How to change the inside color (fill) of a subset of points with the same shape based on a second variable in ggplot2?

Question

I had a hard time to find how to change inside color of geom_point in combined regression plot but I couldn't do it using scale_color_manual()

For site F and F1 I want to have same shape and same outside color but the inside of "F" blanck and inside of "F1" grey

Here is my code and plot

mydata<-read.csv(choose.files(),header=T)
ggplot(mydata,aes(y=density,x=Do, color=color, shape=color,group=site))+geom_point(size=3.5)+
  stat_smooth(method="lm",se=FALSE)+
  scale_shape_manual(values=seq(0,8))+
  scale_y_log10()+
  scale_x_log10()+
  labs(x = "Do", y = bquote('density'~(No.m^-3)))+
  theme_classic()+
  theme(aspect.ratio=1)+
  theme(axis.title.x = element_text(color = 'black',face = 'bold',size = 18, hjust = 0.5))+
  theme(axis.text = font)+
  theme(axis.title.y = element_text(color = 'black',face = 'bold',size=25, hjust = 0.5))+theme(legend.title = element_text(size=8, color = "white"),legend.position=c(0.17, 0.20),

enter image description here

mydata

     site     Do      density     color
1     A  15.32500000 14.6254359     A
2     A   3.46153846  2.5634207     A
3     A   1.37692308 54.5375387     A
4     A   0.52115385 44.4123777     A
5     B   9.71431056  5.4299784     B
6     B   7.60803311  2.9582927     B
7     B   6.61874506  1.3994870     B
8     B   5.38889098  2.1169919     B
9     C  18.95522388  0.6911765     C
10    C  12.23880597  1.2352941     C
11    C   7.91044776  1.1029412     C
12    C   4.02985075  1.0441176     C
13    C   3.43283582  0.7500000     C
14    D  19.85892857  5.7023070     D
15    D  15.21785714  3.2076197     D
16    D   8.00357143  2.3692545     D
17    D   7.67142857  3.0929227     D
18    D   5.75357143  3.8173552     D
19    E  96.26288660  5.5576024     E
20    E 118.41752580  4.3844430     E
21    E  88.97938144  4.3557405     E
22    E  33.16752577  6.1140918     E
23    E   8.27164948  4.8505888     E
24    F  23.54920101  2.7593361     F
25    F  20.52144659  2.3236515     F
26    F  15.97981497  3.0497925     F
27    F  56.09756097  5.6639004     F
28    F  37.93103448  6.2448133     F
29    F   2.80908326  5.5186722     F
30    F   5.53406223 11.0373444     F
31    F   4.85281750  5.2282158     F
32    F   9.92430614  1.8879668     F
33    F   0.36809306  5.6639004     F
34    F   2.73338940  1.8879668     F
35    F   3.94449117  6.3900415     F
36    F   1.59798150  3.9211618     F
37    F   4.39865433  2.6141079     F
38    F   1.21951220 33.5477178     F
39    F  40.84051724  4.3388430     F
40    F  36.96120690  2.4586777     F
41    F  35.02155172  3.3264463     F
42    F  44.71982759  0.8677686     F
43    F  52.47844828  0.8677686     F
44    F   4.41379310  2.0247934     F
45    F   1.45358090  7.2314050     F
46    F   0.25615763 17.6446281     F
47    F   5.08222812  1.1570248     F
48    F   0.25615763 30.8057851     F
49    F   3.26790451  5.7851240     F
50    F   3.84084881  4.4834711     F
51    F   2.02652520  8.8223140     F
52    F   5.17771883  2.0247934     F
53    F   0.05941645 12.5826446     F
54   F1  68.20857863  2.7593361     F
55   F1  64.42388562  2.3236515     F
56   F1  72.91645498  3.3402490     F
57   F1  60.63919260  2.1784232     F
58   F1  59.12531539  3.1950207     F
59   F1  56.09756097  4.2116183     F
60   F1  51.55592935  3.7759336     F
61   F1  50.04205214  2.4688797     F
62   F1  49.28511354 11.0373444     F
63   F1  68.96551724  0.8677686     F
64   F1  76.79352531  0.5785124     F
65   F1  79.59555352  2.8925620     F
66   F1  55.38793103  0.8677686     F
67   F1  51.50862069  1.3016529     F
68   F1  53.44827586  5.9297521     F
69   F1  49.56896552  1.1570248     F
70   F1  43.75000000  0.7231405     F
71   F1  44.71982759  7.9545455     F
72    G  31.69856459  2.0000000     G
73    G  61.60287081  4.0000000     G
74    G  53.22966507  8.0000000     G
75    G   8.97129187 12.0000000     G
76    G  10.16746411  6.0000000     G
77    G   9.56937799  5.0000000     G
78    G   8.97129187  5.0000000     G

Any help would be appreciated

Ian Campbell · Accepted Answer · 2021-04-11T16:25:36.333

One approach is to pre-calculate the color you want to apply ahead of time with mutate and case_when. This saves us from having to manually define each color. TRUE ~ "white" just makes everything else white.

Then, crucially, you must insure the points you want to fill have a shape that is compatible with a fill, specifically shapes 21, 22, 23, 24, or 25. You can insure this by manually defining them with scale_shape_manual. Finally, use scale_fill_manual to define the colors.

If we want to set up the legend, we can define the labels in scale_shape_manual. Then we can call guides to override the shapes to ones that are compatible with fill.

mydata %>%
  mutate(fill.col = case_when(site == "F" ~ "black",
                              site == "F1" ~ "gray",
                              TRUE ~ "white")) %>%
ggplot(aes(y=density,x=Do, color=color, shape=color, group=site, fill = fill.col)) +
  geom_point(size=3.5)+stat_smooth(method="lm",se=FALSE) +
  scale_shape_manual(values = c(A = 3, B = 4, C = 21, D = 22, E = 23, F = 24, F1 = 25, G = 8)) +
  scale_fill_manual(values = c(black = "black",gray = "gray",white = "white"),
                    labels = c(black = "F", gray = "F1", white = "Other"))+
  scale_y_log10() + scale_x_log10()+
  labs(x = "Do", y = bquote('density'~(No.m^-3))) +
  theme_classic() +
  theme(aspect.ratio=1,
        axis.title.x = element_text(color = 'black',face = 'bold',size = 18, hjust = 0.5),
        axis.title.y = element_text(color = 'black',face = 'bold',size=25, hjust = 0.5),
        legend.title = element_text(size=8, color = "white"),
        legend.position=c(0.17, 0.20),
        legend.text = element_text(color = "black",face = 'bold', size=14)) +
  guides(fill = guide_legend(override.aes = list(shape = c(24,24,21),
                                                 linetype = NA)))

Thanks so much, This is really helpful One more question, I would like to remove fill guide_legend, how can I do that? — fahimeh rashidabadi, Apr 11 '21 at 18:26
Add `guide = FALSE` to `scale_fill_manual(...)`, and then you can remove the last part `guides(...)`. — Ian Campbell, Apr 11 '21 at 19:00

score 1 · Answer 2 · answered Apr 11 '21 at 16:30

Here is a solution similar to Ian Campbell's. But it uses named vectors instead. Those vectors are created beforehand based on the unique site's and given appropriate values, note the shapes 23, chosen because that's a value that can be filled. Their names are the sites.

sites <- unique(mydata$site)
shapes <- setNames(c(0:4, 23, 23, 6), sites)
sites <- setNames(rep("white", length(sites)), sites)
sites[names(sites) == "F1"] <- "grey"

ggplot(mydata, aes(Do, density, color = color, shape = color))+
  geom_point(aes(fill = site), size = 3.5, show.legend = FALSE)+
  stat_smooth(method = "lm", formula = y ~ x, se = FALSE) +
  scale_fill_manual(values = sites) +
  scale_shape_manual(values = shapes) +
  scale_y_log10()+
  scale_x_log10()+
  labs(x = "Do", y = bquote('density'~(No.m^-3)))+
  theme_classic()+
  theme(
    aspect.ratio = 1,
    axis.text = element_text(family = "Sans"),
    axis.title = element_text(color = 'black', face = 'bold', hjust = 0.5),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 25),
    legend.title = element_text(size = 8, color = "white"),
    legend.position = c(0.17, 0.20),
    legend.text = element_text(color = "black", face = 'bold', size = 14)
  )

Thank you very much, this code also could work :) For the ggplot part I also included group to have regression lines for all sites : ggplot(mydata, aes(Do, density, color = color, shape = color,group= site)) — fahimeh rashidabadi, Apr 11 '21 at 18:33
@fahimehrashidabadi You don't need `group` to have regression for all sites, any of `color`, `fill`, `shape` and others will do it. — Rui Barradas, Apr 11 '21 at 18:35
Actually I mean for all sites separately, because I tried to keep shapes the same for two groups but have different regression lines for each one ( F blank and F gray should contain different regression lines but here they have one line) — fahimeh rashidabadi, Apr 11 '21 at 18:41

How to change the inside color (fill) of a subset of points with the same shape based on a second variable in ggplot2?

2 Answers2