Vertebrates
    fish
        goldfish
        clownfish
    amphibian
        frog
        toad
    reptiles
        snake
        lizard
        turtle
        tortoise
    birds
        sparrow
        crow
        parrot
    mammals
        dog
        cat
        horse
        whale

Vertebrates
    fish
        goldfish
        clownfish
    amphibian
        frog
        toad
    reptiles
        snake
        lizard
        turtle
        tortoise
    birds
        sparrow
            house
            factory
        crow
        parrot
        crane
    mammals
        dog
        cat
        horse
        whale

dat  = structure(list(V1 = c("Vertebrates", NA, NA, NA, NA, NA, NA,
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
NA), V2 = c(NA, "fish", NA, NA, "amphibian", NA, NA, "reptiles",
NA, NA, NA, NA, "birds", NA, NA, NA, NA, NA, NA, "mammals", NA,
NA, NA, NA), V3 = c(NA, NA, "goldfish", "clownfish", NA, "frog",
"toad", NA, "snake", "lizard", "turtle", "tortoise", NA, "sparrow",
NA, NA, "crow", "parrot", "crane", NA, "dog", "cat", "horse",
"whale"), V4 = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
NA, NA, NA, "house", "factory", NA, NA, NA, NA, NA, NA, NA, NA
)), .Names = c("V1", "V2", "V3", "V4"), class = "data.frame", row.names = c(NA,
-24L))
>

library(zoo)
library(igraph)

# read tab delimited data - keep structure by setting "" to missing
# (would of been great if you had given this in a format easier to use)

dat <- read.table("test.txt", sep="t", header=FALSE, fill=TRUE,
                  na.strings="", strip.white=TRUE, stringsAsFactors=FALSE)

head(dat, 7)
#             V1        V2        V3
#1   Vertebrates      <NA>      <NA>
#2          <NA>      fish      <NA>
#3          <NA>      <NA>  goldfish
#4          <NA>      <NA> clownfish
#5          <NA> amphibian      <NA>
#6          <NA>      <NA>      frog
#7          <NA>      <NA>      toad

# carry forward the last value in first two columns to impute missing
dat[1:2] <- sapply(dat[1:2], na.locf, na.rm=FALSE)
dat <- na.omit(dat)

# get edges for graph - we want two columns (from and to) for each edges
edges <- rbind(dat[1:2],setNames(dat[2:3],names(dat[1:2])))

# create graph
g <- graph.data.frame(edges)

# Plot graph
E(g)$curved <- 0
plot.igraph(g, vertex.size=0, edge.arrow.size=0 ,
                      layout=-layout.reingold.tilford(g)[,2:1])

dat <- structure(list(V1 = c("Vertebrates", NA, NA, NA, NA, NA, NA,
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA), V2 = c(NA,
"fish", NA, NA, "amphibian", NA, NA, "reptiles", NA, NA, NA,
NA, "birds", NA, NA, NA, "mammals", NA, NA, NA, NA), V3 = c(NA,
NA, "goldfish", "clownfish", NA, "frog", "toad", NA, "snake",
"lizard", "turtle", "tortoise", NA, "sparrow", "crow", "parrot",
NA, "dog", "cat", "horse", "whale")), .Names = c("V1", "V2",
"V3"), class = "data.frame", row.names = c(NA, -21L))

# To prepare the data

# carry forward the last value in columns if lower level (col to the right)
# is non-missing
dat2[1] <- na.locf(dat2[1], na.rm=FALSE)

for(i in ncol(dat2):2)  {
  dat2[[i-1]] <-  ifelse(!is.na(dat2[[i]]), na.locf(dat2[[i-1]], na.rm=F),
                                                                   dat2[[i-1]])
      }

# get edges for graph
edges <- rbind(na.omit(dat2[1:2]),
                       do.call('rbind',
                               lapply(1:(ncol(dat2)-2), function(i)
                                  na.omit(setNames(dat2[(1+i):(2+i)],
                                                         names(dat2[1:2])))))
                         )