The Webb’s hyacinth (Bellevalia webbiana Parl., Asparagaceae) is an Italian narrow endemic species, listed as Endangered (EN A2c) in the IUCN Global Red List of Threatened Species. The range of this bulbous perennial herb is restricted to two disjunct areas of Central Italy, separated by the mountain ridge of Northern Apennine. To evaluate the impact of climate change on this species, we used Ecological Niche Modelling (ENM) to predict future potential distribution under different scenarios, through Maximum entropy algorithm. The estimated potential distribution highlights the vulnerability of Webb’s hyacinth to the effects of climate change. The current potential Area Of Occupancy (AOO) (992 2×2 km cells) is forecast to dramatically decrease in the range 2041–2100, under the scenarios SSP3-7.0 (2070 = –249, 2100 = –682) and SSP5-8.5 (2070 = –372, 2100 = –948). In all future scenarios, the predicted potential distribution shifts towards higher elevations, located in the two main areas in which the species currently occurs. This could imply a loss of the current genetic differentiation documented at the population level. To overcome these problems, in addition to assisted colonization, an ex situ conservation programme should be planned.

Area of occupancy, climate change, conservation, endangered, Environmental Niche Modelling, Maxent, potential distribution

Although soil consumption is a major concern for plant conservation, climate change caused by human activities can further exasperate conservation problems (Corlett 2016, 2020). Climate change is already impacting biodiversity and is likely to intensify over the next century (Bellard et al. 2012; Casazza et al. 2014), so that mitigation efforts are urgent to implement. In the case of species endemic to a country, the global chance of survival entirely relies on national policies, so that endemic taxa are key elements for setting national conservation priorities and for assigning conservation tasks (Orsenigo et al. 2018). Italy hosts a high number of endemics, between 19% and 21% of the native flora according to different estimates (Peruzzi et al. 2014, 2015; Bartolucci et al. 2018, 2022), and most of them show a quite restricted distribution range.

The Webb’s hyacinth (Bellevalia webbiana Parl., Asparagaceae, monocots; Fig. 1) is one of the most relevant Italian narrow endemic species (Chiarugi 1949; Borzatti von Loewenstern et al. 2013; Astuti et al. 2017). The distribution range of this bulbous perennial herb is restricted to two relatively small and disjunct areas of Central Italy, separated by the mountain ridge of Northern Apennine. Despite the small distribution range, this species is linked to relatively common habitat types, since it typically grows in open fields and meadows, wood margins, olive groves, and vineyards (Gestri et al. 2010). Recent studies investigated the population genetics (Peruzzi et al. 2021), functional traits (Astuti et al. 2019), and reproductive performance (Astuti et al. 2018) of the five largest populations known for this species. These studies highlighted some differentiation in the population located at the upper NE limit of the species range, around Faenza (Ravenna). The International Union for Conservation of Nature (IUCN) listed the Webb’s hyacinth in the Red List of Threatened Species (Peruzzi and Carta 2011; see also Orsenigo et al. 2018), as Endangered (EN A2c). During the last century, the main threat to this species was represented by the spread of human settlements, which caused local extinctions from several historical localities (Gestri et al. 2010). Climate change can quickly become a further threat for the persistence of the small and localised populations of this species, magnifying its extinction risk in the near future (Casazza et al. 2021). Therefore, to evaluate the impact of climate change on the conservation of Bellevalia webbiana, we modelled its future potential distribution under different scenarios by using Ecological Niche Modelling (ENM) (Khanum et al. 2013; Assis et al. 2018).

The ensembled Maxent model for the Webb’s hyacinth with the selected environmental variables performed particularly well, achieving high evaluation metric scores (AUC_Train = 0.932, AUC_Test = 0.928, CBI_Test = 0.853). In total, 24 single models were combined into a final ensembled model. The current potential Area Of Occupancy (AOO), made up by 992 2×2 km cells, is forecast to decrease in the range 2041–2100 (Fig. 2).

Figure 2.

Predicted potential distribution for Bellevalia webbiana under current and future climatic scenarios. Cells potentially suitable for the species are highlighted in red, while the circumscribed study area is enclosed by the white line A current climatic condition B SSP1-2.6 scenario for 2040–2070 C SSP1-2.6 scenario for 2070–2100 D SSP3-7.0 scenario for 2040–2070 E SSP3-7.0 scenario for 2070–2100 F SSP5-8.5 scenario for 2040–2070 G SSP5-8.5 scenario for 2070–2100.

The SSP1-2.6 scenario highlights limited variations in potential AOO (2070 = –11 potentially occupied cells, 2100 = –95). On the other hand, both SSP3-7.0 and SSP5-8.5 highlight a dramatic reduction of the potential AOO (SSP3-7.0: 2070 = –249, 2100 = –682; SSP5-8.5: 2070 = –372, 2100 = –948) (Table 1).

The feature importance ranking measure (Table 2) points out the great importance of Annual Mean Temperature and Annual Potential Evapotranspiration as predictors, even if Annual Precipitations also give a relevant contribution.

The overall uncertainty of predictions for current climatic condition (Fig. 3), calculated as a range of values, indicated a low level of uncertainty among replicated runs. The mean value of uncertainty for cells showing low and high habitat suitability (< 0.25 and > 0.75) was equal to 0.071 and 0.121, respectively. This indicates a high stability of predictions in the most relevant habitat suitability range. The mean value of uncertainty for cells showing intermediate habitat suitability (0.25 < suitability < 0.75) was equal to 0.216, higher but still low. The analysis of uncertainty for future projections returned results very similar to those calculated with current climatic conditions (not shown).

Figure 3.

Plots of model uncertainty calculated as range of values observed across 24 replicate analyses in modelling the potential distribution of Bellevalia webbiana under current climate conditions. The values of uncertainty (y axis) were plotted against those of habitat suitability (x axis). Dashed lines delimit key areas for which habitat suitability can be associated with absence or presence with a high degree of confidence. Red, yellow, and blue areas of the scatter plot are associated with high, intermediate or low values of uncertainty.

We modelled the potential changes in distribution range of the Italian narrow endemic Bellevalia webbiana under different climate change scenarios, highlighting a significant reduction in its potential area of occupancy. This confirms the estimations made by Casazza et al. (2021), albeit carried out with a different focus and scope. These authors calculated that the assisted colonization of 13–16 new 2×2 km grid cells would be required to halt the estimated loss of range under a pessimistic scenario.

Several authors already evidenced similar future reductions in other stenochorous endemic plant species (Hong-Wa and Arroyo 2012; Wulff et al. 2013; Dülgeroğlu and Aksoy 2019; Khajoei Nasab et al. 2020; Semu et al. 2021), building up a consolidated body of knowledge on the potential risk of climate changes for the conservation of endemic plant species in different biogeographical contexts. The results of this study also reinforce the findings by Walker (2014) and Darrah et al. (2017), who highlighted the high risk of extinction of bulbous monocots, which often have stable local populations but limited dispersal capacity. In addition, the same authors remarked on the general underrepresentation of geophytes in plant conservation projects.

In all future scenarios, the predicted potential distribution tends to shift towards higher elevations, which are located between to the two main areas in which this species currently occurs, but where this species is currently absent for other extra-climatic ecological reasons. We highlight that the Webb’s hyacinth typically occurs at low elevations (on average ca. 200 m a.s.l.). Despite this, we predicted a tendency to shift towards higher elevations, as already evidenced for many species (Randin et al. 2009; Feng et al. 2020). Several other studies returned similar results, but concerning montane plants (Dirnböck, et al. 2011; Engler et al. 2011; He et al. 2019). In addition, important resurvey studies of historical data demonstrated the upslope shift of several species, resulting in a complete transformation of the local assemblages in the same mountain range (Morueta-Holme et al. 2015; Steinbauer et al. 2018).

However, the upslope shift here modelled for Bellevalia webbiana and the connected reduction in potential area of occupancy is only predicted on the basis of climatic niche, and does not consider other potential variables affecting the ecological niche of the species. The most relevant factor potentially making the future scenario much worse than those modelled on the basis of predicted climate is represented by the availability of habitat. This species is strongly linked to open habitats, often with traditional management (Gestri et al. 2010), but these habitats are disappearing in most of the hills and submontane ecosystems of peninsular Italy because of socio-economic changes and of the concentration of population in lower elevation areas (Falcucci et al. 2007). Forest communities are occupying previously grazed or cultivated lands in the higher hills and lower mountains (see e.g., Geri et al. 2010; Amici et al. 2013), promoting more natural habitat types but affecting the persistence of species linked to secondary and cultural landscapes (Amici et al. 2015).

In addition to habitat loss, the limited dispersal ability is a further factor which should be considered to assess the actual extinction threat of B. webbiana. Indeed, given its relatively large seeds (Astuti et al. 2018), the dispersal capacity of this species may be insufficient to cope with the speed needed for the upslope shift requested by the climate change. Then, even admitting that potentially suitable areas at higher elevations could effectively host B. webbiana and that this species could colonise these areas by natural or assisted migration (Brodie et al. 2021), then this would imply an increasing connection between the two currently disjunct population groups. This increasing connection would, in turn, promote a potential loss of the genetic and functional differentiation documented for some populations (Astuti et al. 2018, 2019; Peruzzi et al. 2021). Moreover, considering that the actual AOO documented for Bellevalia webbiana amounts to just 47 2×2 km cells (of which only 24 were confirmed in recent times) and that forest expansion in Apennine mountains could further reduce the availability of open habitats (Gargano et al. 2012), this species shows evident conservation concerns. To overcome these problems, in addition to assisted colonization (Casazza et al. 2021), a proper ex situ conservation programme should be planned, as an insurance policy against extinction (Li and Pritchard 2009).