal distance

Study site B: a long rectangle 4km by 0.5km, with a high population density and lower mean dispersal distance.

It was suggested that I build a model and let the program do random walks in and out of the study sites to mimic dispersing individuals. Then test whether the model correctly represents what is actually happening. However neither of my advisors has any experience in movement ecology. Ideally it would be nice to account for actual home ranges within the study sites since being able to see how theoretical movement would happen in a polygon (or how often individuals may move out of it) based on actual locations of home ranges would be exceedling helpful.

So my questions are

1: is this extra modeling step necessary or is it overkill?

2: what program(s) (open source please) could I use to do this?

3: how do I do this? can you provide tutorials and examples to help me figure this out, or terms to search for?

I have novice R and QGIS skills by the way. Thank you very much in advance!