1: \begin{abstract}

2: We show analytically and numerically that the appearance of lumps and gaps in the 

3: distribution of $n$ competing species along a niche axis is a robust phenomenon 

4: whenever the finiteness of the niche space is taken into account. 

5: In this case depending if the niche width of the species $\sigma$ is above or below 

6: a threshold $\sigma_c$, which for large $n$ coincides with $\frac{2}{n}$, there are 

7: two different regimes. For $\sigma > \sigma_c$ the lumpy pattern emerges directly from 

8: the dominant eigenvector of the competition matrix because its corresponding eigenvalue 

9: becomes negative. For $\sigma \leq \sigma_c$ the lumpy pattern disappears. 

10: Furthermore,  this {\em clumping transition} exhibits critical slowing down

11: as $\sigma$ is approached from above. 

12: We also find that the number of lumps of species vs. $\sigma$ displays a stair-step 

13: structure. The positions of these steps are distributed according to a power-law.

14: It is thus straightforward to predict the number of groups that can be packed along a 

15: niche axis and it coincides with field measurements for a wide range of the model 

16: parameters.

17: \end{abstract}