How to use gradient of a color for scale of discrete integers?

I want to use a gradient along a single color (i.e. from light to dark) for a set of integers that can be considered discrete in the sense that there cannot be any fractional values between them.

But if I use a continuous scale then the legend unnecessarily displays colors for the potential fractional values. And if I use a discrete scale then fundamentally different colors are displayed (as opposed to gradient of shades). What to do?

Consider the following chart:

library(ggplot2) 
set.seed(123) 
(df <-rbind(data.frame(x=1:5, y=rnorm(5, 3), age=1),
            data.frame(x=1:5, y=rnorm(5, 5), age=2),
            data.frame(x=1:5, y=rnorm(5, 7), age=3)))

(p <- ggplot(df, aes(x, y, group=age, color=age)) +
   geom_point() + geom_line())

.
To get rid of the fractional values I attempted the following, which produced different legends showing unnecessary colors for the in-between values.

p + scale_color_continuous(n.breaks=3) 
p + scale_color_binned()

If I use a factor then I don't get a gradient. Here are the legends after two attempts.

(q <- ggplot(df, aes(x, y, group=age, color=factor(age, ordered = T))) +
    geom_point() + geom_line())

q + scale_color_discrete()   

How do I get a legend that shows colors only for the discrete values but auto-selects those colors from a gradient of shades (i.e. light to dark) rather than using fundamentally different colors.

To apply a color gradient to the discrete values you can map factor(age) to the color aesthetic as you tried, and use a palette that provides the gradient appearance you're looking for.

Manually specified palette

You can create a vector of colors manually, and there are many sources that can help generate a suitable gradient.

custom_blues <- c("#5E93CF","#21548D","#002650")

p +
  scale_color_manual(values = custom_blues) +
  labs(color = 'Age')

Paletteer

This package is an interface to a large collection of color palettes.

library(paletteer)

p +
  scale_color_paletteer_d("beyonce::X40") +
  labs(color = 'Age')

RColorBrewer (Original Example)

Using the RColorBrewer package was in my original answer. If you like the Paletteer interface, you can still access and apply palettes from this package (e.g. RColorBrewer::Blues)

library(RColorBrewer)

p +
  scale_color_brewer(palette= "Blues") +
  labs(color = 'Age') +
  theme_bw()

Setup: Data and base plot

library(ggplot2)

set.seed(123) 
(df <-rbind(data.frame(x=1:5, y=rnorm(5, 3), age=1),
            data.frame(x=1:5, y=rnorm(5, 5), age=2),
            data.frame(x=1:5, y=rnorm(5, 7), age=3)))
#>    x        y age
#> 1  1 2.439524   1
#> 2  2 2.769823   1
#> 3  3 4.558708   1
#> 4  4 3.070508   1
#> 5  5 3.129288   1
#> 6  1 6.715065   2
#> 7  2 5.460916   2
#> 8  3 3.734939   2
#> 9  4 4.313147   2
#> 10 5 4.554338   2
#> 11 1 8.224082   3
#> 12 2 7.359814   3
#> 13 3 7.400771   3
#> 14 4 7.110683   3
#> 15 5 6.444159   3

p <- ggplot(df, aes(x, y, group=age, color=factor(age, ordered = TRUE))) +
  geom_point(size = 4) +
  geom_line(linewidth = 2)