Temperature niche difference and interspecic competition determine the parapatric distribution of two congeneric species in Diaphanosoma

Background Diaphanosoma excisum and D. dubium are parapatrically distributed in tropical and subtropical waters, and rarely coexist at local communities in the transition between tropics and subtropics. According to Janzen’s hypothesis and the modern coexistence theory, the specic thermal adaptation and strong interspecic competition between the two congeneric species are assumed to result in their range limits. To explore the mechanisms underlying such parapatric distribution, we modeled their current geographic distribution, and compared temperature niche difference and tness inequality in interspecic competition with their representative clones. Results The species distribution modeling showed that the two Diaphanosoma species had signicantly divergent climate niches. Their life history parameters in monocultures also demonstrated as stable temperature niche difference, while D. excisum had higher tness than D. dubium in dimension of food niche. The competition experiments revealed a strong exploitative competition between the two species, and D. excisum was the superior competitor, and excluded D. dubium in all the competition experiments within overlapping temperature niche regardless of food conditions. Conclusions Our results suggest that stable temperature niche difference overcoming interspecic competition for food plays a critical role in shaping Diaphanosoma species range.


Background
The primary goal of ecology and biogeography is to understand the spatial distribution of species [1][2][3][4]. Species range largely depends on their niche: an n-dimensional hypervolume de ned by axes of environmental conditions and resource use within which populations of a species are able to maintain a positive intrinsic rate of increase [5][6][7]. Temperature affects most key life history traits and has long been recognized as the most fundamental dimension of niche space, which determines species geographical distribution [8,9]. Change in temperature will signi cantly affect species adaptation and spatial distribution [10][11][12]. In tropics, species experience a rather stable thermal condition, and consequently, they are predicted to have narrower temperature niches than their counterparts from higher latitudes and become more sensitive to temperature variation [10,13]. In high latitudes, where annual temperature largely uctuates, species have broad temperature tolerance threshold. Food availability is another fundamental niche dimension, which determines species food utilization rate and realized tness [14].
Individuals of one species that use marginal resources do not exploit resources as e ciently as those for which these resources are nearly optimal [14].
In the modern coexistence theory, species coexistence requires stable niche difference or tness equality [15][16][17]. Congeneric species are considered to be more similar in niche space than species from different genera, resulting in strong competition in overlapping habitats [4,18,19]. Diaphanosoma is a high diversity genus of Cladocera, commonly dominating in tropical and subtropical freshwaters. Two congeneric species, D. dubium and D. excisum are common in tropics or subtropics but rarely coexist at local communities [20][21][22]. In tropics, D. excisum is frequently present as the only representative of Cladocera [23]. D. dubium is widespread in subtropical China while D. excisum is restricted to the coastal islands of the southern China [24]. The mean temperature of the coldest month (January) at the northern limit of the distribution of D. excisum in this region is about 15 °C, while the northward limits of D. dubium reaches the lower Amur river [21]. Pajk et al. [25] measured the life history traits of clones from 16 populations of D. dubium and D. excisum under a large temperature range from 10 °C to 40 °C, and showed the tropical D. excisum had narrower thermal performance curves (TPCs) and higher optimum temperature than the subtropical D. dubium, but failed to reproduce at ≤ 15 °C. The authors concluded that negative r below 15 °C in combination with competition from the cold-adapted D. dubium limited northward distribution of D. excisum. Due to the similarity in body size and feeding by ltration of shared prey -small phytoplankton and bacteria, strong exploitative competition between the two congeneric species is expected when total abundance approaches the carrying capacity, and strong interspeci c competition may constraint the southward distribution of the cold-adapted D. dubium in China. However, the temperature niche and the competition for food between the two species in temperatures above 15 °C remain unclear.
To illustrate the temperature niche difference, we rst calculated climate niche overlap between the two congeneric species under a species distribution model. To check further temperature niche and interspeci c competition between D. dubium and D. excisum, we compared the thermal performance curve and conducted competition experiments at combinations of two temperature and two food concentrations.

Species distribution modeling
Species data were simple presence observations based on records from checklist and our eld sampling.
Occurrence records for D. dubium were mostly referred to [22], our consecutive sampling as well as eld observation. Occurrence records for D. excisum were mostly from the published literature. D. excisum was widely distributed in tropics and subtropics of Australia, Asia and Africa, up to eastern China. No records were reported for both species from South America, North America and Antarctic. In total, 165 populations for D. dubium and 162 populations for D. excisum were used for species distribution models (see Additional le 1).
We used a comprehensive set of 19 bioclimatic variables with a spatial resolution of 2.5 arc min (about 4 km in the study area) as climate predictors available through http://www.worldclim.org [26]. We computed pair-wise spearman rank correlations to assess multi-colinearity of the 19 wordclim variables, wherein species distribution models (SDMs) were developed with a subset of variables with R 2 < 0.75: bio2, bio3, bio5, bio6, bio8, bio13, bio14, bio15, bio18, bio19 (see Additional le 1). For each species we use the modelled association between current climate and present-day distribution to estimate current distributional areas.
We performed spatial predictions of the two species distributions in 'ecospat' written in R language [27], which provided methods and utilities for quantifying climatic niches shifts between two congeneric peripheral distribution species, analyzing niche overlap between them [28]. We used a shape le containing presence-absence records for a species as spatial points generated in QGIS v2. 18. Raster datasets (in grid format) of above ten wordclim variables were used as explanatory variables (predictors). Coupled with selected explanatory variables, the shape les with only occurrence data of both D. excisum and D. dubium were used to perform modeling and evaluation with package 'sdm'. Four models ('rf' (Random Forest), 'brt' (Generalized Boosting Model or usually called Boosted Regression Trees (gbm)), 'bioclim' (Surface Range Envelop (sre), 'glm' (Generalized Linear Model)) were used for model training. Roc curves were used to compare the results for all models. Then we just predicted the habitat suitability into the whole study area, and tted habitat suitability value of each present-absent site for each model were obtained from the output shape les in QGIS v2.18 (see Additional le 1).
For pre-modeling, we investigated spatial autocorrelation of climate predictors by Mantel correlogram to determine the minimal distance between species records. We performed Niche quanti cation and comparison with ordination techniques. The PCA was calibrated on all the sites of the study area to describe the contribution of all original predictors, covering the ranges of the two study species [29].
Niche overlap in terms of Schoener's D was computed based on kernel density estimates for both species occurrence records in PCA space accounting at the same time for the relative frequency of speci c climate conditions within the study area. The results allowed a quanti cation of those parts of niche space occupied by both species as well as the unique parts of two species. The classi cation of Schoener's D was proposed by Rödder and Engler [30].
Randomization tests for niche equivalency and similarity as proposed by Warren et al. [31] were computed to test the hypothesis of niche equivalency and niche similarity [32]. Parameter alternative was set as 'greater' to test for niche conservatism, i.e. whether the overlap is more equivalent/similar than random. For Core Niche Modeling, we accessed the calibration capacity of our models using a repeated split-plot approach [33]. We used a random subsample of 75% of the points of presence to calibrate the models using the approach described above, and the remnant 25% to evaluate the model, repeating the whole procedure 10 times. Presence-only data were used to calculate the Boyce Index for each model [34], as Boyce index measured how much model predictions differ from a random distribution of the observed presences across the prediction gradients [35]. Finally, we ensemble selected small models with the Boyce index more than 0.80 for spatial predictions with current climatic variables. Once the models were calibrated and evaluated, we projected the potential distribution of the species over space. Spatial projections were visualized and plotted in QGIS v2.18. All statistical analyses were performed using R 3.5.3 [27], and the packages being "biomod2", "ecospat", "sdm', "maptools", "raster", "rgdal", "rtiff".
Life history and temperature niche Analysis of temperature niche difference between the two species was conducted with two representative clones, one for D. dubium and another for D. excisum. D. dubium (clone N) was collected from a sh pond on the way to Changtang reservoir, Shaoguan, China (N24°78', E113°50′). D. excisum (clone D) was collected from Hedi reservoir, Zhanjiang, China (N21°85', E110°31'). The two clones chosen were each the best performing clone of its species from our previous life history experiments with 16 Diaphanosoma clones [25]. We measured their life history traits (net reproductive rate (R 0 ), intrinsic rate of population increase (r)) under a temperature range from 10°C to 40°C. 40℃ was estimated to be above the critical thermal maximum (CT max ), while 10℃was estimated to be below the critical thermal minimum (CT min ) [25]. Life table experiments were conducted at 15,20,25,30,34, and 38 °C. There were at least ve replicates per clone by temperature combination; replicates being single individuals in 30 mL of medium. Additional replicates were added in environments with high juvenile mortality (T > 34℃). The detailed life history experiments and parameter calculation were described in Pajk et al. [25]. As temperature curve performance of R 0 and r can be used as representative of temperature niche, they were the tted by mixed linear models with the lme function from the R package "nlme" [36]. The choice of the best t was made through comparison of the AICc.

Interspeci c competition experiments
The competition experiments were conducted with above clones of two species of Diaphanosoma. The 12±12 h old second clutch offspring of the F 3 -generation were used to start the experiments. The experiments were run at two temperatures (20°C and 30°C) and food levels (1.0 mgC/L and 0.5 mgC/L) [37]. The temperatures chosen are representative of the localities of D. dubium and D. excisum, while the food levels are selected from the life history experiments of D. dubium and D. excisum, in which a sound intrinsic rate of increase (r>0.2) was obtained at above 0.5 mgC/L. Animals were kept in 300 mL of medium. Monoculture replicates initially contained 10 neonates of a single species. Competition replicates were started with 10 neonates of each species. The number of animals in each treatment was recorded every six days at 20°C and every four days at 30°C. For each replicate we determined the maximum density and the maximum growth rate N 0 and N t : population densities at two consecutive samplings). We calculated r of the two species in both competition cultures and monocultures [38][39][40]. These parameters were compared between the treatment groups (species×T×food×monoculture/competition culture) with a factorial ANOVA (SPSS 16.0).

Species distribution and climate niche overlap
Climate niche overlap between D. dubium and D. excisum was low (Schoener's D = 0.218) (Fig. 1c). The climatic predictors of Mean Diurnal Range, Min Temperature of Coldest Month, Precipitation of Driest month, Precipitation of warmest Quarter, and Precipitation Seasonality contributed more than the rest variables (see Additional le 2). With a value of p=1/p=0.73, we accepted the hypothesis that the niche overlap between sister species D. dubium and D. excisum (see Additional le 3) is more equivalency /similarity than random.
For four tested models, the Boyce indexes were all listed in Table 1. The rank of predicted expected ratio ordered along habitat suitability axis was plotted (see Additional le 4). Model 'glm' had the best calibration capacity for the dataset (D. dubium: Boyce index=0.926; D. exciusm: Boyce index=0.906), followed by model 'rf' (D. dubium: Boyce index=0.883; D. exciusm: Boyce index=0.805). Based on Boyce index, both 'glm' and 'rf' were used for species distribution projection. We generated a species distribution model as well as climate niche overlap analysis for both D. dubium and D. exciusm populations (Fig. 1a,  b). The results demonstrated a strong niche separation of the two Diaphanosoma species.
Life history and temperature niche D. dubium had a broad temperature niche and was able to reproduce at <20°C (Fig. 1d, e). D. excisum had a narrower temperature niche and failed to reproduce at <20°C, but higher net reproductive rate (R0) and tness (r) than D. dubium at ≥20°C (Fig. 1d, e).

Fitness inequality and interspeci c competition
Temperature, food level and competition all had a signi cant effect on maximum density (Table 2). D. excisum had higher average maximum population density than D. dubium in all monocultures and competition cultures. Average maximum density in monoculture was higher at 20℃ than at 30℃ irrespective of food level for D. excisum, but it was higher at 20℃ than at 30℃for D. dubium only at low food level (Fig. 2a). Maximum growth rate was not directly affected by competition (ANOVA, F 1,32 =2.129, p=0.154).
In both monoculture and competition experiments, D. excisum had higher r than D. dubium in all food and temperature combinations (Fig. 2b). At both higher temperature and food level, two species had higher r, whereas at both lower temperature and food level, two species had higher r (Fig. 2b). In the competition experiments, D. excisum always outcompeted D. dubium irrespective of food concentration at both 20℃ and 30℃.

Page 7/15
The species distribution modeling con rmed that D. dubium dominates in subtropical and temperate regions, and D. excisum in tropical regions, and indeed they have a very narrow climate niche overlap. D. excisum always has higher intrinsic rate of population increase ( tness) and outcompeted D. dubium irrespective of food concentration within overlapping temperature niche.

Temperature niches and species distribution
The species distribution modeling shows the fundamental niches for both D. dubium and D. excisum are larger than their realized niches represented by few occurrence records. The temperature niche for tropical D. excisum was narrower than that of subtropical D. dubium. Failure of occurrence of the two species in America, North Europe, and North Africa may be due to their colonization history, dispersal restrictions, or their competition with temperate zooplankters, e.g. Daphnia [41][42][43].
Generally, temperature is an essential determinant in species' distribution [2,[10][11][12][13][44][45][46][47]. Difference in the critical thermal minimum and maximum appears to be the basis for niche separation between D. dubium and D. excisum [25]. Due to its critical thermal minimum, the extension of D. excisum into subtropics is strongly limited. When the temperature falls below their critical thermal minimum, Diaphanosoma normally produces resting eggs to protect them from extinction and expand their distribution [22,48]. Theoretically, D. excisum can produce resting eggs to preserve its population below 15℃, i.e. a temporal storage effect. If so, D. excisum can colonize new habitats even more northward, which is contradicted with its present distribution. Three reasons may account for D. excisum's inability to expand its northward distribution, one of which is resting eggs viability, because hatching of resting eggs requires stimuli from speci c temperature and photoperiod conditions [49]. Another is that the hatchlings from the potential resting eggs are simply outcompeted by populations of other Diaphanosoma species capable of overwintering in active stage, never allowing D. excisum populations to become established outside of tropics. A third one is that there may be other mechanisms (e.g. hybridization) rather than producing resting eggs for D. excisum to adapt to low temperature. During northward expansion, when encountering with D. dubium, D. excisum produces male clones to hybridize with female D. dubium. This can explain the observed hybrids of D. excisum and D. dubium in their distribution edge [49], which may have a higher tness than their parental species [50][51][52].

Interspeci c competition and species distribution
For D. dubium, its extension to tropics is limited. From our life history experiment (also see [25]) we know that both species can adapt to high temperature and so high temperature does not likely limit the southern extension of D. dubium. D. excisum outcompeted D. dubium in the exploitative competition irrespective of temperature and food in the competition experiments. Exploitation competition or interference competition during expansion is su cient to account for parapatric distribution of two congeneric species [53,54].
The maximum population density in monocultures is much higher for D. excisum than for D. dubium. Such result is expected only when D. excisum utilizes food more e ciently and has higher tness. The competitive superiority of D. excisum in our experiments is explained by its higher r, which allows it to monopolize available resources and thus gain a numerical advantage in the competition. High r is considered characteristic of tropical cladocerans [55]. D. excisum is a tropical species, and does possess higher r than D. dubium in the present study as well as previous experiment with a greater variety of clones of both species. The fast initial increase in abundance of D. excisum is especially critical in face of the lower maximum population density of D. dubium, producing strong suppression on D. dubium population growth.
D. excisum fully outcompeted D. dubium at all food concentrations when temperature ≥ 20 °C. It is thus reasonable to assume that this competition limits D. dubium's spread into the tropics along with sh predation. In addition, high intensity of sh predation pressure in the tropics may aggravate the inferior competitive position of the larger sized D. dubium because visual-hunting predators or lter feeders actively or passively select for larger prey [56]. Achenbach and Lampert [57] also pointed out that the size shift from larger bodied to small bodied zooplankton in lakes occurred with increasing temperature due to the increase selective predation. The similar shift in body size was observed in an introduced goby species preying on large-sized zooplankton and aquatic insects [58]. Although the difference in body size for our congeneric species is rather small, the failure of the larger sized D. dubium when competing with the slight smaller D. excisum in the eld may be attributed in part to such selective predation which persist all the year-round in tropical water.

Conclusions
In conclusion, as predicted by Janzen's hypothesis and the modern coexistence theory [10,14], difference in thermal adaptation causes a narrow overlap of climate niches between two Diaphanosoma species. D. excisum has a narrower temperature niche and higher intrinsic rate of population increase in warm water. At ≤ 15 °C, D. excisum can not maintain its active population, and is therefore limited to tropics. The temperature is not the physiological factor limiting D. dubium's extension into tropics; however at ≥ 20 °C, D. excisum outcompetes D. dubium and limits its spread to the south. Stable temperature niche difference overcoming tness inequality in food dimension explains the parapatric distribution pattern for the two competitive congeneric species.
Declarations BPH and FP designed the study. ZW, PY, and QH performed the experiment. JL and PL analyzed the data with substantial advice and direction from BPH and HD. PL, BPH and HD wrote and corrected the manuscript. PL, PY and ZW conducted all eld work and collected all samples with assistance from FP. All authors read and approved the nal manuscript.

Funding
Financial and infrastructure was supported by National Nature Science Foundation of China (31901098 and 31670460).

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its additional les.  Figure 1 The potential distribution of D. dubium (a) and D. excisum (b) in current climate condition, and their temperature niches represented by net reproductive rate (R0) (d) and the intrinsic rate of population increase (r) (e) for single clones. Their climate niche overlap under species distribution models was also displayed (c). The two maps ( Fig.1a and 1b) were generated with 'ecospat' package in R 3.5.3, and visualized in QGIS v2.1. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.