I need some help with a figure. I need a boxplot in which the y axis is total number of brain cells, and x is ordered by mean body mass, grouped by species and is coloured according to the sociality.

Please find the R code I am using and the data below:

ggplot(Data, aes(x=mean_bodymass, y = Total.brain.cell, group=Species))+
  geom_boxplot (aes(fill=Sociality),width=5)+
  geom_point (aes(fill=Sociality),shape=21,size=4)+
scale_y_continuous(trans='log10', n.breaks = 8,labels = comma)+
  scale_x_continuous(trans='log10',labels = comma)+
theme_classic()+
scale_fill_manual(values=c("#FF1493","#BF3EFF","#FF3030", "#FF7F00","#0066ff"))+
  guides(fill="none")+
  theme(base_size=10, text = element_text(size = 30))+
  labs(x = "body mass(g)(log)", y = "total brain cells(log)")

Small sample of my data:

Data <- structure(list(Sociality = c("Semi-social", "Semi-social", "Semi-social", 
"Solitary", "Solitary", "Solitary", "Solitary", "Solitary", "Solitary", 
"Solitary", "Solitary", "Solitary", "Solitary", "Solitary", "Solitary", 
"Solitary", "Solitary", "Solitary", "Kleptoparasitism", "Kleptoparasitism", 
"Kleptoparasitism"), Species = c("Lipotriches australica", "Lipotriches australica", 
"Lipotriches australica", "Hylaeus  euxanthus", "Hylaeus  euxanthus", 
"Hylaeus  euxanthus", "Leioproctus amabilis", "Leioproctus amabilis", 
"Leioproctus amabilis", "Leioproctus recusus", "Leioproctus recusus", 
"Leioproctus recusus", "Megachile (Hacteriapis)", "Megachile (Hacteriapis)", 
"Megachile (Hacteriapis)", "Pseudoanthidium repetitum", "Pseudoanthidium repetitum", 
"Pseudoanthidium repetitum", "Thyreus nitidulus", "Thyreus nitidulus", 
"Thyreus nitidulus"), Total.brain.cell = c(1468750, 1441250, 
1496250, 906250, 853750, 870000, 961250, 975000, 947500, 855000, 
958750, 988750, 1952500, 1926250, 1993750, 1411250, 1075000, 
1450000, 1781250, 1757500, 1742500), mean_bodymass = c(0.065933333, 
0.065933333, 0.065933333, 0.008333333, 0.008333333, 0.008333333, 
0.0377, 0.0377, 0.0377, 0.027433333, 0.027433333, 0.027433333, 
0.053966667, 0.053966667, 0.053966667, 0.043, 0.043, 0.043, 0.084533333, 
0.084533333, 0.084533333)), row.names = 4:24, class = "data.frame")

My issue is the width of the boxplot. This is the result without adjusting the width of the boxplot:

And this is the result when adjusting the boxplot width (see code below), where the points are not not inline with the boxplot.

Is there any way I can adjust the width of the boxplot without affecting the points? Or do I need to do two separate plots and combine them? Any help would be appreciated.

I need some help with a figure. I need a boxplot in which the y axis is total number of brain cells, and x is ordered by mean body mass, grouped by species and is coloured according to the sociality.

Please find the R code I am using and the data below:

ggplot(Data, aes(x=mean_bodymass, y = Total.brain.cell, group=Species))+
  geom_boxplot (aes(fill=Sociality),width=5)+
  geom_point (aes(fill=Sociality),shape=21,size=4)+
scale_y_continuous(trans='log10', n.breaks = 8,labels = comma)+
  scale_x_continuous(trans='log10',labels = comma)+
theme_classic()+
scale_fill_manual(values=c("#FF1493","#BF3EFF","#FF3030", "#FF7F00","#0066ff"))+
  guides(fill="none")+
  theme(base_size=10, text = element_text(size = 30))+
  labs(x = "body mass(g)(log)", y = "total brain cells(log)")

Small sample of my data:

Data <- structure(list(Sociality = c("Semi-social", "Semi-social", "Semi-social", 
"Solitary", "Solitary", "Solitary", "Solitary", "Solitary", "Solitary", 
"Solitary", "Solitary", "Solitary", "Solitary", "Solitary", "Solitary", 
"Solitary", "Solitary", "Solitary", "Kleptoparasitism", "Kleptoparasitism", 
"Kleptoparasitism"), Species = c("Lipotriches australica", "Lipotriches australica", 
"Lipotriches australica", "Hylaeus  euxanthus", "Hylaeus  euxanthus", 
"Hylaeus  euxanthus", "Leioproctus amabilis", "Leioproctus amabilis", 
"Leioproctus amabilis", "Leioproctus recusus", "Leioproctus recusus", 
"Leioproctus recusus", "Megachile (Hacteriapis)", "Megachile (Hacteriapis)", 
"Megachile (Hacteriapis)", "Pseudoanthidium repetitum", "Pseudoanthidium repetitum", 
"Pseudoanthidium repetitum", "Thyreus nitidulus", "Thyreus nitidulus", 
"Thyreus nitidulus"), Total.brain.cell = c(1468750, 1441250, 
1496250, 906250, 853750, 870000, 961250, 975000, 947500, 855000, 
958750, 988750, 1952500, 1926250, 1993750, 1411250, 1075000, 
1450000, 1781250, 1757500, 1742500), mean_bodymass = c(0.065933333, 
0.065933333, 0.065933333, 0.008333333, 0.008333333, 0.008333333, 
0.0377, 0.0377, 0.0377, 0.027433333, 0.027433333, 0.027433333, 
0.053966667, 0.053966667, 0.053966667, 0.043, 0.043, 0.043, 0.084533333, 
0.084533333, 0.084533333)), row.names = 4:24, class = "data.frame")

My issue is the width of the boxplot. This is the result without adjusting the width of the boxplot:

And this is the result when adjusting the boxplot width (see code below), where the points are not not inline with the boxplot.

Is there any way I can adjust the width of the boxplot without affecting the points? Or do I need to do two separate plots and combine them? Any help would be appreciated.

• r
• ggplot2

• Welcome to SO! Please help yourself by helping us to help you by providing your test data in a more easily accessible form. Posting the result of dput(Data) would be ideal. – Limey Commented Mar 27 at 9:44
• Due to logarithmic scaling on the x-axis the boxplots are very close to each other. Adjusting the width would lead to the boxplots overlapping which I guess they will not do, but rather they nudge each other out of the way it seems. If you want to increase the width, think about scaling the x-axis linearly which will automatically increase the boxplot widths. – Tim G Commented Mar 27 at 11:07
• FumblebeeFae, see how others have helped to format your question to make it easier to try your code: with pictures alone, you're unlikely to get much help (see meta.stackoverflow/a/285557 and xkcd/2116), and while "pretty table" text is a step in the right direction, the use of dput(head(x,20)) for unambiguous representative data makes it the easiest and most direct way for somebody to play with your actual data, not how it may look on the console (the two can be very different). – r2evans Commented Mar 27 at 14:14
• For more examples/discussions on how to format questions well for reproducible "play", see stackoverflow/q/5963269 , minimal reproducible example, and stackoverflow/tags/r/info. Thanks, and good luck! – r2evans Commented Mar 27 at 14:15

1 Answer 1

Note that the default action is geom_boxplot(position="dodge2"), which means it uses position_dodge2(). I couldn't find a width= greater than width=0.05 that kept the points aligned with the boxplots, it always nudged them out of the way.

One alternative is to use position_dodge() instead.

Original,

  geom_boxplot(aes(fill=Sociality), width=5) +

Shifting to position_dodge() and reducing the width:

  geom_boxplot(aes(fill=Sociality), position=position_dodge(), width=0.1) +

position_dodge() allows for overlaps, you can get fairly crazy with it. Because your boxplots are positioned on a continuous scale and you have unique values that are relatively close to each other, you may have to play with the width= to get them exactly how you like.