When separation occurs, it produces turbulent eddies (and in some cases, what is known as the von Karman vortex streets). By the concept of energy conservation, the additional KE of the eddies/vortices must come from somewhere. This is the potential energy (i.e., the pressure of the flow). This additional pressure loss is due to the form or shape of the object (or the duct) and is variously termed as pressure drag or form drag and is in addition to the viscous drag that would occur without separation. It is termed drag because it acts in the same way as viscous drag to cause pressure loss, which can be calculated by equating the pressure force to the shear force (drag force).

Often viscous drag is a small fraction of the total drag (form drag + viscous drag). If you like more clarity, please Google for 'why golf balls have dimples on them'. 

An attached/unseparated flow over the aft portion of a wing produces an increasing pressure that results in a forward force that opposes the drag.  If the flow separates, resulting in recirculation, a low pressure flow results, on the aft portion. This produces a smaller forward thrust. This decrease of forward thrust leads to an increase in the overall drag.. 

When separation occurs, it produces turbulent eddies (and in some cases, what is known as the von Karman vortex streets). By the concept of energy conservation, the additional KE of the eddies/vortices must come from somewhere. This is the potential energy (i.e., the pressure of the flow). This additional pressure loss is due to the form or shape of the object (or the duct) and is variously termed as pressure drag or form drag and is in addition to the viscous drag that would occur without separation. It is termed drag because it acts in the same way as viscous drag to cause pressure loss, which can be calculated by equating the pressure force to the shear force (drag force).

Often viscous drag is a small fraction of the total drag (form drag + viscous drag). If you like more clarity, please Google for 'why golf balls have dimples on them'. 

Prof. Raghavan provides part of the explanation.  Separation of smooth laminar flow, without eddies, produces a drag increase as well.

One must compare the difference between separated and attached flows to understand.  Typically a flow can remain attached to a surface (such as a curved airfoil) as static pressure builds up positively in the direction of flow (a pressure rise usually, but not always, due to surface curvature).  Over the surface with the rising static pressure, the attached flow creates a component of net force opposite to the direction of flow, a "thrust" if you will. Without going into boundary layer theory, the rising pressure creates an instability at the surface which if sustained over a long enough distance or at a high enough pressure rise per distance (pressure gradient), the momentum of the flow  very near the surface will fail to balance the rising pressure, and the flow will separate from the surface.  The separation results in a cessation of the rising pressure at the surface and a subsequent cessation of "thrust' that would have occurred with attached flow.  The result of separation is an imbalance of forces over the body that ordinarily with attached flow would have resulted in no net flow direction force, but which is now replaced with a net force in the flow direction (drag).

Agreed!

The flow in the rear portion of a blunt body must slow down due to obstruction. Yet the faster flow in front has to continue. Hence flow separates with formation of wake.

You have asked a good question. John Denton in his IGTI Scholar lecture addresses the point. The wall shear stress under the bubble is in the forward or favourable direction so why is there increased drag? What Denton points out is that it is dissipation and entropy generation that really matters. He shows us that the fluid inside the bubble is indeed straining and has shear stresses that oppose the motion. So the additional drag arrises inside the recirculating fluid. The second reason why we see loss increases is because the free shear layer over the top of the bubble is highly unstable and liable to Kelvin Helmholz instability so at reattachment the boundary layer is highly unsteady and tends to create a strongly turbulent boundary layer. So the bubble boosts the transition process and helps to create a high shear stress high disipation turbulent boundary layer.

The original question was not regarding a bubble. Why invoke thermodynamics when there is no question of transfer of energy as heat?

my small contribution:

immagine a turbulent flow along a flat plate, the instantaneous velocity field has a wide range of eddies with separated (and reattached) regions at the wall. The averaged velocity profiles shows that the dU/dy at wall is greater than in laminar condition. That is an increasing in viscous drag.

Furhter, you have to add also pressure drag increasing for separated flow on bodies..

The above is true if the upstream flow is turbulent, otherwise the drag is viscous proportional to upstream flow velocity and small.

Dear Professor Jaffar,

Your work seems to be interesting. Could you post it for us?

It is to avoid the several misconceptions (and repetitions that are inaccurately worded) which I see above, that I gave a simple and physically easy to understand explanation. Martin George Rose has not rebutted the comment on the bubble, and on his behalf I like to clarify that the recirculating fluid is often termed "recirculating bubble" or just "bubble" for short. As for Filippo Maria Denaro, what he says would occur with a flat plate of finite thickness or with obstacles on the plate such as, say, a rib roughness element. In the case of an infinitesimally thin and smooth flat plate (both of which are idealizations), the higher drag in the turbulent region as compared to the laminar region is indicative of loss of energy taken by the K.E. of the turbulence. There is no pressure drag here.