The second possibility to consider for the interpretation of PKS 1718-649 is that it belongs to the class of extragalactic radio sources known as GHz Peaked-Spectrum (GPS) radio sources. These are a little understood class but distinct from flat-spectrum, core-jet radio sources in their properties. O'Dea, Baum, and Stanghellini [1991] reviewed the properties of GPS sources and list their defining characteristics: radio spectra peaked near 1 GHz, low radio polarisations, large radio luminosities, and mostly compact radio structure.
PKS 1718-649 meets the criteria set out for candidature of a sample of Northern Hemisphere GPS radio sources [Stanghellini et al. 1990]: 
1 Jy, 0.4 GHz 
5.0 GHz, 
( 
, 
), and |b|> 10 
. However, the radio spectrum of PKS 1718-649 is only an average example of a GPS spectrum. The archetypical GPS sources PKS 1934-638 and 0108+388 have very well defined spectral turnovers and narrow peaks. The PKS 1718-649 radio spectrum is broader but similar to some of the candidate GPS sources listed by O'Dea, Baum, and Stanghellini [1991].
O'Dea, Baum, and Stanghellini [1991] verify that low radio polarisation is a key characteristic of GPS sources. For 38 sources with measurements, the median fractional polarisation at 21 cm is approximately 0.2%. The value of 0.35% at 13 cm for PKS 1718-649 is consistent with this result.
PKS 1718-649 is a strong radio source but cannot rival the luminosities of any of the known GPS sources. PKS 1718-649 would be the lowest luminosity GPS source known. However, it would be by far the lowest red shift GPS source known also. Most known GPS sources lie at red shifts greater than 0.2 and only two are known at red shifts less than 0.1 (1404+286, O`Dea, Baum, and Stanghellini 1991; 1144+352, Snellen et al. 1995). However, the selection criteria for candidature of the GPS class introduces a severe bias favouring luminous, and therefore, high red shift sources. There is no reason to conclude that all GPS sources have very large luminosities. Weaker GPS sources may be prevalent but missed due to observational limitations.
As stated, PKS 1718-649 is a highly compact radio source, fulfilling the third criteria of O'Dea, Baum, and Stanghellini [1991]. Furthermore, the pc-scale morphology of PKS 1718-649 is typical for GPS sources which are found to be associated with galaxies. Phillips and Mutel [1980] first noted that GPS radio sources in galaxies (as opposed to quasars) consist of two pc-scale components of similar flux density and spectral shape. A prominent example of this type of GPS galaxy is PKS 1934-638 [Tzioumis et al. 1989]. The description of GPS galaxy applies very well to the pc-scale morphology of PKS 1718-649. However, the limited spectral information on the two components between 4.8 and 8.4 GHz shows that they probably do not have the same spectral shape.
Five possible physical interpretations for the characteristics of the GPS sources are currently favoured [O'Dea, Baum, and Stanghellini 1991; Stanghellini et al. 1993]. These interpretations are centred around the suggestion that the radio plasma is confined on the scale of the optical narrow-line region by an unusually dense and clumpy interstellar medium which may also be kinematically complicated, or at least that the radio plasma and its environment are interacting heavily. In the case of the GPS quasars, a dense and clumpy protogalaxy environment may be the cause. In GPS galaxies, the dense and clumpy environment may be caused by recent merger activity and a disruption of the nuclear region caused by material infall.
Unfortunately, constraints on the conditions in the narrow-line regions of GPS hosts, to support the above suggestions, are rare. O'Dea, Baum, and Stanghellini [1991] list only two GPS galaxies which have reasonably well determined optical spectra, PKS 1934-638 [Fosbury et al. 1987] and 2021+614 [Bartel et al. 1984]. The data for these two sources support the notion that the environments of GPS sources are unusually dense and highly reddened.
Stanghellini et al. [1993] have obtained optical images of a large sample of GPS hosts and find evidence for a high incidence of mergers. This is perhaps not surprising since most of the objects are at high red shifts and higher merger rates are predicated at high red shifts. It is not clear yet that GPS sources have a stronger tendency for being involved in mergers.
For PKS 1718-649, as mentioned in the introduction, extensive optical imaging and spectroscopic observations have been made. Fosbury et al. [1977], Carswell et al. [1984], Filippenko [1985], and Keel and Windhorst [1991] note the existence of a strong LINER spectrum and the possibility that high electron densities exist in the nuclear region. From high quality data and a careful analysis Filippenko [1985] concludes that the electron density near the nucleus, in the narrow-line region, is very high, approximately 10 
- 10 
cm 
. He also concludes that the densest clouds live close to the nucleus and move most rapidly.
Véron-Cetty et al. [1995] have investigated the large amount of HI gas around PKS 1718-649 and conclude from the overall kinematics that the system probably arose from the recent merger of two galaxies, at least one of which was a gas rich spiral. This conclusion may also explain the optical morphology of the galaxy. Its spiral arms may be the tidally disrupted remains of the gas rich spiral. The r 
behaviour of the inner region of the optical galaxy can also be explained in the merger picture. Barnes and Hernquist [1991] show, with numerical simulations of mergers which account for stars and gas, that the central region of the merger product quickly relaxes into an r profile.
Many properties of the PKS 1718-649 radio source are shared with the class of GPS radio sources associated with galaxies. Although it is not as well defined a GPS source it bears many strong resemblances to PKS 1934-638, one of the best known examples of a GPS source, the major lacking being the broadness of the peaked spectrum of PKS 1718-649. Optical and HI observations of the host galaxy of PKS 1718-649 show strong evidence for a high density radio source environment and a galaxy merger origin. These results are qualitative support for the broader paradigm that the nuclear radio sources in GPS galaxies are confined by a dense and clumpy environment which may have been induced by merger activity.