Conditionally split strings cells in a dataframe R

I have a dataset with species name where some names originally used are now obsolete, so they are noted "old_speciesretired use new_species", whereas correct cells are just noted "new_species".

Here is a sample of the data :

df<- data.frame(species=c("Etheostoma spectabile","Ictalurus furcatus","Micropterus salmoides","Micropterus salmoides","Ictalurus punctatus","Ictalurus punctatus","Ictalurus punctatus","Micropterus salmoides","Etheostoma olmstedi","Noturus insignis","Lepomis auritus","Lepomis auritus","Nocomis leptocephalus","Scartomyzon rupiscartes***retired***use Moxostoma rupiscartes","Lepomis cyanellus","Notropis chlorocephalus","Scartomyzon cervinus***retired***use Moxostoma cervinum","Ictalurus punctatus","Lythrurus ardens","Moxostoma pappillosum","Micropterus salmoides","Micropterus salmoides","Ictalurus punctatus"))

I have tried

sapply(strsplit(df$species, split='***retired***use', fixed = T),function(x) (x[2])))

but the cells for which the data is correct returns NA because they do not contain the split.

Is there a way to make the split just for the cells actually containing it?

Solution

You can change the old names to the new names using gsub plus backreference:

gsub(".*\\*\\*\\*retired\\*\\*\\*use\\s(.*)", "\\1", df$species)

# [1] "Etheostoma spectabile"   "Ictalurus furcatus"      "Micropterus salmoides"   "Micropterus salmoides"  
# [5] "Ictalurus punctatus"     "Ictalurus punctatus"     "Ictalurus punctatus"     "Micropterus salmoides"  
# [9] "Etheostoma olmstedi"     "Noturus insignis"        "Lepomis auritus"         "Lepomis auritus"        
# [13] "Nocomis leptocephalus"   "Moxostoma rupiscartes"   "Lepomis cyanellus"       "Notropis chlorocephalus"
# [17] "Moxostoma cervinum"      "Ictalurus punctatus"     "Lythrurus ardens"        "Moxostoma pappillosum"  
# [21] "Micropterus salmoides"   "Micropterus salmoides"   "Ictalurus punctatus"

Explanation:

.* anything any number of times followed by ...

\\*\\*\\*retired\\*\\*\\*use\\s ... the literal pattern ***retired***use followed by ...

(.*) ... anything any number of times--that's the capturing group that the backreference \\1 in the replacement argument of gsubrefers to