A well-known problem of the \Lambda CDM model is the tension between the relatively high level of clustering , as quantified by the parameter \sigma _ { 8 } , found in cosmic microwave background experiments and the smaller one obtained from large-scale observations in the late Universe .
In this paper we show that coupled quintessence , i.e .
a single dark energy scalar field conformally coupled to dark matter through a constant coupling , can solve this problem if the background is taken to be identical to the \Lambda CDM one .
We show that two competing effects arise .
On one hand , the additional scalar force is attractive , and is therefore expected to increase the clustering .
On the other , in order to obtain the same background as \Lambda CDM , coupled quintessence must have a smaller amount of dark matter near the present epoch .
We show that the second effect is dominating today and leads to an overall slower growth .
Comparing to redshift distortion data , we find that coupled quintessence with \Lambda CDM background solves the tension between early and late clustering .

We find for the coupling \beta and for \sigma _ { 8 } the best fit values | \beta| = 0.079 ^ { +0.059 } _ { -0.067 } and \sigma _ { 8 } = 0.818 ^ { +0.115 } _ { -0.088 } .
These values also fit the lensing data from the KiDS-450 survey .
We also estimate that the future missions SKA and Euclid will constrain \beta with an error of \pm 1.5 \times 10 ^ { -3 } and for \sigma _ { 8 } of \pm 1.8 \times 10 ^ { -3 } at 1 \sigma level .