left_join duplicates even after troubleshooting - r

Question

Sample data:
full<-structure(list(Location = structure(c(1L, 1L, 1L, 1L, 1L, 1L), .Label = c("AKS",
"AOK", "BTX", "GTX", "HKS", "JKS", "LOK", "MKS", "MOK", "PKS",
"SKS", "VTX"), class = "factor"), CT_NT = structure(c(1L, 1L,
1L, 1L, 1L, 1L), .Label = c("CT", "NT"), class = "factor"), Depth = c(5L,
10L, 15L, 5L, 10L, 15L), Site = c(1L, 1L, 1L, 1L, 1L, 1L), PW = c(22.8,
21.5, 18.2, 22.5, 20.5, 19.2), BD = c(1.1, 1.2, 1.1, 1.3, 1.3,
1.5)), .Names = c("Location", "CT_NT", "Depth", "Site", "PW",
"BD"), row.names = c(NA, 6L), class = "data.frame")
osu<-structure(list(Location = structure(c(1L, 1L, 1L, 1L, 1L, 1L), .Label = c("AKS",
"AOK", "BTX", "GTX", "HKS", "JKS", "LOK", "MKS", "MOK", "PKS",
"SKS", "VTX"), class = "factor"), CT_NT = structure(c(1L, 1L,
1L, 2L, 2L, 2L), .Label = c("CT", "NT"), class = "factor"), Depth = c(5L,
10L, 15L, 5L, 10L, 15L), pH = c(5.1, 5.4, 5.9, 5.2, 5.9, 6.2),
N = c(50, 31, 22, 35, 17, 8), P = c(122, 55, 34, 107, 23,
17), K = c(1301, 1202, 1078, 1196, 1028, 948), OM = c(2.3,
1.8, 1.5, 2.1, 1.4, 1.2), NH4 = c(19.3, 14.5, 11.6, 12.3,
8.6, 8.4), Sand = c(22.5, 25, 25, 25, 22.5, 18.8), Silt = c(56.3,
52.5, 50, 51.3, 52.5, 51.3), Clay = c(21.3, 22.5, 25, 23.8,
25, 30)), .Names = c("Location", "CT_NT", "Depth", "pH",
"N", "P", "K", "OM", "NH4", "Sand", "Silt", "Clay"), row.names = c(NA,
6L), class = "data.frame")
I am trying to join two datasets using left_join in dplyr. To my astonishment, I'm getting duplicate rows that are somehow not being identified as such. After reading all the other answers I could get my hands on here that seemed to address "join" issues (at least I'm not the only one who has them...?), I have tried:
Checking the group types of the joining variables in the two datasets
to ensure they match
Checking that I don't have duplicates within f1 or f2
Checking that the categorical columns I'm using to join are, in fact,
the same length and have the same contents. They're EXACTLY the same,
all the way down to the order I put them in
Explicitly specifying to dplyr to use Location, CT_NT, and Depth to
join
Letting dplyr figure out the joining variables itself Joining in both
orders Using inner_join--I ended up with f1 only
I've used left_join before and not had this issue, and it was with a very similar dataset (the pilot data to this full study, in fact). I thought I understood what left_join was doing, but now I'm wondering if I don't actually. I'm trying to get better with using dplyr, but unfortunately it's a lot of me bashing away at things until something works and I can figure out why it worked so I can reproduce it again later as needed.
Given my inexperience, I'm sure the answer is going to be frustratingly straightforward and simple, to the annoyance of everyone involved. Such is the life of learning to code, I guess. Thank you in advance for dealing with a rookie's doofy questions!
Here's my code:
f1<-full %>% #Build pilot_summary. Pipe pilot to...
group_by(Location,CT_NT,Depth,Site) %>% #group_by to work on CT or NT at each site
summarise_at(5:6,funs(mean)) %>% #calculate site means
ungroup(f1)
f1$Depth<-as.factor(f1$Depth)
f1$Site<-NULL
osu$Texture_Class<-NULL#Take out the texture class column
f2<- osu %>%
group_by(Location,CT_NT,Depth) %>% #group because otherwise R tries to crash on the next line of code...
arrange(Location,CT_NT,Depth) %>% #Put everything in order like f1, just in case
ungroup(f2)
f2$Depth<-as.factor(f2$Depth)
full_summary<-left_join(f1,f2)