Combining niche modelling , land use-change , and genetic information to assess the 1 conservation status of Pouteria splendens populations in Central Chile 2 3

To assess the conservation status of a species with little ecological information is usually a challenging process. Pouteria splendens is an endemic shrub of the coastal range of Central Chile currently classified as lower risk (LR) by IUCN (version 2.3). Knowledge about this species is extremely limited. Currently P. splendens is only found in two small and isolated populations, which are thought to be remaining populations of an originally large metapopulation. However, there is no evidence to support this hypothesis, limiting our ability to gauge the real current conservation status of this species. In this study we combine niche modelling, land-use information, and genetic techniques to test the metapopulation hypothesis and reassess the conservation status of P. splendens using the IUCN criteria. We also evaluated the potential effects of climate change in the species distribution. Our results support the hypothesis of a large metapopulation that was recently fragmented. Future climate could increase the range of P. splendens; however the high level of fragmentation would preclude colonization processes. We recommend reclassifying P. splendens as Endangered (EN) and developing strategies to protect the remaining populations. Similar approaches like the presented here could be used to reclassify other species with limited ecological knowledge.

but monthly distribution of precipitation follow the same pattern in the whole range, with 149 July being the wettest month and February the driest. Temperatures are relatively similar 150 for the entire study area, with a mean annual temperature of 13C, where July is the 151 coldest month (10.5C) and January the warmest (17C) (Dirección Meteorológica de   The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/026336 doi: bioRxiv preprint first posted online Sep. 8, 2015; maximum entropy, which is part of the software MaxEnt (Phillips et al. 2006). MaxEnt 172 estimates the probability of occurrence of the species of interest using presence only data 173 and a group of environmental variables. We decided to use MaxEnt because it has been 174 proven to be one of the best performing modelling methods available (Elith et al. 2006), it 175 is reliable even with a small number of samples (Pearson et al. 2007), and is freely 176 available. All the habitat distribution models were performed using MaxEnt version 3.3.3k 177 (http://www.cs.princeton.edu/~schapire/maxent).   (Table 1). it was in our case. Therefore, to determine the optimal parameters to configure MaxEnt we 200 compared different models with a combination of the "feature class" and "regularization  Once the optimal parameters were determined, we analyzed the performance of the model 213 of choice using the method proposed by Pearson et al. (2007)  analyzed using a prediction success rate and a p value using the software "pValuecompute 219 v.1.0" provided by Pearson et al. (2007). 220 221 We produced binary maps using two different thresholds to define the suitable vs. non-222 suitable habitat. We used the "maximum sensitivity plus specificity logistic (MSS)" and the 223 "10 percentile training presence logistic (10PL)" thresholds (values of 0.203 and 0.298 224 respectively). We used these parameters as they are among the two most commonly used 225 thresholds for creating binary suitability maps for species distribution with MaxEnt (e.g.,   Binary maps for projected future potential distribution were generated using the same 243 thresholds and approaches used to build the current conditions maps. To estimate the historical land-use change that has occurred within our study area we used     After the field campaigns we were able to confirm the current presence of P. splendens in 309 only nine sites, which were aligned to the species distribution range reported in the The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/026336 doi: bioRxiv preprint first posted online Sep. 8, 2015; in the coastal range, but also spreading further east through the valleys towards the Chilean 339 central valleys. This increase in distribution also generates an important reduction in the 340 differences between the total potential areas predicted by the two aforementioned 341 techniques when compared to the predicted total area under current climatic conditions 342 (Table 3).  (Table 3). However, in relative terms 352 the percentage of habitat loss was practically the same, with nearly 30% of the potential 353 area of distribution being lost by anthropogenic land-use change. and future scenarios, and were slightly higher than 30% for both future threshold 360 approaches (Table 3).   regeneration and seed dispersion constraints (i.e., lack of seed dispersers). In addition, we 401 incorporated references that mentioned threats to P. splendens and its habitat (i.e. fires).

402
Moreover, we estimated habitat loss in the next decades in the area of occurrence of P.   This is the first attempt to study the distribution of P. splendens throughout its current and 411 historical range. Furthermore, this study could be considered as one of the first efforts to   increase under a climate change scenario, but probably in a magnitude much smaller that is 481 shown by the model.

483
In terms of the effects of land-use change in P. splendens distribution, our results show that 484 the current potential habitat distribution is at least 30% smaller than the original potential 485 distribution. This was regardless of the threshold used to estimate the species distribution.

486
This percentage of potential habitat loss appears not to be so substantial if we take into 487 account that this land-use change started more than 200 years ago, and that today nearly   The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/026336 doi: bioRxiv preprint first posted online Sep. 8, 2015; between populations, such as animal or wind mediated outcrossing pollination and seed 529 dispersal (Luan et al. 2006;Millar et al., 2014). In the case of P. splendens there is no 530 available information about the pollination mechanism, but observation from other 531 Pouteria species report self-pollination and insect-mediated pollination (Jordan 1996).

532
None of these mechanisms could explain the genetic diversity patterns found within and 533 between P. splendens populations. In relation to seed dispersal, the fruit size (2-3 cm in    report what information they used to perform their assessment.

555
We could not calculate or find reliable data to estimate the area of occupancy of the 556 species. Although there is data from the national forest inventory that could help to estimate 557 an area of occupancy, it seems that the inventory did not include accurate data at the 558 species level, and only three sites containing P. splendens were apparently mapped, 559 representing an extension no larger than 400 ha. In contrast, based in our field survey we 560 presume that the area of occupancy is not smaller than 2.000 ha. However, even taking an 561 optimist approach of 3.000 ha for the area of occupancy, this would not change P. 562 splendens to any other conservation status (e.g. from endangered to vulnerable) because 563 other assessment criteria will prevail.   We were also able to update information regarding the current populations, establishing 587 that there are only two main areas with P. splendens populations left, as was also suggested 588 by previous studies. However, we found that these remaining populations are more reduced 589 in extension and density that was previously thought. Moreover, the larger and better  Based on the results from our work we recommend: (1) update the information available to      TCT CTC TCT CTC TRC  54  420-1900  849  GTG TGT GTG TGT GTG TYA  57  410-1700  854  TCT CTC TCT CTC TCT CRG  54  430-2080  856  ACA CAC ACA CAC ACA CYA  59  430-2080  887  888  889  890  891  892  893  894  895  896  897 . CC-BY-NC-ND 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/026336 doi: bioRxiv preprint first posted online Sep. 8, 2015;