Reconstruction of seed dispersal via modeling, seedling recruitment, and dispersal efficiency of Hemiphaga novaeseelandiae in Vitex lucens and Prumnopytis ferruginea in New Zealand

Abstract

My research investigates seed dispersal, post-dispersal stages, and dispersal vector behaviour in native New Zealand forest canopy tree species that have large fruit by asking: (i) What are the local seed ‘shadows’ (distributions) and seedling patterns in Prumnopytis ferruginea (miro, Podocarpaceae, dioecious) and Vitex lucens (puriri, Verbenaceae, hermaphroditic), for which Hemiphaga novaeseelandiae, the New Zealand Pigeon (kereru, Columbidae), is the keystone agent of dispersal? (ii) What are the effects of kereru densities on the quantity of dispersed miro and puriri seeds, and do kereru preferentially disperse large and/or viable seeds? (iii) Are other native large-fruited tree species, which depend on kereru, being actively dispersed or regenerating locally? (iv) Do recruitment patterns in miro and puriri demonstrate seedling ‘escape’, and if so, are there associated advantages? (v) How are the parameters of long distance seed dispersal kernels (probability density functions) in simulated miro and puriri forests affected by changes in kereru densities and spatial patterns of trees? Local seed distributions in miro and puriri were anisotropic and were described best by gamma and Weibull (i.e. leptokurtic) probability density functions. Seed shadows across both seasons had less steep slopes in Hunua miro compared with Waitakere miro, but increased production of actively dispersed depulped seeds did not increase dispersability. In both tree species, passive dispersal of seeds with mesocarp occurred mostly to the north and occasionally to the south, but active dispersal was less defined. Canopy openness, canopy size, and fruit production did not influence local mean dispersal distances in both tree species. Seedling distributions were inversely spatially concordant with seed patterns (demonstrating escape) only in miro in the first season. Seedlings were confined to north to north-easterly sectors for both tree species in both seasons, which mirrored the direction of passive seed rain. Seedling abundance for all miro trees decreased by c. 56% over 21 months, but there was lower seedling loss overall for puriri at c. 25%. Seedling survival and growth rates in either tree species were not significantly different whether under or away from the canopy. Kereru counts and densities were not significantly different between regions. There was a marginally significant correlation between kereru densities and the percentage of actively dispersed depulped seeds in seed shadows across both tree species in season one only. For puriri and miro trees overall, actively dispersed depulped seeds were significantly longer than seeds from passively dispersed intact fruit, demonstrating vector preference for larger fruit. Removal of miro fruit and seeds by mammals was higher overall at Waitakere compared with Hunua; Waitakere puriri had higher predation overall than Wenderholm puriri, and higher predation of actively dispersed depulped seeds vs. intact fruit. Actively dispersed depulped miro seeds had significantly lower endosperm integrity rates than seeds from passively dispersed intact miro fruit. iv In the study regions, miro, puriri, Corynocarpus laevigatus (karaka), Beilschmiedia tarairi (taraire), and Beilschmiedia tawa (tawa) trees were regenerating locally since they all had seedlings of their own species growing under their own canopies. There was also satisfactory active dispersal since the observed deviations from expected values of the presence of seedlings of other large-fruited tree species were generally significant. There were preferential patterns of seed dispersal: taraire seedlings were found most often under puriri trees and tawa seedlings were found most often under female miro trees. However, there was an absence of tawa seedlings under karaka trees, karaka seedlings below female miro trees, and only a very few miro seedlings under taraire trees. Miro seedlings were consistently absent under puriri and karaka trees, and no taraire seeds were dispersed at female miro trees in Waitakere. At Pelorus Bridge, Marlborough, miro seeds were preferentially dispersed under a miro tree and to a lesser degree under a hinau tree, across five seasons. There was virtually no active dispersal of tawa or hinau seeds under their own canopies or under co-fruiting tree species during this time. A rimu tree received only a few miro or tawa seeds, and rimu seeds occurred only in small numbers under a female miro and tawa tree. Kahikatea and matai were the main actively dispersed species under trees. Damage of bird-dispersed depulped seeds by mammals was high for tawa and low in miro. Experiments using an original spatially-explicit in silico simulation model showed that scale (mean dispersal distance), shape (kurtosis-1), and percentage of seeds dispersed beyond the nearest fruiting tree neighbour all decreased with increasing tree aggregation in simulated puriri and miro forests. There were no significant changes in the parameters when kereru density changed, except in some puriri forests at high kereru densities. Total mass of dispersed seeds increased with increasing kereru density, but the % of seeds dispersed beyond the nearest fruiting tree neighbour did not vary appreciably. My research suggests that low kereru densities may reduce the quantity of dispersed seeds and the rates of dispersal of larger seeds, especially if alternative dispersers are not available. Kereru are particularly important for dispersal in miro, since dioecious trees may be more susceptible to dispersal failure than co-sexual species.