A. halimus is mainly considered a monoecious species which is occasionally dioecious. However, some authors have found individuals that present unisexual and hermaphrodite flowers so this species could be polygamous or, more precisely, trimonoecious (Talamali et al. 20012003). Our study observed no other flower type than the typical male and female flowers, although one female flower was found to contain abortive stamen primordia. The male-to-female flower ratio was 2.75/1, consistent with results obtained by other authors for this species (Freeman et al. 2007).
Flowering started in July in both years of our study. However, on Porquerolles Island close to Toulon (S France), plants originating from Algeria and Tunisia started flowering in September (Talamali et al. 2003), and in Morocco, Abbad and Benchaabane (2004) found that flowering commenced in mid-August in the Safi and Marrakech regions, and around the beginning of September in Ouarzazate. These two authors justified differences in flowering phenology by the variety of climatic conditions; high temperature, low humidity and periods without rainfall accelerate the formation of flower buds. Climate data show that the station in France is colder and rainier than the location we studied, with an annual mean temperature of 14.6°C with rainfall of 925.4 mm (Cuers Station: Rivas-Martínez and Rivas-Sáenz 1996–2009), but the locations under study in Morocco are hotter and drier, with an annual mean of 18.5°C and 317.0 mm in Safi, 19.9°C and 241.0 mm in Marrakech and 19.4°C and 101.4 mm in the Ouarzazate region (Rivas-Martínez and Rivas-Sáenz 1996–2009), so the climatic conditions cited do not explain the phenological delay observed in Morocco. This difference could be due to yearly variations in meteorological or microclimatic conditions.
The Atriplex species seems to be self-compatible (Heyligers 2001), and A. halimus is considered to be highly outbreeding (Haddioui and Baaziz 2001; Ferchichi et al. 2006), indicating that this species has effective mechanisms to prevent self-crossing. Protandry had been observed in this species by Abbad and Benchaabane (2004), who established a delay of nearly a month between the appearance of male and female flowers. In our study, we observed that male and female flowers seemed to open in the same weeks at population level (Figure 3), but the highest number of male flowers opened is reached one week before that of the female flowers, and that the upper, middle and lower parts of the inflorescence showed a similar pattern of behaviour (Figure 4). At plant and glomerule level (Figure 5), we see that male flowers opened one week before the females. This is in line with observations by Talamali et al. (2001), because female flowers are distributed along the distal parts of the dichasium and so open later than the males, which are mainly found on the proximal parts.
The start of fruit set and ovule development began when the stigmas were totally receptive. In 2006 the number of fruits per glomerule was significantly higher than in 2007, but seeds and bracts were much smaller in the first year. Abbad and Benchaabane (2004) observed that drought could cause a reduction in the number of female flowers in this species, so yearly variations in the number of fruits per glomerule could be explained by differences in rainfall between these years (Figure 1); the average accumulated rainfall from January to July in the location studied is 255.1 mm, while in 2006 and 2007 this value reached 327.0 and 175.6 mm respectively. This environmental effect has been observed in other species of Atriplex (McArthur 1977; McArthur and Freeman 1982).
The yearly variation in fruit size might be the result of physiological constraints including positional, temporal and environmental effects during offspring development; but most often offspring size is determined by an equal distribution of resources among the offspring (McGinley et al. 1987), so seed number could determine its weight (Smith and Fretwell 1974; Vaughton and Ramsey 1998; Sakai and Sakai 2005; Sadras 2007; Gambín and Borrás 2009; Kosiński 2010).
These size differences cannot be considered a case of seed heteromorphism, which occurs frequently in Chenopodiaceae (Imbert 2002), but it would be interesting to determine if these differences affect seed germination requirements, as occurs in species of Atriplex that show seed heteromorphism (Khan and Ungar 1984;1986), or if they influence individual plant performance, as in Atriplex triangularis Willd. (Ellison 1987. Kheiria et al. (2007) showed the existence of fruit polymorphism in Tunisian populations, the fruits from southern Tunisia being smaller than those collected from the central and northern areas. Fruit size recorded in our study for 2006 was similar to data collated by these authors in populations from central and northern Tunisia, but the size recorded in 2007 was bigger.
A. halimus fruits suffer high levels of pre-dispersal predation by insects. Larvae of Coleophoridae (Lepidoptera) were the most abundant predators of fruits and seeds in both years of the study. Huertas Dionisio (2007) found eight species of Lepidoptera on A. halimus plants, four of which are common in inflorescences: Coleophora gaviaepennella Toll, C. granulatella Zeller, Goniodoma auroguttella Zeller (fam. Coleophoridae) and Gymnancyla sfakesella Chrétien (fam. Pyralidae). This author also found pupals of the first and second species in the inflorescences in November, and Gymnancyla sfakesella larvae in September and October.
Huertas Dionisio (2005) describes the biological cycle of Goniodoma auroguttella in A. halimus plants in Huelva (SW Spain), including locations in the “Marismas del Odiel” Natural Park. Adults emerge in June, July and August and larvae feed on Atriplex fruits until October, gaining access to the ovary/achene by a hole made in the side of the valves, and moving along the inflorescence camouflaged by fruit valves. In October, the larvae go down the stem and bore a tunnel to the pith, where they transform into chrysalides. This cycle synchronizes with the previously described phenological phases (female flowering, fruit development and predation) observed in the present study. The anthesis of pistilate flowers occurs mainly in July and August, when adult insects are in flight; the first fruit predation event in both years was observed as the fruiting bracts reached maximum size, in the second half of September; peak predation is in October when larvae stop eating and the fruits begin to drop, which points to behavioural adaptation by the predator to exploit seed resources.
The yearly differences in levels of fruit predation could be explained by factors such as population size of the predators or variations in fruit and seed size. In 2006, seeds were smaller and the predation level was 62%, while in 2007 seeds were approximately 20% bigger and the predation level was 43% which suggests that when seeds are smaller, predators compensate by attacking more fruits. This is in line with the satiation mechanism proposed by Bonal et al. (2007), in which the satiation of small predators at seed level ensures that a proportion of the seeds survive predation.
In plants with sequential flowering, seed predation may be affected by the position of the fruit within the plant (Ehrlén 1996; Kudo et al. 2001; Sánchez et al. 2008). In A. halimus there were no significant differences in predation levels between different parts of the inflorescence. Mass fruiting appears to favour seed predators and allows them to attack the fruit on different parts of the inflorescence with the same frequency, but this strategy reduces fruiting time and ensures that some fruits avoid total seed destruction (Augspurger 1981; Pías and Guitián 2007; Gomes da Silva and Bezerra Pinheiro 2009). This strategy seems most appropriate when greater seed loss is caused by a specialist predator (Ims 1990), as is the case of the Atriplex population we have studied.