I am afraid your problem is precipitation of the OVA by ionic aluminium species onto the Alhydrogel. Maybe washing of the Alhydrogel with water close the isoelectric point before adsorption, use of a small concentration of background electrolyte may change the picture.

Alhydrogel is a colloid gel suspension. Thus the method you use to separate the gel from the protein solution and also the volume ratio between the solution and the gel are important. Ovalbumin tends to form molecular associates in water and the associates may also form on or near the surface of gel, especially at high concentrations. Could it be that a fraction of the protein was removed from the solution by a kind of entrapment, each time during the gel separation from the supernatant? There were some reports on Freundlich type protein adsorption (without saturation) on ion-exchangers, and alhydrogel is probably a weak ion-exchanger, too. Thus the whole OVA adsorption pattern should be dependent on the conditions such as pH, ionic strength and buffer composition. I would presume that bi- and three-valent anions like phosphate may influence adsorption phenomena on alhydrogel, because of their interaction with the surface. It is worth trying variations in the background salt concentration or pH as Dr. Sobisch suggested.